ECONOMIC EVALUATION OF THE COMMUNITY-LEVEL INTERVENTIONS FOR PRE-ECLAMPSIA (CLIP) IN SINDH, PAKISTAN by Asif Raza Khowaja BSc. Nursing, Aga Khan University Pakistan, 2003 MSc. Health Policy and Management, Aga Khan University Pakistan, 2007 A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Reproductive and Developmental Sciences) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2018 © Asif Raza Khowaja, 2018 ii Abstract Background Hypertensive disorders of pregnancy, particularly pre-eclampsia and eclampsia (PE/E), can lead to life-threatening complications or even the death of a mother or a newborn. Because there are few clinical trials of community-based interventions for PE/E, there is little evidence about the cost-effectiveness of these interventions. The aim of this dissertation is to conduct an economic evaluation of the Community-Level Interventions for Pre-eclampsia (CLIP) combined with routine pregnancy care, compared to routine pregnancy care alone, in Sindh, Pakistan. Methods A mixed-methods (i.e., qualitative and quantitative) approach to economic evaluation was undertaken alongside the CLIP Trial. A literature review of published epidemiological and economic studies was carried out to document evidence on PE/E interventions and guide the design of the economic model. Data were collected through focus groups, a structured questionnaire embedded into CLIP Trial surveillance, a cross-sectional survey of health facilities, and program budgetary reviews. The cost-effectiveness analysis was performed using a societal perspective. Probabilistic analysis was applied to estimate incremental cost-effectiveness ratios (ICERs), and sensitivity analysis was done to characterize uncertainties. Results The literature review found economic studies mainly in developed countries and focused only on costs to the health system. Focus groups revealed a large burden of out-of-pocket spending and productivity losses to pregnant women and families. Health care providers and decision makers identified upfront technology costs as a key challenge for the health system. Maternal and newborn care costs varied significantly between and within public and private iii sectors. In the probabilistic analysis of the base case, the incremental cost of the intervention as compared to control was $20,438, while the years of life lost was -37 (i.e., negative health gains), indicating a wide range of statistical uncertainty around ICERs. Overall, the probability that ICERs fell below the country-specific threshold was less than 30%. Conclusion This dissertation highlights knowledge gaps for costs and cost-effectiveness of PE/E interventions in low- and-middle-income countries (LMICs). The economic analysis indicates that CLIP is not a cost-effective strategy, compared to routine pregnancy care. More research is needed to conduct the process evaluation to inform policy decisions on resource allocation in Sindh, Pakistan. iv Lay Summary Pakistan has the highest rates of maternal and newborn mortality in South Asia. The Community-Level Interventions for Pre-eclampsia (CLIP) Trial introduced a mobile health (mHealth) intervention to provide home-based screening of hypertensive disorders of pregnancy, case management, and referrals by the Lady Health Workers in Sindh, Pakistan. In this study, we conducted an economic evaluation of CLIP interventions compared to routine pregnancy care alone. The economic literature on PE/E interventions mainly focused on costs to the health system. Focus groups with pregnant women revealed a large burden of out-of-pocket spending and productivity losses to families. Health care providers and decision makers identified challenges of initial investments in mHealth technologies. The cost-effectiveness analysis of CLIP interventions demonstrated higher costs relative to health benefits, indicating a wide range of statistical uncertainty in the results. Further research is needed to evaluate the cost-effectiveness of surveying households in community settings. v Preface Asif contributed significantly to all elements of research presented in this dissertation. Asif conducted the literature review, contributed to the study design, prepared data collection instruments, and was involved in the process of data collection (focus groups and cross-sectional surveys) onsite. Asif developed the data analysis plans and performed both qualitative, as well as quantitative analyses. The design of an economic model for CLIP was guided by the literature review and qualitative analysis undertaken by Asif. The work presented in the thesis has led to the following publications/manuscript submitted: 1. Khowaja AR, Mitton C, Bryan S, Magee LA, Bhutta ZA, von Dadelszen P. Economic evaluation of Community Level Interventions for Pre-eclampsia (CLIP) in South Asian and African countries: a study protocol. Implementation Science. 2015 May 26;10(1):76. This work is associated with Chapter 3 and outlines methodological undertakings for economic analysis of the CLIP Trials. Asif was responsible for literature review, developed study protocol, and served as the primary author on the publication. 2. Khowaja AR, Mitton C, Qureshi R, Bryan S, Magee LA, von Dadelszen P, Bhutta ZA. Societal perspective on cost drivers for health technology assessment in Sindh, Pakistan. International Journal of Technology Assessment in Health Care. 2017 July:1-7. This work is associated with Chapter 4 and describes qualitative research to explore cost drivers to guide the design of quantitative instruments for health resource utilization. Asif prepared the data collection tools, worked on ethics application, vi trained data collectors, involved in the primary data collection onsite, performed qualitative data analysis and served as the primary author of the publication. 3. Khowaja AR, Mitton C, Qureshi R, Bryan S, Magee LA, von Dadelszen P, Bhutta ZA. A Comparison of Maternal and Newborn Health Services Costs in Sindh Pakistan. PLOS ONE. Submitted August 2017. This work is associated with Chapter 5 and outlines the cross-sectional survey of public and private health providers/facilities. Asif prepared the data collection tools, trained data collectors, involved in the primary data collection onsite, performed quantitative data analysis and served as the primary author of the publication. Ethics A written consent was obtained from every participant before conducting the study. All study participants were assigned a unique participant identification numbers to ensure their confidentiality. This study received ethical approval from the Ethics Review Committee (ERC) of the Aga Khan University, Karachi, Pakistan (ERC # 3230-OBS-ERC-14), and the Institutional Review Board of the University of British Columbia in Vancouver, Canada (ETHICS # H12-00132), as the central coordinating site. vii Table of Contents Abstract .................................................................................................................................... ii Lay Summary ........................................................................................................................... iv Preface ....................................................................................................................................... v Table of Contents ....................................................................................................................vii List of Tables ............................................................................................................................. x List of Figures..........................................................................................................................xii List of Abbreviations ............................................................................................................. xiii Acknowledgements .................................................................................................................. xv Chapter 1: Introduction ............................................................................................................ 1 1.1 Background ....................................................................................................................1 1.2 Definition of pre-eclampsia and eclampsia......................................................................1 1.3 Situation analysis: global context of PE/E .......................................................................2 1.4 Community-level interventions for pre-eclampsia (CLIP) trials ......................................3 1.5 The purpose ....................................................................................................................5 1.6 Study hypothesis.............................................................................................................5 1.7 Research questions .........................................................................................................5 Chapter 2: Evidence of costs and cost-effectiveness of pre-eclampsia / eclampsia interventions: a literature review ............................................................................................. 8 2.1 Background ....................................................................................................................8 2.2 Objectives ......................................................................................................................8 2.3 Methods .........................................................................................................................9 2.4 Results .......................................................................................................................... 11 viii 2.5 Discussion .................................................................................................................... 17 2.6 Conclusions .................................................................................................................. 20 Chapter 3: Economic evaluation of Community Level Interventions for Pre-eclampsia (CLIP) in South Asian and African countries: a study protocol ........................................... 27 3.1 Background .................................................................................................................. 27 3.2 Objectives .................................................................................................................... 28 3.3 Methods ....................................................................................................................... 29 3.4 Possible study limitations ............................................................................................. 36 Chapter 4: Societal perspective on cost drivers for health technology assessment in Sindh, Pakistan ................................................................................................................................... 41 4.1 Background .................................................................................................................. 41 4.2 Objectives .................................................................................................................... 42 4.3 Methods ....................................................................................................................... 43 4.4 Results .......................................................................................................................... 46 4.5 Discussion .................................................................................................................... 50 4.6 Conclusions .................................................................................................................. 52 Chapter 5: A comparison of maternal and newborn health services costs in Sindh, Pakistan ................................................................................................................................................. 57 5.1 Background .................................................................................................................. 57 5.2 Objectives .................................................................................................................... 57 5.3 Methods ....................................................................................................................... 58 5.4 Results .......................................................................................................................... 60 5.5 Discussion .................................................................................................................... 62 ix 5.6 Conclusions .................................................................................................................. 65 Chapter 6: Cost-effectiveness of the Community Level Interventions for Pre-eclampsia (CLIP) in Sindh, Pakistan ....................................................................................................... 69 6.1 Background .................................................................................................................. 69 6.2 Objectives .................................................................................................................... 70 6.3 Methods ....................................................................................................................... 70 6.4 Results .......................................................................................................................... 73 6.5 Discussion .................................................................................................................... 79 6.6 Conclusions .................................................................................................................. 84 Chapter 7: Conclusions ......................................................................................................... 103 7.1 Summary of findings .................................................................................................. 103 7.2 Implications ................................................................................................................ 106 7.3 Conclusions ................................................................................................................ 109 Bibliography .......................................................................................................................... 110 Appendices ............................................................................................................................ 121 x List of Tables Table 1. List of articles included in the literature review ............................................................ 21 Table 2. Methodological approaches on selected economic studies ............................................ 25 Table 3. Methods for collecting resource utilization and cost information.................................. 37 Table 4. Focus group participants’ characteristics ...................................................................... 53 Table 5. Descriptive coding list for thematic analysis ................................................................ 54 Table 6. Lump sum out-of-pocket costs related to healthcare sought for HDP ........................... 55 Table 7. Cost of maternal health services in public and private health sectors ............................ 66 Table 8. Cost of newborn health services in public and private health sectors ............................ 67 Table 9. Model input parameters and type of distributions ......................................................... 85 Table 10. Baseline characteristics at enrollment ........................................................................ 86 Table 11. Maternal health resource utilization ........................................................................... 87 Table 12. Newborn health resource utilization ........................................................................... 89 Table 13. CLIP handbag items in Sindh, Pakistan...................................................................... 90 Table 14. Societal costs comparing intervention and control ...................................................... 91 Table 15. Regression-adjusted coefficients for study variables .................................................. 92 Table 16. Regression-adjusted cost estimates for resource utilization (in US$) .......................... 93 Table 17. Years of life lost (YLL) in the intervention and control groups .................................. 94 Table 18. Years lost due to disability (YLD) in the intervention and control groups .................. 95 Table 19. Probabilistic cost-effectiveness estimates (10,000 iterations) ..................................... 97 Table 20. Regression-adjusted effect size for maternal and perinatal outcomes .......................... 98 Table 21. Sensitivity analysis for health resource utilization .................................................... 121 Table 22. Multi-collinearity diagnostic outputs ........................................................................ 122 xi Table 23. Correlation co-efficient matrix ................................................................................. 123 Table 24. Multi-collinearity diagnostic outputs ........................................................................ 124 Table 25. Focus group guide: Pregnant women identified with HDP ....................................... 125 Table 26. Focus group guide: Husband of pregnant women identified with HDP ..................... 126 Table 27. Focus group guide: LHWs/LHS in the intervention clusters ..................................... 127 Table 28. Focus group guide: District level health decision/policy makers .............................. 128 Table 29. Unit costing proforma .............................................................................................. 129 Table 30. Unit costing proforma health facilities ..................................................................... 132 Table 31. Disability weights: maternal morbidities in the CLIP Trial ....................................... 134 Table 32. Disability weights: newborn morbidities in the CLIP Trial....................................... 136 xii List of Figures Figure 1. CONSORT diagram ................................................................................................... 26 Figure 2. Research plan for economic evaluation of the CLIP interventions .............................. 38 Figure 3. Societal costs: model inputs........................................................................................ 39 Figure 4. Decision analytic tree model....................................................................................... 40 Figure 5. Conceptual framework and approach for societal perspective ..................................... 56 Figure 6. Healthcare providers and facilities in the public and private health sector ................... 68 Figure 7. Cost-effectiveness plane showing statistical uncertainty ............................................. 99 Figure 8. Cost-effectiveness acceptability curve for the CLIP Trial ......................................... 100 Figure 9. One-way sensitivity analysis .................................................................................... 101 Figure 10. Time trends for costs and rate of adverse pregnancy outcomes avoided .................. 102 xiii List of Abbreviations ANC, antenatal care; ANM, auxiliary nurse midwives; APE, agente polivalente elementar, ASHA, accredited social health activist; CEA, cost-effectiveness analysis; CEAC, cost-effectiveness acceptability curve; CEmOC, comprehensive emergency obstetric care; cHCPs, community health care providers; CHEW, community health extension workers; CI, confidence interval; CISM, manhiça health research centre; CLIP, community level interventions for pre-eclampsia; cRCT, cluster randomized control trial; DALY, disability-adjusted life years; FG, focus group; GDP, gross domestic product; HDP, hypertensive disorders of pregnancy; HLCF, higher-level care facility; HRU, health resource utilization; HTA, health technology assessment; ICER, incremental cost-effectiveness ratio; IRB, institutional review board; xiv ISPOR, international society of pharmacoeconomics and outcomes research; LHWs, lady health workers; LMIC, low-and-middle-income countries; MDG, millennium development goals; MDs, medical doctors; mHealth, mobile health; MMR, maternal mortality ratio; MNH, maternal and newborn health; OOP, out-of-pocket; OR, odds ratio; PE, pre-eclampsia; PHC, primary health centre; PIERs, pre-eclampsia integrated estimate of risk; PKR, Pakistani rupees; POM, piers on the move application; PRE-EMPT, pre-eclampsia/eclampsia, monitoring, prevention and treatment; QALY, quality-adjusted life years; TBA, traditional birth attendants; UBC, University of British Columbia; UNICEF, United Nations Children’s Emergency Fund; WHO, World Health Organization; WRA, women of reproductive age. xv Acknowledgements I express my sincere gratitude to my faculty, supervisors, and academic advisor at the University of British Columbia (UBC), for their continuous support, and guidance throughout my academic journey: Peter von Dadelszen, Craig Mitton, Laura Magee, Stirling Bryan, and Dan Rurak. I owe particular thanks to site collaborators for their input on study design, data collection tools, and coordination with stakeholders in the Ministry of Health, Pakistan: Zulfiqar Bhutta, Rahat Qureshi, Sadiqa Naeem Jafarey, Sana Sheikh, and Zahra Hoodbhoy. I thank to field supervisory team members in Pakistan for organizing training of data collectors, logistic/transport arrangements for my visits to health facilities, field-level monitoring of data collection, and data transfer to the UBC: Imran Ahmed, Farrukh Raza, and Yaar Muhammad. I am also grateful to Government of Sindh Health Department officials for their support in the overall economic evaluation of CLIP Pakistan trial: Dr. Iqbal Hussain Durrani (Secretary Health), Dr. Hassan Murad Shah (Director General Health), and Dr. Pir Ghulam (Coordinator Lady Health Workers Program, Sindh). A special thanks to my fellow students, friends and central office CLIP team members in Vancouver: Tania Conte, Graham Shaw, Selva Bayat, Neale Smith, Penny Brasher, Vivian Ukah, Tatenda Makanga, Chirag Kariya, Beth Payne, Larry Li, Jeffrey Bone, Marianne Vidler, Sharla Drebit, Sumedha Shgroupa, Michelle La, Maggie Kinshella, Tang Lee, Mansun Lui, Domena Tu, and Roshni Nair. I would like to acknowledge the time contributions and welcoming attitude of all study participants in the communities and health facilities: pregnant women, husbands, fathers’ in-law, mothers’ in-law, LHWs, medical doctors at referral health facilities, and district health management team members in Matiari and Hyderabad Sindh Pakistan. Their support really xvi helped in timely data collection and provided local insight on societal perspective reported in this thesis. Most importantly, I would like to thank my parents, Yasmin Khowaja (spouse), and Aareer Khowaja and Aimen Khowaja (children) for their affection, prayers and constant support. Having them in my life is a true blessing, and I am forever grateful for their sacrifices and helping me cope-up with academic stressors or time pressures. This work is part of the University of British Columbia PRE-EMPT (Pre-eclampsia/Eclampsia, Monitoring, Prevention, and Treatment) initiative supported by the Bill & Melinda Gates Foundation. I also acknowledge, and feel honored to have received several prestigious awards: Canadian Institute of Health Research, Vanier doctoral scholarship, Killam pre-doctoral fellowship, UBC’s Four Year Fellowship, Aga Khan Foundation International Scholarship, Vancouver Coastal Health Rising Star award, UBC’s Ibn-e-Battuta award for field research, and UBC’s Warren George Povey award in global health. 1 Chapter 1: Introduction 1.1 Background Reduction of adverse pregnancy outcomes has received much scholarly attention in the scientific reproductive, maternal, and perinatal health literature over the last two decades (1,2). Recent advancements in the diagnostic, preventive, and therapeutic interventions have contributed to a significant decline in maternal and perinatal mortality in developed countries; however, very little change is evident in low-and-middle-income countries (LMICs) (3,4). The hypertensive disorders of pregnancy (HDP), namely pre-eclampsia and eclampsia (PE/E), are a leading cause of high maternal and perinatal mortality in LMICs (5). 1.2 Definition of pre-eclampsia and eclampsia PE is an acute onset of high blood pressure in pregnancy, when the systolic blood pressure is 140 mmHg or higher, or diastolic blood pressure is 90 mmHg or higher (6). Another definition describes PE as when a normotensive woman experiences a systolic blood pressure increase of 30 mmHg or more, or a diastolic blood pressure increase of 15 mmHg or more, at any point between week 20 of gestation and two weeks postpartum (7). Several clinical studies have suggested that PE is commonly reported towards the end of pregnancy, whereas a few studies reported that PE could be observed as early as week 20 of gestation (8). According to the classic definition, a woman identified with PE will present with the protein in the urine (proteinuria), swelling in the extremities (peripheral edema), headache, or a sudden increase in body weight, and may experience impaired vision (9). Eclampsia is the more serious condition. It is characterized by uncontrolled seizures and may result in coma or death. In the absence of early identification and timely case management of PE, women are at higher risk of disseminated intravascular coagulation, pulmonary edema, 2 placental abruption, acute renal failure, circulatory collapse, and cerebral hemorrhage. Unmanaged PE may also result in severe fetal problems, including intrauterine growth retardation, birth asphyxia, premature birth, and low birth weight (10). Review of the literature also revealed variations in international and research guidelines in terms of specific definitions for PE used in clinical or research contexts (11,12). Given such variations in the guidelines, trends of incidence for PE/E are not consistent in clinical settings versus population settings and may result in gross underestimation of the condition (13). 1.3 Situation analysis: global context of PE/E Globally, PE is diagnosed in 10 million women, and results in 76,000 maternal deaths and 500,000 perinatal/neonatal deaths, every year (14). The situation analyses revealed that the risk of developing PE/E and maternal mortality varies greatly depending on the geographical setting where a woman experiences her pregnancy. In a hospital-based retrospective cohort study of 97,270 pregnant women, from 35 hospitals in central and northern Alberta, Canada, the incidence of PE was 1.7%, and there was a significant difference in the mean birth weight of babies (ranging from 547.5 grams to 239.5 grams) born to women identified with PE (15). In contrast, epidemiological studies conducted in LMICs report that women in these settings are seven times more likely to develop PE compared to women in developed settings (14). Over 99% of PE/E-associated maternal and perinatal deaths are clustered in South Asia and sub-Saharan Africa (16). Because the outcomes of PE/E are largely dependent on timely interventions, many affected women in rural and hard-to-reach settings suffer severe disability or lose their lives due to delays in early identification, triage, transport, and treatment (16). The odds of dying from PE/E are approximately 300 times higher for women living in the LMICs (17). 3 Given the biological plausibility for PE/E, the variation in the number of new cases reported per annum (incidence) has intrigued researchers. Some suggest that the variability of PE incidence is the consequence of different case definitions. Others attribute it to differences in diagnostic procedures or laboratory tests in various regions. It is likely that the burden of disease is changing in developed countries that have more sophisticated interventions for monitoring and treating PE/E. 1.4 Community-level interventions for pre-eclampsia (CLIP) trials The CLIP Trials are three independently-powered cluster randomized control trials (cRCTs) that aimed to reduce pre-eclampsia-related and all-cause maternal and perinatal mortality and major morbidity by 20% or more in intervention clusters in Sindh Province (Pakistan), Maputo and Gaza provinces (Mozambique), and Karnataka State (India). The CLIP interventions introduced a mobile health (mHealth) platform to facilitate triage and treatment by community health workers, along with community engagement to improve care-seeking practices (18). The CLIP package of care was supplemented with the routine pregnancy care, and consisted of:  Provision of HDP-oriented antenatal care through home visits by community health care providers (cHCPs) who carried electronic tablets programmed with a mHealth application for identifying women at risk of PE [Pre-eclampsia Integrated Estimate of Risk (PIERS) (19), on the Move (POM) (20)].  Community-based case management and referral recommendation(s) for women based on the CLIP “triggers.” The recommendations consisted of one of the following, based on the risk profile of the woman: 4 o Treatment with oral methyldopa (antihypertensive therapy), intramuscular magnesium sulphate (i.m. MgSO4) (anticonvulsant therapy), and urgent transport (within 4 hours) to a health facility. o Treatment with i.m. MgSO4 and urgent transport (within 4 hours) to a health facility. o Urgent transport (within 4 hours) to a health facility. o Non-urgent transport (within 24 hours) to a health facility. o Continuation of routine care.  Community engagement through one-on-one sessions with pregnant women, and group sessions with dyadic household decision makers (husbands, fathers-in-law) and community leaders about HDPs, signs/symptoms, potential consequences, pre-permissions for maternal transport, and fundraising activities for transport and treatment costs. In the intervention group, the cHCPs were trained to inquire about the woman’s symptoms, assess blood pressure using a digital device, and check for evidence of protein in urine using a dipstick on the first visit and on any subsequent visits where the systolic blood pressure was 140 mmHg or higher. This clinical assessment informed the diagnosis of and risk assessment for PE/E. More detailed description of the CLIP trial methods/protocol has been published elsewhere (18). In the control group, pregnant women continued to receive routine pregnancy care at least every four weeks, either at a health facility or from cHCPs at home, as per the standard of care. The routine pregnancy care comprised of: 1) health education related to pregnancy care and birth preparedness, 2) psychological counselling on reproductive health and family planning, and 3) 5 advice/reminder on vaccinations. Those cHCPs conducting home visits were not trained to assess or counsel on HDP and did not have any components of the CLIP package of care to diagnose, triage, or treat PE/E in the control group. In addition, the household surveillance for collecting pregnancy outcomes and health system capacity enhancement through workshops/seminars for facility-level care providers were undertaken in both intervention and control groups. 1.5 The purpose The aim of this study was an economic evaluation of the CLIP intervention to inform policy decisions about resource allocation for post-trial programme scale-up in Sindh, Pakistan, should the primary trial be found clinically effective. 1.6 Study hypothesis The CLIP package combined with routine pregnancy care, compared to routine pregnancy care alone, will result in a low incremental cost-effectiveness ratio (ICER) in reducing maternal and perinatal mortality and major morbidities using a country-specific willingness to pay (WTP) threshold of USD$1,468 (i.e. GDP per capita as of 2016) in Pakistan. 1.7 Research questions This dissertation is based on four key research questions aimed at understanding the appropriate design of the economic model and common reporting metrics necessary for an economic evaluation of the CLIP interventions in Sindh, Pakistan. 1. What is known about the costs and cost-effectiveness of PE/E interventions for risk reduction/complication prevention, screening/diagnosis, and appropriate case management in the developed countries and elsewhere in LMICs? 6 A comprehensive review of published epidemiological and economic studies that focus on PE/E interventions was done. Methodological approaches used in the cost-effectiveness studies guided the design of the economic model for the CLIP. 2. What are the costs to the health system and to pregnant women and their families related to mHealth interventions in LMICs? This question was explored through mixed-methods research to assess societal perspective (including costs to the health system and families) of cost drivers for health technology assessment in Sindh, Pakistan. 3. What is the cost of maternal and newborn health services at various levels of health facilities in the public and private health sectors? A cross-sectional survey of health providers/facilities was undertaken in the study catchment areas to determine the unit costs of various maternal and newborn health services in Sindh, Pakistan. 4. What are the costs relative to benefits for CLIP combined with routine pregnancy care, compared to routine pregnancy alone in Sindh, Pakistan? The incremental societal costs and years of life lost (YLL) were compared in the intervention and control groups. The model-based economic analysis was undertaken drawing input parameters from the CLIP Trial, and mixed-methods research in this study. This thesis is organized into seven chapters: Chapter 1 describes the PE/E definitions, global context, study hypothesis, and research questions; Chapter 2 reports the review of epidemiological and economic literature on PE/E interventions; Chapter 3 presents the overview of study methodology and data collection plans for economic analysis in three CLIP countries. 7 The next four chapters focus on data, analyses, and findings related to economic evaluation of the CLIP Pakistan Trial: Chapter 4 explains the qualitative research on the societal perspective for cost drivers; Chapter 5 compares the maternal and newborn health services costs in the public and private health sectors; Chapter 6 illustrates the cost-effectiveness analysis of CLIP interventions, and Chapter 7 summarizes the main study findings and directions for future research. 8 Chapter 2: Evidence of costs and cost-effectiveness of pre-eclampsia / eclampsia interventions: a literature review 2.1 Background The trajectory of PE/E could impact utilization of health services and may lead to life-threatening complications or even death of a mother or newborn in the absence of appropriate management (10). An economic study conducted in the United Kingdom reported that PE/E is one of the most common reasons for antenatal admission to hospitals (20%) and obstetric admissions to intensive care units (25%) (21). Studies in LMICs that evaluated the economic consequences of health care expenditures found that out-of-pocket (OOP) spending combined with the loss of household income due to ill health could plunge families into a vicious cycle of poverty (22). Understanding the cost and cost-effectiveness of alternative strategies for PE/E identification and appropriate case management is essential to inform decisions about resource allocation and guide the economic appraisal of recent initiatives such as CLIP in the context of LMICs. 2.2 Research question What is known about the costs and cost-effectiveness of PE/E interventions for risk reduction/complication prevention, screening/diagnosis, and appropriate case management in the developed countries and elsewhere in LMICs? 2.3 Objectives The primary objectives of this study were to describe the economic burden of PE/E, identify clinical interventions and strategies for PE/E, and document the cost-effectiveness of these interventions based on the recent literature. The secondary objective was to appraise 9 methodological approaches used in economic studies to guide the design of an economic model for the CLIP Trial. 2.4 Methods 2.4.1 Study design and search strategy A literature review was conducted using research databases to identify epidemiological and economic studies related to PE/E interventions. Three electronic databases were searched: PubMed, EconLit, and Health Economics Evaluation Database (HEED), developed by the UK National Health Service. The literature identified several clinical practice guidelines for HDP that describe PE/E differently, and no standard PE/E definition was recommended (12). This review used the definition of PE/E employed by the Society of Obstetrics and Gynecologists of Canada (SOGC) in reference to the CLIP Trial (23). The SOGC definition defines PE as an acute onset of hypertension, when the systolic blood pressure is 140 mmHg or higher, diastolic blood pressure is 90 mmHg or higher, and proteinuria is 1 or higher on dipstick or the PrCr ratio is 30 mg/mmol or higher on random urine. Eclampsia was defined as a serious condition characterized by uncontrolled seizures for women with signs and symptoms of PE after week 20 of gestation, which may result in a coma or death. The developing countries were defined as countries with gross national income per capita below US $11,905, according to the World Bank (24). The key search expressions used in the databases were: (“cost of hypertensive disorders of pregnancy intervention in developing countries” OR “pre-eclampsia interventions in developing countries” OR “eclampsia interventions in low-and-middle income-countries” OR “cost-effectiveness of pre-eclampsia interventions in low-and-middle income countries” “cost minimization interventions for pre-eclampsia in developing countries” OR “economic evaluation 10 of pre-eclampsia interventions in low resource settings” OR “economic appraisal of pre-eclampsia and eclampsia measures in developing countries”). These search expressions returned a very limited number of epidemiological and economic studies in the developing countries, so terms related to “low-and-middle-income countries, developing countries, and/or low-resource settings” were omitted in subsequent searches. 2.4.2 Inclusion and exclusion criteria Studies were considered eligible for review if they met the following inclusion criteria:  Focus on health resource utilization or cost of illness related to PE/E  Epidemiological investigations of PE/E interventions with focus on risk reduction/prevention, screening/diagnosis, and case management  Economic evaluations that report on costs relative to health benefits of PE/E interventions  Full-text articles available in English language  Published between January 2002 and December 2017 Conference abstracts and non-research articles were excluded. Duplicate studies were identified and removed. 2.4.3 Screening and data extraction Applying these search expressions to the electronic databases produced a list of articles for title screening. Abstracts of related studies were reviewed, and studies were excluded based on the criteria described above. During the review of full-text articles, we found references to two related studies conducted by The National Institute for Health and Care Excellence (NICE) in the UK. These two studies were included in the full-text review. 11 Data were extracted from full-text articles using a PICO (population, intervention, comparator, key outcomes) format to report on study characteristics (25). The summary table lists studies in chronological order according to the publication year. In the epidemiological publications, we reviewed point estimates in reference to the odds ratio (OR), relative risks (RR), and 95% confidence intervals (CI). Among the economic evaluation publications, we reviewed costs of interventions, health benefits as quality-adjusted life years (QALYs), net monetary benefit, cost savings, and ICER of comparators. Costs data included in this review represent United States dollars (US$), UK pounds (£), or Euro (€). This review also reports on methodological approaches in the economic studies, such as sources of cost and outcome data, modeling technique, outcome parameters, perspective adopted, discounting, time horizon, and methods for dealing with uncertainty. 2.5 Results Of the 473 articles retrieved from PubMed, Econ.Lit, and HEED searches, 447 were excluded because they were duplicates (n=6), did not focus on PE/E interventions/cost-effectiveness (n=420) or were not original research studies (n=21). The remaining 28 studies (26 articles retrieved plus 2 NICE articles) were reviewed. Of these 28, 15 publications (54%) represent a wide range of economic studies such as cost analysis, cost-benefit analysis, cost-effectiveness analysis, and budget/financial impact analysis. The epidemiological studies (n=13) include cross-sectional surveys, case-control, prospective cohort, and clinical trials. Only summary findings from the meta-analyses were reviewed, as some primary studies reported in the meta-analyses neither met the inclusion criteria nor showed-up in the literature search (Figure 1, pg 26). 12 2.5.1 Economic burden of PE/E Three studies on the topic of cost of illness were included in this review. These studies examined PE/E-related costs and the financial impact on the health system mainly in a public-payer context. Studies from the US reported that hospitalization costs for PE management and associated complications averaged $11,208 per patient (26). This translated into an aggregated cost to the US health care system of $1.03 billion in maternal health care costs and $1.15 billion in costs for infants born to mothers with PE (27). Another follow-up study reported excessive maternity services resulting in higher costs of PE (average €5,243 per case), compared to €2,452 per case for an uncomplicated pregnancy (28). This study also estimated the annual national costs ranging from €6.5 to 9.1 million, as the prevalence of PE increased from 5% to 7%. Economic studies related to the financial burden of PE/E to society, and particularly OOP spending to pregnant women and their families, were not available in the literature (Table 1, pg 21). 2.5.2 Interventions for risk reduction/complication prevention of PE/E Six epidemiological studies and one economic study on the topic of risk reduction and prevention of PE/E were included in this review. A recent meta-analysis on the prophylactic use of aspirin reported significant risk reduction (OR = 0.84, 95% CI: 0.11 to 0.35) of PE in the East Asian population (29). A cost-benefit analysis on the prophylactic use of low-dose aspirin (81 mg) compared with no prophylaxis demonstrated cost savings of $377 million in direct medical care in the US annually. In this study, low-dose aspirin was found to have over 90% probability of cost-effectiveness at the WTP threshold of $100,000 per neonatal QALYs gained (30). Another meta-analysis on calcium supplementation during pregnancy found 45% risk 13 reduction of PE in the LMICs (31). We did not find any publications that support the cost-effectiveness of calcium supplementation during pregnancy. Furthermore, several observational studies have suggested lifestyle or dietary modifications for preventing PE/E, but no evidence on the cost-effectiveness was found in the published literature. A longitudinal study from rural Congo found dietary intake of less than three servings of vegetables per day increased PE risk by 33%, whereas moderate physical activity reduced the risk by 37% (32). There is an abundance of literature on the role of vitamins supplementation in PE risk reduction. In Uganda, a randomized placebo-control trial evaluated the use of vitamin C (1,000 milligrams as ascorbic acid) compared to placebo. In this study, investigators did not find a significant difference in the incidence of PE between the intervention and control groups (33). Another study from Bangladesh reported higher odds of developing PE/E (OR= 5.14, 95% CI: 1.98 to 13.37) in women with vitamin D deficiency (34). Administration of folic acid together with multivitamins was found to reduce the risk of developing PE by 63% (95% CI: 0.18 to 0.75) (35) (Table 1, pg 21). 2.5.3 Interventions for screening/diagnosis of PE/E Three epidemiological studies and six economic studies about screening and diagnostic tests were included in this review. Most studies focused on PE diagnosis in the antepartum period, and very few addressed prognostic investigations in the postpartum period. A recent study reported an early prediction of PE through spot urinary albumin-to-creatinine ratio of 9.8mg or more per gram of creatinine. The sensitivity of this test was as low as 67%, and specificity was as high as 76% (36). The economic analysis evaluated three alternative diagnostic practices: (i) Protein-creatinine ratio (PCR) alone; (ii) Automated urine analysis followed by PCR; and (iii) Automated urine analysis followed by 24-hour urine collection. In a cohort of 14 60,000 pregnant women with mild to moderate hypertension, PCR alone was less expensive (i.e., £226,800) and generated higher QALYs compared to other two alternatives (37). Another economic study of women with gestational hypertension compared visual dipstick urine analysis versus automated urine analysis using a reagent-strip reading device. This study found an incremental cost of £51,540 for the automated procedure and incremental QALYs of 415 in a cohort of 60,000 women with moderate gestational hypertension (38). The economic study evaluated Doppler tests for identification of PE and concluded no evidence of cost-effectiveness for high-risk versus low-risk pregnancies (39). Literature also revealed an emerging scope of the mHealth technologies and novel serum tests of placental biomarkers for PE identification and risk stratification of adverse pregnancy outcomes. A recent study reported the high accuracy of a handheld semi-automated blood pressure device (Microlife 3AS1-2) used for PE screening, which costs less than $20 in LMICs (40). Others have reported the role of potential placental biomarkers, namely alpha-fetoprotein, placental growth factor, soluble tumor necrosis factor receptor-1, and retinol binding protein-4 in the prediction of early onset of PE (41). A study from Israel evaluated the economic benefit of first-trimester screening of multiple placental biomarkers compared with no screening. This study found a cost per QALY of less than $10,000 for screening, given the prevalence of pre-eclampsia at 3% (42). Another recent study, from the Italian national health service, evaluated the budget impact of placental biomarker screening compared to standard practice for prediction of PE. This study reported health care costs as low as €1,714 in women who received biomarker screening, compared to €2,384 in the group that underwent standard diagnosis. This demonstrated substantial cost savings for placental biomarkers through improvement in diagnostic accuracy and reduction in unnecessary maternal hospitalization (43). Similarly, 15 placental biomarkers (SFlt-1 and PlGF), compared to standard diagnosis, were attributed to cost savings of £945 per pregnancy in the UK (44) (Table 1, pg 21). 2.5.4 Interventions for appropriate case management of PE/E Four epidemiological studies and five economic studies on the topic of case management of PE/E were included in this review. A number of clinical trials have evaluated the choice of surgical and therapeutic interventions for case management or prevention of complications associated with PE/E. An open-randomized controlled trial evaluated immediate delivery versus expectant monitoring for HDP between 34 and 37 weeks of gestation (HYPITAT-II). Investigators found that immediate delivery reduces the risk of adverse maternal outcomes (RR=0.36, 95% CI: 0.21 to 1.11) in women with non-severe hypertensive disorders at 34 to 37 weeks of gestation (45). The trial-based economic evaluation compared costs of labor induction and expectant monitoring in women identified with PE at term. A study from the Netherlands found that inducing labor for immediate birth resulted in cost savings of €831 (95%CI: €144 to €1,561) per woman identified with mild or moderate levels of hypertension at 34 to 37 weeks of gestation (46). Another multicenter trial on the Control of Hypertension in Pregnancy Study (CHIPS) evaluated less-tight (i.e., target diastolic blood pressure 100 mmHg) versus tight control (target diastolic blood pressure 85 mmHg) with most commonly used antihypertensive drugs (either methyldopa or labetalol). In this study, women in the less-tight control group were found to have significantly higher maternal hypertension (incremental prevalence 13.1%, p-value <0.001) (47,48). In British Columbia, Canada, the cost analysis of the CHIPS Trial indicated higher health system costs of woman-and-infant hospitalization in the less-tight control strategy ($30,593) compared to tight control ($24,776) (49). 16 Furthermore, the use of magnesium sulphate (MgSO4) compared to a placebo was evaluated in a Magpie Trial. This study found that prophylactic use of MgSO4 prevented deaths or serious morbidity (RR= 0.84, 95% CI: 0.60 to 1.18) (50). Similarly, prophylactic use of MgSO4 was evaluated in a randomized control trial (PIPES) in Dhaka, Bangladesh, which found that a low-dose regime reduced the incidence of eclampsia by 1.48% (51). The economic analysis of the Magpie Trial reported ICERs for preventing each case of eclampsia with MgSO4 intervention as low as $456 in low-income countries, compared to $21,202 in high-income countries (52). In Pakistan, investigators reported lower costs of $0.56 for a single loading dose of MgSO4, compared to a cost of $2.40 for the standard prophylaxis regimen (53). Another study reported that a carefully designed outpatient management program, including patient education, blood pressure monitoring, and urine protein measurement for women with PE saved an average of $2.50 per woman compared to standard pregnancy care (54) (Table 1, pg 21). 2.5.5 Methodological approaches on economic studies Six economic studies included in this review evaluated the cost-effectiveness of PE/E interventions using the decision analytic tree model. Only two studies used trial-based economic evaluation, and a few reported on cost-benefit or budget impact analysis. The model parameters for health outcomes were mainly drawn from previously published literature, Cochrane review, and interviews/consultation with clinical experts or health care payers. The cost inputs were derived from multiple sources including national health services/administrative databases, health insurance providers, and health facility records. Almost all studies reported from the health system perspective (i.e., public payer); one study reported on societal perspective in the Netherlands. Only four studies reported on discounting; a 3% discount rate was applied. The majority of economic studies (n=7) modeled a time horizon of one year or less, and sensitivity 17 analysis was undertaken in many studies to address parameter uncertainties. The health outcome parameters varied in many studies, such as incremental cost per PE case identified, the incremental cost per E case averted, the incremental cost per PE case managed, and cost savings per pregnancy (Table 2, pg 25). 2.6 Discussion This chapter reports on the epidemiological and economic literature that focuses on costs and cost-effectiveness of PE/E interventions. Our review identified a number of PE/E interventions that appear to be effective (e.g. diet, and physical activity, prophylactic administration of aspirin and treatment with magnesium sulphate). The epidemiological evidence on PE risk reduction or complication prevention strategies is largely fragmented, as few randomized controlled trials exist. More emphasis is given to the early identification and case management of PE/E to avert disease-related complications and mortality. The economic literature on the cost-effectiveness of PE/E interventions only represents work done in developed (western) countries, and knowledge gaps exist for cost-effectiveness of PE/E interventions in the context of LMICs. All studies identified in this review investigated either clinical or laboratory related intervention and lacked information on comprehensive community-based interventions for PE/E. The CLIP package of care focused more on the community-based HDP screening, case management and community engagement aspects for the design of the intervention. The rising cost of care and alarmingly high OOP expenses in health care are important challenges to health systems in LMICs (22). The economic studies in this review focused on health system costs and included only the public payer’s perspective. Ironically, the literature on societal costs related to PE/E is almost nonexistent in parts of the world where most PE/E-related deaths occur. Incorporating a societal perspective (i.e., including costs to pregnant women and 18 their families and to the health system) provides a complete reflection of the financial burden and health outcomes. Recent guidelines on the economic evaluation strongly recommended societal perspective for cost-effectiveness analyses (55). Further research is needed to understand and explore the societal perspective of cost drivers related to PE/E care in low-resource settings. Such in-depth information is critical to inform the economic analysis of the CLIP Trial. The findings from this review highlight a potential issue of contextual variation for costs and health outcomes, as reported in the economic studies. The cost input parameters were derived from multiple sources including published literature, active clinical trials, national health services databases, and consultation with clinical experts or health care payers. Such an inclusive approach to data collection, although beneficial to capture a wide range of information, may introduce the uncertainty of costs parameters available from multiple sources. Investigators did not explain how they dealt with possible cost uncertainties, particularly in four economic studies in which sensitivity analysis was not performed. This finding suggests that it is important to include sensitivity analysis in the economic evaluation of the CLIP Trial. Moreover, the health outcome parameters in the selected economic studies were inconsistent and reflected on surrogate outcomes rather than the tangible outcomes (e.g., number of lives saved, or life years saved) commonly used in the economic literature in LMICs (56). Thus, policy decisions related to program allocative efficiency—that is, choosing an alternative strategy with lower cost and more health gains—are severely compromised when health outcomes are inconsistently reported for alternative strategies (57). This finding further informed the inclusion and modeling of life years saved as the standard health outcome metric in the economic evaluation of the CLIP Trial. In LMICs, the health care system includes the private and public health sectors—a major divide that is seldom included in economic analyses. The costs of diagnostic tests, drugs, their 19 administration, and health facility-level admissions vary substantially between the public and private health sectors (58). It is imperative to assess, particularly in the societal perspective, because individuals may pay differently for similar health services. Further research on ascertaining unit costs of maternal and newborn health services at public and private health facilities (i.e., a real-world scenario) will be useful to inform the economic analysis of the CLIP Trial. Furthermore, the contextual variations can be anticipated in the price of mobile health technology, Internet infrastructure, and health care costs between countries in the same region, which suggest country-specific economic analysis in the selected CLIP countries. 2.6.1 Strengths and limitations This literature review reports on the epidemiological and economic evidence on costs and cost-effectiveness of PE/E interventions evaluated in the last 15 years. The knowledge gaps highlighted in this review generated important areas for future research on PE/E interventions, particularly relevant in the context of LMICs. The methodological approaches in the cost-effectiveness studies included in this review provided important directions in the design of economic appraisals of future PE/E interventions, such as the CLIP Trials. This review includes four limitations that warrant discussion: (i) the literature review focused on epidemiological and economic studies published in 2002 or later, and there is a possibility of missing information published before 2002; (ii) the literature review examined only English-language publications, so there is a possibility of missing information published in other languages; (iii) many studies were conducted in the developed countries, so findings may not apply generally to LMICs; and (iv) there is the possibility that this review is affected by publication bias, i.e., authors selectively publishing epidemiological studies with direction and strength of findings that favor clinically effective interventions (59). 20 2.7 Conclusions Reduction and prevention of PE/E risks and complications has received little attention in the current epidemiological and economic literature. The evidence on cost and cost-effectiveness of P/E interventions concentrated on the health system perspective and lacked information on societal costs. Most of these studies were conducted in developed countries, and knowledge of cost-effective PE/E interventions is very limited in the context of LMICs, where the majority of the disease burden and associated mortality occur. Ascertaining costs of health services from a mix of public and private health facilities and consistent use of health outcome parameters will be beneficial in the future. 21 Table 1. List of articles included in the literature review Scope Publication, year Study methodology Study population (P) Intervention (I) Comparator (C) Study outcomes (O) Risk-reduction or complication prevention Jie Gan et al. 2016 A systematic review and meta-analysis Women at risk of pre-eclampsia (cohort of East Asian and non-East Asian pregnancies) Low-dose Aspirin Placebo or no-treatment Low-dose Aspirin reduced pre-eclampsia risk (OR = 0.20, 95% CI: 0.11–0.35) in East-Asian; and non-East Asians (OR = 0.84, 95% CI: 0.77–0.92). Werner et al. 2015 Cost-benefit analysis Pregnant women in the United States of America Universal prophylaxis of low-dose Aspirin No prophylaxis Universal prophylaxis offers net monetary benefit of $365 million assuming four million births each year in US. Agrawal et al. 2015 Cross-sectional survey Indian women of reproductive age (15-49 years) who have had live birth in the last five years NA NA Iron and folic acid supplementation reduced frequency of pre-eclampsia or eclampsia symptoms. Kiondo et al. 2014 A randomized placebo controlled clinical trial Uganda pregnant women aged 15-42 years living 15KM or less from the hospital 1000mg of vitamin C (as ascorbic acid) Placebo No difference in vitamin and placebo groups on the incidence of pre-eclampsia (3.1% versus 4.1%; RR 0.77; 95% CI: 0.37-1.56), and severe pre-eclampsia (1.2% versus 1.0%; RR 1.25; 95% CI: 0.34-4.65). Ullah et al. 2013 Case control study Women identified with pre-eclampsia, eclampsia, and normotensive pregnancies in Bangladesh NA NA Vitamin D insufficiency (i.e. 25 (OH) D levels < 30ng/ml) was the risk factor for pre-eclampsia and eclampsia (OR 3.9 (95% CI=1.18-12.87). Imdad et al. 2011 Meta-analysis All pregnancies in developing countries Calcium supplementation Placebo Calcium supplementation in pregnancy led to risk reduction of pre-eclampsia (RR 0.41; 95 % CI 0.24-0.69), neonatal mortality (RR 0.70; 95 % CI 0.56-0.88), and pre-term births (RR 0.88, 95 % CI 0.78-0.99). Longo-Mbenza et al. 2008 Hospital-based longitudinal assessments African rural black pregnant women hospitalized NA NA Diet rich in vegetables and physical activity prevents pregnancy-induced hypertension. 22 Scope Publication, year Study methodology Study population (P) Intervention (I) Comparator (C) Study outcomes (O) Screening/ diagnosis Frusca et al. 2017 Budget impact analysis Women with suspected pre-eclampsia in Itlay Placental biomarker (i.e. sFlt-1/PlGF ratio) Standard pre-eclampsia screening Lower healthcare costs €1714 in women screened with placental biomarker, compared to €2384 in standard screening. Nupur Gupta et al. 2017 A retrospective study of urine samples Early pregnancies i.e., before 20 weeks of gestation. NA NA Spot urinary albumin-to-creatinine ratio (i.e., > 9.6 mg/g of creatinine) was strong predictor of pre-eclampsia, indicating 67% sensitivity and 76% specificity. Nevalainen et al. 2017 A retrospective study of maternal serum obtained/analyzed at the hospital setting Pregnant women in the first trimester in Finland NA NA Biomarkers such as Alpha fetoprotein (AFP), placental growth factor (PIGF), soluble tumor necrosis factor receptor-1 (sTNFR1) and retinol binding protein-4 (RBP4) were found strong predictor of early onset of pre-eclampsia. Hannah L et al. 2015 Prospective observational study Pregnant women in low-and-middle income countries. NA NA Microlife 3AS1-2 BP device was recommended for use in pregnancy, and preeclampsia screening. Shmueli et al. 2012 Cost-effectiveness analysis Pregnant women in Israel Routine screening using placental biomarkers Uterine Doppler as the standard care The ICER for intervention was USD$24,723 per pre-eclampsia case averted at 5% prevalence. The cost per quality of life-adjusted life year with screening was 140mmHg) during a CLIP pilot cRCT visit; and (iii) available for a discussion lasting at least 60 minutes. The literature suggested that household decision makers traditionally make decisions about women’s health issues in LMICs(71). As 44 inclusively as possible, household decision makers (e.g. husbands, mothers-in-law, fathers-in-law) of hypertensive pregnant women were eligible if they also expressed availability for at least 60 minutes of discussion. LHWs, and medical doctors (MDs) were eligible, if they participated in the implementation of the CLIP pilot cRCT. The health decision-/policy-makers were eligible if they were involved in decisions related to the execution of the CLIP pilot cRCT. Participants’ characteristics varied in terms of age, schooling years, and occupation (Table 4, pg 53). 4.3.4 Sampling procedures A purposive sampling strategy was imperative to this study design as we needed individuals or groups with lived experiences (i.e. information-rich cases) who could provide insight based on their own experiences related to HDP care sought/provided during the CLIP pilot cRCT. The sampling framework included 4 intervention clusters (~400 pregnancies per cluster in one year), in which we estimated 80 pregnant women would experience HDP (i.e. 5% of all pregnancies). Pregnant women and household decision-makers were identified through the trial recruitment logs, while health care providers and decision-/policy-makers were identified through health facility networks. The project-trained RAs approached eligible participants in their respective settings (i.e. either at their home or workplace) and invited them to take part in the study. We had anticipated 5 FGs (at least one FG per each group); however, we increased the number of FGs (two FGs per group; except health decision-/policy-makers) to achieve data saturation. In total, nine FGs were conducted: two with pregnant women/mothers-in-law (n=19 in total), two with husbands/fathers-in-law (n=17), two with LHWs (n=19), two with medical doctors at health facilities (n=20), and one with district-level health decision-/policy-makers (n=10). Participants were reimbursed for study-related transportation costs. 45 4.3.5 Data collection The FG guides were developed from the literature review. Key constructs included out-of-pocket (OOP) costs related to care seeking during pregnancy (86), and health service utilization in the context of LMICs (87). Participants of the study were asked to reflect on lived experiences and interpret situations related to financial costs as a result of their participation in a one-year pilot phase of the CLIP cRCT. FG guides were developed in English as the main language of literature review. FG guides were translated into Sindhi; and back translated into English. They were pilot-tested for comprehension and cultural sensitivity. Native Sindhi-speaking and project-trained RAs moderated the FGs. The RAs were local residents, with undergraduate degrees, who had experience as data collectors in previous maternal health research. Given cultural tradition related to the veil system, women are not allowed to participate in public meetings in the presence of men. To respect the cultural values and participants’ preferences, FGs were held separately with women and men. Likewise, separate FGs were organized for LHWs, MDs and health decision-/policy-makers given the logistics and ease of care providers. Data saturation was determined through a review of FG transcripts for new emerging codes/ideas, and the saturation point was deemed to have been reached when transcripts returned no new codes. 4.3.6 Data analysis and quality control FGs were taped using a digital voice-recorder and transcribed verbatim in Sindhi. The transcripts were imported to NVivo version 11 software [QSR, Doncaster Vic, Australia] for data analysis. The analysis of FG data was conducted in the same language [Sindhi], using the new version of NViVo that allows for coding in languages other than English. This increased rigor in the data analysis and prevented meaning loss from translation. The participants’ attributes were 46 analyzed, and their responses were coded on a hierarchy of tree nodes (i.e. branches of relevant constructs). A combined approach to data analysis inclusive of inductive and deductive reasoning was used to interpret emerging themes/sub-themes (88). The descriptive coding list provided a comprehensive understanding of thematic areas for subsequent interpretations (Table 5, pg 54). Data quality was ensured through random observations of FGs by field coordinators and a public health scientist, 10% (audit-trail) verifications of the content of manual transcripts by audio-recording reviews, and weekly debriefing sessions with moderators/transcribers. The FG moderator recorded a self-reflection after each session to describe personal thoughts and impressions to better contextualize the data, as well as to protect against self-bias. 4.4 Results Participants’ discussions were categorized into emerging themes and sub-themes as HDP-related health resource utilization, and/or as perceived costs to family, health system and program implementation. 4.4.1 Health care resource utilization for HDP The theme on resource utilization suggested patterns and types of care sought from health facilities in the catchments. Almost all pregnant women reported having sought health care after they were identified as hypertensive and referred to a health facility by LHWs. The majority visited public secondary and tertiary health facilities that were usually located outside study clusters, requiring families to arrange transport. In most cases, women were delivered by Cesarean delivery (C-section). One woman described her experience of visiting a distantly located public health facility in the following quote: 47 “We traveled to Shadadpur government hospital for the treatment. Doctors told … [me] … that my blood pressure was extremely high and I had to deliver the baby. They said that normal delivery is unlikely because of high blood pressure, therefore, they did C-section to deliver my baby”. Participant 6, FG 1, pregnant women and mothers-in-law A few pregnant women stated that they visited private general practitioners because of easy access, family preferences, and past unfavorable experiences with public sector health facilities. 4.4.2 Perceived cost drivers The theme on perceived cost drivers contextualized a broad range of costs to care receivers, health care providers and decision-/policy-makers.  OOP costs Pregnant women (i.e. care-receivers) emphasized OOP expenditure as the main barrier to seeking timely care. The OOP costs were described as in- and out-patient care, diagnostic procedures, informal care received from traditional healers, self-medication at home, childbirth, and transport to health facilities. The reported lump sum OOP costs ranged from 50 Pakistani rupees (PKR) (US$0.5) for diagnostic procedures to 10,000 PKR ($100) for in-patient hospitalization for childbirth (Table 6, pg 55). The OOP costs were usually perceived as a large financial burden for the family of a pregnant woman, and they had a substantial impact on their economic conditions and wellbeing. One husband described this in the following quote: “We don’t have savings for emergencies related pregnancy and childbirth. [We had to sell] family assets and borrowed money from friends/neighbors… It took us several months to pay back the borrowings.” Participant 7, FG 1, husbands and fathers-in-law 48 Alternatively, costs deterred families from seeking care: “Because I did not have money, we didn’t go to hospital when my wife was identified as hypertensive during the current pregnancy….we cannot afford health care expenses!”. Participant 7, FG 2, husbands and fathers-in-law Participants stated that husbands (in most cases) and/or mothers-in-law were responsible for accompanying pregnant women when they were transported to a health facility. Additionally, blood relatives, and/or family friends were often reported to have stayed with, and remained involved in patient care, when a pregnant woman was hospitalized for HDP management and for childbirth. A vast majority of family members were paid a lump sum wage of 150 to 500 PKR ($1.5 to 5) per day, or a fixed salary of 8000 to 13000 PKR ($80 to 130) per month. Consequently, missed wages for family members were considered to be a double economic burden for the family and others involved.  Health system costs LHWs and MDs (i.e. care providers) revealed health system costs associated with human resources, transport, and communication. LHWs described making once-monthly home visits to provide basic antenatal care to pregnant women in their catchment areas. Home visits were described to include screening pregnant woman for HDP, initial case management (if the pregnant woman was identified at risk for adverse birth outcomes), and referrals to health facilities. The duration of a completely routine visit, where the pregnant woman was found to be in good health, was 20 to 30 minutes. A more typical visit would take 30-60 minutes to complete if the pregnant woman required medications and/or the family needed counseling about referral to a health facility. One LHW described her typical home visit experience in the following quote: 49 “A pregnant woman would often speak about her problems. Most women have a problem of anemia and complain of pain. They share domestic worries related to their husband’s lack of support and misunderstandings with mothers-in-law. Therefore, it sometimes takes one hour to complete a visit.” Participant 6, FG 1, LHWs Moreover, a few LHWs mentioned families’ expectations that they (LHWs) accompany pregnant women to health care facilities and exchange their cell phone numbers to maintain communication for emergencies. MDs practicing at primary and secondary health facilities considered performing basic diagnostic tests such as blood pressure monitoring, blood sugar testing, proteinuria measurement, and ultrasound imaging. Prescribing patterns revealed frequent use of oral tablet methyldopa (aldomet) to treat high blood pressure, injection [of] diazepam to control seizures, and referral to higher-level health facility for severe complications during pregnancy. MDs practicing at tertiary-level health facilities considered advanced maternal and fetal tests, such as 24-hour urine collection, blood tests (i.e., serum uric acid, prothrombin time, activated partial thromboplastin time, electrolyte profile, complete blood count, lipid profile, liver function tests), and fetal imaging. All recommended close monitoring as an in-patient and interventions that included intravenous administration of magnesium sulfate, blood transfusion, and Cesarean delivery.  Program implementation costs Health decision-/policy-makers recognized program implementation costs (such as mobile health infrastructure, staff training, and monitoring/supervision) as major investments for the health system. All health decision-/policy-makers believed that mHealth technologies have potential to improve maternal and perinatal health in LMICs. They discussed the prerequisites for effective program implementation: (i) a major initial investment in establishing the mHealth 50 platform (e.g., purchase of electronic tablet devices for app-guided clinical care, computers for downloading and data-sharing, and high speed internet to facilitate data synchronization); and (ii) patient screening equipment (e.g., digital blood pressure devices, pulse oximeters, and urine dipsticks). Many health decision-/policy-makers reported the need to train all existing staff through refresher programs and to provide ongoing monitoring and supervision. 4.5 Discussion This paper reports a societal perspective of cost drivers relevant to stakeholders considering the CLIP interventions, and has demonstrated that care-seeking practices vary between public and private health sectors in Sindh, Pakistan. Our findings further highlighted contextual aspects of resource utilization that guided the design of a comprehensive questionnaire for quantitative ascertainment of individual level health resource utilization during the trial period. Given ever-increasing reliance on technology use in health care, there are implications for incremental costs to society (89). We found that referral of pregnant women requiring treatment at health facility resulted in the OOP costs; and productivity loss/missed wages further added financial burden on families and/or primary caregivers. Our findings are corroborated with other studies from developing countries that indicate over 50% of total health spending as OOP (90). Ironically, OOP costs are often omitted from cost analyses because such costs are assumed to be irrelevant for program implementation. A recent cost analysis of peer health workers and mHealth support interventions for improving AIDS care in Uganda lacked OOP costs in their analysis (91). We also interpreted from FGs that the sicker a patient was, in terms of severity and complication, and/or if the decisions to seek care were delayed, the higher the OOP costs 51 incurred. Thus, our findings point to the need to consider OOP costs when conducting economic evaluation of health technologies in LMICs. The program implementation as indicated in this study suggests incremental costs to the provincial public health department, as health technology procurement and trainings require substantial investments. Program costs are often feared to increase the burden on existing health budgets in LMICs. It is likely that community-based screening and/or interventions using mHealth platform will influence care-seeking behaviours. Our argument is supported by a randomized control trial that demonstrated that short message services (SMS) and cell phone reminders (i.e. the intervention group), compared with no reminders (i.e. the control group), were associated with significantly higher attendance rates at health promotion centres (92). Another cost-effectiveness study from Pakistan reported higher costs related to personnel time, equipment and supplies in community-based response stimulation and nutrition interventions on early child development (93). The economic appraisal of emerging health technologies plays a pivotal role in policy advocacy and resource allocation for post-trial program scale-up (94). We found that health decision-/policy-makers supported mHealth initiative, and expressed their interest in finding cost-effectiveness of the CLIP interventions in Sindh, Pakistan. These findings are similar to those observed in a qualitative study in the UK, where health authorities strongly recommended economic evaluation to advise policy decisions in health care (95). 4.5.1 Strengths and limitations The International Society of Pharmacoeconomics and Outcome Research (ISPOR) recommend in-depth assessment of jurisdiction-specific costs and outcome parameters for decision-analytical models of multi-national trials. In this study, we used a pragmatic approach 52 to assess cost drivers necessary for the economic analysis alongside the CLIP trial in Sindh, Pakistan. FGs with a wide range of stakeholders in relation to the CLIP trial further add methodological rigor to the overall study. We recognize that combine FGs for pregnant women- and mothers-in-law; husbands- and fathers-in-law may have resulted in social desirability bias and it is the main limitation of current study. The transferability of our study findings is limited to care-receivers, care-providers, and program implementers in relation to the CLIP trial in Sindh, Pakistan. The methodological approaches as reported in this study may guide future health economics studies evaluating MNCH interventions in Pakistan and other LMICs. 4.6 Conclusions A thorough understanding of care-seeking practices revealed financial implications for families of pregnant women, and program costs for the health system during implementation of CLIP trial. The societal perspective provided contextual information, and revealed a more comprehensive description of cost drivers that can be used to design an economic model to fulfill the needs of health decision-/policy-makers considering CLIP interventions in Sindh, Pakistan. 53 Table 4. Focus group participants’ characteristics Groups # of focus groups Number of participants Age in years Mean (+SD) Schooling years Mean (+SD) Occupation (n, frequency of participants) Pregnant women with pregnancy hypertension 02 12 24.5 (3.8) 3.1 (4.7) 8 Housewife 4 unskilled labour work Mothers-in-law 07 44.2 (4.2) 0 (-) 6 Housewife 1 unskilled labour work Husband of pregnant women with pregnancy hypertension 02 13 26.3 (5.2) 3.7 (4.1) 9 unskilled labour work 3 Farmers 1 Employed in grouped services Fathers-in-law 04 43.8 (2.1) 0.75 (1.5) 2 Farmers 1 unskilled labour work 1 Business Community care providers 02 19 31.2 (4.8) 11.1 (2.0) 19 Employed as LHW Medical doctors at referral health facilities 02 20 33.7 (5.6) 17.8 (1.1) 11 OBGYN specialist 9 General physician District health decision/policy makers 01 10 53.4 (3.3) 17.1 (1.2) 8 General physician 1 Specialists 1 Administration Total 09 85 LHW, Lady health worker OBGYN, Obstetrics and gynecologist 54 Table 5. Descriptive coding list for thematic analysis Program implementation Thematic areas related to cost drivers using societal perspective Out-of-pocket costs Health system costs Program implementation costs Patient screening, and out-patient visits Diagnostic procedures Human resource [personnel time] Procurement of technologies for patient screening Out-patient doctor’s visit Laboratory tests Trainings for technology-use mHealth platform to support technology implementation Referral and transport to health facility Transport to-and-from health facility Transport for home-based visits Cellular communication Missed wages by family and/or care-givers Accompanying women to hospital Access to computer and internet Treatment In-patient hospitalization, other than childbirth Initial case-management for sick women Monitoring and feedback In-patient hospitalization for childbirth Clinical procedures Medications Medications Care received from traditional care-providers Specialist consultations Community engagement Opportunity costs to attend sessions Personnel time Materials, and logistic support 55 Table 6. Lump sum out-of-pocket costs related to healthcare sought for HDP Health resource utilization Cost in PKR (US$) Minimum Maximum Diagnostic procedures 50 (0.5) 3500 (35.0) Out-patient doctor’s visit 150 (1.5) 1500 (15.0) In-patient hospitalization, other than childbirth 2500 (25.0) 7000 (70.0) In-patient hospitalization for childbirth 3000 (30.0) 10000 (100.0) Transport to-and-from health facility 500 (5.0) 3500 (35.0) Childbirth at home [Conducted by traditional birth attendant] 1000 (10.0) 1500 (15.0) Self-medication 800 (8.0) 2000 (20.0) Missed wages (inclusive of woman and caregiver) 900 (9.0) 5000 (50.0) US$ currency exchange rate of 2015 56 Figure 5. Conceptual framework and approach for societal perspective 57 Chapter 5: A comparison of maternal and newborn health services costs in Sindh, Pakistan 5.1 Background The cost of care influences decisions related to care seeking, and utilization of MNH services (96). Data related to health expenditures in Pakistan revealed a declining trend of GDP spending on health over the past decade; currently as low as 2.4%. It is further estimated that over 80% of healthcare spending is out-of-pocket, and predominately in the private sector (97). This shift towards the beneficial use of HT integration has implications for incremental costs to patients, health systems and society at large (98). However, a lack of compressive knowledge of the cost of MNH services at different levels of care confounds policy decisions about introducing existing interventions, and impedes economic appraisal of emerging HT in Pakistan. The assessment of cost-effectiveness of the CLIP trials requires a thorough understanding of maternal and newborn costs at health facilities in the CLIP countries. In designing an economic model for the CLIP trial in Pakistan, similar challenges were faced, as health facility costs were unknown for care received during pregnancy, delivery and early newborn stages in both private and public sectors. 5.2 Objectives The primary objective of this study was to estimate the cost of MNH services. The secondary objective was to compare cost of MNH services within and/or between public and private health sectors in Pakistan. 58 5.3 Methods 5.3.1 Study design A cross-sectional survey of health provider/facilities was conducted during January to March 2016, as part of a large economic evaluation of the CLIP Trial in Sindh, Pakistan. The provision of basic, as well as comprehensive, emergency obstetric and newborn care (EmONC) services, were available in the private health sector (i.e. 100% out-of-pocket costs to patients), and the public health sector (i.e. nearly 100% health system costs, and very little OOP costs for patients [transport, food and diagnostics]). Health facilities were clustered into three broad categories: primary, secondary and tertiary levels. Categorization was based on the population served, hospital size (usually, number of beds), and the provision of clinical subspecialty services and intensive care. The primary level health facilities provide health services to less than 50,000 people, with inpatient capacity of 0 to 10 beds, and focus on basic obstetric and newborn care. The secondary level hospitals provide health services to over 1 million people, with inpatient capacity of 40 to 60 beds, and focus on basic and EmONC services. The tertiary level hospitals serve as the referral point, provide multispecialty clinical services, and offer intensive care facilities to a wider population (80). 5.3.2 Inclusion and exclusion criteria The healthcare providers and facilities were considered eligible if they met these criteria: (i) geographical location within study catchments; (ii) secondary and tertiary level hospitals in the public health sector, to which pregnant women are referred by LHWs under the National Program; and (iii) private healthcare providers and facilities, where pregnant women self-refer for pregnancy care and childbirth. Primary level facilities in the public health sector; and healthcare providers and facilities that declined to participate were excluded (Figure 6, pg 68). 59 5.3.3 Sampling procedures Healthcare providers and facilities were selected from a sampling frame, inclusive of public and private health sector in the study catchments, as previously described (99). Public health facilities were identified through the National LHWs Program and private healthcare providers and facilities were identified through the trial network. The project field coordinator approached the administrative staff at health facilities and invited them to participate in the study. 5.3.4 Methods of data collection A structured questionnaire was used to collect costing data of MNH services. The key variables included: routine antenatal care (ANC), ambulatory visits (AV) for maternal and newborn care, diagnostic tests and imaging, overnight inpatient admission, childbirth and blood transfusion. Project RAs; registered midwives who were bachelor degree holders with MNH research experience) were trained by a Senior Scientist, who is a native speaker of the local Sindhi language, and had experience in MNH research in Sindh. RAs visited healthcare providers and facilities and ascertained unit costs from billing departments as they were charged to patients in the form of fee-for-service at the private facilities. However, a simultaneous allocation method was used to estimate the unit costs in the public facilities because resources were shared between service departments (94). The allocation of costs was based on an estimate of resource-use for shared costs (i.e. staffing, utility, laundry, housekeeping, repair-and-maintenance, and patient food), and independent costs (medication, supplies, equipment, and furniture/fixture). The department-level costs were later divided by the average number of patients attending clinics and the number of beds to calculate the unit costs 60 for ambulatory visits and inpatient admissions, respectively. The costs were estimated in the local currency, Pakistani Rupee (PKR), and later converted into US$ [1 US$ = 104.7 PKR]. 5.3.5 Data analysis Descriptive analyses were performed to calculate mean and standard deviations for cost estimates from public and private hospitals, except tertiary level hospitals where point-estimates were reported. The analyses of variance (ANOVA) tests compared overall mean costs within and between public and private health sectors, and statistically significant differences were interpreted with p-value (two-tailed) <0.05. Analyses were performed using SPSS version 24. 5.4 Results A total of 43 eligible healthcare providers and facilities (n=36 private; n=7 public) were approached within the study catchments. The refusal rate varied from 14% (n=1/7) to 27.8% (n=10/36) in the public and private health sectors, respectively. The final cost analysis included 25 facilities in the private health sector (19 primary-, five secondary- and one tertiary-level facilities), and six facilities in the public health sector (five secondary- and one tertiary-level facilities). The public facilities were geographically scattered, whereas private healthcare providers and facilities were in close proximity. 5.4.1 Cost of maternal health services An AV for routine ANC from a medical doctor costs $3.6 (+SD 2.1) at secondary level; and $0.9 at tertiary level public facilities. In the private health sector, the mean cost of AV for routine ANC were $2.2 (+1.9) in medical clinics, $2.8 (+2.6) in secondary level, and ($6) in tertiary level facilities. The AV was less costly for routine ANC provided by midwives ($0.6 + 0.3) and traditional birth attendants ($0.5 + 0). The costs of many diagnostic tests or imaging 61 were similar within all levels of public facilities. In the private health sector, costs of diagnostic tests were much higher at the tertiary level. The cost of inpatient general ward admission was higher in the tertiary level, compared with secondary-level public facilities ($57.5 versus, 3.4 + 1.7). The costs of delivery were low at secondary level facilities for spontaneous childbirth ($45.4 +SD 30.7 versus, 79), and assisted childbirth, compared with the tertiary level public facility ($50.2 +SD 33.8 versus, 86), respectively. The cost of Caesarean delivery was less in the public tertiary facility, compared with the private tertiary facility ($223.8 versus, 400). Significant differences were found in the overall mean costs for maternal health services within/between public and private facilities at p-values < 0.05 (Table 7, pg 66). 5.4.2 Cost of newborn health services The mean cost of newborn AV in the secondary-level hospitals was $2.7 + 1.8 compared with $0.4 at the public sector tertiary-level hospital. Newborn AV costs were lowest ($0.5 + 0) when provided by midwife and traditional birth attendants in the private health sector. The cost of fetal ultrasound, newborn x-ray, and blood grouping differed substantially at all levels of health facilities in public and private sectors. Phototherapy was only available at tertiary-level facilities, and cost ranged from $1 to 2 in the public and private sectors, respectively. The cost of newborn admission into a nursery was low in the public compared with private tertiary-level facilities ($10.3 versus, 13.8 + 3.7). However, the cost of newborn intensive care admission was higher in the public compared with private tertiary-level facilities ($22.3 versus, $19 + 7.8). Significant differences were found in the overall mean costs for newborn health services within/between public and private facilities (Table 8, pg 67). 62 5.5 Discussion This paper reports the costs of a wide range of MNH services relevant to families (i.e. out-of-pocket) and health systems, and has demonstrated that these costs vary significantly at different levels of health facilities within and between public and private sectors in Sindh, Pakistan. Our findings provide a robust measure of unit-costs by service type and levels of health facility for future economic studies aiming to estimate cost-of-illness in the area of maternal and newborn care in Sindh, Pakistan. PE/E are serious conditions during pregnancy requiring frequent AVs for routine antenatal care, diagnostic tests and health facility admissions in the event of disease complications. The costs of health services at private health facilities as reported in this study suggest a large financial burden related to out-of-pocket payments on families (tables 7 and 8, pg 66-67). We found a progressive trend in the costs for AV dependent on the level of private health care (i.e. costs lower at medical clinics and higher at the tertiary level). Similarly, the costs of diagnostic tests and imaging were higher in private facilities. A previous study evaluated role of public spending on healthcare across 11 Asian countries. Authors in that study found that distribution of public health infrastructure is mainly biased towards the provision of services for the wealthy (pro-rich) in many LMICs, and that transition towards pro-poor healthcare requires limiting the user fees, protecting the poor from catastrophic expenditure on health, and creating a wide network of public health facilities (100). The private sector is expanding within the health industry in LMICs, and employs social marketing techniques to attract patient volume (58). In a previous qualitative study, authors found that many women sought pregnancy-related care from private health facilities (101). Despite higher costs, decisions to seek care are often dominated by the perception of quality of 63 care at such facilities (102). In another study, authors assessed patients’ perception about service quality in Pakistan, and reported that private health facilities deliver better quality of care to patients compared with public facilities (103). Guided by our local field observations, we determined that the private health sector tends to offer cutting-edge diagnostic technologies (i.e. latest and/or expensive imported machines from abroad) to meet the growing demand for precision medicine in Pakistan. We assume that the private health sector could offer lower inpatient prices possibly through cost containment (i.e. controlling overhead expenditures, rational distribution of lower- and higher- cadre of medical staff based on patient volume), and operating with low profit margins to attract greater patient volumes. More efforts are needed to regulate the private sector and promote strategic purchasing to be able to lower health services costs. This study does not estimate profit margins on the cost of MNH services in private sector. Our findings further highlight the policy implications for health systems, as AV for maternal and newborn care are more costly at the secondary-, compared with tertiary-, level public facilities in Pakistan. These findings are similar to those observed in the Northern Province of Pakistan, where AV costs were higher (i.e. $4.1) at basic health units, and probably reflect ambulatory costs being mainly dependent upon the number of patients attending outpatient clinics (i.e., fewer patients, more cost per AV) (104). Upon reviewing records of patient registry at secondary-level facilities, we found that fewer pregnant women were attending outpatient clinics (range: 25 to 50 per day), compared with 250 to 500 per day) at the tertiary level public facility. Previous studies indicated several factors such as inadequate staffing, poor facilities for in-patient admissions, and a poorly-coordinated referral system are responsible for low patient-volumes at lower-level health facilities (105,106). Furthermore, we found that costs of inpatient admissions, other than for delivery, were much higher in the public facilities. The 64 cost allocation exercise indicated higher operating costs (i.e., staff salaries, equipment, medication, and overheads) at inpatient departments. Others also reported increased cost burden in public health facilities, assuming that patients are presenting with severe illnesses and/or disease complications (requiring longer inpatient stays) as a result of delaying care (107). The economic appraisal of emerging HT requires robust data on costs and health outcomes using a societal perspective, inclusive of care receivers, care providers and health system (108). Our findings are critical to calculate real-world cost-inputs representing public and private health sectors, as we embark upon analysing MNH resource utilization (end-user data) reported from the intervention and control groups in the CLIP Trial in Pakistan. In addition, the methodological approaches and our findings may guide future health economics studies evaluating MNH interventions in Pakistan and other LMICs. 5.5.1 Strengths and limitations To our knowledge, this is the first study to estimate the cost of MNH services at different levels of health facilities in the public and private sectors in Pakistan. We benefitted from high participation (86%) of public secondary- and tertiary-level facilities. The primary-level health facilities were omitted from our analysis because they did not serve as a referral point for EmONC services in the community. Also, primary health facilities were large in numbers, which would have required additional project expenses to capture all possible facilities. We recognize that costing information from primary-level facilities may have been important for comparative analysis across three tiers of public health system in Pakistan and it is the main limitation of current study. Collecting evidence-based information on costs and health outcomes is often challenging in the absence of electronic medical records and poor data keeping in LMICS (109). We faced 65 similar challenges in public and private health sectors. It was a resource-intensive exercise, requiring frequent long-distance travel to public hospitals and several in-person meetings with people in the clinical, budget and finance departments. In the private health sector, people were reluctant to share costing information on MNH services and hospital budgets (i.e. revenue and expenses) due to fear of litigation. Also, some private health facilities were only open in the evening, which created operational difficulties for the project RAs and data collection. We encountered a high refusal rate in the private health sector because of operational difficulties and reluctance to share costing information with persons other than patients. We increased the sample size particularly in the private sector by three-fold, compared to the public health sector, to address the possibility of non-response bias in data collection. 5.6 Conclusions This economic appraisal of MNH services revealed cost disparities within public health sector suggesting higher costs for AV at the secondary-level, and inpatient admissions at tertiary-level health facilities in Sindh, Pakistan. The private sector stands-out as an expensive choice of care provider for diagnostics and delivery. An understanding of MNH costs is critical to guide resource allocation within the public sector and for risk mitigation against excessive out-of-pocket costs through third party payer for services in the private sector. 66 Table 7. Cost of maternal health services in public and private health sectors Variables Public Health Sector Costs (USD) Private Health Sector Out-of-pocket Costs (USD) Mean Difference (Overall Private minus Public) p-value Secondary hospitals (n= 5) Mean (+SD) Tertiary hospital* (n=1) Overall public (n=6) Mean (+SD) Clinics/care provider (n= 19) Mean (SD) Secondary hospitals (n=5) Mean (+SD) Tertiary hospital* (n=1) Overall private (n=25) Mean (+SD) Ambulatory visits for routine antenatal care (cost per visit) Medical doctor Nurse Midwife Traditional birth attendant 3.6 (2.1) NA NA NA 0.9 NA NA NA 2.3 (2.1) - - - 2.2 (1.9) 1.7 (2.1) 0.6 (0.3) 0.5 (0) 2.8 (2.6) NA NA NA 6 NA NA NA 3.7 (2.2) 1.7 (2.1) 0.6 (0.3) 0.5 (0) 1.4 1.7 0.6 0.5 <0.0001 Diagnostic tests (cost per test) Pregnancy ultrasound Complete blood count Culture Creatinine Serum albumin Aspartate Aminotransferase Alanine aminotransferase Urine dipstick Urine microscopy Chest x-ray CT scan 0.5 (0) 0.9 (0.9) 1.3 (1.1) 0.4 (0.2) 0.4 (0.2) NA NA 0.3 (0) 0.3 (0) 0.9 (0.6) NA 0.5 0.4 0.6 0.4 0.2 NA 0.8 NA 0.2 0.5 15 0.5 (0) 0.7 (0.6) 0.9 (0.8) 0.4 (0.1) 0.3 (0.1) - 0.8 0.3 (0) 0.3 (0.1) 0.7 (0.3) 15 2.8 (1.1) 2 (1.4) NA NA NA NA NA 0.8 (0.5) NA NA NA 3.9 (3.2) 3.8 (2.2) 6.9 (6.6) 3.4 (2.8) 3.2 (0.8) 6 3.8 (2.7) 1.3 (1.2) 1.8 (1.5) 2.5 (1.4) NA 6 5.2 4.2 4.3 3.2 NA 1.9 NA 2.6 6 20 4.2 (2.4) 3.7 (2.2) 5.6 (4.9) 3.9 (2.2) 3.2 (0.8) 6 2.9 (1.3) 1.1 (0.9) 2.2 (1.4) 4.3 (2.1) 20 3.7 3 4.7 3.5 2.9 6 2.1 0.8 1.9 3.6 5 <0.0001 In-patient admissions (cost per overnight stay) General ward Intensive care 3.4 (1.7) NA 57.5 151.2 30.5 (38.3) 151.2 4.2 (1.4) NA 8.5 (11.1) NA 15.2 36.5 9.3 (8.7) 36.5 -21.2 -114.7 0.0136 Delivery and blood transfusion (cost per procedure) Spontaneous vaginal delivery Assisted vaginal delivery C-section Blood transfusion 45.4 (30.7) 50.2 (33.8) NA NA 79.1 86.6 223.8 20.1 62.3 (30.7) 68.4 (33.8) 223.8 20.1 33.3 (14.8) 35.8 (23.7) NA NA 86 (92.4) 106 (108.8) 304 (167.2) 17.8 (22.3) 150 200 400 25.8 89.8 (93.1) 113.9 (104.6) 352 (154.3) 21.8 (19.6) 27.5 45.5 128.2 1.7 <0.0001 * Mean and standard deviation not calculated for one facility; NA= not available at designated health facility 67 Table 8. Cost of newborn health services in public and private health sectors Variables Public Health Sector Costs (USD) Private Health Sector Out-of-pocket Costs (USD) Mean Difference (Overall Private minus Public) p-value Secondary hospitals (n= 5) Mean (+SD) Tertiary hospital* (n=1) Overall public (n=6) Mean (+SD) Clinics/care provider (n= 19) Mean (SD) Secondary hospitals (n=5) Mean (+SD) Tertiary hospital* (n=1) Overall private (n=25) Mean (+SD) Ambulatory visits for newborn care (cost per visit) Medical doctor Nurse Midwife Traditional birth attendant 2.7 (1.8) NA NA NA 0.4 NA NA NA 1.6 (1.7) - - - 1(0) 2.3 (2.5) 0.5 (0) 0.5 (0) 2.6 (1.1) NA NA NA 2 NA NA NA 1.9 (1.1) 2.3 (2.5) 0.5 (0) 0.5 (0) 0.3 2.3 0.5 0.5 0.0001 Diagnostic tests (cost per test) Fetal ultrasound Fetal x-ray Blood grouping Arterial blood gases Phototherapy 0.5 (0) 1 (0) 0.3 (0) NA NA 0.5 0.5 0.2 NA 1 0.5 (0) 0.5 (0.3) 0.3 (0.1) - 1 4 (0) 1.5 (0) NA NA NA 6.3 (6.4) 3.8 (2.3) 2.2 (2.8) 12.9 (0) NA 6 6 3.4 8.5 2 5.4 (4.7) 3.8 (2.4) 2.8 (2.3) 10.7 (3.1) 2 4.9 3.3 2.5 10.7 1 <0.0001 In-patient admissions (cost per overnight stay) Nursery General ward Intensive care NA 4.5 (0) NA 10.3 8.2 22.3 10.3 6.4 (2.6) 22.3 NA NA NA 11.2 (0) 5.9 (1.28) 17.8 (10.6) 16.4 12.1 20.2 13.8 (3.7) 9 (3.6) 19 (7.8) 3.5 2.6 -3.3 0.0011 * Mean and standard deviation not calculated for one facility; NA= not available at designated health facility 68 Figure 6. Healthcare providers and facilities in the public and private health sector Figure 1: Care providers/facilities in the public and private health sector Total number of health providers/facilities [N= 179] Private Health Sector [n=95] Eligible health providers/facilities [n= 36] Not eligible health providers/facilities [n=59] Consented to participate [n=25] Refused [n=10] Public Health Sector [n=84] Eligible health providers/facilities [n=7 ] Not eligible health providers/facilities [n=77] Consented to participate [n=6] Refused [n=1] Level of care Clinics [n= 19]; Secondary [n=5]; Tertiary [n=1] Level of care Secondary [n=5]; Tertiary [n=1] 69 Chapter 6: Cost-effectiveness of the Community Level Interventions for Pre-eclampsia (CLIP) in Sindh, Pakistan 6.1 Background Several epidemiological studies have demonstrated the positive impact of community-based interventions, specifically those involving LHWs in various parts of Pakistan. Bhutta et al. evaluated a package of community-based interventions (i.e., use of clean delivery kits, promotion of routine ANC and maternal health education, facility births, and newborn care) delivered by LHWs in Sindh, Pakistan. In that study, investigators found reductions of 21% for stillbirth and 15% for neonatal mortality (110). In another cluster randomized equivalence trial, LHWs provided mothers with oral amoxicillin for children aged 2 to 59 months with severe pneumonia in the intervention group, compared to oral cotrimoxazole and referral to a health facility (standard of care). This study reported a significant reduction in the treatment failure (i.e., risk difference of -8.9%; 95% CI: -12.4% to -5.4%) in the intervention group in the Punjab province of Pakistan (111). Given the ever-expanding role of LHWs in pregnancy and newborn care, the ministry of health is considering additional investments to promote LHWs as frontline care providers for preventive and community-based management or referral services in Pakistan (81). As discussed in Chapter 3, the CLIP introduces the mHealth innovation for HDP screening, risk stratification, and case management initiated by LHWs in community settings. As with the introduction of new technologies, there may be financial implications to the health system with post-trial scale-up. Economic evaluation of the CLIP interventions was imperative to inform the decisions on resource allocation for programme scale-up, should the trial find evidence on clinical-effectiveness. Therefore, the economic analysis was conducted to estimate costs relative to benefits associated with the CLIP package of care. 70 6.2 Objectives The primary objective was to estimate the incremental costs and health benefits of CLIP plus routine pregnancy care, compared with routine pregnancy care alone. The secondary objective was to perform the temporal analysis of program-related costs and rate of pregnancy outcomes over the study period in the intervention and control group. 6.3 Methods 6.3.1 Study design Economic evaluation was conducted alongside the trial to assess the costs relative to benefits for the CLIP package of care combined with routine pregnancy care, in comparison to routine pregnancy care alone. The decision analytic model was used, drawing input parameters from the CLIP Trial and mixed-methods research used in this study. Gamma distribution was applied to cost inputs, and lognormal distribution was applied to pregnancy outcome inputs (Table 9, pg 85). 6.3.2 Data analysis The costs to the health system included resources used to implement the CLIP Trial. The capital costs comprised of clinical devices and mHealth technology used in home-based screening for HDPs. Recurrent operating costs included community-based patient management, referral and follow-up (i.e., drugs, supplies, patient transport), and community engagement (i.e., personnel, venue, refreshment, transport allowance for participants). LHW costs included costs of training, the incentive for pregnancy registration, and transport for home visits. Because health system capacity enhancement and surveillance for the ascertainment of pregnancy outcomes occurred simultaneously in both groups, these costs were equally distributed in the intervention and control groups. Costs to women and families included maternal and newborn 71 HRU such as ambulatory visits, overnight hospitalization, diagnostic tests and procedures, childbirth, transport to and from a health facility, and productivity losses. Descriptive analysis was done to calculate the frequency and proportion of maternal and newborn HRU in the intervention and control groups. A generalized linear mixed-model regression was performed to compare mean cost, standard errors, 95% CI and p-values between two groups. The gamma distribution was used in the regression analysis to address the skewed nature of cost data (112). Tests for correlations and multicollinearity were applied on the total costs and explanatory variables to fit the model. The fixed variables included age, years of schooling, comorbidities, and health care-seeking behaviors. The random variables included baseline neonatal mortality ratio and population index. Two variables (permission for routine ANC and permission for obstetric emergencies) were highly correlated, with suggested Pearson Correlation value = 0.93 and Variance Inflation Factor-IVF= 28.2. After discarding one variable (permission for obstetric emergencies), IVF score reduced to 1 for all explanatory variables. The societal cost was calculated as the sum of the costs: maternal and newborn HRU, the CLIP package of care, LHWs incentive/transport, surveillance for ascertaining pregnancy outcomes, and health system capacity enhancement. The annual equivalent costs in Pakistani rupees (PKR) was converted to US dollars using average currency exchange rates of 2014-2016 (104.7 PKR = 1 US$). DALY is a common outcome metric widely used in the economic literature, and this represents the sum of YLL and YLD. A recent publication from the Global Burden of Disease (GBD) provides instructions and tools for calculating YLL and YLD in LMICs (113). The average age of maternal and newborn deaths was used to calculate YLL with respect to the country-specific life expectancy tables for Pakistan (114). Similarly, we calculated YLL for stillbirths using the value on a standard life table for life expectancy at birth. The YLD were 72 derived taking into account the epidemiological rates for maternal and newborn morbidities observed in the trial period, and were multiplied with disability weights (DWs) as reported on the recent publication from Global Burden of Diseases (56,113). However, we faced some challenges in determining YLD for maternal and newborn morbidities, as DWs are not established for several severe conditions. Thus, we restricted our analysis only to include YLL in the economic model, as the principal component of DALYs. Both costs and health outcomes were discounted at a rate of 3%, and estimated incremental cost per life-year saved. In the base-case analysis, the ICERs were calculated first from a health care system perspective, and then from a societal perspective. A probabilistic analysis (a Bayesian approach that allows the effects of joint uncertainty across all input parameters) of the base case was performed using Monte Carlo simulations to select values at random from 10,000 iterations in Microsoft Excel. Results of the probabilistic analysis were plotted on a cost-effectiveness plane to represent the distribution of the ICER. The results were interpreted through the use of a cost-effectiveness acceptability curve (CEAC) showing the probability that ICERs fell below the acceptable WTP threshold. In this study, WTP threshold represents the GDP per capita in 2016 in Pakistan (i.e., $1,468) (115). We applied one-way sensitivity analyses by varying the parameter estimates using a plausible range to test their influence on the ICER. The parameter estimates included the percentage of effect size as 1%, 20%; percentage of discount rate as 0%, 6%; currency exchange rate 2014, 2016; and cost of CLIP implementation +10%. The outputs from sensitivity analysis were plotted on a tornado diagram. For the secondary objective, an exploratory analysis was done to compare quarterly trends for costs and trial-based composite adverse pregnancy outcome (combined maternal and 73 perinatal mortality and severe morbidity) over the study period. The capital costs of electronic tablets and phones were annualized over the expected useful life of 5 years; and digital clinical devices (blood pressure monitor and pulse-oximeter) were annualized over the expected useful life of 10 years. The recurrent operating costs such as clinical supplies, drugs, and transport were based on the number of women who delivered in each quarter. The rate of composite adverse pregnancy outcomes was translated into health benefits as “rate of adverse outcomes avoided” (1 minus the rate of adverse pregnancy outcomes). The average cost per pregnancy and rate of health benefits were compared in the intervention and control groups over the study period. 6.4 Results 6.4.1 Trial recruitments and participants’ characteristics In all, 20,264 pregnancies were included in the 10 clusters randomly assigned to receive CLIP interventions, and 19,180 pregnancies were included in the 10 control clusters. Only seven women withdrew from the study; three in the intervention group, and four in the control group. A very small proportion of pregnant women was lost to follow-up; 752 in the intervention group (3.7%) and 864 in the control group (4.5%). At the trial closeout period, roughly 5% of pregnancies in both the intervention (n=1,063) and control (n=967) groups were undelivered or still on follow-up. The mean (+SD) maternal age in the intervention group was 28 years (+4.7), compared to 27.6 years (+4.4) in the control group. Other characteristics such as mean gestational age, maternal primary education (> 5 years of schooling), husbands with primary education, and parity did not significantly differ between two groups. A large majority of pregnant women were unemployed or engaged in unpaid domestic services: 16,180 (80%) in the intervention group, compared to 14,159 (74%) in the control group. In the intervention group, about one-third of 74 women (n=6,846) reported their husband’s occupation as skilled manual laborer, and another 19% (n=3,859) reported work in agriculture. Fewer women (n=5,784, 30%) reported skilled manual labor as the occupation of their husband in the control group. Other medical conditions did not differ between two groups, except that anemia was higher in the control group (n=4,508, 24%) than in the intervention group (n=3,890, 19%). Care-seeking preferences revealed that a higher proportion of women in the control group were allowed by their husbands to seek care for both routine ANC (n=16,128, 84%) and pregnancy-related emergencies (n=15,674, 82%). A slightly higher proportion of women in the intervention group (n=5,517, 27%) reported having funds saved for pregnancy-related emergencies, compared to women in the control group (n=4,487, 23%) (Table 10, pg 86). 6.4.2 Trial-based maternal and newborn health resource utilization The maternal and newborn HRU estimates are reported on for followed-up pregnancies to compare costs and health benefits in cases where we have complete information over the full term of pregnancy. There were 18,449 followed-up pregnancies of 20,264 (91%) in the intervention group, and 17,349 of 19,180 (90.5%) in the control groups. Of these, 16,013 women (87%) reported routine ANC visits in the intervention group, as compared to 14,266 women (82%) in the control group. Approximately 90% women sought routine ANC from medical doctor in both groups. The mean (+SD) routine ANC visits in the intervention group was 4.4 visits (+4.1), compared to 3.3 visits (+3.2) in the control group. The ambulatory visits at HLCF for pregnancy-related illnesses were slightly higher in the control group compared to the intervention group, with a mean difference of 0.5 visits. All diagnostic tests were reported at higher rates in the control group except urine dipstick, which was lower in the control group (n=6,843, 39%) than in the intervention group (n=8,010, 43%). There were some differences in 75 the maternal hospital admissions between the two groups, indicating higher intensive care admissions in the intervention group. The mean (+SD) length of stay in the intensive care unit was 2.4 (+1.0) in the intervention group, compared to 2.1 (+1.1) in the control group. Approximately 24% of women in both groups reported having a home birth. A higher proportion of women in the intervention group (n=7,756, 43.5%) reported childbirth at a private health facility compared to the control group (n=6,566, 39%). Of births taking place at a health facility, a higher proportion of women in the control group reported spontaneous vaginal birth (9,094 of 12,726, 71%) compared to the intervention group (9,107 of 13,558, 67%). The proportions for Caesarean section were almost similar in the intervention group (n=3,227, 24%) and control group (n=2,715, 22%). Transport trips to health facility were slightly higher in the intervention group, with a mean difference of 0.7 trips. The hired wageworker to assist pregnant woman in the household chores, however, was as high as 9.4% (n=1,738) in the intervention group, compared to only 3.3% (n= 580) in the control group (Table 11, pg 87). Fewer women reported ambulatory visits for newborn illnesses in the intervention group (1,430 of 18,449, 8%) compared to the control (1,901 of 17,349, 11%). Over 90% women sought newborn care from medical doctor in both groups. The mean (+SD) ambulatory visits for newborn care in the intervention group was 1.9 visits (+1.5), compared to 1.8 visits (+1.4) in the control group. The newborn diagnostic tests were almost similar in both groups. There were some differences in the newborn hospital admissions between the two groups, indicating higher intensive care admissions in the intervention group. Overall, the mean (+SD) length of newborn hospital stay in the intervention group was 1.9 (+1.4), compared to 1.7 (+1.2) in the control group. The mean transport trips to health facility and missed daily wages due to newborn illness were higher in the intervention group (Table 12, pg 89). 76 6.4.3 Costs to health system The LHWs in CLIP clusters were supplied with a handbag containing clinical supplies, a Microlife VSA blood pressure device ($22), a PX-43-B Kenek pulse oximeter ($20), and either a Mobile Q-Noir A7 phone ($126) or a Q-300 8GB tablet ($143) to carry out the home visits (Table 13, pg 90). The total financial cost of the intervention (including CLIP package of care, LHWs, surveillance, and health system capacity enhancement) was $530,755, compared to $313,602 for usual practice in the control group. The cost of the CLIP package of care (excluding surveillance, LHWs, and health system capacity costs) was $137,756. Of this total, community engagement accounted for 52% ($71,942), home-based HDP screening for 44% ($60,949), and community-based HDP case management for only 4% ($4,864) (Table 14, pg 91). The community engagement session targeted male members of the community (number of sessions = 820), and cost approximately $88 per session (cost $4 per person). Home-based HDP screening was performed on 11,461 women (cost $5.30 per woman), and the women received a total of 58,409 screening visits (cost $1.10 per visit). The cost of providing CLIP trainings, incentive and transport to make follow-up home visits by LHWs was $79,398 (cost $4 per woman). The compliance with CLIP-POM recommendations were as follows: 1) Of 406 visits requiring non-urgent referrals, 311 (76.6%) were accepted; 2) Of 220 visits requiring urgent referrals, 169 (76.8%) were accepted; 3) Of 168 visits requiring MgSO4 treatment, 98 (58.3%) were accepted; 4) Of 155 visits requiring Methyldopa treatment, 150 (96.8%) were accepted. 77 6.4.4 Costs to pregnant women and their families The total financial cost of maternal HRU was $2,153,346 in the intervention group, and $1,879,995 in control group. Regression analysis revealed statistically significant differences in the mean costs for most of the explanatory variables (such as the maternal age, maternal primary education, husbands with primary education, parity, savings for pregnancy-related emergencies, and medical conditions including high blood pressure, seizures, and diabetes) with p-values <0.05 (Table 15, pg 92). The regression-adjusted mean cost of maternal care was $163.40 (95% CI: $51 to 529) in the intervention group compared to $156.70 (95% CI: $49 to 506) in the control group, with p-value: 0.958. The total financial cost of newborn health resource utilization was $14,260 in the intervention group, compared to $17,961 in the control group. The regression-adjusted mean cost of newborn care was $4.00 (95% CI: $0.70 to 23.41) in the intervention group compared to $4.2 (95% CI $0.60 to 24.60) in the control group, with p-value: 0.964. The regression-adjusted mean cost of combined maternal and newborn care was $164 (95% CI: $51 to 528) in the intervention group and $158 (95% CI: $49 to 507) in the control group, with p-value: 0.962, suggesting no significant difference overall between two groups (Table 16, pg 93). 6.4.5 Effectiveness (YLL and YLD) Maternal mortality rates were estimated at 2.71 (per 1,000 identified pregnancies) in the intervention group and 2.66 in the control group. Maternal deaths were translated into 76 and 64 YLL in the intervention and control groups, respectively. The early neonatal mortality rate was estimated at 40.5 (per 1,000 identified pregnancies) in the intervention group, compared to 42.7 in the control group. The YLL resulting from early neonatal death was calculated at 1,214 and 1,237 in the intervention and control groups, respectively. Stillbirth rate was estimated at 46.2 78 (per 1,000 identified pregnancies) in the intervention group, compared to 49.6 in the control group. Stillbirths resulted in a low of 1,373 YLL in the intervention group and 1,439 YLL in the control group. The combined YLL were slightly higher in the intervention group (YLL=2,968) than in the control group (YLL=2,957) (Table 17, pg 94). Similarly, the combined maternal YLD were higher in the intervention group (YLD=1,264) than in the control group (YLD=1,244). The results are summarized in Table 18, pg 95. 6.4.6 Incremental cost-effectiveness estimates In the probabilistic analysis of the base case, the incremental cost of the intervention group compared to the control group was $14,592, while the YLL was -37 (i.e., negative health gains in the intervention group compared to the control group) using a program perspective. Similarly, the societal perspective revealed the incremental cost of the intervention group compared to the control group was $20,438, while the YLL was -37 (Table 19, pg 97). The cost-effectiveness plane demonstrated statistical uncertainty around ICER, indicating a wider distribution extending from the northeast quadrant to the northwest quadrant. The mean ICER fell into the northwest quadrant, suggesting higher costs and negative health gains (health loss) (Figure 7, pg 99). The CEAC quantifies the uncertainty by demonstrating the probability of ICERs at a given WTP. The probability of ICERs falling under the standard WTP threshold of $1,468 was 26%. In a conservative scenario with WTP threshold of $489 (i.e., one-third of GDP per capita), the probability of ICERs falling under this threshold further decreased to 16% (Figure 8, pg 100). Upon characterizing uncertainty of individual variables, the sensitivity analysis revealed an inverse relationship between trial-based effect size and estimated ICER, suggesting a high degree of uncertainty associated with this variable. When the effect size was assumed to be as 79 low as 1%, the ICER increased to $420 per life-year saved. The ICER substantially declined to $21 per life-year saved when the effect size was hypothetically increased to 20%. Several other scenarios, including varying the discount rate (0% versus 6%), varying the currency exchange rate (2014 versus 2016) and varying the implementation costs (from -10% to +10%), did not dramatically change the results (Figure 9, pg 101). 6.4.7 Time trends for costs relative to outcomes The annualized program-related costs were estimated at $178,199 in the intervention group and $106,706 in control group. The average cost in the intervention group was as high as $71 per pregnancy in the first quarter of 2015 and as low as $29 per pregnancy in the first quarter of 2017 (Figure 10-A, pg 102). The average costs were $52 in the first quarter of 2015 and $18 in the first quarter of 2017 in the control group. The rate of adverse pregnancy outcomes avoided steadily improved over the study period in both groups. The gain was relatively higher in the control group, compared to the intervention group (Figure 10-B, pg 102). 6.5 Discussion This chapter reports the incremental cost relative to health benefit for CLIP combined with routine pregnancy care compared to routine pregnancy care alone. The mean ICER in this study suggests that the intervention requires additional cost but may result in lower or even zero health benefit. In the economic literature, this scenario is referred to as the intervention being dominated by the control. The probabilistic analysis demonstrated a wide range of statistical uncertainty around ICERs and highlights low probability for the intervention to be acceptable at a standard WTP threshold in Pakistan. There is a plethora of scientific evidence in support of community-based interventions in LMICs. Previously, several economic studies have reported favorable (i.e., lower) ICERs for 80 similar community-based interventions in South Asia and Africa. A recent study on newborn home visits by trained volunteers, compared to routine home visits by untrained volunteers, in rural Ghana, reported an ICER of $379 (95%CI: $227 -873) per life-year saved (116). Similarly, other studies evaluated women’s group interventions compared to the current practice of individual-level pregnancy care and found ICERs of $375 and $138 in Bangladesh and Nepal, respectively (117,118). A few studies also reported much lower ICERs, such as newborn home visits by community care provider (ICER= $8) in India (119), training traditional birth attendants (ICER= $23) in Zambia (120), and women’s group interventions (ICER= $112) in Malawi (121). More importantly, the intervention effect size ranged from 40% to 73% in these studies, which likely is an important driver for finding more favorable ICERs. In the CLIP Trial, the economic findings were driven by a statistically insignificant effect size and wide confidence intervals for maternal and perinatal outcomes (Table 20, pag 98). The ICERs falling in the Northwest quadrant on the cost-effectiveness plane suggested higher cost and negative health consequence. It is debatable that a screening-cum-educational intervention (such as the CLIP package) may have resulted in the unintended health consequences or potential harm to some study participants. In particular, higher rates of adverse pregnancy outcomes related to severe maternal and newborn morbidities were important drivers of negative health consequences. Further investigation is needed to perform the disaggregated analysis of mortality and morbidity outcomes taking into account the socio-demographic characteristics, contextual factors (e.g. access to the health system, and ability to pay for services), intervention coverage, and dose-response comparing intervention and control groups. However, smaller sample size for mortality outcomes, lower statistical precision, and wider uncertainty levels will be major limitations for sub-group analysis. 81 The time trend analysis suggests lower costs to rates of adverse outcomes avoided in the control group compared to the intervention group. Although the rates of adverse outcomes avoided improved in both groups, they improved more in the control group indicating a possibility of a ‘Hawthorne effect’. During the trial period, the control group may have benefited from household surveillance and activities related to health system capacity enhancement. Clearly, our study findings raise an important research question: are simple health interventions both less costly and more effective than complex health interventions? The Medical Research Council in the UK describes that complex interventions contain several interacting components, which makes impact evaluation more difficult because there are multiple process-related outcomes (122). Howe et al. reported that cluster trials involving complex health interventions often yield non-significant results, and are costly to implement (123). CLIP meets the classic definition of a complex health intervention (124), so it is plausible that higher costs relative to benefits in the intervention group actually reflect inherent methodological limitation(s) associated with the cluster design or the complex nature of the intervention. We did not find any supporting literature to differentiate the cost-effectiveness of simple versus complex health interventions, and future research in this area is highly recommended. It is worth noting that the temporal trends also indicated a learning curve with a possibility of finding incremental health gains in the intervention group, should the time-to-effect be lengthened. More research is needed to assess the long-term benefits of the CLIP interventions, compared to household surveillance embedded into the routine pregnancy care. This study finding highlights the substantial burden of OOP costs to pregnant women and their families. The mean cost of maternal health resource utilization was higher in the intervention group. The increased number of intensive care admissions and relatively more 82 deliveries taking place in the private health sector could explain incremental maternal costs in the intervention group. Although the protocol-driven health resource utilization remained less than 2%, it may also have contributed towards higher costs in the intervention group. The majority (~71%) of these costs were out-of-pocket. Only 27% (n=5,517) of women reported fund savings at the time of pregnancy registration; higher cost of care associated with obstetric emergencies implied greater financial burden for families. This was further triangulated with our initial FGs, where participants revealed poverty and unaffordability of health services as the main reasons for delaying care. In the absence of a safety net, catastrophic expenditures resulting from pregnancy complications are likely to plunge families into a vicious cycle of poverty. Community-based strategies such as creating village committees for pooling of resources in Rwanda (125), voucher schemes for patient transport in Bangladesh (126), and active community participation on a photo-voice project in Contra Costa (127) have improved care-seeking practices, increased demand for skilled births, and reduced maternal and newborn mortalities elsewhere. 6.5.1 Strengths and limitations The prospective data collection on a large pregnancy cohort alongside the CLIP Trial was a major strength, and a pragmatic approach to real-world costs from the private and public health sectors added methodological rigor to our subsequent cost-effectiveness analysis. Our cost-effectiveness estimates are robust, as we have accounted in our analysis for possible uncertainties related to costs and health outcomes. We faced several challenges in calculating YLD component of the DALYs, as the disability weights (DWs) for many severe maternal and newborn conditions evaluated in the CLIP Trial are not established. Severe conditions include maternal coma, disseminated intravascular coagulation, newborn seizures (excluding epilepsy), newborn coma, and bleeding, 83 which may enormously contribute to disability burden in this vulnerable population. A few other acute conditions, such as feeding/breathing difficulty in a newborn, neonatal jaundice, and newborn infections could possibly result in long-term motor or cognitive impairments, are not considered in DW calculations by GBD. We also noticed some inconsistencies in DWs for a few conditions. For instance, DW of maternal hemorrhage (loss of more than 1 litre of blood) is reported as 0.324, which is more than the DW of 0.316 for stroke, a much more severe lifelong disability. Other limitations, such as premature home births and congenital malformations, where we often lack clinical newborn assessments by a certified health professional, may grossly underestimate the disability burden at a population level. Clearly, more research is needed for a comprehensive understanding of DWs, to capture the most prevalent and likely severe conditions during the pregnancy and postpartum periods. The economic analysis was undertaken on the completed follow-up cases captured in the trial closeout surveillance data as of August/September 2017, and additional data cleaning may result in random invalidation of few cases in both groups. Some limitations related to health resource utilization data include women lost to follow-up and unavailability of medical charts at the health facility. In the CLIP Trial, only a small fraction of women were lost to follow-up and still on follow-up. Schafer, in his paper on statistical methods in medical research, highlights that list-wise deletion, when the data is missing completely at random, provides estimates that are unbiased (128). Bouhlila et al. recommend using multiple imputations using a chained equation approach when the sample size is fairly small and missing data exceed 20% (129). The health facility admission charts were available only for 1% of pregnancies recruited in the trial (416 of 39,444), a consequence of poor medical record-keeping practices at the health facilities. This might have underestimated the intensity of HRU for women who were hospitalized or sought 84 emergency care after they were referred from the community. Lack of medical chart verification is an important limitation to health resource utilization, particularly for women who were very sick and may have received intensive care at the health facility. 6.6 Conclusions The economic analysis suggests that the CLIP package combined with routine pregnancy care is not a cost-effective strategy when compared to routine pregnancy care alone. More research is needed to explore the contextual aspects (i.e. enabling and impeding factors) during trial-implementation period through subsequent process-evaluation design. Interventions involving mHealth technologies also have financial implications of start-up costs for health care system, so budget impact analysis is imperative to inform policy decisions about resource allocation. This also relates to finding an extra funding envelope, or possible areas for disinvestment (i.e. opportunity cost) to support future similar initiatives in Pakistan, and elsewhere in LMICs. 85 Table 9. Model input parameters and type of distributions Model inputs parameters Value 95% confidence intervals Source Distribution Probabilities Maternal mortality, T1 0.0027 0.002, 0.004 CLIP Trial Log-normal Maternal morbidity, T1 0.1088 0.104, 0.113 Early neonatal mortality, T1 0.0404 0.037, 0.043 Late, neonatal mortality, T1 0.0096 0.008, 0.011 Early/late neonatal morbidity, T1 0.0992 0.009, 0.010 Stillbirths, T1 0.0462 0.043, 0.049 Composite adverse outcomes, T1 0.2655 0.258, 0.272 Screened for HDP, T1 0.5655 0.553, 0.576 HDP prevalence, T1 0.0501 0.045, 0.055 Sought care, T1 0.8902 0.877, 0.903 Maternal mortality, T0 0.0026 0.002, 0.003 Maternal morbidity, T0 0.0895 0.085, 0.094 Early neonatal mortality, T0 0.0426 0.039, 0.045 Late, neonatal mortality, T0 0.0078 0.006, 0.009 Early/late neonatal morbidity, T0 0.0651 0.061, 0.068 Stillbirths, T0 0.0496 0.046, 0.052 Composite adverse outcomes, T0 0.2187 0.210, 0.234 Sought care, T0 0.8827 0.864, 0.896 Average cost (in USD$) CLIP-home based screening for HDP, T1 $5.3 4.7, 5.8 Financial reports, budgets, and mixed-methods research alongside CLIP trial Gamma CLIP-community engagement, T1 $4 3.6, 4.4 CLIP-LHWs training and support, T1 $4 3.6, 4.4 Surveillance and health capacity enhancement, T1 $16 14.4, 17.6 Surveillance & health capacity enhancement, T0 $17 15.3, 18.7 Maternal HRU, T1 $163 51.0, 529.0 Newborn HRU, T1 $4 0.7, 23.4 Maternal HRU, T0 $157 49.0, 506.0 Newborn HRU, T0 $4.2 0.6, 24.6 Years of life lost Maternal mortality 44 38, 50 CLIP Trial and GBD Log-normal Perinatal, early/late neonatal mortality, and stillbirths 68 - T1 = Intervention; T0= control; HRU = Health resource utilization; HDP = Hypertensive disorders of pregnancy; GBD = Global Burden of Disease 86 Table 10. Baseline characteristics at enrollment Variables Intervention [N=20,264] Control [N=19,180] Socio-demographic profile Maternal age in years, mean (+SD) 28 (4.7) 27.6 (4.4) Gestational age in weeks at enrollment, mean (+SD) 21 (8.3) 21.3 (8.2) Maternal primary education, number (%) 3972 (19.6) 3276 (17.1) Husbands with primary education, number (%) 9532 (47.2) 8246 (43.2) Parity, mean (+SD) 2.5 (2.4) 2.7 (2.5) Maternal occupation, number (%)  Unemployed / unpaid domestic services  Unskilled manual labour  Skilled manual labour  Agricultural work  Others (Professional, sales, student)  Don’t know / Missing 16180 (80) 601 (3) 585 (2.8) 458 (2.3) 151 (0.8) 2289 (11.1) 14159 (74) 1599 (8.3) 721 (3.8) 444 (2.3) 71 (0.4) 2186 (11.2) Husband’s occupation, number (%)  Unemployed  Domestic services  Unskilled manual labour  Skilled manual labour  Agricultural work  Sales and services  Professionals / managerial  Others (clerical, student)  Don’t know / Missing 393 (2) 287 (1.4) 3595 (17.7) 6846 (34) 3859 (19) 1253 (6.3) 1310 (6.5) 398 (2) 2248 (11) 241 (1.3) 114 (0.6) 6485 (34) 5784 (30) 2449 (12.3) 823 (4.3) 769 (4) 514 (2.5) 2129 (11) Medical conditions/illnesses, number (%) High blood pressure 1184 (6) 1005 (5) Seizures 214 (1) 110 (1) Diabetes 99 (0.5) 102 (1) Tuberculosis 115 (0.6) 72 (0.4) Anemia 3890 (19) 4508 (24) Worms 49 (0.2) 56 (0.3) Care-seeking preferences, number (%): Allowed to seek routine antenatal care 15728 (78) 16128 (84) Allowed to seek care in the event of health emergency 15388 (76) 15674 (82) Require care from family/friend, who would otherwise work for wages 2806 (14) 2232 (12) Funds saved for pregnancy related emergencies 5517 (27) 4487 (23) Cluster-level variables Baseline neonatal mortality rate (per 1,000 live births) in last 12 months 29.74 27.96 Population density 4.10 5.28 87 Table 11. Maternal health resource utilization Variables Intervention [N=18,449] Control [N=17,349] Mean difference (Int. – Cont.) n (%) Mean (+SD) n (%) Mean (+SD) Received routine antenatal care (ANC), from*:  Medical doctor  Nurse  Midwife  Lady Health Worker  Any provider-pilot [n=16013, 87%] 14449 (90) 269 (2) 546 (3) 2864 (18) 936 (6) 4.4 (4.1) [n=14266, 82%] 13192 (91) 206 (2) 98 (1) 69 (0.5) 835 (6) 3.3 (3.2) 1.1 Ambulatory care for pregnancy related ailments, other than routine ANC Primary Health Centre (PHC) visit 2992 (16) 2.5 (1.6) 3152 (18) 2.5 (1.7) - Higher Level Care Facility (HLCF) visit 2631 (14) 2.8 (1.9) 2509 (15) 2.3 (1.6) 0.5 Diagnostic tests/procedures* Imaging / scanning:  Ultrasound  Chest x-ray  CT scan [n=18041, 98%] 17829 (99) 2866 (16) 2817 (16) 2.5 (1.7) [n=16931, 97%] 16688 (99) 3392 (20) 3364 (20) 3 (2.1) -0.5 Blood tests:  Complete blood count (CBC)  Other blood tests: (Culture, Creatinine, AST/ALT) [n=10120, 55%] 9635 (95) 3676 (34) 0.7 (0.9) [n=10732, 62%] 10630 (99) 3630 (35) 0.8 (0.9) -0.1 Urine tests:  Dipstick  Lab/microscopy [n=8010, 43%] 7734 (97) 3960 (49) 0.5 (0.7) [n=6843, 39%] 6552 (96) 3678 (54) 0.3 (0.6) 0.2 In-patient admissions at higher level health facility-HLCF In-patient admissions HLCF setting:  General ward (GW)  Intensive care unit (ICU)  Both, GW and ICU  GW or ICU-pilot [n=365, 2%] 123 (34) 120 (33) 4 (1) 118 (32) 1.9 (1.1) 2.4 (1.0) - 3.4 (4.5) [n=542, 3%] 386 (71) 49 (9.6) 2 (0.4) 105 (19) 2.2 (1.2) 2.1 (1.1) - 3.1 (2.6) -0.3 0.3 - 0.3 88 Variables Intervention [N=18,449] Control [N=17,349] Mean difference (Int. – Cont.) n (%) Mean (+SD) n (%) Mean (+SD) Delivery related information Deliveries, excluding miscarriage Place of delivery  Home  Public health facility  Private health facility  En-route [n=17,891, 97%] 4277 (24) 5802 (32.2) 7756 (43.5) 56 (0.3) - [n=16,849, 97%] 4072 (24) 6160 (36.7) 6566 (39) 51 (0.3) - - Type of delivery [of those delivered at health facility]  Spontaneous vaginal delivery  Assisted vaginal delivery  Caesarean-section [n=13,558, 76%] 9107 (67) 1224 (9) 3227 (24) - [n=12,726, 76%] 9094 (71) 917 (7) 2715 (22) - - Others: Transport, productivity losses due to illness and wage-worker Transport to health facility 16046 (87) 4.6 (3.2) 14860 (86) 3.9 (2.6) 0.7 Productivity/time losses (# of days missed wages due to ill health/hospitalization) 5719 (31) 2.3 (5.1) 5638 (33) 2.3 (5.4) - Hired wage-worker for household chores 1738 (9.4) 0.3 (2.2) 580 (3.3) 0.2 (1.3) 0.1 Protocol driven resource utilization Non-urgent (24 hours) referrals to health facility 287 (1.6) - - - - Urgent (4 hours) referrals to health facility 155 (0.8) - - - - * Represents multiple response question 89 Table 12. Newborn health resource utilization Variables Intervention [N=18,449] Control [N=17,349] Mean difference (Int. – Cont.) n (%) Mean (+SD) n (%) Mean (+SD) Ambulatory care* Newborn illness and/or complications  Medical doctor  Nurse  Midwife  Lady Health Worker  Traditional healer [n=1430, 8%] 1332 (93) 15 (1) 9 (0.6) 8 (0.5) 55 (4) 1.9 (1.5) [n=1901, 11%] 1858 (97) 16 (1) 12 (0.6) 13 (0.7) 33 (2) 1.8 (1.4) 0.1 Diagnostic tests/procedures* Diagnostic tests:  Phototherapy  Fetal imaging  Blood tests [n=164, 1%] 23 (14) 75 (46) 145 (88) 0.5 (1.1) [n=205, 1.2%] 11 (5) 101 (49) 172 (84) 0.6 (1) -0.1 In-patient admissions at higher level health facility-HLCF In-patient admissions HLCF setting:  General ward (GW)  Intensive care unit (ICU)  Both, GW and ICU [n=241, 1.3%] 147 (61) 80 (33) 14 (6) 1.9 (1.4) [n=324, 2%] 227 (70) 75 (23) 22 (7) 1.7 (1.2) 0.2 Others: Transport, productivity losses due to illness and wage-worker Transport to health facility 976 (5.3) 2.1 (1.4) 1037 (6) 1.9 (1.2) 0.2 Productivity/time losses (# of days missed wages due to ill health/hospitalization) 683 (3.7) 4.2(8.4) 752 (4.3) 3.2 (7.6) 1 * Represents multiple response question 90 Table 13. CLIP handbag items in Sindh, Pakistan Items Quantity procured Unit costs (US$) Microlife VSA blood pressure monitoring device 390 devices 22.2 Large size cuff for blood pressure device 197 cuffs 11.2 PX-13-B Kenek pulse oximeter device 377 devices 20.1 Mobile device (Model Q-Noir A7 phone) 92 phones 125.6 Mobile device (Model Q-300 tablet 7.0”) 566 tablets 143 Urine specimen cups 30 packs (250 cups/pack) 25.3 Urine dipsticks 105 packs (100 sticks/pack) 16.7 Dipstick colour-code sheet 70 packs (100 sheets/pack) 18.7 Adhesive bandages 12 packs (100 bandages/pack) 0.7 Alcohol swabs 15 packs (100 swabs/pack) 1.7 Syringes (10ml) 14 packs (100 syringes/pack) 7.5 Needles (23 gram) 9 packs (100 needles/pack) 1.6 Hand-gloves 8 packs (100 gloves/pack) 2.9 Needle cutters 11 packs (3 cutters/pack) 7.2 Disposal/sharp boxes 42 boxes 0.6 CLIP-card with polythene bag 16 packs (1000 cards/pack) 26.7 Magnesium sulphate (10ml) injection 242 packs (5 vials/pack) 0.6 Methyldopa (250mg) tablet 19 packs (100 tablets/pack) 3.5 1 US$ = 104.7 PKR (average exchange rate of 2014-2016 http://www.sbp.org.pk/); figures are rounded off to the nearest tenth. 91 Table 14. Societal costs comparing intervention and control Costs in, US$* Intervention Control (A) Costs to health system (program-related) CLIP package of care: home-based screening for HDP, community-based HDP management/referral, and community engagement) $137,756 - Incentive, trainings and transport for Lady Health Workers $79,398 - Health system capacity enhancement (workshop, and seminar) $3,573 $3,573 Surveillance for ascertainment of trial outcomes (personnel, mobile technology and transport) $309,848 $309,848 Sub-total (A) $530,755 $313,602 (B) Costs to pregnant women and their families Maternal health resource utilization and productivity losses $2,153,346 $1,879,995 Newborn health resource utilization and productivity losses $14,260 $17,961 Sub-total (B) $2,167,606 $1,897,956 Total societal costs (A + B) $2,698,361 $2,211,558 HDP = Hypertensive disorders of pregnancy; Exchange rate: 1 US$ = 104.7 PKR (average exchange rate of 2014-2016 http://www.sbp.org.pk/). * Discounted at 3% rate 92 Table 15. Regression-adjusted coefficients for study variables 93 Table 16. Regression-adjusted cost estimates for resource utilization (in US$) Variables Mean Std. Error 95% Confidence Interval Lower Upper Maternal health resource utilization, (p-value = 0.958) Intervention $163.4 98 $51 $529 Control $156.7 94 $49 $506 Newborn health resource utilization, (p-value = 0.964) Intervention $4 3.6 $0.7 $23.4 Control $4.2 3.8 $0.6 $24.6 Combined maternal and newborn health resource utilization, (p-value = 0.962) Intervention $164 98 $51 $528 Control $158 94 $49 $507 Intracluster correlation coefficient (ICC)= 0.02 Adjusted for maternal age, maternal primary education, husbands with primary education, parity, savings for pregnancy-related emergencies, and medical conditions including high blood pressure, seizures, and diabetes. 94 Table 17. Years of life lost (YLL) in the intervention and control groups Outcomes Intervention Control n Rate* YLL n Rate* YLL Maternal deaths 55 2.71 76 51 2.66 64 Early neonatal deaths 820 40.5 1,214 818 42.7 1,237 Late neonatal deaths 196 9.7 305 151 7.9 217 Stillbirths 937 46.2 1,373 952 49.6 1439 Combined YLL for maternal, perinatal and neonatal deaths 2,968 2,957 * Rate per 1000 identified pregnancies; CI= Confidence intervals 95 Table 18. Years lost due to disability (YLD) in the intervention and control groups Outcomes Intervention Control n Rate* (95% CI) YLD n Rate* (95% CI) YLD Maternal morbidities (excluding maternal deaths) Eclampsia 29 1.43 (0.90, 2.12) 0.0 16 0.79 (0.49, 1.3) 0.0 Stroke 63 3.11 (2.41, 3.92) 489 85 4.19 (3.5, 5.9) 660 Antepartum haemorrhage 639 31.53 (2.90, 3.43) 72 483 23.84 (22.7, 27.6) 54 Obstetric sepsis 1,472 72.64 (69.1, 76.4) 193 1,189 58.68 (56.5, 65.3) 156 Vesicovaginal or rectovaginal fistula 56 2.76 (2.11, 3.61) 470 39 1.92 (1.4, 2.9) 328 Dialysis 1 0.05 (0.02, 0.24) 14 2 0.10 (0.07, 0.34) 28 Blood transfusion 235 11.60 (10.1, 13.2) 26 161 7.95 (5.2, 9.7) 18 Coma 26 1.28 (0.85, 1.84) NR 22 1.09 (0.73, 1.8) NR Disseminated intravascular coagulation (DIC) 4 0.20 (0.62, 0.47) NR 6 0.30 (0.21, 0.43) NR CPR 4 0.20 (0.62, 0.47) NR 5 0.25 (0.18, 0.33) NR Mechanical ventilation 10 0.49 (0.25, 0.88) NR 8 0.39 (0.27, 0.52) NR Intervention for major post-partum haemorrhage 1 0.05 (0.02, 0.24) NR 4 0.20 (0.11, 0.31) NR (A) Sub-total YLD for maternal morbidities 1,264 1,244 Newborn morbidities (excluding newborn deaths) Feeding difficulty 722 37.64 (33.1, 39.3) 0.0 188 9.80 (8.4, 11.3) 0.0 Breathing difficulty 230 11.99 (9.9, 13.2) 0.0 174 9.07 (7.7, 10.5) 0.0 Lethargy 464 24.19 (20.8, 26.7) 0.0 455 23.72 (21.6, 25.7) 0.0 Hypothermia 271 14.13 (11.8, 16.2) 0.0 170 8.86 (7.6, 10.3) 0.0 Jaundice 324 16.89 (14.3, 18.8) 0.0 327 17.05 (15.2, 18.9) 0.0 Seizures 161 8.39 (6.78, 10.2) NR 134 6.99 (5.8, 8.2) NR Coma 11 0.57 (0.28, 0.94) NR 13 0.68 (0.37, 1.13) NR 96 Outcomes Intervention Control n Rate* (95% CI) YLD n Rate* (95% CI) YLD Newborn morbidities (excluding newborn deaths) continued Umbilical cord infection 273 14.23 (11.9, 16.2) NR 232 12.10 (10.6, 13.7) NR Skin infection 2 0.10 (0.90, 0.33) NR 0 0.00 - NR Bleeding 55 2.87 (2.1, 3.6) NR 26 1.36 (0.90, 1.95) NR (B) Sub-total YLD for newborn morbidities - - Total (A+B) 1,264 1,244 * Rate per 1000 identified pregnancies; NR = not reported by Global burden of Diseases; Zero YLD represents acute conditions 97 Table 19. Probabilistic cost-effectiveness estimates (10,000 iterations) Outcomes* Intervention Control Increment# ICER (95% CI) Program perspective: Combined YLL for maternal, perinatal, and early/late neonatal outcomes Mean cost (in US$) $31,695 $17,103 $14,592 Intervention is dominated Mean YLL 2,962 2,925 -37 Societal perspective: Combined YLL for maternal, perinatal, and early/late neonatal outcomes Mean cost (in US$) $179,670 $159,232 $20,438 Intervention is dominated Mean YLL 2,962 2,925 -37 * per 1000 identified pregnancies # Incremental costs = Intervention, minus control; Incremental YLL= Control, minus intervention 98 Table 20. Regression adjusted effect size for maternal and perinatal outcomes Clinical outcome parameters Adjusted odds ratio (95% confidence intervals) % Protective effectiveness (1 – adjusted odds ratio) Maternal mortality 1.08 (0.69 – 1.72) -8% Maternal morbidity 1.10 (0.57 – 2.14) -10% Perinatal and late neonatal mortality 0.94 (0.86 – 1.03) 6% Neonatal morbidity 1.22 (0.77 – 1.95) -22% Composite adverse pregnancy outcomes 1.20 (0.84 – 1.72) -20% Adjusted for maternal age, maternal primary education, husbands’ with primary education, parity, and baseline neonatal mortality. 99 Figure 7. Cost-effectiveness plane showing statistical uncertainty This graph represents probabilistic analysis using Monte Carlo simulation for 10,000 iterations. On X-axis, we have incremental years of life lost averted, and Y-axis we have incremental cost in thousands. Each do represents a single ICER from the probabilistic model. The highlighted dot represents the mean ICER. Data points falling northwest of the origin indicate the possibility that (positive) costs will incur relative to no or even negative health benefit. 100 Figure 8. Cost-effectiveness acceptability curve for the CLIP Trial This graph represents probability of ICERs at given willingness to pay threshold. On X-axis, we have willingness to pay to life years saved; and Y-axis we have probability around ICERs. LIC = Low-income countries; LMIC = Low-and-Middle income countries. Attractive (LICs $223) Highly cost-effective (Pakistan $489) Highly cost-effective (LMICs $1417) Cost-effective (Pakistan $1468) Cost-effective (LMICs $4252) 101 Figure 9. One-way sensitivity analysis Dark green bars indicate the direction and magnitude of change of the ICER when the given input variable is at its minimum plausible value, whereas light green bars indicate the direction and magnitude of change of the ICER when the same input variable is at its maximum plausible value. Variables listed towards the top of this diagram contribute more to the overall uncertainty in the cost-effectiveness ratio than do those towards the bottom, which contribute relatively little to the uncertainty in the cost-effectiveness ratio. 102 Figure 10. Time trends for costs and rate of adverse pregnancy outcomes avoided 10A. 10B. Rate of adverse outcomes avoided Rate of adverse outcomes avoided 103 Chapter 7: Conclusions 7.1 Summary of findings The objective of this dissertation was an economic appraisal of the CLIP package of care combined with routine pregnancy care, compared to routine pregnancy care alone, in Sindh, Pakistan. We investigated the evidence about costs and cost-effectiveness of CLIP interventions to inform policy decisions about resource allocation for post-trial scale-up, should the intervention findings provide evidence of clinical effectiveness. The literature review in Chapter 2 identified knowledge gaps for societal costs, and contextual limitation of scientific evidence about cost-effective PE/E interventions only to developed (western) countries, as not many economic studies exist in the area of PE/E identification and appropriate case management in LMICs. The cost-effective measures as reported in previous economic studies, such as automated urine analysis, parental administration of MgSO4, and immediate delivery are mainly available at the health facility level. Thus, evidence about costs and cost-effectiveness of comprehensive community-based PE/E interventions is grossly lacking. Chapter 3 provided an overview of a mixed-methods approach to economic evaluation, illustrated contextual relevance to collect costs and health outcomes data alongside the trial, and reflected on the important considerations for economic model design (i.e., use of decision tree analysis, societal perspective, appropriate discount rates, methods of dealing with uncertainties, and relevant health outcome measures), as indicated in the literature review. The FGs with pregnant women and dyadic decision makers in the community indicated a large burden of OOP costs and productivity losses due to ill health of the mother or newborn, as presented in Chapter 4. The sociocultural circumstances in which a woman requires her 104 husband’s permission for care-seeking, lacks financial savings for pregnancy-related emergencies, anticipates higher transportation costs for visit to the health facility, or requires a family member or friend to accompany her during hospitalization together make women and their families vulnerable to catastrophic expenditure on health. Much attention is needed in the areas of community-based provision of basic health services, micro-financing schemes for obstetric emergencies, local transport arrangements, health education, community engagement and interaction, risk stratification, and timely referral of sick woman to a health facility, to avert long-term financial consequences and improve maternal and newborn health in LMICs. The survey of health care providers and facilities highlighted cost disparities within and between public and private health sectors, as presented in Chapter 5. Despite lower costs for inpatient care and delivery-related hospitalizations in the public health sector, many women sought care from private facilities. Close geographic distance, long operating hours, and shorter wait times (as identified in FGs) could have contributed to the higher rates of private health facility utilization. The private sector can also provide cutting-edge diagnostic and therapeutic technologies that may not be available in the public sector, albeit at relatively higher cost. Considering that the role of the private sector is rapidly expanding to almost all rural and urban settings, the economic principle of supply-induced demand could explain this preference for private care providers. This was relevant, as the rate of Caesarean section was higher in the private sector. Unless pregnant women have community-based safety nets or some financial savings, they and their families will continue to face the higher burden of OOP spending as a result of health resource utilization in the private sector. Economic evaluation of the CLIP implementation revealed financial implications of incremental costs to the health care system, as shown in Chapter 6. The cost of mHealth 105 infrastructure, particularly upfront costs to purchase digital clinical devices and smartphones or tablets, requires a huge initial financial investment from the health care system. Health benefits of such interventions yield a long-term return on investments, so strong scientific evidence, constant policy advocacy, ever-increasing political will, and winning public support are essential elements to promote mHealth initiatives in LMICs. This supports our argument in the literature review to have more research done on epidemiological and economic aspects of such initiatives so that decisions about resource allocations are informed by robust evidence. The cost-effectiveness of the CLIP interventions, as indicated in Chapter 6, suggests incremental costs and no evidence of health benefits. The probabilistic analysis demonstrated greater uncertainty around ICERs, which is largely explained by statistically insignificant effect size and wide confidence intervals. The economic evaluation is based on a utilitarian philosophy implying “good for larger society,” in which decisions are mainly weighted in reference to societal WTP for a given intervention. For example, those interventions that produce a DALY for $1,468 or less (i.e., GDP per capita in Pakistan) are a bargain, whereas those that require more than $1,468 are considered unaffordable. In this study, the probability of ICERs falling below the acceptable WTP threshold was less than 30%. Such a low probability implies “bigger buck for lower bang,” a common phrase used in the economic literature to explain interventions yielding in low value for money. This relates to an emerging debate in the economic literature on the fallacy of the threshold in the cost-effectiveness analysis. GDP per capita can be quite volatile given the unpredictability of economic situations in low-income countries. An arbitrary, high WTP threshold can be extremely welcoming to many new interventions. But these interventions may be less cost-effective than expected and may become a drain on the health budget rather than being a good use of it. More research is needed to determine an appropriate 106 threshold, and to investigate the effect of changing threshold levels (i.e., high or low) on existing and future maternal and newborn health interventions in LMICs. Advantages of using a mixed-methods approach in economic analysis alongside the trial include real-time data collection, qualitative exploration of community preferences for care utilization and cost drivers that are not typically measured in the quantitative design, capturing real-world costs including health system and patient perspectives, consistency and rigor in data collection and analysis, triangulation of findings, and timely results for decision making. 7.2 Implications This dissertation highlights several implications relevant to the national program of LHWs, health care providers, communities, and future research. The economic analysis findings are useful to inform policy decisions on future investments and/or disinvestments when planning maternal and child health services in Pakistan. The routine pregnancy care demonstrated a strong case for program-related efficiency gains in the delivery of MNH services in Pakistan. Shortage of publicly funded community-based health services further affects the way individuals seek care, assuming that many patients and families will have no choice but to seek care from the private health sector resulting in higher OOP costs. Therefore, the role of public-private partnership, financial protection programs through micro-insurance or community-based financial safety nets, and health services regulations can be instrumental in bridging the service delivery gaps for routine pregnancy care. On other hand, integration of mHealth component into routine pregnancy care requires a careful consideration of incremental costs to health system. For instance, it is estimated that countrywide, CLIP implementation would require additional funding of about $28.5 million per year, given an incremental cost of $284 per each of 100,000 LHWs in the national program. Such an incremental cost may have a detrimental effect on other 107 public health programs (such as the expanded program on immunization, MNCH, blindness prevention, and hepatitis control) funded by the federal government in Pakistan. According to the official budget release for the year 2016-17, $231million was earmarked for the national health services, of which less than 5% was allocated for public health programs. This mismatch of funding should be seen as a significant barrier to health innovation that could lead to a public health crisis in Pakistan. A long-term fiscal solution could be possible with price negotiation with technology partners and the active involvement of civil society representatives. Such an inclusive approach will foster more avenues for technology development and applications in the area of global maternal and newborn health. This study also draws the attention of policymakers to individual components of the CLIP package of care, as some of these could be feasibly and affordably integrated into the existing LHWs service delivery platform. Guided by our local field observations, we determined that: (i) community engagement had the potential to maximize the program reach/coverage through active participation of women and household decision makers; (ii) equipping community health workers with blood pressure devices improved HDP diagnosis and initiated timely patient referral; and (iii) training care providers led to compliance with a standard management regime as more women received antihypertensive and MgSO4 treatments. The successful implementation of these interventions has greater implications for task shift to LHWs in Sindh, Pakistan. The World Health Organization has strongly suggested shifting specific tasks (e.g., clinical screening and management) normally performed by high-level care providers to lower-level care providers, to address inefficiencies in the health system and to improve population health outcomes. 108 The use of a mixed-methods approach alongside the trial is a resource intensive exercise but offers several advantages for collecting data, interpretation of findings, perform data triangulation, and strategizing ways of communicating and interpreting findings to all stakeholders in the study. The involvement of community stakeholders and health policy decision makers at the initial stage also reflected patient-oriented research, where study participants are engaged throughout the process of research. Holding FGs with all stakeholders was an important step in understanding cost drivers relevant and important to health care providers and pregnant women and their families. Although subsequent interactions and participation in the annual meetings were limited to health decision makers in the LHWs program, we plan on organizing a knowledge translation seminar to keep the wider communities informed about the study results. Economic evidence of cost-effectiveness is only one necessary factor under consideration in the decision-making process. Other factors such as system innovation, the burden of disease, unmet need, political willingness, and feasibility aspects also come into play. In this connection, a higher-level policy dialogue with the ministry of health, stakeholders from other preventive programs, and donor agency representatives will provide an opportunity to discuss and reflect on key study findings and indicate areas of future research in Pakistan. I hope to facilitate this policy discussion, and share evidence-based findings, and my field experience in maternal and newborn health in Sindh, Pakistan. This dissertation indicates several areas that would benefit from future research: (i) process evaluation to explore contextual aspects (i.e. enabling and impeding factors) during trial implementation, and sub-group analysis to assess health benefits comparing intervention and control groups; (ii) real-time economic evidence of the costs and cost-effectiveness of novel 109 strategies to identify and manage PE/E using mHealth applications, clinical digital devices, and placental biomarkers in LMICs; (iii) landscape analysis on the determinants for the use of private health facilities despite higher OOP costs; (iv) comparative analysis of maternal and newborn outcomes between private and public health facilities; (v) exploration of DWs for severe newborn conditions; (vi) determining appropriate WTP threshold and effects of changing threshold levels on existing MNCH interventions; (vii) cost-effectiveness of household surveillance and health system capacity enhancement; and (viii) potential cost savings associated with task shift to LHWs using a societal perspective. 7.3 Conclusions This dissertation identified critical knowledge gaps related to cost and cost-effectiveness of PE/E interventions in LMICs. The societal perspective on cost drivers in particular highlighted a large burden of OOP spending and productivity losses for pregnant women and their families. The health care providers and district-level decision and policy makers critically appraised the merits of health technology use and identified financial challenges related to large initial investments into mHealth devices and Internet technology. Economic analysis revealed that CLIP is not a cost-effective strategy compared to routine pregnancy care alone. Further research is needed to understand the contextual aspects of trial implementation through process evaluation. Also, future economic studies evaluating simple versus complex health interventions will add value to the existing body of literature and guide policy decisions on resource allocation in the area of MNH in Pakistan and elsewhere. 110 Bibliography (1) Chong YS, Arulkumaran S. Obstetricians perhaps it’s time to change lenses. Pregnancy and Childbirth 2013;17:19. (2) Black RE, Levin C, Walker N, Chou D, Liu L, Temmerman M, et al. Reproductive, maternal, newborn, and child health: key messages from Disease Control Priorities 3rd Edition. The Lancet 2016;388(10061):2811-2824. (3) Hogan MC, Foreman KJ, Naghavi M, Ahn SY, Wang M, Makela SM, et al. Maternal mortality for 181 countries, 1980–2008: a systematic analysis of progress towards Millennium Development Goal 5. The lancet 2010;375(9726):1609-1623. (4) Donnay F. Maternal survival in developing countries: what has been done, what can be achieved in the next decade. International Journal of Gynecology & Obstetrics 2000;70(1):89-97. (5) Ekele BA, Bello SO, Adamu AN. Clusters of eclampsia in a Nigerian teaching hospital. International Journal of Gynecology & Obstetrics 2007;96(1):62-66. (6) Higgins JR, de Swiet M. Blood-pressure measurement and classification in pregnancy. The Lancet 2001;357(9250):131-135. (7) Levine RJ, Ewell MG, Hauth JC, Curet LB, Catalano PM, Morris CD, et al. Should the definition of preeclampsia include a rise in diastolic blood pressure of≥ 15 mm Hg to a level< 90 mm Hg in association with proteinuria? Obstet Gynecol 2000;183(4):787-792. (8) Gupte SA, Wagh G. Classification and Diagnosis of Hypertensive Disorders of Pregnancy: A Practical Approach. Hypertensive Disease in Pregnancy 2013. (9) Ødegård RA, Vatten LJ, Nilsen ST, Salvesen KÅ, Austgulen R. Risk factors and clinical manifestations of pre‐eclampsia. BJOG: An International Journal of Obstetrics & Gynaecology 2000;107(11):1410-1416. (10) Yücesoy G, Özkan S, Bodur H, Tan T, Çalışkan E, Vural B, et al. Maternal and perinatal outcome in pregnancies complicated with hypertensive disorder of pregnancy: a seven year experience of a tertiary care center. Arch Gynecol Obstet 2005;273(1):43-49. (11) What we have learned about preeclampsia. Seminars in perinatology: Elsevier; 2003. (12) Gillon TE, Pels A, von Dadelszen P, MacDonell K, Magee LA. Hypertensive disorders of pregnancy: a systematic review of international clinical practice guidelines. PLoS One 2014;9(12):e113715. 111 (13) Wallis AB, Saftlas AF, Hsia J, Atrash HK. Secular trends in the rates of preeclampsia, eclampsia, and gestational hypertension, United States, 1987–2004. American journal of hypertension 2008;21(5):521-526. (14) Health Information about Preeclampsia. 2013; Available at: https://www.preeclampsia.org/health-information. Accessed Oct 29, 2017. (15) Xiong X, Demianczuk NN, Saunders LD, Wang F, Fraser WD. Impact of preeclampsia and gestational hypertension on birth weight by gestational age. Am J Epidemiol 2002;155(3):203-209. (16) Hezelgrave NL, Duffy SP, Shennan AH. Preventing the preventable: pre-eclampsia and global maternal mortality. Obstetrics, Gynaecology and Reproductive Medicine 2012;22(6):170-172. (17) ENGENDER HEALTH. Balancing the scales: A report on barriers and solutions to treat pre-eclampsia and eclampsia. 2007; Available at: http://www.engenderhealth.org/files/pubs/maternal-health/engenderhealth-eclampsia-report.pdf. Accessed October 29, 2017. (18) Payne B, von Dadelszen P, Bhutta Z, Magee L, Adetoro O, Sotunsa J, et al. Protocol 13PRT/9313: The Community Level Interventions for Pre-eclampsia (CLIP) Trials: four prospective cluster randomised controlled trials comparing a package of interventions directed towards improving maternal and perinatal outcomes related to pre-eclampsia with current standards of care (NCT01911494). Lancet 2015. (19) Hutcheon JA, Lisonkova S, Joseph KS. Epidemiology of pre-eclampsia and the other hypertensive disorders of pregnancy. Best Pract Res Clin Obstet Gynaecol 2011 August 01;25(4):391-403. (20) Dunsmuir DT, Payne BA, Cloete G, Petersen CL, Grges M, Lim J, et al. Development of mHealth applications for pre-eclampsia triage. IEEE journal of biomedical and health informatics 2014;18(6):1857-1864. (21) Rosenberg K, Twaddle S. 6 Screening and surveillance of pregnancy hypertension—an economic approach to the use of daycare. Baillière's clinical obstetrics and gynaecology 1990;4(1):89-107. (22) McIntyre D, Thiede M, Dahlgren G, Whitehead M. What are the economic consequences for households of illness and of paying for health care in low- and middle-income country contexts? Soc Sci Med 2006 Feb;62(4):858-865. (23) Magee LA, Pels A, Helewa M, Rey E, von Dadelszen P, SOGC Hypertension Guideline Committee. Diagnosis, evaluation, and management of the hypertensive disorders of pregnancy: executive summary. J Obstet Gynaecol Can 2014 Jul;36(7):575-576. 112 (24) World Bank. World Bank Group - International Development, Poverty, & Sustainability. Available at: http://www.worldbank.org/. Accessed Dec 20, 2017. (25) Methley AM, Campbell S, Chew-Graham C, McNally R, Cheraghi-Sohi S. PICO, PICOS and SPIDER: a comparison study of specificity and sensitivity in three search tools for qualitative systematic reviews. BMC health services research 2014;14(1):579. (26) Agency for Healthcare Research and Quality. Health Care Utilization Project Data Source. 2005; Available at: http://hcup.ahrq.gov. (27) Tsigas EZ, Li R, Callaghan WM. Health and economic burden of preeclampsia: no time for complacency. American Journal of Obstetrics and Gynecology 2017 Sep 1,;217(3):235. (28) Fox A, McHugh S, Browne J, Kenny LC, Fitzgerald A, Khashan AS, et al. Estimating the Cost of Preeclampsia in the Healthcare System: Cross-Sectional Study Using Data From SCOPE Study (Screening for Pregnancy End Points). Hypertension 2017 Dec;70(6):1243-1249. (29) Gan J, He H, Qi H. Preventing preeclampsia and its fetal complications with low-dose aspirin in East Asians and non-East Asians:A systematic review and meta-analysis. Hypertens Pregnancy 2016 August 01;35(3):426-435. (30) Werner EF, Hauspurg AK, Rouse DJ. A Cost-Benefit Analysis of Low-Dose Aspirin Prophylaxis for the Prevention of Preeclampsia in the United States. Obstet Gynecol 2015 December 01;126(6):1242-1250. (31) Imdad A, Jabeen A, Bhutta ZA. Role of calcium supplementation during pregnancy in reducing risk of developing gestational hypertensive disorders: a meta-analysis of studies from developing countries. BMC Public Health 2011 April 13;11 Suppl 3:S18. (32) Longo-Mbenza B, Tshimanga KB, Buassa-bu-Tsumbu B, Kabangu MJ. Diets rich in vegetables and physical activity are associated with a decreased risk of pregnancy induced hypertension among rural women from Kimpese, DR Congo. Niger J Med 2008 August 01;17(3):265-269. (33) Kiondo P, Wamuyu-Maina G, Wandabwa J, Bimenya GS, Tumwesigye NM, Okong P. The effects of vitamin C supplementation on pre-eclampsia in Mulago Hospital, Kampala, Uganda: a randomized placebo controlled clinical trial. BMC Pregnancy Childbirth 2014 Aug 21,;14:283. (34) Ullah MI, Koch CA, Tamanna S, Rouf S, Shamsuddin L. Vitamin D deficiency and the risk of preeclampsia and eclampsia in Bangladesh. Horm Metab Res 2013 September 01;45(9):682-687. (35) Agrawal S, Fledderjohann J, Vellakkal S, Stuckler D. Adequately diversified dietary intake and iron and folic acid supplementation during pregnancy is associated with reduced occurrence of symptoms suggestive of pre-eclampsia or eclampsia in Indian women. PLoS One 2015 March 18;10(3):e0119120. 113 (36) Gupta N, Gupta T, Asthana D. Prediction of Preeclampsia in Early Pregnancy by Estimating the Spot Urinary Albumin/Creatinine Ratio. J Obstet Gynaecol India 2017 August 01;67(4):258-262. (37) National Collaborating Centre for Women's and Children's Health. Cost effectiveness of quantifying proteinuria in women with gestational hypertension. : RCOG Press; 2010. (38) National Collaborating Centre for Women's and Children's Health. Cost effectiveness of automated urinalysis compared with visual urinalysis in screening for proteinuria in women with gestational hypertension. : RCOG Press; 2010. (39) Meads CA, Cnossen JS, Meher S, Juarez-Garcia A, ter Riet G, Duley L, et al. Methods of prediction and prevention of pre-eclampsia: systematic reviews of accuracy and effectiveness literature with economic modelling. Health Technol Assess 2008 Mar;12(6):270. (40) Nathan HL, de Greeff A, Hezelgrave NL, Chappell LC, Shennan AH. Accuracy validation of the Microlife 3AS1-2 blood pressure device in a pregnant population with low blood pressure. Blood Press Monit 2015 October 01;20(5):299-302. (41) Nevalainen J, Korpimaki T, Kouru H, Sairanen M, Ryynanen M. Performance of first trimester biochemical markers and mean arterial pressure in prediction of early-onset pre-eclampsia. Metabolism 2017 October 01;75:6-15. (42) Shmueli A, Meiri H, Gonen R. Economic assessment of screening for pre-eclampsia. Prenat Diagn 2012 January 1,;32(1):29-38. (43) Frusca T, Gervasi M, Paolini D, Dionisi M, Ferre F, Cetin I. Budget impact analysis of sFlt-1/PlGF ratio as prediction test in Italian women with suspected preeclampsia. J Matern Fetal Neonatal Med 2017 Sep;30(18):2166-2173. (44) Hadker N, Garg S, Costanzo C, Miller JD, Foster T, Van WdH, et al. Financial impact of a novel pre-eclampsia diagnostic test versus standard practice: a decision-analytic modeling analysis from a UK healthcare payer perspective. J Med Econ 2010;13(4):728-737. (45) Broekhuijsen K, van Baaren G, van Pampus MG, Ganzevoort W, Sikkema JM, Woiski MD, et al. Immediate delivery versus expectant monitoring for hypertensive disorders of pregnancy between 34 and 37 weeks of gestation (HYPITAT-II): an open-label, randomised controlled trial. Lancet 2015 Jun 20,;385(9986):2492-2501. (46) Vijgen SMC, Koopmans CM, Opmeer BC, Groen H, Bijlenga D, Aarnoudse JG, et al. An economic analysis of induction of labour and expectant monitoring in women with gestational hypertension or pre-eclampsia at term (HYPITAT trial). BJOG 2010 Dec;117(13):1577-1585. (47) Magee LA, von Dadelszen P, Rey E, Ross S, Asztalos E, Murphy KE, et al. Less-tight versus tight control of hypertension in pregnancy. N Engl J Med 2015 Jan 29,;372(5):407-417. 114 (48) Magee LA, von Dadelszen P, Singer J, Lee T, Rey E, Ross S, et al. Control of Hypertension In Pregnancy Study randomised controlled trial-are the results dependent on the choice of labetalol or methyldopa? BJOG 2016 Jun;123(7):1135-1141. (49) Ahmed RJ, Gafni A, Hutton EK, Hu ZJ, Pullenayegum E, von Dadelszen P, et al. The Cost Implications of Less Tight Versus Tight Control of Hypertension in Pregnancy (CHIPS Trial). Hypertension 2016 Oct;68(4):1049-1055. (50) The Magpie Trial: a randomised trial comparing magnesium sulphate with placebo for pre-eclampsia. Outcome for women at 2 years. BJOG 2007 Mar;114(3):300-309. (51) Keepanasseril A, Maurya DK, Manikandan K, Suriya JY, Habeebullah S, Raghavan SS. Prophylactic magnesium sulphate in prevention of eclampsia in women with severe preeclampsia: randomised controlled trial (PIPES trial). J Obstet Gynaecol 2017 October 03:1-5. (52) Simon J, Gray A, Duley L. Cost-effectiveness of prophylactic magnesium sulphate for 9996 women with pre-eclampsia from 33 countries: economic evaluation of the Magpie Trial. BJOG 2006 Feb;113(2):144-151. (53) Shoaib T, Khan S, Javed I, Bhutta SZ. Loading dose of magnesium sulphate versus standard regime for prophylaxis of pre-eclampsia. J Coll Physicians Surg Pak 2009 Jan;19(1):30-33. (54) Barton JR, Istwan NB, Rhea D, Collins A, Stanziano GJ. Cost-savings analysis of an outpatient management program for women with pregnancy-related hypertensive conditions. Dis Manag 2006 Aug;9(4):236-241. (55) Murray CJ, Evans DB, Acharya A, Baltussen RM. Development of WHO guidelines on generalized cost‐effectiveness analysis. Health Econ 2000;9(3):235-251. (56) World Health Organization. Metrics: Disability-Adjusted Life Year (DALY). Available at: http://www.who.int/healthinfo/global_burden_disease/metrics_daly/en/. Accessed October 29, 2017. (57) Donaldson C, Currie G, Mitton C. Cost effectiveness analysis in health care: contraindications. BMJ 2002;325(7369):891-894. (58) Mills A, Brugha R, Hanson K, McPake B. What can be done about the private health sector in low-income countries? Bull World Health Organ 2002;80(4):325-330. (59) Franco A, Malhotra N, Simonovits G. Social science. Publication bias in the social sciences: unlocking the file drawer. Science 2014 Sep 19,;345(6203):1502-1505. (60) Mitton C, Donaldson C. Health care priority setting: principles, practice and challenges. Cost effectiveness and resource allocation 2004;2(1):3. 115 (61) Michalow J, Chola L, McGee S, Tugendhaft A, Pattinson R, Kerber K, et al. Triple return on investment: the cost and impact of 13 interventions that could prevent stillbirths and save the lives of mothers and babies in South Africa. BMC pregnancy and childbirth 2015;15(1):39. (62) Adam T, Lim SS, Mehta S, Bhutta ZA, Fogstad H, Mathai M, et al. Cost effectiveness analysis of strategies for maternal and neonatal health in developing countries. BMJ 2005;331(7525):1107. (63) Saftlas AF, Olson DR, Franks AL, Atrash HK, Pokras R. Epidemiology of preeclampsia and eclampsia in the United States, 1979-1986. Obstet Gynecol 1990;163(2):460-465. (64) Gnen O, Cicek N, Grkemli H, Celik C, Acar A, Akyrek C. The effect of methyldopa treatment on uterine, umblical and fetal middle cerebral artery blood flows in preeclamptic patients. Arch Gynecol Obstet 2002;266(3):141-144. (65) Jowett M. Safe motherhood interventions in low-income countries: an economic justification and evidence of cost effectiveness. Health Policy 2000;53(3):201-228. (66) Bhutta ZA, Ali S, Cousens S, Ali TM, Haider BA, Rizvi A, et al. Interventions to address maternal, newborn, and child survival: what difference can integrated primary health care strategies make? The Lancet 2008;372(9642):972-989. (67) von Dadelszen P, Magee LA. Pre-eclampsia: an update. Curr Hypertens Rep 2014;16(8):454. (68) Ramsey S, Willke R, Briggs A, Brown R, Buxton M, Chawla A, et al. Good research practices for cost-effectiveness analysis alongside clinical trials: the ISPOR RCT-CEA Task Force report. Value in health 2005;8(5):521-533. (69) ISPOR. nternational Society of Pharmacoeconomics and Outcome Research (ISPR. Available at: http://www.ispor.org/. Accessed October 29, 2017. (70) Freeman T. ‘Best practice’in focus group research: making sense of different views. J Adv Nurs 2006;56(5):491-497. (71) Osubor KM, Fatusi AO, Chiwuzie JC. Maternal health-seeking behavior and associated factors in a rural Nigerian community. Matern Child Health J 2006;10(2):159. (72) Evans DB, Edejer TT, Adam T, Lim SS. Methods to assess the costs and health effects of interventions for improving health in developing countries. BMJ 2005;331(7525):1137-1140. (73) Rushby JF, Hanson K. Calculating and presenting disability adjusted life years (DALYs) in cost-effectiveness analysis. Health Policy Plan 2001;16(3):326-331. 116 (74) Lawn JE, Blencowe H, Kinney MV, Bianchi F, Graham WJ. Evidence to inform the future for maternal and newborn health. Best practice & research Clinical obstetrics & gynaecology 2016;36:169-183. (75) Sachs JD. From millennium development goals to sustainable development goals. The Lancet 2012;379(9832):2206-2211. (76) Lee SH, Nurmatov UB, Nwaru BI, Mukherjee M, Grant L, Pagliari C. Effectiveness of mHealth interventions for maternal, newborn and child health in low–and middle–income countries: Systematic review and meta–analysis. Journal of global health 2016;6(1). (77) Hanney SR, Gonzalez-Block MA, Buxton MJ, Kogan M. The utilisation of health research in policy-making: concepts, examples and methods of assessment. Health research policy and systems 2003;1(1):2. (78) Clinical Trials. Health technology assessments. Available at: https://www.clinicaltrials.gov/ct2/results/map?term=health+technology+assessment. Accessed October 29, 2017. (79) World Bank. Country income groups. Available at: http://chartsbin.com/view/2438. Accessed October 29, 2017. (80) World Health Organization. Health system in Pakistan. (81) Hafeez A, Mohamud BK, Shiekh MR, Shah SAI, Jooma R. Lady health workers programme in Pakistan: challenges, achievements and the way forward. JPMA: Journal of the Pakistan Medical Association 2011;61(3):210. (82) National Institute of Population Studies (NIPS) and ICF International. Pakistan demographic and health survey 2012-13. 2014; Available at: https://dhsprogram.com/pubs/pdf/fr290/fr290.pdf. Accessed October 29, 2017. (83) Merriam SB. Types of qualitative research: phenomenology. Qualitative research. A guide to design and implementation: John Wiley & Sons; 2014. p. 25-29. (84) Sayed B. District Matiari. 2015; Available at: http://matiari.net/district-matiari-2/. Accessed October 29, 2017. (85) Population Welfare Department Government of Sindh. District Hyderabad. 2015; Available at: http://www.pwdsindh.gov.pk/districts/hyderabad.htm. Accessed October 29, 2017. (86) Sarowar MG, Medin E, Gazi R, Koehlmoos TP, Rehnberg C, Saifi R, et al. Calculation of costs of pregnancy-and puerperium-related care: experience from a hospital in a low-income country. J Health Popul Nutr 2010;28(3):264. 117 (87) Kruk ME, Galea S, Prescott M, Freedman LP. Health care financing and utilization of maternal health services in developing countries. Health Policy Plan 2007;22(5):303-310. (88) Fereday J, Muir-Cochrane E. Demonstrating rigor using thematic analysis: A hybrid approach of inductive and deductive coding and theme development. International journal of qualitative methods 2006;5(1):80-92. (89) Gerber T, Olazabal V, Brown K, Pablos-Mendez A. An agenda for action on global e-health. Health Aff 2010;29(2):233-236. (90) Leive A, Xu K. Coping with out-of-pocket health payments: empirical evidence from 15 African countries. Bull World Health Organ 2008;86(11):856C. (91) Chang LW, Kagaayi J, Nakigozi G, Serwada D, Quinn TC, Gray RH, et al. Cost analyses of peer health worker and mHealth support interventions for improving AIDS care in Rakai, Uganda. AIDS Care 2013;25(5):652-656. (92) Chen Z, Fang L, Chen L, Dai H. Comparison of an SMS text messaging and phone reminder to improve attendance at a health promotion center: a randomized controlled trial. Journal of Zhejiang University-Science B 2008;9(1):34-38. (93) Gowani S, Yousafzai AK, Armstrong R, Bhutta ZA. Cost effectiveness of responsive stimulation and nutrition interventions on early child development outcomes in Pakistan. Ann N Y Acad Sci 2014;1308(1):149-161. (94) Khowaja AR, Mitton C, Bryan S, Magee LA, Bhutta ZA, von Dadelszen P. Economic evaluation of Community Level Interventions for Pre-eclampsia (CLIP) in South Asian and African countries: a study protocol. Implement Sci 2015 May 26;10:5. (95) Hoffmann C, Stoykova BA, Nixon J, Glanville JM, Misso K, Drummond MF. Do health-care decision makers find economic evaluations useful? The findings of focus group research in UK health authorities. Value in Health 2002;5(2):71-78. (96) Simkhada B, Teijlingen ERv, Porter M, Simkhada P. Factors affecting the utilization of antenatal care in developing countries: systematic review of the literature. J Adv Nurs 2008;61(3):244-260. (97) World Bank. Health expenditure in Pakistan. 2014; Available at: http://www.tradingeconomics.com/pakistan/health-expenditure-private-percent-of-gdp-wb-data.html. Accessed October 29, 2017. (98) Anderson I, Axelson H, Tan BK. The other crisis: the economics and financing of maternal, newborn and child health in Asia. Health Policy Plan 2010;26(4):288-297. 118 (99) Khowaja AR, Qureshi RN, Sawchuck D, Oladapo OT, Adetoro OO, Orenuga EA, et al. The feasibility of community level interventions for pre-eclampsia in South Asia and Sub-Saharan Africa: a mixed-methods design. Reproductive health 2016;13(1):56. (100) O'Donnell O, Van Doorslaer E, Rannan-Eliya RP, Somanathan A, Adhikari SR, Harbianto D, et al. The incidence of public spending on healthcare: comparative evidence from Asia. The World Bank Economic Review 2007;21(1):93-123. (101) Qureshi RN, Sheikh S, Khowaja AR, Hoodbhoy Z, Zaidi S, Sawchuck D, et al. Health care seeking behaviours in pregnancy in rural Sindh, Pakistan: a qualitative study. Reproductive health 2016;13(1):34. (102) Abrahams N, Jewkes R, Mvo Z. Health care–seeking practices of pregnant women and the role of the midwife in Cape Town, South Africa. Journal of Midwifery & Women’s Health 2001;46(4):240-247. (103) Irfan SM, Ijaz A. Comparison of service quality between private and public hospitals: Empirical evidences from Pakistan. Journal of Quality and Technology Management 2011;7(1):1-22. (104) Malik MA, Gul W, Iqbal SP, Abrejo F. Cost of primary health care in Pakistan. Journal of Ayub Medical College Abbottabad 2015;27(1):88-92. (105) Shaikh BT, Hatcher J. Health seeking behaviour and health service utilization in Pakistan: challenging the policy makers. Journal of public health 2004;27(1):49-54. (106) Siddiqi S, Kielmann AA, Khan MS, Ali N, Ghaffar A, Sheikh U, et al. The effectiveness of patient referral in Pakistan. Health Policy Plan 2001;16(2):193-198. (107) Barnum H, Kutzin J, Johansson S, Nygren O, Kincaid DL, Rimon JG, et al. Public hospitals in developing countries: resource use cost financing. POPULATION AND DEVELOPMENT REVIEW 1991;17(1):35-51. (108) Jnsson B. No title. Ten arguments for a societal perspective in the economic evaluation of medical innovations 2009. (109) Electronic medical records: A review comparing the challenges in developed and developing countries. Hawaii International Conference on System Sciences, Proceedings of the 41st Annual: IEEE; 2008. (110) Bhutta ZA, Soofi S, Cousens S, Mohammad S, Memon ZA, Ali I, et al. Improvement of perinatal and newborn care in rural Pakistan through community-based strategies: a cluster-randomised effectiveness trial. The Lancet 2011;377(9763):403-412. 119 (111) Bari A, Sadruddin S, Khan A, Khan IuH, Khan A, Lehri IA, et al. Community case management of severe pneumonia with oral amoxicillin in children aged 2-59 months in Haripur district, Pakistan: a cluster randomised trial. Lancet 2011 Nov 19,;378(9805):1796-1803. (112) Kilian R, Matschinger H, Loffler W, Roick C, Angermeyer MC. A comparison of methods to handle skew distributed cost variables in the analysis of the resource consumption in schizophrenia treatment. Journal of Mental Health Policy and Economics 2002;5(1):21-32. (113) Hay SI. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. 2017. (114) World Health Organization. Global Health Observatory: Life tables by country - Pakistan. 2017; Available at: http://apps.who.int/gho/data/?theme=main&vid=61230. Accessed Oct 31, 2017. (115) Trading Economics. Pakistan GDP per capita | 1960-2017 | Data | Chart | Calendar | Forecast. 2017; Available at: tradingeconomics.com/pakistan/gdp-per-capita. Accessed Oct 31, 2017. (116) Pitt C, Tawiah T, Soremekun S, ten Asbroek AH, Manu A, Tawiah-Agyemang C, et al. Cost and cost-effectiveness of newborn home visits: findings from the Newhints cluster-randomised controlled trial in rural Ghana. The Lancet Global Health 2016;4(1):e56. (117) Fottrell E, Azad K, Kuddus A, Younes L, Shaha S, Nahar T, et al. The effect of increased coverage of participatory women’s groups on neonatal mortality in Bangladesh: A cluster randomized trial. JAMA pediatrics 2013;167(9):816-825. (118) Borghi J, Thapa B, Osrin D, Jan S, Morrison J, Tamang S, et al. Economic assessment of a women's group intervention to improve birth outcomes in rural Nepal. The Lancet 2005;366(9500):1882-1884. (119) Bang AT, Bang RA, Reddy HM. Home-based neonatal care: summary and applications of the field trial in rural Gadchiroli, India (1993 to 2003). Journal of Perinatology 2005;25:S122. (120) Gill CJ, Guerina NG, Mulenga C, Knapp AB, Mazala G, Hamer DH. Training Zambian traditional birth attendants to reduce neonatal mortality in the Lufwanyama Neonatal Survival Project (LUNESP). International Journal of Gynecology & Obstetrics 2012;118(1):77-82. (121) Lewycka S, Mwansambo C, Rosato M, Kazembe P, Phiri T, Mganga A, et al. Effect of women's groups and volunteer peer counselling on rates of mortality, morbidity, and health behaviours in mothers and children in rural Malawi (MaiMwana): a factorial, cluster-randomised controlled trial. The Lancet 2013;381(9879):1721-1735. 120 (122) Medical Research Council. Developing and evaluating complex interventions: new guidance. Available at: https://www.mrc.ac.uk/documents/pdf/complex-interventions-guidance/. Accessed October 29, 2017. (123) Hawe P, Shiell A, Riley T. Complex interventions: how “out of control” can a randomised controlled trial be? BMJ 2004;328(7455):1561-1563. (124) Sharma S, Adetoro OO, Vidler M, Drebit S, Payne BA, Akeju DO, et al. A process evaluation plan for assessing a complex community-based maternal health intervention in Ogun State, Nigeria. BMC health services research 2017;17(1):238. (125) Logie DE, Rowson M, Ndagije F. Innovations in Rwanda's health system: looking to the future. The Lancet 2008;372(9634):256-261. (126) Ahmed S, Khan MM. A maternal health voucher scheme: what have we learned from the demand-side financing scheme in Bangladesh? Health Policy Plan 2010;26(1):25-32. (127) Wang CC, Pies CA. Family, maternal, and child health through photovoice. Matern Child Health J 2004;8(2):95-102. (128) Schafer JL. Statistical Methods in Medical Research. 8th ed.; 1999. (129) Bouhlila DS, Sellaouti F. Multiple imputation using chained equations for missing data in TIMSS: a case study. Large-scale Assessments in Education 2013;1(1):4. 121 Appendices Table 21. Sensitivity analysis for health resource utilization (A) Summary of routine antenatal care costs Group Excluding ANC visits in pilot phase Assuming 50% of ANC visits to MD in pilot phase Assuming 80% of ANC visits to MD in pilot phase Intervention Min. 0.0 0.0 0.0 Mean 9.3 10.3 10.7 Max. 99 99 99 Control Min. 0.0 0.0 0.0 Mean 8.2 9.0 9.4 Max. 87 87 87 Mean diff. 1.1 1.3 1.3 MD = Medical doctor (B) Summary of in-patient hospitalization costs Group Excluding hospitalization in pilot phase Assuming 50% of hospitalization to GW in pilot phase Assuming 80% of hospitalization to GW in pilot phase Intervention Min. 0.0 0.0 0.0 Mean 118 180 156 Max. 385 641 641 Control Min. 0.0 0.0 0.0 Mean 54 87 74 Max. 339 539 539 Mean diff. 64 93 82 GW = General ward 122 Table 22. Multi-collinearity diagnostic outputs Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics B Std. Error Beta Tolerance VIF 1 (Constant) 64.411 13.948 4.618 .000 maternal_mortality 112.245 44.792 .042 2.506 .012 .985 1.015 matmorbid 20.898 4.222 .085 4.949 .000 .943 1.061 pr_di_age -.665 .402 -.028 -1.656 .098 .952 1.051 PW_edu_cat 46.094 4.787 .170 9.628 .000 .882 1.133 parity_cat 53.055 4.201 .215 12.628 .000 .955 1.047 husb_edu_cat 19.232 3.737 .091 5.146 .000 .890 1.124 Arm 6.460 3.676 .030 1.757 .079 .975 1.025 per_obtained_RANC 33.590 20.991 .143 1.600 .110 .034 28.986 per_obtained_emerg_PC -19.209 20.840 -.082 -.922 .357 .035 28.886 funds_saved_OE 9.321 4.009 .040 2.325 .020 .940 1.063 Hasshebeentoldthatshehasseizures .115 .234 .009 .490 .624 .772 1.296 Hasshebeentoldthatshehasdiabetes -.041 .163 -.005 -.251 .802 .790 1.266 HassheeverbeentoldshehasTB -.222 .241 -.016 -.919 .358 .873 1.146 Doessheneedcarefromfamilyorfriendswhowouldotherwiseworkforwagesa 3.269 4.876 .012 .671 .503 .898 1.113 a. Dependent Variable: G_total_Mat_NB_costs 123 Table 23. Correlation co-efficient matrix 124 Table 24. Multi-collinearity diagnostic outputs Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics B Std. Error Beta Tolerance VIF 1 (Constant) 64.386 13.948 4.616 .000 maternal_mortality 112.234 44.791 .042 2.506 .012 .985 1.015 matmorbid 20.962 4.222 .085 4.965 .000 .943 1.061 pr_di_age -.664 .402 -.028 -1.654 .098 .952 1.051 PW_edu_cat 46.150 4.787 .170 9.641 .000 .882 1.133 parity_cat 53.137 4.200 .215 12.651 .000 .956 1.046 husb_edu_cat 19.214 3.737 .091 5.142 .000 .890 1.124 Arm 6.376 3.675 .029 1.735 .083 .976 1.025 per_obtained_RANC 14.624 4.150 .062 3.524 .000 .883 1.133 funds_saved_OE 9.365 4.008 .040 2.336 .020 .941 1.063 Hasshebeentoldthatshehasseizures .113 .234 .009 .481 .630 .772 1.296 Hasshebeentoldthatshehasdiabetes -.037 .162 -.004 -.225 .822 .791 1.265 Doesshehaveahistoryofworms -.654 .331 -.037 -1.979 .048 .770 1.298 Doessheneedcarefromfamilyorfriendswhowouldotherwiseworkforwagesa 3.213 4.875 .012 .659 .510 .899 1.113 a. Dependent Variable: G_total_Mat_NB_costs 125 Table 25. Focus group guide: Pregnant women identified with HDP Topic I. Costs drivers and health resource utilization as result of the CLIP package of care 1. What were you preferences for health facility and why, after you received referral advice (non-urgent) from Community Health Worker? 2. What were you preferences for health facility and why, after you received referral advice (urgent) from Community Health Worker? 3. What cost(s) you were to pay, after you received referral advice (urgent or non-urgent) from Community Health Worker? 4. What cost(s) you were to pay, after you received medication (MgSO4 or Methyldopa) from Community Health Worker? 5. Explain with examples of any additional out-of-pocket expense(s) i.e., costs that you did not anticipate as result of accepting referral or medication advice from Community Health Worker? Topic II: Perceived cost benefits of the CLIP package of care and task-sharing to community health care workers 1. What is your opinion about the benefits of equipping community health workers with the CLIP Package of care (blood pressure monitoring device, urine dipstick, cell phone PIERS on the move application, and pulse-oximeter)? a. Her knowledge or capacity development b. Strengthening her rapport with the community c. Bridging program or service delivery gap at the community 2. How do you think, the PIERS on the move recommended (referral or medication) advice from community health worker would have benefited other pregnant women or family in the intervention areas? a. Improving health condition of woman b. Decreasing financial load on the family as result of pregnancy complications 3. Reflecting on your experiences, how do you think your decision would have affected you or family? a. Impact of your decision of accepting or refusing referral advice? b. Impact of your decision of accepting or refusing medication advice? 4. Explain your general feelings or thoughts (having participated) for the CLIP package of care in your area. 126 Table 26. Focus group guide: Husband of pregnant women identified with HDP Topic I: Costs drivers and health resource utilization as result of the CLIP package of care 1. What were you preferences for health facility and why, after your wife received referral advice (non-urgent) from Community Health Worker? 2. What were you preferences for health facility and why, after your wife received referral advice (urgent) from Community Health Worker? 3. What cost(s) you were to pay, after your wife received referral advice (urgent or non-urgent) from Community Health Worker? 4. What cost(s) you were to pay, after your wife received medication (MgSO4 or Methyldopa) from Community Health Worker? 5. Explain with examples of any additional out-of-pocket expense(s) i.e., costs that you did not anticipate as result of your wife accepting referral or medication advice from Community Health Worker? Topic II: Perceived cost benefits of the CLIP package of care and task-shifting to community health care workers 1. What is your opinion about the benefits of equipping community health workers with the CLIP Package of care (blood pressure monitoring device, urine dipstick, cell phone PIERS on the move application, and pulse-oximeter)? a. Her knowledge or capacity development b. Strengthening her rapport with the community c. Bridging program or service delivery gap at the community 2. How do you think, the PIERS on the move guided recommendation (referral or medication) from community health worker would have benefited other pregnant women or family in the intervention areas? a. Improving health condition of woman b. Decreasing financial load on the family as result of pregnancy complications 3. Reflecting on your experiences, how do you think your decision would have affected your wife / daughter in-law or family? a. Impact of your decision of accepting or refusing referral advice? b. Impact of your decision of accepting or refusing medication advice? 4. Explain your general feelings or thoughts (having your wife /daughter-in-law participated) for the CLIP Trial in your area. 127 Table 27. Focus group guide: LHWs/LHS in the intervention clusters Topic I: Costs drivers and health resource utilization as result of the CLIP package of care 1. How would you describe your experience of home-visit for CLIP package of care assessments? a. Patient assessment (blood pressure monitoring, urine dipstick testing, oxygen saturation measurement, and putting data on cell phone application) b. Explaining the recommendations (transport or medications) c. Administering the medication (methyldopa or MgSO4) 2. How much additional time, you think it may have taken to complete CLIP package of care assessments for each pregnant woman, other than routine visit activities? 3. What were the expectations of pregnant woman or her family from you for emergency or on-request visits for any of the CLIP package of care assessments, other than routine visits? 4. What were the numbers of extra home-visit did you make for CLIP package of care assessments, other than routine visits? 5. Explain with examples of any additional out-of-pocket expense(s) for you i.e., costs that you did not anticipate as result of your visit for CLIP package of care assessments, referral, and medication advice to pregnant women? Topic II: Perceived cost benefits of the CLIP package of care and task-shifting to community health care workers 1. What is your opinion about the benefits of equipping community health workers with the CLIP Package of care (blood pressure monitoring device, urine dipstick, cell phone PIERS on the move application, and pulse-oximeter)? a. Knowledge or capacity development b. Strengthening rapport with the community c. Bridging program or service delivery gap at the community 2. How do you think, the PIERS on the move recommended (referral or medication) advice from you would have benefited pregnant women or family in the intervention areas? a. Improving health condition of woman b. Decreasing financial load on the family as result of pregnancy complications 3. Explain your general feelings or thoughts (having participated) for the CLIP package of care in your area. 128 Table 28. Focus group guide: District level health decision/policy makers Topic I: Costs drivers as result of the CLIP package of care 1. What are the program and/or health system costs that you would ascribe to LHW’s home-visit for CLIP package of care? d. Patient assessment (blood pressure monitoring, urine dipstick testing, oxygen saturation measurement, and putting data on cell phone application) e. Explaining the recommendations (transport or medications) f. Administering the medication (methyldopa or MgSO4) 2. What additional resources (that may have implications on program costs), you think it may have taken for LHW s to complete CLIP package of care visits for each pregnant woman, other than routine visit activities? 3. Explain with examples of any additional program expense(s) i.e., costs that you did not anticipate as result of your visit for CLIP package of care assessments, referral, and medication advice to pregnant women? Topic II: Perceived cost benefits of the CLIP package of care and task-shifting to community health care workers 1. What is your opinion about the benefits of equipping community health workers with the CLIP Package of care (blood pressure monitoring device, urine dipstick, cell phone PIERS on the move application, and pulse-oximeter)? a. Her knowledge or capacity development b. Strengthening her rapport with the community c. Bridging program or service delivery gap at the community 2. How do you think, the PIERS on the move guided recommendation (referral or medication) from community health worker would have benefited pregnant women or family in the intervention areas? a. Improving health condition of woman b. Decreasing financial load on the family as result of pregnancy complications 3. Explain your feelings or thoughts for the CLIP package of care, in general. 129 Table 29. Unit costing proforma National LHWs Program S# Section 1: Manpower support per cluster during definitive phase of the CLIP Trial Cluster number  1 2 3 4 5 6 7 8 9 10 11 12 1 # of health workers (LHWs) serving in the cluster 2 # of supervisor(s) to health workers (LHWs) serving in the cluster 3 Average # of households in each health worker’s catchments 4 Average # of pregnant women in health worker’s catchments (at any given time) 5 Total # of women enrolled for the CLIP Trial 6 Total # of women enrolled in POM 7 Total # of community engagement sessions conducted in intervention cluster 8 Total # of community engagement sessions attended by health workers (LHWs) in intervention cluster 9 Average time to travel to conduct POM visit 10 Average time to conduct one POM visit per woman: first visit 11 Average time to conduct one POM visit per woman: subsequent visits 12 Total # of CLIP training sessions conducted for health workers (LHWs) 13 Total # of CLIP training sessions conducted for health workers’ (LHWs) supervisors 130 Section 2: Program costing Category Specification Cost in INR. [If not applicable, put NA] Working notes i. Incentive of a health worker (LHWs) per month Health workers (LHWs) ii. Stipend/incentive of a health worker (LHWs) for CLIP work (per PW) iii. Average transport cost for a POM visit (per PW) (avg. of inside and outside cluster) iv. Average transport cost accompanying pregnant woman for routine ANC care to PHC (one-side trip per person) (avg. of pub and private vehicle) v. Average transport cost accompanying pregnant woman for routine ANC care to higher health facility (one side trip per person) (avg. of pub and private vehicle) vi. Average transport cost accompanying pregnant woman for emergency referral within 24 hours by CLIP triggers (one side trip per person) (avg. of pub and private vehicle) vii. Average transport cost accompanying pregnant woman for emergency referral within 4 hours by CLIP triggers (one side trip per person) (avg. of pub and private vehicle) Supervision & mentoring i. Salary of a government appointed supervisor to health workers (LHWs) (per month) ii. Stipend/incentive of a Supervisor to health workers (LHWs) for CLIP work (per PW) iii. Average monthly cost of a supervisory visit by CLIP staff for monitoring purposes per health worker (LHWs) Training & support for CLIP i. Cost of training manual per health worker (LHWs) ii. Cost of training manual per health worker (LHWs) iii. Cost of stationary / supplies per health worker (LHWs ) iv. Average cost of POM supplies per health worker (LHWs) in any given time Mobile devices MgSO4 Methyldopa Tablets 131 Section 2: Program costing Training & support for CLIP (continued) Urine dipstick Handbag Gloves plastic (for health worker while doing urine dipstick) Gloves surgical (for health worker while giving injection) Syringe10ml Ampoule cutter, Urine cups Disposal box Batteries Per CLIP kit (inclusive of two 10ml MgSO4 ampules, two 10ml syringes, two swabs, and a box with instructions, labels, and two surgical gloves) Per CLIP kit Methyldopa (inclusive of three Methyldopa tablets each 250mg strength, a label and a A8 size envelope, a moisture-free inner envelope) Other items (specify) v. Total cost of POM kit maintenance (inclusive of Table and CLIP injection kit) per health worker (LHWs) vi. Average cost of community engagement visual-aids or posters per health worker (LHWs) to be displayed in community in the overall trial period vii. Average cost of facility posters per health worker (LHWs) to be displayed in referral facility in the overall trial period viii. Average cost of central team visit for training of health workers (LHWs) ix. Average cost of incentive for attending one training session per health worker (LHWs) x. Average cost of incentive for attending one training session per health worker (LHWs) xi. Average cost of refreshment during training session (per session) 132 Table 30. Unit costing proforma health facilities HEALTH FACILITY UTILIZATION Section 1: Health Facility Identification Name of health facility Type of health facility [Mark X] Country/District / Site Date of visit Form completed by: Form checked by: Section 2: Maternal out-patient services Category Specification Cost [If not applicable, put NA; Round off to nearest integer] Routine antenatal care (per visit) viii. First visit ix. Subsequent visit Investigations iv. Pregnancy ultrasound v. Blood test: complete blood count (CBC) vi. Blood test: Culture vii. Blood test: LFT viii. Blood test: Blood urea ix. Blood test: Serum creatinine x. Blood test: Coagulation profile xi. Urine dipstick test for urine albumin and sugar xii. Urine routine test (inclusive of urine albumin, sugar and microscopy) xiii. X-ray: chest xiv. Biophysical profile xv. Non-stress test (NST) Section 3: Maternal in-patient services, other than delivery Category Specification Cost. [If not applicable, put NA; Round off to nearest integer] Overnight admission in this facility i. General ward (per night) ii. Intensive care unit (per night) Blood transfusion (per unit) i. pRBC ii. Platelets: RDP iii. Platelets: SDP iv. Fresh Frozen Plasma v. Whole blood i. Mechanical ventilation (per day) Private clinic Private secondary hospital Private tertiary hospital 133 Section 3: Maternal in-patient services, other than delivery Maternal interventions ii. Dialysis (per day) iii. Brace Sutures (per procedure) iv. Internal iliac artery ligation (per procedure) v. Hysterectomy (per procedure) Section 4: Delivery related admissions Category Specification Cost [If not applicable, put NA; Round off to nearest integer] Cost of delivery in this facility i. Spontaneous delivery ii. Assisted vaginal delivery iii. Caesarean delivery Section 5: Newborn out-patient services Category Specification Cost [If not applicable, put NA; Round off to nearest integer] Newborn visits at this facility (per visit) i. First visit ii. Subsequent visit Investigations i. Fetal imaging: ultrasound ii. Fetal imaging: x-ray chest iii. Echocardiography iv. Blood test: Serum bilirubin v. Blood test: Blood group vi. Blood test: Arterial blood gases Section 6: Newborn in-patient services, other than delivery Category Specification Cost [If not applicable, put NA; Round off to nearest integer] Overnight admission in this facility (per night stay) i. Neonatal special / intensive care unit Newborn interventions i. Mechanical ventilation (per day) ii. CPAP (per day) iii. Phototherapy: single surface (per day) iv. Phototherapy: double surface (per day) 134 Table 31. Disability weights: maternal morbidities in the CLIP Trial CLIP Trial outcome Description of condition Disability weight (Range) Assumption (citation) Eclampsia Occurrence of generalised convulsions during pregnancy, labour or within 42 days of delivery in the absence of epilepsy or another condition predisposing to convulsions 0.00 This outcome is acute, and women either die or survive without significant sequelae. (Global Burden of Disease (GBD)-2016/17; Table 5, page# 809) Stroke Hemiparesis and/or blindness developed during pregnancy or in the 42 days postpartum lasting greater than 48 hours 0.316 (0.206 – 0.437) Acute ischemic stroke. (GBD-2016/17; Table 5, page# 820) Coma Prolonged unconsciousness ≥12 hours Not reported GBD does not list this condition. Antepartum haemorrhage Vaginal bleeding ≥ 15 mL with or without pain before the onset of labour 0.114 (0.078 – 0.159) Maternal haemorrhage (< 1 L blood lost). (GBD-2016/17; Table 5, page# 809) Disseminated intravascular coagulation (DIC) Abnormal bleeding from mucosa (mouth and/or ears) Not reported GBD does not list this condition. Obstetric sepsis Fever and one of: abdominal/uterine tenderness, foul smelling vaginal discharge/lochia, productive cough and shortness of breath, dysuria or flank pain, headache and neck stiffness. 0.133 (0.088 – 0.19) Puerperal sepsis (GBD-2016/17; Table 5, page# 809) Vesicovaginal or rectovaginal fistula Continuous loss of urine and/or faeces after delivery 0.342 (0.227 – 0.478) Vesicovaginal fistula (GBD-2016/17; Table 5, page# 810) CPR A set of emergency procedures including chest compressions and lung ventilation applied in cardiac arrest victims. 0.00 GBD does not list this condition. We assume that woman undergoing this life saving intervention will either die or survive with no disability. 135 CLIP Trial outcome Description of condition Disability weight (Range) Assumption (citation) Dialysis Haemodialysis and/or peritoneal dialysis 0.571 (0.398 – 0.725) End stage renal disease on dialysis due to hypertension. (GBD-2016/17; Table 5, page# 829) Mechanical ventilation Intubation and ventilation not related to anaesthesia 0.00 GBD does not list this condition. GBD does not list this condition. We assume that woman undergoing this life saving intervention will either die or survive with no disability. Blood transfusion Require ≥1 unit 0.114 (0.078 – 0.159) We are using a conservative estimate for maternal haemorrhage (< 1 L blood lost). (GBD-2016/17; Table 5, page# 809) Interventions for post-partum haemorrhage Brace sutures, external and internal uterine compression, anti-shock garment use, internal iliac artery ligation and/or hysterectomy with or without transfusion. 0.00 GBD does not list this condition. GBD does not list this condition. We assume that woman undergoing this life saving intervention will either die or survive with no disability. 136 Table 32. Disability weights: newborn morbidities in the CLIP Trial CLIP Trial outcome Disability weight (range) Assumption (Source) Feeding difficulty 0.00 This outcome is acute- and primary based on subjective recall. Newborns may survive without any significant sequelae. Breathing difficulty 0.00 Seizures Not reported GBD does not list this condition. Lethargy 0.00 This outcome is acute- and primary based on subjective recall. Newborns may survive without any significant sequelae. Coma Not reported GBD does not list this condition. Fever 0.00 This outcome is acute- and primary based on subjective recall. Newborns may survive without any significant sequelae. Hypothermia 0.00 Umbilical cord infection 0.00 This outcome is acute- and primary based on subjective recall. Newborns may survive without any significant sequelae. Skin infection 0.00 Bleeding Not reported GBD does not list this condition. Jaundice 0.00 This outcome is acute- and primary based on subjective recall. Newborns may survive without any significant sequelae. Vomiting / diarrhea 0.00