Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The evaluation of potential weight-estimation methods in a primarily HIV positive cohort in Botswana… Wozniak, Roberta 2012

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2013_spring_wozniak_roberta.pdf [ 1.69MB ]
Metadata
JSON: 24-1.0073523.json
JSON-LD: 24-1.0073523-ld.json
RDF/XML (Pretty): 24-1.0073523-rdf.xml
RDF/JSON: 24-1.0073523-rdf.json
Turtle: 24-1.0073523-turtle.txt
N-Triples: 24-1.0073523-rdf-ntriples.txt
Original Record: 24-1.0073523-source.json
Full Text
24-1.0073523-fulltext.txt
Citation
24-1.0073523.ris

Full Text

THE EVALUATION OF POTENTIAL WEIGHT-ESTIMATION METHODS IN A PRIMARILY HIV POSITIVE COHORT IN BOTSWANA FOR USE IN RESOURCE LIMITED SETTINGS  by Roberta Wozniak B.Sc., The University of British Columbia, 2010  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in The Faculty of Graduate Studies (Human Nutrition) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2012 © Roberta Wozniak, 2012  i  ABSTRACT Appropriate drug dosing in children should be calculated based on a child’s weight. In resource constrained settings however, inaccurate drug dosing is common due to the absence of working weight scales. Existing proxies for weight, such as those based on age or height, have been shown to be problematic, especially in populations in developing countries. Long bone measurements in children, such as ulna and tibia lengths, have yet to be studied as surrogate measures for weight. The purpose of this study was 1) to examine the association between weight and a series of proxy anthropometric measurements including height, ulna and tibia lengths, mid-upper arm circumference (MUAC), and triceps skinfolds in a primarily HIV positive population of Botswana children (18 months – 12 years); 2) to determine what percentage of the study population has a predicted weight within 10% of their actual weight; and 3) to determine a simple weight-prediction method that would most accurately predict a child’s weight (18 months – 12 years). This study was a cross-sectional survey carried out in a clinical setting at the BotswanaBaylor Children’s Clinical Center of Excellence in Gaborone, Botswana. We measured weight, height, mid-upper arm circumference (MUAC), triceps skinfolds, ulna length, and tibia length in 777 children between the ages of 18 months and 12 years. Univariate linear regression and multiple linear regression analysis were performed using SPSS and coefficients of determination (R2) were calculated. Accuracy of the weight-prediction method was defined as having a predicted weight within 10% of the child’s actual weight.  ii  The MUAC-Tibia and the MUAC-Ulna weight-prediction models had the highest accuracy for predicting a child’s weight with adjusted R2 values of 0.95 and 0.94, respectively. Of the participants, 82% of weights were predicted to within 10% using the MUAC-Tibia method and 79% using the MUAC-Ulna method. Due to the high degree of accuracy, the MUAC-Tibia or MUAC-Ulna weight-prediction methods could potentially be used to estimate a child’s weight. Studies are needed to confirm these findings in other resource poor settings where there is no access to working scales.  iii  PREFACE This thesis is comprised of research conducted by myself, Roberta Wozniak, under the supervision of my graduate supervisor, Dr. Tim Green, as well as my supervisory committee members Dr. Judy McLean and Dr. Charles Larson. The study design was completed by Dr. Charles Larson, Dr. Tim Green, Dr. Judy McLean, and myself. I, along with Dr. Mia Pradinuk, completed the data collection that took place in Gaborone, Botswana. My additional contributions to the research include setting up the study, compiling the survey questionnaire and training materials, and coordinating the data collection. The staff from the Botswana-Baylor Children’s Clinical Center of Excellence was responsible for recruitment of participants and the translation of the consent form and survey into the local language of Setswana. The data analysis was primarily completed by me under the guidance and supervision of Dr. Tim Green. The writing in this manuscript is comprised primarily of my work. There will be sections of this thesis submitted to peer reviewed journals for publication. Ethical approval for this study was provided by the UBC Children’s and Women’s Research Ethics Board (Certificate Number: H11-01052).  iv  TABLE OF CONTENTS  ABSTRACT ................................................................................................................................... ii PREFACE ..................................................................................................................................... iv TABLE OF CONTENTS ............................................................................................................. v LIST OF TABLES ..................................................................................................................... viii LIST OF FIGURES ...................................................................................................................... x LIST OF ACRONYMS ............................................................................................................... xi ACKNOWLEDGMENTS .......................................................................................................... xii Chapter 1: INTRODUCTION ..................................................................................................... 1 1.1. Background .......................................................................................................................... 1 1.2. Purpose of the Study ............................................................................................................ 2 1.3. Research Objectives ............................................................................................................. 2 1.3.1.  Primary Objectives.................................................................................................... 2  1.3.2.  Secondary Objectives................................................................................................ 2  Chapter 2: LITERATURE REVIEW ......................................................................................... 3 2.1. Importance of Measuring Weight and Height in Children .................................................. 3 2.1.1.  Low Birth Weight ..................................................................................................... 3  2.1.2.  Growth and Nutrition Monitoring............................................................................. 4  2.1.3.  Drug Dosing .............................................................................................................. 7  2.2. Drug Dosing Problems in Children...................................................................................... 7 2.2.1.  Adult Dose Extrapolation ......................................................................................... 8  2.2.2.  Dispersible Tablets.................................................................................................... 9  2.2.3.  Weight Banding ...................................................................................................... 10  2.3. When Weight Measurements are Unavailable ................................................................... 11 2.3.1.  Emergencies ............................................................................................................ 11  2.3.2.  Resource Limited Settings ...................................................................................... 11 v  2.4. Proxies for Weight ............................................................................................................. 12 2.4.1.  Existing Methods for Estimating Weight ............................................................... 12  2.4.2.  Other Proxy Measures for Weight .......................................................................... 18  2.5. Summary ............................................................................................................................ 20 Chapter 3: METHODOLOGY .................................................................................................. 22 3.1. Overview of Study Design ................................................................................................. 22 3.2. Ethics.................................................................................................................................. 22 3.3. Setting ................................................................................................................................ 23 3.4. Subjects .............................................................................................................................. 23 3.5. Recruitment ........................................................................................................................ 24 3.6. Data Collection .................................................................................................................. 24 3.6.1.  Procedures ............................................................................................................... 24  3.6.2.  Health and Demographic Data ................................................................................ 25  3.6.3.  Anthropometric Measures ....................................................................................... 25  3.7. Data Analysis ..................................................................................................................... 28 Chapter 4: RESULTS ................................................................................................................. 31 4.1. Recruitment ........................................................................................................................ 31 4.2. Participant Characteristics ................................................................................................. 31 4.3. Inter-rater and Intra-rater Variability ................................................................................. 39 4.4. Correlation between Caliper and Tape Measurements ...................................................... 41 4.5. Univariate Analysis Using Weight and Anthropometric Measurements ........................... 42 4.6. Accuracy of the Univariate Weight-Prediction Models..................................................... 43 4.7. Multivariate Analysis Using Weight, Anthropometric Measures, and Other Potential Confounding Variables ............................................................................................................. 44 4.8. Accuracy of the Multivariate Weight-Prediction Models .................................................. 50 4.9. Validation of MUAC-Ulna and MUAC-Tibia Weight-Prediction Equations ................... 56 4.10. Comparison of Existing Weight-Prediction Methods and the MUAC-Ulna and MUAC-Tibia Weight-Prediction Models ................................................................................. 63 Chapter 5: DISCUSSION........................................................................................................... 65 5.1. Overview ............................................................................................................................ 65 5.2. Participant Characteristics ................................................................................................. 65 5.3. Devised Weight-Prediction Methods ................................................................................. 67 5.4. Comparison of Existing Weight-Prediction Methods ........................................................ 70 vi  5.5. Correlation between Caliper and Tape Measurements ...................................................... 73 5.6. Study Limitations ............................................................................................................... 74 5.7. Generalizability .................................................................................................................. 75 5.8. Implications of Findings .................................................................................................... 76 5.9. Directions for Further Research ......................................................................................... 76 5.10.  Conclusion .................................................................................................................. 77  REFERENCES ............................................................................................................................ 78 APPENDICES ............................................................................................................................. 88 Appendix A. Participant Consent Form .................................................................................... 88 Appendix B. Survey Question .................................................................................................. 90 Appendix C. MUAC-Ulna Weight-Estimation Chart ............................................................... 91 Appendix D. MUAC-Tibia Weight-Estimation Chart .............................................................. 93  vii  LIST OF TABLES  1.  Existing age-based formulas for estimating weight in children………………........ 23  2.  Participant characteristics………………………………………………………….  3.  Age range and number of participants by weight group…………………………... 45  4.  Prevalence of low weight-for-age (underweight) in predominately HIV positive children (18 mo to 10 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana……………………………………………………..  46  Prevalence of low length/height-for-age (stunting) in predominately HIV positive children (18 mo to 12 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana……………………………………………………..  47  5.  42  6.  Prevalence of low BMI-for-age in predominately HIV positive children (18 mo to 10 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana……………………………………………………………....................... 48  7.  Total percent of the study population who are underweight (low weight-for-age), stunted (low height-for-age), and low BMI-for-age……………………………….  48  8.  Inter-rater variability……………………………………………………………….  50  9.  Intra-rater variability……………………………………………………………….  50  10. Univariate analysis of prediction equations for weight estimation………………... 53 11. Percentage agreement between the predicted weight and the actual weight for each univariate weight-prediction model………………………………….............. 54 12. Examination of the effects of cofactors and covariates on the relationship between proxy measures for weight and actual weight………………………........  55  13. Regression equations for predicting weight using weight-estimation models.......... 57 14. Comparison of Adjusted R2 values between the univariate weight-prediction model and the multivariate weight-prediction model………………………….......  58  15. Percentage agreement between the predicted weight and the actual weight for each multivariate weight-prediction model………………………………………..  63  viii  16. Percentage agreement within 10% between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models, separated by weight group……………………………………………....................  66  17. Regression equations for predicting weight using MUAC-Ulna and MUAC-Tibia weight-prediction models on a random sample of approximately 50% of the data (Sample 1, n=386)…………………………………………………………………. 67 18. Percentage agreement between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models………………………..  70  19. Percentage agreement between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models and three existing weight-prediction methods based on age…………………………………………..  74  ix  LIST OF FIGURES  1.  Distribution of participants by age (a), weight (b), and height (c)………………...  43  2.  Correlation between ulna lengths using a caliper vs. a tape……………………….  51  3.  Correlation between tibia lengths using a caliper vs. a tape……………………….  52  4.  Linear regression equations for the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models………………………………………………………….. 59  5.  Modified Bland-Altman plots for predictive accuracy of six weight-prediction models. The dotted lines represent a residual weight of 2.0 kg and -2.0 kg………. 61  6.  Percentage agreement between predicted weight and actual weight for the MUAC-Ulna weight-prediction model…………………………………………….  64  Percentage agreement between predicted weight and actual weight for the MUAC-Tibia weight-prediction model……………………………………………  65  Linear regression equations for the MUAC-Ulna (a) and MUAC-Tibia (b) weight-prediction models using approximately half the data set (Sample 2, n=391)……………………………………………………………………………...  67  Modified Bland-Altman plots for predictive accuracy of the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models using approximately half the data set (Sample 2, n=391). The dotted lines represent a residual weight of 2.0 kg and -2.0 kg……………………………………………………………………………...  69  10. Percentage agreement between predicted weight and actual weight for the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models (Sample 2, n=391)………………………………………………………………………...........  71  7.  8.  9.  x  LIST OF ACRONYMS LBW  low birth weight  WHO  World Health Organization  MDG  Millennium Development Goal  UN  United Nations  SAM  severe acute malnutrition  MUAC  mid-upper arm circumference  BMI  body mass index  IMCI  integrated management of childhood illness  ARV  antiretroviral  EML  essential medicines list  UNICEF  United Nations Children’s Fund  HIV  human immunodeficiency virus  APLS  Advanced Pediatric Life Support  ARC  Australian Resuscitation Council  NCHS  National Center for Health Statistics  UK  United Kingdom  DWEM  devised weight estimation method  COE  Botswana-Baylor Children’s Clinical Center of Excellence  HAART  highly active antiretroviral therapy  FANTA  Food and Nutrition Technical Assistance  USAID  United States Agency for International Development  PMTCT  prevention of mother-to-child transmission  xi  ACKNOWLEDGMENTS First and foremost I would like to extend my sincerest gratitude to my supervisor, Dr. Tim Green, and my committee members Dr. Judy McLean and Dr. Charles Larson. Your expertise and encouragement were much appreciated. Thank you Dr. Green for all the countless hours you put into teaching me SPSS, I could not have done it without you. Thank you to the staff at the Botswana-Baylor Children’s Clinical Center of Excellence in Gaborone, Botswana. Your guidance and support throughout 10 weeks of data collection made it possible for us to reach our goal of measuring 800 children. Dr. Mia Pradinuk, my research partner, I could not have done it without your help and laughter. And to all the beautiful children we measured: your strength and optimism is inspiring. It is hard to believe that my journey over the last 3 years began with an undergraduate course in international nutrition. Dr. Judy McLean introduced me to and ignited my passion for the fascinating world of international nutrition, and for that I will be forever grateful. From there, Judy gave me the opportunity to be involved in a study in Rwanda and assist her on a UN project in Cambodia. Judy’s generosity, compassion, and support are endless and I am so thankful for everything she has done for me. Finally, the support of my parents, Marilyn and Graham, has always been continual and unconditional. Whether it has been pursuing a professional career in dance or helping malnourished children in rural Rwanda, they have always encouraged me to follow my passion, work hard, and persevere.  xii  Chapter 1: INTRODUCTION 1.1.  Background Weight measurements in children are important for determining and tracking nutritional  status as well as for appropriate drug dosing. A drug dose that is dependent on a child’s weight should be calculated using a mg/kg body weight formula to determine the correct amount of medication the child should be administered. In ideal settings, a child is weighed on a calibrated weight scale in order to calculate the appropriate dose. Weight scales, however, are not always available to obtain an accurate body weight measurement for a child, such as in emergency situations or in resource limited settings. Therefore, a proxy measure for weight that can be used to estimate a child’s weight is a necessity when weight scales are impractical or unavailable. Proxy measures based on age or height have been developed and validated mostly on Western populations. Tools based on height like the Broselow tape may be more effective at estimating a child’s weight compared to age-based formulas. However, there is concern of its validity in resource poor settings where malnutrition may be prevalent. Long bone measurements like ulna and tibia length are simple surrogate measures of height and may be more appropriate proxies for weight in populations where malnutrition is prevalent. Ulna and tibia lengths, as well as mid-upper arm circumference, should be examined as potential proxy measures for weight, especially in resource limited settings.  1  1.2.  Purpose of the Study The purpose of this study was to determine if a child’s weight could be accurately  estimated using a proxy anthropometric measure for weight in the absence of a working scale in children 18 months to 12 years in Botswana.  1.3.  Research Objectives  1.3.1. Primary Objectives 1) To examine the association between the weight of children aged 18 months to12 years and a series of proxy anthropometric measurements including height, ulna and tibia lengths, mid-upper arm circumference, and triceps skinfolds in a primarily HIV positive population in Botswana. 2) To determine what percentage of the study population would have a predicted weight within 10% of their actual weight. 3) To develop a simple weight-prediction method that would predict a child’s weight with an acceptable degree of accuracy (18 months – 12 years). 1.3.2. Secondary Objectives 1) To compare the accuracy of the devised weight-prediction method against common existing methods used to estimate weight in children. 2) To determine the accuracy of a tape compared to a validated caliper for measuring ulna and tibia lengths. 3) To describe the nutritional status of the study population in terms of rates of stunting, underweight, and body mass index-for age.  2  Chapter 2: LITERATURE REVIEW 2.1.  Importance of Measuring Weight and Height in Children Measuring a child’s weight and height is something that is often done from birth.  Determining and tracking a child’s weight and height allows health care providers to gain valuable information regarding the health of the child. There are many reasons for measuring a child’s weight or height, such as for monitoring growth, determining nutritional status, and for formulating appropriate drug dosages. 2.1.1. Low Birth Weight A low birth weight (LBW) baby is defined by the World Health Organization (WHO) as an infant at birth who weighs less than 2,500 grams. (1) A LBW baby can have severe, if not fatal, health outcomes and they are at a much greater risk of dying than a baby born at a healthy weight. (1) Globally, 14% of children born each year are LBW, but they account for around 70% of all neonatal deaths. (2) In addition, being born with a low birth weight is closely linked to inhibited growth, reduced cognitive development, and an increased risk of chronic disease later in life. (1) Therefore, weighing a child at birth is important for determining whether a baby has an increased risk of death or complications, and thus appropriate actions can be taken. On a population level, it is important to have data on birth weight for monitoring indicators for goals such as reducing child mortality rates as part of the Millennium Development Goals (MDGs), and goals set in ‘A World Fit for Children’, from the United Nations Declaration and Plan of Action. (1) While birth weight is one key health indicator, tracking height and weight over childhood can provide further information regarding health and nutrition status of the child.  3  2.1.2. Growth and Nutrition Monitoring Determining a child’s height, weight, and age are important measurements for use in growth monitoring and indicators of malnutrition. On a global scale, the use of growth charts can help governmental, non-governmental, and United Nations (UN) agencies track the nutritional status of populations. This is important for formulating interventions and health policies, and being able to monitor the programs and assess their effectiveness. (3) On an individual level, anthropometric measurements can be monitored over the course of childhood to assess whether a child is faltering in their growth. (4) The WHO Child Growth Standards, which were developed in 2006, are commonly used to identify children less than 5 years of age with malnutrition, including undernutrition and obesity. (5) In addition, the WHO established a growth reference for school-aged children and adolescents in 2007 for use in children 5-19 years. (6) Indices for undernutrition including height-for-age, weight-for-height, and weight-for-age are used to identify nutritional conditions such as stunting, wasting, and underweight, respectively. Cut-off values used to classify low height-for-age (stunting), low weight-for-height (wasting), and low weight-for-age (underweight), are defined as a child being two standard deviations below the reference mean based on data from the WHO Child Growth Standards for children under 5 and the WHO growth reference for school-aged children and adolescents. (3,6) Severe undernutrition is defined as a Z-score value less than 3 standard deviations of the reference mean. (3)  4  2.1.2.1.  Low Length/Height-for-Age (Stunting)  Stunting, or a low length/height-for-age, indicates a child with previous or chronic undernutrition. (7) The term length-for-age is used for children under 2 years of age, while height-for-age refers to children over 2 years of age. (3) A low length/height-for-age is due to a slowed growth in the fetus and into early childhood meaning the child does not reach their expected height compared to a child of the same age who is healthy and well-nourished. (7) Stunting is a result of growth failure due to an insufficient diet, frequent or recurring infections, or a combination of both. (7) The lasting effects of stunting are often detrimental to a child’s health and intellectual capacity, and it puts the child at an increased risk of illness or even death. (7) Moreover, the effects of stunting are often irreversible once a child has reached 2 years of age. (7) As stunting is not a measure of current nutrition status, it should not be used to measure short-term changes in a child’s nutritional status. (7) 2.1.2.2.  Low Weight-for-Height (Wasting)  Wasting, or low weight-for-length/height, indicates a child with current or acute undernutrition. (5) A child who is wasted has most likely either had severe weight loss or has experienced a failure to gain weight. (5) The child has a weight much lower than what a healthy child of the same height or length would be expected to have. (7) Unlike children experiencing stunting, wasting can occur quickly in children and is sensitive to changes in seasonal food availability, or shortages, and disease prevalence. (7) Whereas stunting is often irreversible once a child reaches 2 years, wasting is commonly short-term and can be treated. (5)  5  Severe acute malnutrition (SAM) is defined as a child being less than 3 standard deviations of the WHO growth standards for weight-for-height. (5) SAM can also be defined as a mid-upper arm circumference (MUAC) of less than 115 mm. (5) Being able to measure children’s height and weight, and even MUAC in field settings, is important for the identification of SAM in populations. (5) The ability to determine the prevalence of SAM can help inform organizations when and where therapeutic feeding programs and other interventions are needed. (5) Body mass index (BMI) is another index that uses weight and height (BMI= weight, kg/ height, m2) and is often used for adults. (8) BMI in children, however, varies by age so the reference data are specific to age. (9) The weight-for-height index targets children under the age of 5, while the BMI-for-age index can be used for children 5-19 years. 2.1.2.3.  Low Weight-for-Age (Underweight)  Underweight, or low weight-for-age, is an indicator of both past and present undernutrition but is more difficult to interpret. (3) Underweight is a composite measure of stunting and wasting, but it cannot distinguish between the two measures. (7) Underweight is used as the indicator of undernutrition in the first goal of the UN’s MDGs, but has stirred up controversy. (10) As mentioned above, there is more than one indicator using anthropometric data to assess nutrition status in children. (11) There can also be overlap among the 3 indicators. For instance, a child could be underweight and wasted, or a child could be stunted but not underweight, or a child could be all three: stunted, wasted, and underweight. (10) Therefore, only using one indicator may not accurately reflect the true presence of  6  undernutrition in populations, thus all three indicators are important for assessing nutrition status in children. (10) 2.1.3. Drug Dosing It is important that appropriate and correct administration of pediatric medications be given to children in order to treat illness and disease, and prevent death. Drug metabolism is correlated with lean body mass but for many pediatric drugs, correct dosages are calculated based on the child’s weight. (12) The World Health Organization (WHO) and the integrated management of childhood illness (IMCI) guidelines state that drug dosing for pediatric medicines should be administered according to a child’s weight or age. (13,14) Ideally, a child is weighed on an accurate scale and then a milligram per kilogram (mg/kg) body weight dose is administered, usually in the form of a syrup or portion of an adult fixed-dose. (14) Therefore, it is imperative for health workers to have a current and accurate weight for the child who is being prescribed a medication based on weight.  2.2.  Drug Dosing Problems in Children Globally, an estimated 25-75% of antibiotic dosages prescribed in teaching hospitals are  inaccurate. (15) Recent studies conducted in Nigeria and other parts of sub-Saharan Africa found that over 50% of pediatric medications were being under or over dosed. (16) If recorded weights for children are inaccurate or unavailable, there is a risk of potentially under or over dosing medications. Over dosing of many pediatric medications can lead to toxicity. Underdosing a pediatric medication increases may result in the child not being treated for their illness or disease, or it could mean the child develops resistance to that drug. (17) While an inaccurate weight measurement may result in under or over dosing of a child, there are other problems that can arise when attempting to appropriately dose a child. 7  2.2.1. Adult Dose Extrapolation The WHO formulated an essential medicine list (EML) that could be adopted and reformulated to address the particular public health concerns of individual countries (Quick, 2002). The EML has been called a “successful public health initiative” (15); however, the EML concentrates on adult populations with minimal focus on pediatric formulations. (18) Many pediatric medicines are scaled down from adult doses. (19) Extrapolation of adult doses may be safe and acceptable for children with any non-toxic drugs or drugs with wide therapeutic ranges, but drugs with narrower toxicity margins could be harmful to children if improperly dosed. (20) Children are physiologically different from adults and children absorb and metabolize drugs at different rates compared to adults. (21) For example, children under the age of 3 have low levels of acid in the stomach, which can affect the absorption of acid-sensitive drugs like penicillin by increasing absorption. (20) Absorption of rifampicin on the other hand, a drug used to treat tuberculosis, may be reduced in the low-acid environment of the child’s stomach. (20) Therefore, extrapolating adult doses for children can increase the risk of inappropriately dosing a child. (22) Manipulation of adult tablets like cutting or crushing the tablets can have other potentially harmful effects. (23) A protective coating usually ensures the contents of tablets are released in the small intestine (enteric coating) rather than the stomach; therefore, crushing or cutting tablets may increase a child’s risk of stomach ulcers. (23) Cutting or crushing adult tablets also exposes the child to the unfavourable or bitter taste of the tablet’s contents. (23) In response to the problems associated with extrapolation from adult doses, work has been done to ascertain essential medicines and their optimal doses specific for children. The  8  WHO passed a resolution in 2007 called ‘Better Medicines for Children’, which includes addressing the need for improved dosage forms. (20) The resolution resulted in an EML for children, which was first published in 2007 with the latest version released in 2009. (23) Currently, the United Nations Children’s Fund (UNICEF) and the WHO are advocating the use of dispersible tablets as the optimal dosage form for delivering drugs to children. (20,23) 2.2.2. Dispersible Tablets Dispersible tablets are solid tablets that disintegrate and dissolve when mixed with a small amount of water or breast milk. (24) Dispersible tablets are packaged in single doses in a solid state. When a dose is ready to be administered, it is ready within a few minutes after being added to liquid. (23) As a contrast to using adult doses, dispersible tablets have been specifically dosed for children ensuring that they are safe and appropriate for use in children of various ages and weights. (23) Some current drugs available in a dispersible tablet form for children include: rifampicin/isoniazid (tuberculosis), zinc sulfate (diarrhea), artemether/lumefantrine (malaria), lamivudine/stavudine (human immunodeficiency virus, HIV), amoxicillin (pneumonia), and paracetamol (pain and fever). (23) One advantage of dispersible tablets is that they can be administered to infants who otherwise would not be able to swallow tablets. (24) Syrups are another way to administer drugs to very young children who have difficulty swallowing tablets. Syrup formulations, however, are difficult to administer at a correct dose. (24) Caregivers need to measure out a specific amount of syrup that has been calculated based on the child’s weight, but measurement errors are common and can lead to an increased risk of over or under dosing. (24) On the contrary, dispersible tablets are dispensed as a fixed dose and thus require minimal manipulation which reduces the risk of dosage errors. (24,25) 9  Syrups are also difficult to transport and store due to refrigeration requirements. (24) Dispersible tablets are easily transported and are stable in blister packaging. (24) Another problem with syrups is that they are commonly in low concentrations so children may be required to swallow large volumes of syrup resulting in low compliance. (17,26) Dispersible tablets only require approximately 5-10 mL of liquid per dose, so minimal liquid is consumed even at higher doses. (24) In addition, dispersible tablets are relatively inexpensive in comparison to syrups. (24,26) Dispersible tablets are easy to administer to children as they are prepared as fixed doses; however, this could also be seen as a potential disadvantage. One advantage of syrups over dispersible tablets is that a syrup dose can be easily altered depending on the weight of the child because syrup doses are calculated as a mg/kg body weight dose. (27) Dispersible tablets have a dose regime with only set doses (e.g. a 100 mg tablet of paracetamol divided into 4 doses: 0.5 tablet, 1 tablet, 1.5 tablets, 2 tablets), which means children of different weights could be administered the exact same dose. In order to guide a health care professional to which dose a child of a particular weight should receive, weight bands should be calculated for each drug being administered as dispersible tablets. (25) 2.2.3. Weight Banding The fixed-dose, scored tablets can be based upon weight bands in which a child who falls within a certain weight range would receive that fixed dose. (28) The use of weight bands with dispersible tablets eliminates dosage calculations, which can be time-consuming and susceptible to calculation errors. (26)  10  More research is needed to further understand optimal dosing formulations for pediatric populations including safety margins and therapeutic ranges for each pediatric drug. However, one thing remains constant: an accurate weight measurement is needed every time a weightbased drug dose is prescribed to a child.  2.3.  When Weight Measurements are Unavailable  2.3.1. Emergencies Weighing a child using a scale is considered the ‘gold standard’ because it is the most accurate way to obtain the weight of a child. (29) However, calibrated weight scales are not always available or easily accessible in certain situations or settings. In emergency situations, obtaining an accurate weight for the child is one of the crucial first steps the clinician needs in order to progress with treatment. (30) Children undergoing resuscitation need the appropriate equipment size, defibrillation energy, fluid replacement rate, and medication dosage. (30-34) These critical steps in resuscitating a child require an accurate weight measurement. (28,30-34) However, delaying resuscitation in order to take a weight measurement using a scale is impractical, and often an estimated weight is used. (28,30-34) In addition, it may be difficult to move children to be weighed on a scale who are in the intensive care unit with severe trauma or pain. (12,35) Therefore, the need for a fast and reliable weight-estimation method is a necessity in pediatric emergencies and trauma when using a weight scale in unfeasible. (28,30-34) 2.3.2. Resource Limited Settings In many settings with limited resources, like rural health centers in developing countries, functioning or calibrated scales are lacking. (25) The WHO IMCI programs have been introduced in many developing countries to improve national health care systems, but recent  11  evaluations of these programs have exposed problems. (36) The Multi-Country Evaluation of IMCI Effectiveness, Cost and Impact identified problems in maintaining equipment, such as weight scales. (36,37) The cost of a simple electronic floor scale recommended by UNICEF is around US$90 and while some clinics may be able to afford the initial cost of the scale, maintaining the functionality of the scale does not occur. (7,25) Therefore, there is urgency for the use of a proxy measure for weight that can be used in developing countries or other resource scarce settings in the absence of working scales.  2.4.  Proxies for Weight  2.4.1. Existing Methods for Estimating Weight There is evidence suggesting visual assessment of weight by parents or clinicians is not accurate and methods to estimate a child’s weight need to be based on a proxy for weight. (30,38-40) There have been many methods developed for estimating weight, mostly using age or height. (32-34) 2.4.1.1.  Age-Based Proxies for Weight  Aged-based methods for estimating weight have been developed using formulas that allow for health care workers to simply put the child’s age in years into a validated formula to get an estimated weight. (34) An advantage of using an age-based formula is that a health care worker can calculate a drug dose for a child without the child being present; so when the child visits the health clinic, a drug dose has already been prepared and is ready to be administered. (12,34) Another advantage of aged-based formulas is that it does not require any equipment, like a length board or tape measure. (12,34) In developing countries it can be difficult to purchase  12  and maintain expensive equipment used for anthropometric measurements. (34) A number of existing age-based formulas are listed in Table 1. Table 1. Existing age-based formulas for estimating weight in children Source  Year  Location  Age Range, yrs  Calculation  Advanced Pediatric Life Support (APLS) (41)  2005  UK  1 – 10  W = 2 (a + 4)  Luscombe (35)  2007  UK  1 – 10  W = 3a + 7  Argall (29)  2003  UK  1 – 10  W = 3a + 6  Theron (42)  2005  New Zealand  1 – 10  W = e[2.197099 + (0.175571a)]  Shann (42)  2005  New Zealand  1–9  W = 2a + 9  10+  W = 3a  1–6  W = 2a + 8  7 – 12  W = 0.5 (7a – 5)  1–9  W = 2a + 8  10+  W = 3.3a  Nelson (43)  Australian Recuscitation Council (ARC) (44)  2004  2006  USA  Australia  W = weight, kg a = age in years  One of the most widely used age-based formulas for children 1-10 years is the advanced pediatric life support (APLS) formula. The original APLS formula was derived in 1977 using the National Center for Health Statistics (NCHS) data in the United States and is taught on the APLS course in the United Kingdom (UK). (12,45) The formula uses the child’s age in years from their most recent birthday, and is as follows:  13  Weight estimation, kg = 2 x (age in years + 4) The APLS formula has been a part of resuscitation training and guidelines for treatment in many countries globally. (46) For instance, the guidelines for the IMCI produced by the WHO recommend the use of the APLS formula for calculating drug doses and fluid volumes in children. (46,47) However, with the average weight of children on the rise and obesity rates increasing, there is concern in the scientific community that the APLS formula currently underestimates weight in pediatric populations in developed countries. (29,35,42,46,48) Luscombe and Owens (35) conducted a large-scale study in 2005 with 17 244 children from the UK from 1-10 years of age. When the researchers tested the current APLS formula (weight = 2 x (age + 4)), they found that weight was underestimated by an average of 18.8%. (35) The authors worried that this underestimation would result in underdosing of pediatric drugs and fluids, which could result in complications and additional risks to the child’s health. (35) Therefore, Luscombe and Owens derived a new, updated formula to correct for the increasing weights of the ‘modern day child’, which only found to underestimate weight by 2.5%. (35) The Luscombe and Owens formula is as follows: Weight estimation, kg = (3 x age in years) + 7 Similar findings were observed in studies conducted in Australia and the UK in that the APLS formula underestimated weight in children. (48-51) On the contrary, weights are not increasing in children in developing countries like they are in children in developed countries, and a number of researchers have questioned the validity of the APLS formula in developing countries. (12,30,32,34,46,52,52) Several studies have validated the APLS formula, but they have been in Western countries including the study by Luscombe and Owens. (30,46,52) Only a 14  few studies have examined how accurate age-based formulas are at estimating weight in developing countries. Varghese et al (53) found that the APLS formula overestimated weight by a mean of 2-3 kg in a study of 500 Indian children between the ages of 1 month and 12 years. A study conducted in South African children aged 1-10 years in 2010 found that the original APLS formula predicted weight better than the revised Luscombe and Owens formula. (12) Moreover, the authors found that the Luscombe and Owens formula overestimated weight by a mean of 12.4% in the study population. (12) If the APLS formula has been developed in populations of well-nourished children, many researchers have hypothesized it will overestimate weight in populations in developing countries where children are more likely to have a low weight-for-age. (32,46,52,53) In addition to a possible overestimation of weight, another problem associated with age-based weight estimation methods in developing countries is that parents or caregivers may be unsure of the child’s age, which is common in many low income countries. (32) The use of length-based methods for estimating weight has also been investigated as a proxy for weight and may be more accurate than age-based formulas. 2.4.1.2.  Length/Height-Based Proxies for Weight  More and more countries are switching from recommending the use of age-based methods of estimating weight, such as the APLS formula, to length-based methods, like the Broselow tape. (12) Other methods using height include the devised weight estimation method (DWEM) (54), Traub-Johnson (55), Traub-Kichen (56), and Oakley (57). However, the Broselow tape is one of the most common weight estimation methods used for children. (30)  15  The Broselow tape was developed by James Broselow in 1986 using the National Center for Health Statistics (NCHS) data. (58) The NCHS data collected between 1963 and 1975 is from a nationwide sample of over 20 000 children up to age 18 living in the United States. (58) Broselow used this data and calculated the 50th percentile for weight-for-height. (58) The result was the Broselow tape: a tape measure for children between 46 cm and 143 cm with nine colourcoded zones representing ranges of weight estimates. (58) To use the Broselow tape, the child lays horizontal and one end of the tape is placed at the head with the colour-coded weight zones placed at the feet. (31) Each weight range provides drug dosages and the appropriate equipment sizes for children who fall into that particular weight range. (32) Multiple studies have been conducted comparing different methods for estimating weight in children and results have shown that the Broselow tape more accurately predicts children’s weights over age-based methods, such as the APLS formula. (28,30,31) Moreover, researchers have stated that age-based formulas should not be used when another method is available. (30,31) However, as obesity rates in children have increased, some recent studies conducted in developed countries have found that the Broselow tape underestimated weight, much like the APLS formula did in similar populations. (42,51,59-61) Most of these studies have had small sample sizes and have been in specific populations (e.g. First Nations children), (59) so further research is needed to determine if the Broselow tape has a tendency to underestimate weight. (32) While the Broselow tape has been validated in many developed countries, there have been only 4 studies that have been conducted in children living in developing countries. (32) The first study, conducted in India by Varghese et al (53) and published in 2006, included a sample of 500 children between the ages of 1 month and 12 years. Overall, the Broselow tape accurately 16  predicted weight over the entire weight range and was a better predictor than the Argall, Nelson, and APLS formulas. (53) Although the Broselow tape performed the best, there was a slight tendency for it to overestimate weight in children weighing over 15 kg. (53) The second study published in 2008 by Ramarajan et al (30), was also conducted in India. The 548 participants in the study were mostly poor and undernourished children between the ages of 1 month to 12 years. (30) The Broselow tape, on average, overestimated weight in children over 10 kg by at least 10%. (30) The authors added a 10% correction factor in order to adjust for the overestimation and better predict the weights of the children over 10 kg in their sample. (30) The third study, published in 2011 by Geduld and colleagues (12), was conducted in 2 832 children aged 1-10 years in Cape Town, South Africa. Approximately 64% of the children had an estimated weight within 10% of their actual weight when the Broselow tape was used to estimate weight. (12) There was a slight tendency to underestimate weight, but overall the authors concluded that the Broselow tape was the most accurate method for estimating weight in their population. (12) However, the study took place in one hospital in Cape Town where average incomes are most likely higher than other parts of the country; therefore, the sample may not be representative of the entire country of South Africa. (12) The final study to be conducted in a developing country took place in Kenya. House et al (32) published a study in 2012 that compared the Broselow tape to two age-based methods, APLS and Nelson’s formula in 967 children under the age of 14 years. Findings were similar to those from previous studies in which the Broselow tape predicted weight better than the agebased formulas. (32) Approximately 66% of children were classified within the appropriate  17  colour zone for weight. (32) There was a small overestimation of weight using the Broselow tape with 24% of the children being misclassified by one colour zone, and this was mostly observed in children over 18 kg. (32) However, House et al recommended the use of the Broselow tape in the Kenyan population. In summary, the Broselow tape was found to be the most accurate method for estimating weight in children in developing countries. The tape is easy to use and requires minimal training and education. (32) However, a potential obstacle of its acceptability and use is the cost of the tape. (30,53,32) Also, some populations showed a tendency for the Broselow tape to overestimate weight in pediatric populations. The Broselow tape was derived from weight-toheight correlations in well nourished, Western populations; therefore, its use may not be applicable to undernourished populations. If children are undernourished and have a low weightfor-height, the Broselow tape may overestimate weight and therefore adjustments or alternate methods should be developed and validated for these populations. 2.4.2. Other Proxy Measures for Weight 2.4.2.1.  Long Bone Measurements  Other proxy measures for weight may be needed as an alternative to the Broselow tape or age-based formulas; one that is not expensive or derived from Western populations. In populations where stunting (low height-for-age) is prevalent, height-based methods for estimating weight is most likely going to be a better predictor of weight compared to age-based formulas. When taking height measurements, it is recommended to use a calibrated stadiometer or recumbent length board in order to obtain accurate measurements. In developing countries however, access to these pieces of equipment is limited. If height is difficult to measure in  18  developing countries but is a strong predictor of weight, a possible proxy for height could be used. Proxy measures for height have been examined in pediatric populations where a height measurement is difficult or impossible to measure like in children with cerebral palsy, spinal deformity, or children with an amputated lower limb. (62) Arm span has commonly been used as an alternative for height, but measuring a child’s arm span can be difficult as it depends on exact positioning. (62) Long bone measurements have been examined as an alternative to arm span for predicting height in children with cerebral palsy and children without disability. (62-64) Distal limb measurements including ulna and tibia lengths are easily accessible, and have landmarks that are easy to palpate meaning greater reproducibility. (63) Studies have found that ulna and tibia lengths are highly correlated with height and could be used as a proxy measure when height measurements are unavailable. (62-64) There have been many studies exploring ulna and tibia lengths as proxies for height, but there have been no studies examining their role in estimating weight. In developing countries, using lone bone measurements as an alternative for height could potentially be more acceptable in weight-estimation methods based on height. 2.4.2.2.  Mid-Upper Arm Circumference  Mid-upper arm circumference (MUAC) is commonly used as a quick screening tool for assessing nutritional status in children. (32,33) A child with a MUAC measurement below the cut-off of 11.5 cm is classified as being severely malnourished (5). Currently, MUAC is not being recommended or used for estimating children’s weight in clinical settings. (33) However, a 19  recent study conducted in Hong Kong by Cattermole et al (33) determined that MUAC correlated the strongest with weight compared to other measurements like age, height, and foot-length. Cattermole et al (33) derived a weight-estimation formula, called the MAC formula, based on a study population of 1 391 children aged 1-11 years: weight, kg = (MUAC, cm – 10) x 3. This formula was found to be a better predictor of weight in older children (≥ 8 years) compared to the Broselow tape and the APLS formula. (33) Overall however, the MAC formula predicted weight within 10% in 44% of the population whereas the Broselow tape estimated weight within 10% of actual weight in 58% of the population. (33) The Cattermole et al study is the first to test a MUAC-based formula. (33) More studies are needed in other populations to explore the potential of using MUAC as a predictor for weight in children. (32,33)  2.5.  Summary Weight measurements in children are imperative for tracking and monitoring a child’s  growth and nutritional status, as well as appropriately dosing medications based on body weight. However, taking a child’s weight on a calibrated weight scale is not always practical or accessible. In particular, access to working equipment, like weight scales, is limited in developing countries; therefore, a simple and accurate proxy measure for weight is vital. Age-based formulas like the APLS and the Luscombe formula have been shown to be ineffective at accurately estimating weight in children and the Broselow tape may be more effective. However, because the Broselow tape has been derived based on Western data, there has been some disagreement amongst researchers on its accuracy in populations in developing countries. Alternative proxy measures for weight should be examined in populations where  20  undernutrition is prevalent. Long bone measurements, like ulna and tibia lengths, as well as MUAC, have yet to be validated in developing countries as potential proxy measures for weight.  21  Chapter 3: METHODOLOGY 3.1.  Overview of Study Design The present study was a cross-sectional survey of children (18 months to 12 years)  conducted in a clinical setting at the Botswana-Baylor Children’s Clinical Center of Excellence (COE), in Gaborone, Botswana. The research project was a non-experimental study and each participant was surveyed once during the two-month data collection time-frame from July to August, 2011. The sampling method was a convenience sample of the Baylor clinic population and their families as the first 800 children, whose caregiver agreed to have them participate in the survey, were selected for our sample. Caregivers who agreed to have their child participate were asked to: a) give health and demographic information of the child participating; and b) allow for a series of 6 anthropometric measurements to be taken of the child participant.  3.2.  Ethics Ethics approval was obtained from the three institutions involved in the study: the  Institutional Review Board from UBC (H11-01052); the Baylor College of Medicine in Texas; and the Ministry of Health’s Health Research Development Committee in Botswana. The anthropometric measurements carried out on the children were minimally invasive and the study was deemed very low risk. Verbal consent was obtained from the caregiver of the child by a trained study nurse in the preferred language of the caregiver. Remuneration equivalent to an average half of a day’s income was given to the caregiver, which went towards travel and meal costs related to voluntarily participating in the study. Remuneration was established according to the Botswana Ministry of Health research guidelines and was fixed at 30 Pula for the caregiver  22  and another 30 Pula for the child if they were over 5 years old (60 Pula equals approximately $8.00 CDN). See Appendix A for the consent form.  3.3.  Setting Participants in this study were recruited from the Botswana-Baylor Children’s Clinical  Center of Excellence (COE) in Gaborone, the capital city of Botswana. The COE facility opened in 2003 following the launch of the country’s nation-wide antiretroviral (ARV) treatment program. The COE provides free highly active antiretroviral therapy (HAART) to children with HIV. The COE provides care and treatment for over 4000 children and families infected with HIV. HIV-exposed infants are tested for HIV at 18 months of age and if they test positive, the infant is enrolled in a treatment program at the COE. Regular check-up appointments are scheduled at 3 month intervals, on average. Standard appointments usually consist of weighing the child and recording their height, an account of remaining medication, refilling medications as necessary, and an appointment with a pediatrician. In addition, patients of the COE receive social support, adherence information classes for parents and caregivers, and nutrition and growth monitoring.  3.4.  Subjects The proposed study sample size of 800 was decided upon based on the time period during  which the data collection could occur at the clinic and the average number of patients visiting the clinic per day. Children were eligible to participate in the study if they were between the ages of 18 months and 12 years, inclusive. This age range was targeted because the IMCI guidelines are 23  provided for children up to 12 years, and children are screened for HIV at the Baylor clinic at age 18 months. Although the population of children at the clinic is primarily HIV positive, we did not exclude any children based on their HIV status. Both children with HIV and children without HIV were included in the study and their HIV status was recorded. Children were excluded from the study if their caregiver did not provide verbal consent, or if one or more measurements could not be completed.  3.5.  Recruitment Recruitment for study participants occurred between July 5, 2011 and August 24, 2011.  Participants were recruited by a clinic nurse trained for our study in one of two ways. Participants were either recruited from the waiting room of the clinic or contacted from a patient list provided by the clinic. Children within the appropriate age range who were recruited from the waiting room were either a clinic patient, were accompanying a family member who was a clinic patient, or their caregiver had heard about the study from other clinic patients and they wanted their child to participate in the study. Caregivers contacted from a patient list and asked to participate had children within the targeted age range who were either current or past patients of the clinic, or were family members of current or past clinic patients.  3.6.  Data Collection  3.6.1. Procedures Caregivers of children attending the COE were approached by a study nurse in the waiting room upon arrival and following being checked-in at the reception desk. The potential participants were asked if they had a child with them between 18 months and 12 years and if they would be willing to hear more about the study. If the caregiver and their child agreed, the study 24  nurse would bring them into the study procedures room and describe the study and the study objectives. A consent form was read to the caregiver in either English or Setswana, the local language, and verbal informed consent was given. The study nurse and one of the two study investigators signed the form as witnesses, confirming verbal consent was given. Once the caregiver and their child agreed to participate, children were prepared for anthropometric measurements by removing shoes and outdoor or excess clothing. The caregiver was asked health and demographic survey questions pertaining to the child participant while one of the study investigators performed a series of six anthropometric measurements on the child. Upon completion of the survey, the caregiver and child participant received their remuneration from the study treasurer and provided either a signature or finger print as confirmation of receiving the funds. 3.6.2. Health and Demographic Data A short survey consisting of seven health history and demographic questions were asked of each caregiver whose child was participating (see Appendix B). Information obtained included the child’s birthdate, gender, HIV status, ARV use, caregiver’s age, number of people living in the household, and the main reason for attending the clinic on the day of the study. The questions provided on the survey were obtained from the Demographic Health Survey, and have thus been validated. 3.6.3. Anthropometric Measures The anthropometric measurements included weight, height, ulna length, tibia length, midupper arm circumference (MUAC), and triceps skinfolds. All step-by-step procedures were 25  strictly followed according to the 2003 Anthropometric Indicators Measurement Guide published by Food and Nutrition Technical Assistance (FANTA) (7), which is a program funded by the US Agency for International Development (USAID); thus, all measurement techniques performed in this study have been previously validated. All measurements were performed at least twice by the same investigator to assess intra-observer reliability. In addition, inter-observer reliability was assessed on 30 participants to test the agreement of measuring between the two study investigators on the same child. 3.6.3.1.  Weight  Weight was obtained from a calibrated digital floor scale to the nearest 0.1 kg. No shoes or excess clothing was worn during weighing. If a child could not stand on the scale unassisted due to physical conditions or deformities, or if they were too young or frightened, the child was weighed indirectly. Indirect weighing was done by getting the caregiver to hold the child in his/her arms and recording a combined weight. The child was then passed to another family member or the study investigator and the caregiver was weighed on their own. The difference between the two weights was recorded as the child’s weight. (7) 3.6.3.2.  Length/Height  Length was taken using a recumbent length-board for children under 2, or if the child could not stand upright long enough to be accurately measured. The recumbent length-board could measure children up to 120 cm. Children were laid flat on the center of the board with their head at the top of the board and their feet pressed flat against the sliding base. (7) Height was measured using a standing stadiometre for children over 2 years or those who could stand upright with minimal assistance. Children stood with their head, shoulders, back, and  26  feet against the board and legs straight with the child looking straight ahead. (7) Measurements were recorded to the nearest 0.1 cm. If the two repeat measurements were not within 0.5 cm of each other, a third measurement was taken. 3.6.3.3.  Ulna Length  Ulna length measurements were taken with a long bone caliper as well as a measuring tape. The purpose of using two different measuring tools was to validate the use of the tape against the caliper, which is considered the gold standard for long bone measurements. The child’s left arm was placed in front of the body on a flat surface with approximately a 90 degree bend in the elbow. With the fingers extended and together and palm flat against the surface, the length of the ulna was measured from the proximal end of the ulna to the tip of the styloid process at the wrist. (62) Measurements with each the tape and the caliper were recorded to the nearest 0.1 cm. If the two repeat measurements were not within 0.5 cm of each other, a third measurement was taken. 3.6.3.4.  Tibia Length  Tibia length measurements were also taken with a long bone caliper as well as a measuring tape. The child’s right leg was crossed over the left with the right ankle flexed over the left knee. The tibia length was measured from the proximal aspect of the right tibial plate and the distal end of the tibia at the ankle bone. Measurements with each the tape and the caliper were recorded to the nearest 0.1 cm. If the two repeat measurements were not within 0.5 cm of each other, a third measurement was taken.  27  3.6.3.5.  Mid-Upper Arm Circumference  Mid-upper arm circumference (MUAC) was taken using a validated head circumference measuring tape. The midpoint of the child’s left upper arm was located by measuring the length from the tip of the child’s shoulder to the base of the elbow, found by bending the child’s arm at a 90 degree angle. (7) With the arm relaxed at the child’s side, the tape was wrapped with correct tension around the midpoint, marked at the triceps with a pen. (7) Measurements were recorded to the nearest 0.1 cm. If the two repeat measurements were not within 0.5 cm of each other, a third measurement was taken. 3.6.3.6.  Triceps Skinfolds  Triceps skinfolds were taken by a validated caliper. With the left arm relaxed at the child’s side, a vertical fold of skin was gently pulled away from the muscle right above the mark at the triceps made during the MUAC measurement. The caliper jaws were applied to the marked midpoint at a right angle. Measurements were recorded to the nearest millimetre. If the two repeat measurements were not within 3 mm of each other, a third measurement was taken.  3.7.  Data Analysis Statistical analyses were performed using SPSS, version 20.0.0 for Windows (SPSS Inc.,  Chicago, IL). Data was entered into Microsoft Excel 2010 by both study investigators on site, following each day data collection occurred. Participant responses were number coded prior to analysis and any identified entry or measurement errors were corrected or removed before any statistical analyses were conducted. The multiple values for each of the anthropometric measurements were consolidated and the average for each measurement was used during analysis. Intra-rater variability was  28  determined by assessing the average variability within each of the two study investigators measurements. The data from the first 30 participants was used to determine inter-rater variability between the measurements of the two study investigators, as these 30 participants were measured by each of the two investigators. The following forms of analyses were executed:   Descriptive and summary statistics on the demographics of the study sample.    Descriptive statistics using the World Health Organization’s (WHO) Anthro Plus statistics program (version 1.0.4) to determine rates of underweight (low weight-for-age), stunting (low height-for-age), and BMI-for-age. Participants with a Z-score below 2 standard deviations of the mean were classified as moderately undernourished, and those below 3 standard deviations of the mean were classified as severely undernourished.    Univariate General Linear Models were produced to determine weight-prediction regression equations using each of the 5 anthropometric measurements as independent variables and measured, or actual, weight as the outcome (dependent) variable.    The coefficient of determination, R2, was used to see how much of the total variability of the outcome (weight) could be accounted for by the predictor variable. A high R2 value indicates that a large proportion of the variability in weight can be explained by the predictor variable being investigated.    Multivariate General Linear Models were produced to determine if the addition of potential covariates like age and cofactors such as gender, HIV status, and ARV treatment significantly improved the weight-prediction models. Pairing of anthropometric measures was also included in the models to see if two predictor variables could give a better estimation of actual weight (e.g. MUAC and ulna length). 29    Bland-Altman analysis (65) was conducted to determine the strength of agreement between the actual weight and the predicted weight using each of the weight-prediction models.    Accuracy of the weight-prediction models were defined as having a predicted weight within 10% of the child’s actual weight.    The weight-prediction equations were validated by randomly dividing the study population into 2 samples (Sample 1 and Sample 2), and testing the accuracy of the regression equation developed from Sample 1 on Sample 2.    The accuracy of existing weight-estimation methods were determined using the present study sample and compared against the accuracy of the weight-estimation methods derived in this study.  All correlation analyses and other results were considered significant if the p-value was less than 0.05 (p<0.05).  30  Chapter 4: RESULTS 4.1.  Recruitment A total of 817 children were invited to take part in the study, of which 807 completed the  survey. Nine caregivers declined consent to have their child measured, and one child was not surveyed due to a suspected measles infection. Therefore, the response rate was 807/817 or 98.8%. The pilot testing included the first 30 participants enrolled in the study during the first two days of the surveying period, and this data was used to measure inter-rater reliability between the two study investigators. These 30 participants were not included in the overall analysis. Seven-hundred and seventy-seven participants were included in the analysis.  4.2.  Participant Characteristics Participant characteristics, including gender and age demographics, participant HIV  status and Antiretroviral (ARV) therapy use, relationship of caregiver, and the participant’s reason for attending the clinic are presented in Table 2. The majority of participants were over 5 years (74%), and almost 32% of participants were between 10 and 12 years. About 80% of the study population was HIV positive, with 95% of those children receiving ARV therapy treatment. Almost three quarters of the participants were recruited on a day he/she had a scheduled follow-up appointment with the medical staff at the COE. Other reasons for the participant attending the clinic during the study period included: accompanying a family member (n=62); heard about the study and wanted to participate (n=54); were phoned and invited to participate (n=71); refilling of medications (n=10); and screening for HIV (n=10).  31  Table 2. Participant characteristics Frequency, n  Percent  Male  403  51.9%  Female  374  48.1%  Total  777  100%  <24 months  30  3.9%  24 - <60 months  173  22.3%  60 - <120 months  328  42.2%  120 - <156 months  246  31.7%  Negative  129  16.6%  Positive  625  80.4%  Unknown  23  3.0%  No  31  5.0%  Yes  594  95.0%  Mother  481  61.9%  Father  64  8.2%  Aunt/Uncle  100  12.9%  Grandparent  59  7.6%  Sibling/Cousin  57  7.4%  Foster Parent  12  1.5%  Not Specified  4  0.5%  Check-up/Follow-up  570  73.4%  Other  207  26.6%  Gender  Age  HIV Status  If HIV positive, on Antiretroviral therapy  Relationship of caregiver  Reason for attending clinic  32  The distribution of participants by age, weight, and height are represented in Figure 1. Ages ranged from 18 months to 155 months (mean=93 months), weight ranged from 6.9 kg to 50.3 kg (mean=21.7 kg), and height ranged from 61.4 cm to 162 cm (mean=117.7 cm). The number and age distribution of participants categorized into eight weight groups is shown in Table 3. Figure 1. Distribution of participants by age (a), weight (b), and height (c) (a) Age  Mean=93.1 mo SD=39.3 mo  33  (b) Weight  Mean=21.7 kg SD=7.3 kg  (c) Height  Mean=117.7 cm SD=19.2 cm  34  Table 3. Age range and number of participants by weight group Weight group n (kg)  Number in group  Median age in months  Minimum age in months  Maximum age in months  ≤10.0  28  21.65  17.16  37.22  10.1-15.0  139  40.96  18.15  91.43  15.1-20.0  164  70.83  26.73  137.26  20.1-25.0  193  107.08  52.11  152.02  25.1-30.0  165  126.74  56.88  155.64  30.1-40.0  80  139.96  111.06  155.84  ≥40.1  8  138.94  128.88  154.45  Weight-for-age (Table 4), length/height-for-age (Table 5), and BMI-for-age (Table 6) statistics were analyzed using the WHO Anthro Plus program (version 1.0.4). Data on the children less than 61 months of age are values based on the WHO standards and for children 61 months to 12 years, values based on the 2007 WHO reference data are used. Moderate undernutrition is defined as a Z-score of less than two standard deviations. Severe undernutrition is defined as a Z-score of less than three standard deviations. The number and percent of the study population who have a low weight-for-age (underweight), length/height-for-age (stunted), and BMI-for age are shown in Table 7. The severity of underweight was medium (19%) but was very high for stunting (42%). The severity of BMI-for-age was relatively low at 10%. For children under 5, approximately 11% were moderately stunted, 8% were underweight, and 12% had a low BMI-for-age.  35  Table 4. Prevalence of low weight-for-age (underweight) in predominately HIV positive children (18 mo to 10 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana Number (N)  % < -3 SD  95% CI (%)  % <-2 SD  95% CI (%)  Mean  Standard Deviation (SD)  12-23  31  0  (0, 1.6)  0  (0, 1.6)  6.38  2.67  24-35  48  2.1  (0, 7.2)  6.3  (0, 14.1)  3.12  3.54  36-47  64  1.6  (0, 5.4)  6.3  (0, 13)  2.37  2.61  48-60  64  9.4  (1.5, 17.3)  10.9  (2.5, 19.4)  0.99  2.67  61-71  48  22.9  (10, 35.8)  33.3  (19, 47.7)  -0.76  2.34  72-83  48  31.3  (17.1, 45.4)  35.4  (20.8, 50)  -0.82  2.67  84-95  62  25.8  (14.1, 37.5)  32.3  (19.8, 44.7)  -1.06  2.23  96-107  73  35.6  (23.9, 47.3)  42.5  (30.4, 54.5)  -1.8  2.19  108-119  95  1.1  (0, 3.6)  1.1  (0, 3.6)  -0.11  0.57  Age in months  36  Table 5. Prevalence of low length/height-for-age (stunting) in predominately HIV positive children (18 mo to 12 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana Number (N)  % < -3 SD  95% CI (%)  % <-2 SD  95% CI (%)  Mean  Standard Deviation (SD)  12-23  31  0  (0, 1.6)  0  (0, 1.6)  12.2  4.87  24-35  48  4.2  (0, 10.9)  6.3  (0, 14.1)  6.19  5.69  36-47  64  9.4  (1.5, 17.3)  9.4  (1.5, 17.3)  4.61  4.86  48-60  64  15.6  (5.9, 25.3)  20.3  (9.7, 31)  2.19  4.36  61-71  48  33.3  (19, 47.7)  33.3  (19, 47.7)  -0.46  3.84  72-83  48  33.3  (19, 47.7)  35.4  (20.8, 50)  -0.67  3.7  84-95  62  29  (16.9, 41.1)  35.5  (22.8, 48.2)  -1.26  3.35  96-107  73  35.6  (23.9, 47.3)  42.5  (30.4, 54.5)  -1.99  2.91  108-119  95  38.9  (28.6, 49.3)  51.6  (41, 62.2)  -2.97  3.1  120-131  93  59.1  (48.6, 69.7)  73.1  (63.6, 82.7)  -4.1  2.87  132-143  85  69.4  (59, 79.8)  85.9  (77.9, 93.9)  -4.75  2.75  Age in months  37  Table 6. Prevalence of low BMI-for-age in predominately HIV positive children (18 mo to 12 yrs) who are patients of the Botswana-Baylor Children’s Clinic in Gaborone, Botswana Age in months  Number (N)  % < -3 SD  95% CI (%)  % <-2 SD  95% CI (%)  Mean  Standard Deviation (SD)  12-23  31  0  (0, 1.6)  6.5  (0, 16.7)  -0.43  1.07  24-35  48  2.1  (0, 7.2)  8.3  (0, 17.2)  -0.5  1.27  36-47  64  1.6  (0, 5.4)  6.3  (0, 13)  -0.03  2.73  48-60  64  0  (0, 0.8)  1.6  (0, 5.4)  -0.09  1.04  61-71  48  0  (0, 1)  8.3  (0, 17.2)  -0.31  1  72-83  48  0  (0, 1)  0  (0, 1)  -0.17  0.81  84-95  62  1.6  (0, 5.6)  4.8  (0, 11)  -0.13  1.5  96-107  73  0  (0, 0.7)  5.5  (0, 11.4)  -0.53  0.91  108-119  95  1.1  (0, 3.6)  11.6  (4.6, 18.5)  -0.72  1.03  120-131  93  3.2  (0, 7.4)  18.3  (9.9, 26.7)  -1.16  0.97  132-143  85  2.4  (0, 6.2)  24.7  (14.9, 34.5)  -1.4  0.86  Table 7. Total percent of the study population who are underweight (low weight-for-age), stunted (low height-for-age), and low BMI-for-age Indicator  Total N  % <-3SD (n)  % <-2SD (n)  Weight-for-age  533  14.4% (77)  18.6% (99)  Height-for-age  711  34.3% (244)  41.9% (298)  BMI-for-age  711  1.3% (9)  10.0% (71)  38  4.3.  Inter-rater and Intra-rater Variability Inter-rater variability was assessed on 30 participants. The 30 participants were surveyed  twice, once by each of the two study investigators (R, M). Inter-rater variability is presented in Table 8. The average variability between the two investigators did not exceed 1% for weight, height, tibia (tape), and the mid-upper arm circumference (MUAC) measurement. Tibia (caliper), ulna (caliper), and ulna (tape) measurements were under an average of 2.5% variability. The triceps skinfolds measurement using a caliper had the greatest inter-rater variability with an average variability of 5.67%. The intra-rater variability, presented in Table 9, assessed the variability within each of the two study investigators (R, M). Seven of the measurements were measured at least twice by the same investigator. Weight was not analyzed for intra-rater variability because only one weight measurement was taken for each child. The average variability did not exceed 1% for height, ulna (tape and caliper), tibia (tape and caliper), and MUAC measurements for both investigators. The triceps skinfolds measurement had a higher variability for both investigators compared to the other six measurements, but the average variability was still under 5%.  39  Table 8. Inter-rater variability Average variability (%)  Minimum variability (%)  Maximum variability (%)  Weight  0.14  0  0.49  Height  0.19  0  0.71  Ulna (tape)  1.47  0  5.32  Ulna (caliper)  2.17  0  6.54  Tibia (tape)  0.92  0  4.40  Tibia (caliper)  1.20  0  8.43  MUAC  0.98  0  3.23  Triceps skinfolds  5.67  0  28.57  Measurement  Table 9. Intra-rater variability [R] Average (range) % variability  [M] Average (range) % variability  Weight  -  -  Height  0.28 (0-1.13)  0.16 (0-6.48)  Ulna (tape)  0.31 (0-3.75)  0.76 (0-4.60)  Ulna (caliper)  0.53 (0-7.79)  0.78 (0-5.13)  Tibia (tape)  0.39 (0-38.73)  0.46 (0-6.57)  Tibia (caliper)  0.41 (0-5.61)  0.83 (0-31.81)  MUAC  0.20 (0-3.00)  0.42 (0-5.92)  Triceps skinfolds  3.70 (0-28.57)  Surrogate Measure  4.08  0-50.00)  40  4.4.  Correlation between Caliper and Tape Measurements Using univariate regression, a coefficient of determination (R2) was calculated to  determine the strength of the relationship between a tape and a caliper, the gold standard, for taking ulna and tibia lengths. Both ulna tape and caliper and tibia tape and caliper measurements were strongly correlated with an R2 value of 0.993 for the ulna measurements, and R2= 0.988 for the tibia measurements. The graphs and regression equations are depicted in Figures 2 and 3. Due to the high correlation between the tape and caliper, ulna and tibia measurements using the tape values were chosen to be used for the remainder of the analyses. Figure 2. Correlation between ulna lengths using a caliper vs. a tape  R2 = 0.993  41  Figure 3. Correlation between tibia lengths using a caliper vs. a tape  R2 = 0.988  4.5.  Univariate Analysis Using Weight and Anthropometric Measurements Using univariate linear regression analysis, prediction equations for weight with each of  the anthropometric measurements were determined. The weight-prediction equations, along with the adjusted R2 values and the root mean square of the error are presented in Table 10. Height, ulna length, and tibia length, were moderately strong predictors of weight with adjusted coefficients of determination (R2) equating to 0.87, 0.86, and 0.84, respectively. The triceps skinfolds measure did not predict weight (p=0.152).  42  Table 10. Univariate analysis of prediction equations for weight estimation Prediction equation for weight  Adjusted R2  RMSE  Age  W=0.162A+6.589  0.764  12.533  Height  W=0.355H-20.116  0.870  6.885  Ulna  W=2.026U-17.056  0.859  7.494  Tibia  W=1.215T-10.869  0.841  8.439  MUAC  W=3.271M-33.684  0.656  18.251  Triceps  W=0.126S+20.647  0.001  53.029  Surrogate Measure  RMSE, root mean square of the error; W, weight (kg); A, age (months); G, gender; H, height (cm); U, ulna length (cm); T, tibia length (cm); M, mid-upper arm circumference (cm); S, triceps skinfold (mm)  4.6.  Accuracy of the Univariate Weight-Prediction Models How well the univariate weight-prediction model accurately predicted actual weight was  measured by percentage agreement. The proportion of the study population who had a predicted weight within 5%, 10%, and 15% of their actual weight, using each of the six univariate linear regression models, are presented in Table 11. The weight-prediction models using height, ulna length, and tibia length had the greatest percentage agreement with 60.5% of the children having a predicted weight within 10% of their actual weight using the height model; 61.3% using the ulna model; and 55.7% using the tibia model.  43  Table 11. Percentage agreement between the predicted weight and the actual weight for each univariate weight-prediction model  Weight-Prediction Model (Univariate)  Percentage in agreement within: 5% of actual weight (n)  10% of actual weight (n)  15% of actual weight (n)  Age  26.9 (209)  52.2 (406)  69.5 (540)  Height  35.1 (273)  60.5 (470)  81.3 (632)  Ulna  30.9 (240)  61.3 (476)  80.3 (624)  Tibia  30.4 (236)  55.7 (433)  76.8 (597)  MUAC  18.4 (143)  36.2 (281)  56.0 (435)  Triceps  10.6 (82)  20.7 (161)  4.7.  32.2  (250)  Multivariate Analysis Using Weight, Anthropometric Measures, and Other Potential Confounding Variables Additional potential covariates like age, and cofactors like gender, HIV status, and ARV  therapy use, were included in the linear regression models for each of the anthropometric measurements (height, ulna length, tibia length, MUAC, and triceps skinfolds) to determine if accuracy of the weight prediction could be improved. The β values (slopes) and the p-values for each of the added confounders are shown in Table 12. HIV status and ARV therapy use were not significant predictors of weight for any of the anthropometric measures (p-values>0.05). Age was found to be a significant predictor of weight in the ulna, tibia, MUAC, and triceps skinfolds weight-prediction models. In addition to age, gender was also a significant predictor of weight for the MUAC and triceps skinfolds models.  44  Table 12. Examination of the effects of cofactors and covariates on the relationship between proxy measures for weight and actual weight Surrogate Measure  β values (95% CI)  P value  1) Height (per cm)  0.362 (0.333, 0.390)  <0.0001*  -0.001 (-0.016, 0.013)  0.849  0.264 (-0.106, 0.634)  0.162  0.838 (0.661, 1.014)  0.090  -0.317 (-1.189, 0.554)  0.475  1.773 (1.622, 1.925)  <0.0001*  0.025 (0.012, 0.039)  <0.0001*  -0.207 (-0.591, 0.176)  0.289  Age (per month) Gender Male (n=403) vs. Female (n=374) HIV Negative or Unknown (n=152) vs. Positive (n=625) ARV No (n=183) vs. Yes (n=594) 2) Ulna Length (per cm) Age (per month) Gender Male (n=403) vs. Female (n=374) HIV Negative or Unknown (n=152) vs. Positive (n=625) ARV No (n=183) vs. Yes (n=594)  0.755 (-0.225, 1.735)  0.131  -0.191 (-1.093, 0.712)  0.678  1.055 (0.949, 1.162)  <0.0001*  0.026 (0.011, 0.041)  0.001*  0.353 (-0.054, 0.761)  0.089  0.850 (-0.190, 1.891)  0.109  -0.339 (-1.299, 0.620)  0.487  1.887 (1.784, 1.989)  <0.0001*  Age (per month)  0.114 (0.108, 0.119)  <0.0001*  Gender Male (n=403) vs. Female (n=374)  0.372 (0.068, 0.675)  0.017*  3) Tibia Length (per cm) Age (per month) Gender Male (n=403) vs. Female (n=374) HIV Negative or Unknown (n=152) vs. Positive (n=625) ARV No (n=183) vs. Yes (n=594) 4) MUAC (per cm)  45  HIV Negative or Unknown (n=152) vs. Positive (n=625) ARV No (n=183) vs. Yes (n=594)  -0.183 (-0.962, 0.696)  0.645  0.243 (-0.471, 0.957)  0.504  5) Triceps Skinfolds (per mm)  0.578 (0.500, 0.655)  <0.0001*  Age (per month)  0.173 (0.167, 0.180)  <0.0001*  0.740 (0.292, 1.188)  0.001*  0.911 (-0.213, 2.035)  0.112  -0.271 (-1.308, 0.765)  0.608  Gender Male (n=403) vs. Female (n=374) HIV Negative or Unknown (n=152) vs. Positive (n=625) ARV No (n=183) vs. Yes (n=594) *Significant at p<0.05  The insignificant confounders were excluded from further analysis and only the significant confounders (p<0.05) were included in the models. The linear regression equations and adjusted R2 values for each of the weight-prediction models, also separated by gender, are presented in Table 13. A comparison of adjusted R2 values between the simple univariate regression model and the model including the significant confounders are shown in Table 14. Age improved the model only a minimal amount for both the ulna and tibia models. The addition of age and gender into the MUAC and triceps skinfolds regression equations improved both models: a 26% improvement in the MUAC R2 value, and an 82% improvement in the triceps skinfolds R2 value.  46  Table 13. Regression equations for predicting weight using weight-estimation models Prediction equation for weight  Adjusted R2  RMSE  W=0.355H-20.116  0.870  6.885  Ulna  W=1.784U+0.022A-14.495  0.861  7.393  Tibia  W=1.059T+0.023A-8.870  0.843  8.347  MUAC  W=1.888M+0.113A+0.366G-21.018  0.913  4.611  Triceps  W=0.587S+0.171A+0.751G  0.817  9.735  W=0.345H-18.799  0.873  6.240  Ulna  W=1.580U+0.038A-12.170  0.859  6.904  Tibia  W=0.959T+0.034A-7.038  0.848  7.450  MUAC  W=1.792M+0.115A-19.175  0.907  4.546  Triceps  W=0.494S+0.170A+2.125  0.818  8.937  Height  W=0.360H-20.800  0.855  8.334  Ulna  W=2.048U-17.332  0.864  7.807  Tibia  W=1.244T-11.859  0.840  9.218  MUAC  W=1.962M+0.112A-22.198  0.919  4.669  Triceps  W=0.662S+0.170A  0.817  10.526  Surrogate Measure Both genders (n=777) Height  Males (n=403) Height  Females (n=374)  RMSE, root mean square of the error; W, weight (kg); A, age (months); G, gender; H, height (cm); U, ulna length (cm); T, tibia length (cm); M, mid-upper arm circumference (cm); S, triceps skinfold (mm)  47  Table 14. Comparison of Adjusted R2 values between the univariate weight-prediction model and the multivariate weight-prediction model Simple univariate model, Adjusted R2  Model including significant confounding variables, Adjusted R2  Improvement in Adjusted R2  Height  0.870  -  -  Ulna  0.859  0.861  0.002  Tibia  0.841  0.843  0.002  MUAC  0.656  0.913  0.257  Triceps  0.001  0.817  0.816  Measure  Replacing the age and gender variables with ulna length in the MUAC weight-prediction model resulted in a similar improvement to the adjusted R2 value. The weight-prediction regression equation including ulna length in the MUAC weight-prediction model is as follows: W=1.511U + 1.495M – 32.508 (R2=0.941). The relationship between the predicted weight using the MUAC-Ulna model and the actual weight is represented graphically in Figure 4a. Tibia length also had a similar effect as ulna length on the MUAC weight-prediction model (W=0.892T + 1.637M -29.918; R2=0.946). The relationship between the predicted weight using the MUAC-Tibia model and the actual weight is shown in Figure 4b.  48  Figure 4. Linear regression equations for the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models (a) MUAC-Ulna  W=1.511U + 1.495M – 32.508 R2=0.941  (b) MUAC-Tibia  W=0.892T + 1.637M - 29.918 R2=0.946  49  4.8.  Accuracy of the Multivariate Weight-Prediction Models Modified Bland-Altman analysis was conducted to assess the agreement between the  actual weight and the predicted weight using six of the weight-prediction models (height, MUAC, ulna length, tibia length, MUAC-Ulna, and MUAC-Tibia) (shown in Figure 5). The difference between the predicted weight and actual weight (actual – predicted), or the residuals, were plotted against the actual weight. The upper and lower “limits of agreement” were set at 2.0 kg and -2.0 kg.  50  Figure 5. Modified Bland-Altman plots for predictive accuracy of six weight-prediction models. The dotted lines represent a residual weight of 2.0 kg and -2.0 kg (a) Height (a) Height  (b) MUAC (b) MUAC  (c) Tibia  (d) Ulna  (e) MUAC-Ulna  (f) MUAC-Tibia  51  Because a misclassification of 2 kg is more critical in a smaller child who weighs 10 kg compared to one that weighs 40 kg, percentage agreement was also calculated. The percentage agreement of the predicted weight within 5%, 10%, and 15% of the actual weight for the multivariate linear regression models are shown in Table 15. The weight-prediction models that had the greatest percentage agreement were the MUAC-Ulna model and the MUAC-Tibia model. A total of 79% of the children were classified as having a predicted weight within 10% of their actual weight when using the MUAC-Ulna model, and 82% of children for the MUACTibia model. Weight was underestimated by greater than 10% in 10.6% of the population using the MUAC-Ulna model and in 10.0% of the population using the MUAC-Tibia model. Weight was overestimated by greater than 10% in 10.8% of the population using the MUAC-Ulna model and in 8.5% of the population using the MUAC-Tibia model. Graphical representation of the percentage agreement of the MUAC-Ulna and MUACTibia models are shown in Figures 6 and 7. As the variance changes over the range of weights, the percentage agreement within 10% for the MUAC-Ulna and the MUAC-Tibia weightprediction models are separated by weight group in Table 16. The weight group of 25.1-30 kg had the highest percentage of children with a predicted weight within 10% of their actual weight. Approximately 95% of children between 25.1 kg and 30 kg had a predicted weight within 10% of their actual weight using the MUAC-Ulna model and 93% of children were within 10% when using the MUAC-Tibia model.  52  Table 15. Percentage agreement between the predicted weight and the actual weight for each multivariate weight-prediction model Weight-Prediction Model (Multivariate)  Percentage in agreement within: 5% of actual weight (n)  10% of actual weight (n)  15% of actual weight (n)  Height  35.1 (273)  60.5 (470)  81.3 (632)  Ulna  32.9 (256)  62.0 (482)  80.8 (628)  Tibia  30.6 (238)  56.6 (440)  77.0 (598)  MUAC  42.1 (327)  72.1 (560)  87.6 (681)  Triceps  30.5 (237)  55.5 (431)  75.3 (585)  MUAC-Ulna  45.6 (362)  78.8 (612)  91.2 (709)  MUAC-Tibia  51.1 (397)  81.5 (633)  93.7 (728)  53  Figure 6. Percentage agreement between predicted weight and actual weight for the MUAC-Ulna weight-prediction model = ± 15% agreement = ± 10% agreement  54  Figure 7. Percentage agreement between predicted weight and actual weight for the MUAC-Tibia weight-prediction model  = ± 15% agreement = ± 10% agreement  55  Table 16. Percentage agreement within 10% between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models, separated by weight group Weight group n (kg)  Number in group, n  MUAC-Ulna, % (n)  MUAC-Tibia, % (n)  ≤10.0  28  28.6 (8)  17.9 (5)  10.1-15.0  139  65.5 (91)  69.8 (97)  15.1-20.0  164  76.8 (126)  82.3 (135)  20.1-25.0  193  86.0 (166)  89.1 (172)  25.1-30.0  165  94.5 (156)  93.3 (154)  30.1-40.0  80  80.0 (64)  86.3 (69)  ≥40.1  8  12.5 (1)  12.5  4.9.  (1)  Validation of MUAC-Ulna and MUAC-Tibia Weight-Prediction Equations In order to validate the MUAC-Ulna and MUAC-Tibia weight-prediction equations  depicted in Figure 4, a random sample of approximately half of the data set was removed from analysis (Sample 2, n=391). The remaining data (Sample 1, n=386) underwent linear regression analysis to produce regression equations for predicting weight (Table 17). These equations were used on a ‘new’ sample (Sample 2, n=391) that were used to validate the prediction equations and see how accurate the equations are at predicting weight. The predicted weight versus actual weight is plotted in Figure 8 for the MUAC-Ulna and MUAC-Tibia weight-prediction equations using Sample 2 (n=391).  56  Table 17. Regression equations for predicting weight using MUAC-Ulna and MUAC-Tibia weight-prediction models on a random sample of approximately 50% of the data (Sample 1, n=386) Regression Equation  Adjusted R2  RMSE  MUAC-Ulna  W=1.443U+1.596M-32.963  0.942  3.266  MUAC-Tibia  W=0.860T+1.715M-30.426  0.950  2.799  Surrogate Measure  RMSE, root mean square of the error; W, weight (kg); U, ulna length (cm); T, tibia length (cm); M, mid-upper arm circumference (cm)  Figure 8. Linear regression equations for the MUAC-Ulna (a) and MUAC-Tibia (b) weightprediction models using approximately half the data set (Sample 2, n=391)  (a) MUAC-Ulna  W=1.443U + 1.596M – 32.963  57  (b) MUAC-Tibia  W=0.860T + 1.715M – 30.426  To determine how accurate the MUAC-Ulna and MUAC-Tibia regression equations were at predicting actual weight over the range of weights, modified Bland-Altman analyses were conducted for Sample 2 (n=391). The two plots are shown in Figure 9 with the difference between the predicted weight and actual weight (residuals) plotted against the actual weight. The upper and lower “limits of agreement” were set at 2.0 kg and -2.0 kg.  58  Figure 9. Modified Bland-Altman plots for predictive accuracy of the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models using approximately half the data set (Sample 2, n=391). The dotted lines represent a residual weight of 2.0 kg and -2.0 kg (a) MUAC-Ulna  (b) MUAC-Tibia  59  Percentage agreement for the MUAC-Ulna and MUAC-Tibia weight-prediction models within 5%, 10%, and 15% of the actual weight are shown for both samples (Sample 1, n=386; and Sample 2, n=391) in Table 18. Eighty-one percent (81%) of the children had a predicted weight within 10% of their actual weight using both models (MUAC-Ulna and MUAC-Tibia) for Sample 2 (n=391). The percentage agreements for the two models are presented graphically in Figure 10. Table 18. Percentage agreement between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models Percentage in agreement within: Weight-Prediction Model  5% of actual weight (n)  10% of actual weight (n)  15% of actual weight (n)  MUAC-Ulna  46.4 (179)  76.4 (295)  91.5 (353)  MUAC-Tibia  50.6 (199)  80.1 (309)  94.3 (364)  MUAC-Ulna  48.3 (189)  80.8 (316)  91.6 (358)  MUAC-Tibia  50.6 (198)  81.1 (317)  93.6 (366)  Sample 1 (n=386)  Sample 2 (n=391)  60  Figure 10. Percentage agreement between predicted weight and actual weight for the MUAC-Ulna (a) and the MUAC-Tibia (b) weight-prediction models (Sample 2, n=391)  (a) MUAC-Ulna = ± 15% agreement = ± 10% agreement  61  (b) MUAC-Tibia = ± 15% agreement = ± 10% agreement  62  4.10. Comparison of Existing Weight-Prediction Methods and the MUAC-Ulna and MUAC-Tibia Weight-Prediction Models A number of validated weight-prediction methods were tested to assess the accuracy of each method on the population being examined, including the advanced pediatric life support (APLS) formula, the Luscombe formula, the Theron formula, and the Broselow tape. The percentage agreement for the MUAC-Ulna and MUAC-Tibia, as well as three previously validated weight-prediction methods based on age, are presented in Table 19. Using the APLS weight-estimation method, approximately 37% of the children in the sample had a predicted weight within 10% of their actual weight. The Luscombe and Theron formulas only estimated weight within 10% of actual weight in approximately 7% and 9% of the study population, respectively. The APLS formula overestimated actual weight by more than 10% in 55% of the population. The Luscombe and Theron formulas both had a strong tendency to overestimated actual weight by more than 10%: 92% of the population using the Luscombe formula and 91% of the population using the Theron formula. The MAC formula (weight = (MAC – 10) + 3) devised by Cattermole et al predicted weight within 10% of actual weight in approximately 32% of the population. The MAC formula had a very slight tendency to underestimate weight with 41% of the population having an estimated weight less than their actual weight by more than 10%, and 26% of the population had an overestimated weight greater than 10%.  63  Table 19. Percentage agreement between the predicted weight and the actual weight for the MUAC-Ulna and MUAC-Tibia weight-prediction models and three existing weightprediction methods based on age Percentage in agreement within: Weight-Prediction Method  N  MUAC-Ulna  5% of actual weight  10% of actual weight  15% of actual weight  777  45.6  78.8  91.2  MUAC-Tibia  777  51.1  81.5  93.7  APLS  531  21.3  37.3  55.0  Luscombe  531  3.6  7.2  13.6  Theron  531  3.8  8.5  12.8  MAC  777  15.7  32.2  49.4  In addition to the three age-based weight-prediction methods and the MAC formula, the accuracy of the Broselow tape was also determined. The Broselow tape correctly predicted approximately 55% (420/758) of the population within the correct colour zone for children between 46 cm and 143 cm. In addition, the Broselow tape overestimated actual weight in 39% (n=295) of the population by at least one colour zone and underestimated actual weight in 6% (n=43) of the population.  64  Chapter 5: DISCUSSION 5.1.  Overview The purpose of this study was to determine if a child’s weight could be accurately  estimated using a proxy anthropometric measure for weight in the absence of a working scale in children 18 months to 12 years in Botswana. In this chapter, the results of the study are discussed and the weight-estimation methods are examined with relation to accuracy, practicality, and generalizability. A review of the current methods used for estimating weight in children is also discussed and a comparison to the weight-estimation methods from this study is examined.  5.2.  Participant Characteristics The majority of the study population was HIV positive because recruitment of  participants took place in a clinic for children receiving treatment for HIV. In addition, the HIV rate in Botswana is one of the highest in the world, with over 16 000 children under the age of 14 infected with HIV. (66) Furthermore, it is highly likely that many of the participants who were HIV negative were HIV exposed, meaning they were born to HIV-infected mothers. The high proportion of children infected with HIV or exposed to HIV may contribute to the high rate of malnutrition observed in this study. Numerous studies have reported that children infected with HIV have poor growth including stunting and wasting. (67-72) Reasons for weight loss, failure to gain weight, and poor linear growth in HIV positive children are related to malabsorption of nutrients, increased energy expenditure, or decreased food intake. (73) In addition, children who were born to HIV positive mothers have been found to have mean birth weights and lengths significantly lower than those infants born to mothers who were HIV negative. (74) 65  On the other hand, some studies have shown that children who are receiving highly active antiretroviral therapy (HAART) have improved growth with regards to both weight and height. (75-79) In the present study, the majority of the children with HIV were receiving ARV treatment, yet many children were stunted, underweight, or both. However, the benefits of HAART on growth are observed when children receive the treatment early. (67) Most likely, many participants in the study did not receive a timely initiation of HAART, which could have contributed to the stunted growth and low weights-for-age observed in the study population. (67) Study participants tended to be in the older age range, with the majority of the participants over the age of 5. A likely reason for the age demographics of the study population is that the incidence rate of mother-to-child transmission of HIV in Botswana is declining due to the widespread use of HAART. (80) In Botswana, everyone has access to free HIV treatment programs, a first of its kind in Africa. (81) Moreover, the Ministry of Health has prioritized the prevention of mother-to-child transmission (PMTCT) program by increasing CD4+ cell count testing and treatment referrals for HAART initiation in pregnant women. (81) As part of the PMTCT program, mothers receive HIV prophylaxis (Zidovudine) and a 12-month supply of formula for the HIV exposed infant. (82) The scaling up of the PMTCT program has resulted in the drastic reduction of the transmission of HIV since 2003. (80) In 2003, 20.7% of infants born to HIV infected mothers were born HIV positive, and that percentage dropped to 4.8% by 2008 and 3.8% by 2010. (82) Therefore, this drop in the mother-to-child transmission rate could mean that fewer younger children are attending the Botswana-Baylor Children’s Clinical Center of Excellence (COE).  66  5.3.  Devised Weight-Prediction Methods The weight-prediction models devised in this study with the highest accuracy were the  MUAC-Ulna and the MUAC-Tibia models, which support the recommendation of using heightbased methods of estimating weight in children in developing countries compared to age-based formulas. (12,32,53) Height-based methods for estimating weight have shown to be less variable with body habitus compared to age-based formulas, especially in populations with prevalent stunting and low weights-for-age. (12,28,31,32,83) Proxy measures of height, such as ulna and tibia length, were used in this study because they have been shown to be well correlated with height and are simple to measure. (62) Also, height is a difficult measure to execute with accuracy in resource limited settings without proper equipment. (63) Moreover, the addition of MUAC to the ulna and tibia length weight-estimation equations greatly improved the accuracy of predicting children’s weight. Mid-upper arm circumference (MUAC) has previously been reported as a strong correlate of weight in a study conducted by Cattermole et al in 2010, (33) but it is not currently being used in any weightestimation methods. Our study suggests that the combination of a proxy for height, like ulna or tibia length, and MUAC, an indicator of muscle mass and subcutaneous fat, may be used to accurately estimate a child’s weight. In addition to their accuracy level, another benefit of the MUAC-Ulna and MUAC-Tibia weight-prediction models is their simplicity of use. Tibia length, ulna length, and MUAC are easy areas to measure with a tape measure. In a clinical or field setting where scales may be unavailable, the only equipment needed to estimate a child’s weight is a tape measure and a MUAC-Ulna or MUAC-Tibia weight-estimation chart (Appendices C and D). A health worker simply measures the child’s ulna or tibia length to the nearest centimeter and MUAC to the 67  nearest half a centimeter, and finds the estimated weight value in kilograms on the chart. Similar to the Broselow tape, there are no calculations required meaning the risk of calculation errors are minimal. Age was found to be a statistically significant predictor of weight; however, age was not included in the MUAC-Ulna and MUAC-Tibia models because increasing the number of variables would complicate the model and make it more difficult to use in a clinical or field setting. More importantly, age may have been statistically significant, but it was not clinically significant as it did not impact the accuracy of the models. In particular, the MUAC-Ulna and MUAC-Tibia weight-prediction models predicted weight with high accuracy for the children weighing 20-30 kg. The majority of children within this weight range had a predicted weight within 10% of their actual weight. Some studies from developing countries showed that particular existing weight-estimation methods poorly predicted weight in children in the higher weight or age ranges. For example, in the Ramarajan et al study (30) in a population of undernourished Indian children, the authors found the Broselow tape to be less accurate at predicting weight in children over 18 kg. Only 34% of the population weighing more than 18 kg had an estimated weight within 10% of their actual weight compared to 53% of those weighing less than 10 kg. (30) However, the Broselow tape was devised based on Western populations and the MUACUlna and MUAC-Tibia models have been devised on a highly stunted population. This might explain why the Broselow tape has been shown to be less accurate in older children in studies conducted in developing countries, where there is a significant divergence in height-for-weight correlations from Western populations. (30,53) Stunting was more pronounced in the older children (≥ 9 years) compared to the younger children (≤ 5 years) in our study; therefore, the MUAC-Ulna and MUAC-Tibia weight-prediction models may be more accurate for use in 68  developing countries where stunting in children is more widespread, particularly in children 2030 kg. However, findings from this study show that the two weight-prediction equations did not predict weight with the same accuracy across the entire weight range; therefore, the MUAC-Ulna and MUAC-Tibia weight-estimation charts should be used with caution. A high proportion of children under 15 kg did not have an estimated weight within 10% of their actual weight. In addition, there were significant outliers amongst this weight group, especially in the children weighing less than 10 kg. Therefore, the devised weight-prediction models may not be applicable for children weighing less than 15 kg. In previous studies conducted in developing countries, there were contradictory findings with regards to children in the lower weight categories. For instance, a study conducted in India by Varghese et al (53) that examined the accuracy of the Broselow tape, found that it predicted weight better in children weighing less than 15 kg compared to children over 15 kg. However, more than 90% of the study population weighed less than 15kg, making it difficult to draw conclusions on children weighing over 15 kg. (32) Similarly, another study in India conducted by Ramarajan et al (30) found that the accuracy of the Broselow tape decreased as weight increased in participants. Weight was predicted with the highest accuracy in children weighing less than 10 kg with 53% of children in this weight category having an estimated weight within 10% of their actual weight. (30) These findings suggest that the Broselow tape may predict weight better in smaller children in developing countries compared to the MUAC-Ulna and MUAC-tibia weightprediction methods. However, a possible reason for the low accuracy in the small weight group in our study could be because of the small sample size. As mentioned earlier, it was difficult to recruit many children under the age of 5 from the COE and therefore there are fewer participants 69  in the lower weight range. Furthermore, the weight-prediction models did not accurately predict weight in children weighing more than 40 kg. Again, the sample size was too small to draw any conclusions. Before any conclusions can be drawn on which weight-estimation method is the most accurate for smaller children, further studies need to be conducted in children in developing countries who weigh less than 15 kg.  5.4.  Comparison of Existing Weight-Prediction Methods The Broselow tape, three age-based weight-prediction formulas, and the MAC formula,  were applied to this study population to compare the accuracies of the different weightprediction methods. The Broselow tape, the APLS formula, and the MAC formula predicted weight better than the Luscombe and Theron formulas. A study conducted in South Africa by Geduld et al (12) found that the APLS method predicted weight within 10% of actual weight in 58% of the population. The APLS formula was more accurate in predicting weight in the South African population compared to our Botswana study population. However, participants in the Geduld study were from a range of socioeconomic backgrounds and therefore rates of undernutrition may not have been as high as in our study. The APLS formula was devised based on Western data and therefore it may be more applicable to the population in the Geduld study. The children in the Geduld study were most likely more similar to Western populations compared to the moderately underweight population in our Botswana study. Also, it was to be expected that the Luscombe formula would be less accurate in our study population compared to the APLS formula. The Luscombe formula was an ‘improvement’ to the APLS formula to adjust for the growing obesity rates in Western populations, an issue not applicable in most undernourished populations.  70  A study conducted in India in a malnourished population found that the APLS method tended to overestimate weight. (53) Results from our study support these findings as the APLS formula poorly predicted weight in children with a tendency to overestimate weight. The reason for the overestimation is most likely due to the fact that the APLS formula is devised based on weight-for-age data from the United States and our study population had much lower weightsfor-age with approximately 19% of our study population classified as underweight. Results from our study were also similar to a study conducted by House et al (32) in children in Kenya. The APLS method had a high degree of variability and tended to overestimate weight. (32) Similar to the Varghese study in India and our Botswana study, children in the Kenyan study were also malnourished. (32) On the contrary, even though the Broselow tape outperformed the age-based formulas, results from the study did suggest that the APLS formula could be used to estimate weight in Kenyan children. (32) The reason the APLS formula may have worked slightly better in the Kenyan population compared to our Botswana population or the Varghese study population, could be because the Kenyan children were malnourished in a different context. (32) The population in the Kenyan study was most likely suffering from protein energy malnutrition with edema. (32) The presence of edema might have caused the children to be not as underweight compared to children suffering from malnutrition in the absence of edema, such as in our study and the Varghese et al study. (32) The Broselow tape predicted weight with the greatest accuracy out of all the existing weight-estimation methods examined in this study population. As previously mentioned, this comes as no surprise as height-based methods have been repeatedly shown to be better predictors of weight compared to age-based methods. (12,30,32,53) However, some researchers have expressed concern over the potential overestimation of weight in developing country populations 71  where stunting may be prevalent. (30) Results from this study support this concern as a significant percent of children had an overestimated weight by at least one colour zone in our study. There have been no studies validating the MAC formula devised by Cattermole et al (33) in children in Hong Kong. Our study is the first to examine the accuracy of the MAC formula on an independent sample. The MAC formula predicted weight in children with low accuracy in our study and was outperformed by the APLS method and the Broselow tape. However, our univariate weight-prediction model using only MUAC as the independent variable provided similar results to the MAC formula devised by Cattermole et al. (33) The addition of ulna length or tibia length into the MUAC formula in our study greatly increased the accuracy of predicting weight. This suggests that a MUAC measurement on its own is not a sufficient predictor of weight in children, but can be used together with a surrogate measure for height like ulna or tibia length to better predict weight in children aged 18 months to 12 years. The MUAC-Ulna and MUAC-Tibia weight-prediction models were the best predictors of weight compared to any other weight-estimation method explored – both existing and devised in this study. However, interpretation should be exerted with caution as the MUAC-Ulna and MUAC-Tibia weight-prediction regression equations are best fitted for this study population, thus correlations are expected to be strong. In order to validate the MUAC-Ulna and MUAC-Tibia regression equations in this study population, the data was split into two random samples. The ‘independent’ sample predicted weight within 10% of actual weight with similar accuracy as the sample from which the regression equations were derived from. This suggests that the MUAC-Ulna and MUAC-Tibia  72  weight-prediction models could potentially predict weight to similar degrees of accuracy in other similar populations. This is a unique step in our study as none of the researchers conducting studies examining weight-prediction methods included this step in their analyses. However, it is understood that while the data from the ‘independent’ sample was not used to devise the regression equations, the sample is not truly independent as the children were selected from the same population. Validation of the MUAC-Ulna and MUAC-Tibia models should be done in child populations in other developing countries.  5.5.  Correlation between Caliper and Tape Measurements Ulna and tibia lengths were measured with both a long bone caliper and a tape measure to  validate the use of the tape in clinical or research study settings. The tape measurements for both the ulna and tibia lengths correlated almost perfectly with the caliper measurements, meaning the tape measure could be used in place of the caliper in future studies. In addition, inter-rater (degree of agreement between investigators) and intra-rater (degree of agreement among multiple measurements conducted by a single investigator) of the tape and caliper measurements were similar with minimal average variability between raters and measures. Calipers are considered the ‘gold standard’ and have metal prongs that are perpendicular to a ruler that can be adjusted to the exact length of the ulna or tibia to the nearest millimeter. However, these pieces of measuring equipment are difficult to transport and are extremely expensive (Rosscraft Campbell 20 caliper, CND$524). The use of a tape measure to take long bone measurements instead of a caliper would significantly reduce costs.  73  5.6.  Study Limitations It is important to acknowledge and discuss limitations of the study and the interpretation  of the results, such as issues with the study design and participant demographics. The study was a convenience sample and was conducted at a single facility: the Botswana-Baylor Children’s Clinical Center of Excellence (COE). Therefore, the study sample may not be representative of the entire population of Botswana. Another limitation of the study was that the participants’ ages were not evenly distributed. For instance, the majority of the participants were over the age of 5, particularly between the ages of 9 and 13. Very few participants were under the age of 5, particularly under the age of 2 years. In addition, because there were fewer children under 5, there were fewer children were in the lower weight range. Due to the small sample size for children under 15 kg, it was difficult to draw any conclusions for the accuracy of the weight-prediction methods in children pertaining to this weight group. Furthermore, the estimated weights listed in the MUAC-Ulna and MUAC-Tibia weightprediction charts should be interpreted with caution as they are most likely not accurate across all the measurements. The two charts should be validated in other populations before they can be considered as replacements for current weight-estimation methods. Moreover, limited demographic and health data was collected in the survey. Without additional information, we were unable to describe the severely misclassified children or describe the outliers. Also, children who were HIV negative were most likely born to HIV positive mothers given the demographics of the patients at the COE; however, it was unknown in our sample whether an HIV negative child was HIV exposed because they were not differentiated in the study. As similar poor growth outcomes has been observed in both HIV positive children and HIV exposed children, it would have been beneficial to separate HIV 74  negative from the HIV exposed children. Also, the sample size for HIV negative children was small and direct comparisons to the children who were HIV positive should be limited. Similar sample sizes are needed for a comparison between how well the MUAC-Ulna and MUAC-Tibia weight-prediction methods performed in HIV positive children versus HIV negative children. In addition, most studies validating weight-estimation methods, especially those based on age did not include children over 10 years. (31) In our study however, we included children who were in their twelfth year and some children therefore may have already entered into puberty. Pubertal status was not recorded and these children may have negatively influenced results because growth spurts and a large increase in weight variability can be observed in children in this age category. (31) Lastly, despite splitting the population and internally validating the weight-prediction equations, results from this study need to be externally validated. The MUAC-Ulna and MUACTibia weight-prediction equations should be tested in other populations in developing countries before any conclusions can be made on the accuracy and generalizability of the two equations.  5.7.  Generalizability The results of this study primarily apply to well-monitored HIV positive children  receiving ARV treatment in Botswana who are moderately underweight with moderate to severe stunting. It should be noted that the HIV positive children in this study had access to the best medical facility for the treatment of HIV in the country with some children receiving treatment since birth. Therefore, results from this study may not be transferable to other HIV positive cohorts of children, especially in populations with no or limited access to treatment. In addition, measurements were taken with trained study investigators in a facility fully equipped with  75  calibrated measuring equipment. Therefore, study results may differ in settings with untrained staff or improper measuring equipment.  5.8.  Implications of Findings There is no single weight-estimation method that is recommended for use across all  populations. (12) In particular, there is concern regarding the applicability of existing weightestimation methods in developing countries that have been derived from Western populations. Results from this study suggest that using ulna or tibia length along with MUAC in children over 15 kg could potentially predict weight in children in developing countries. Furthermore, implications of these findings could mean that health clinics in limited resource settings without access to working scales may be able to dose pediatric medications with an accurate, simple, and inexpensive weight-prediction method that would reduce dosing errors.  5.9.  Directions for Further Research Validation of the MUAC-Ulna and MUAC-Tibia weight-prediction methods is needed in  various populations in other developing countries. In particular, repeat studies need to be conducted in children weighing less than 15 kg, or under the age of 5 years including children less than 18 months. Future studies also need to be conducted in less ideal settings to examine the effectiveness of the MUAC-Ulna and MUAC-Tibia weight-prediction charts. These studies are needed to assess the usability and practicality of the charts in real-life settings. Further research also needs to include applying weight-bands to the MUAC-Ulna and MUAC-Tibia weight-prediction charts for use in drug dosing of fixed-dose dispersible tablets. Colour-coding the charts into weight-bands for each fixed-dose would make dosing pediatric  76  medications easy and quick. Weight-bands would need to be devised and corresponding dosages would then be determined for each band. In addition, ulna and tibia length were only two long bone measurements that were examined in this study; however, addition long bone measurements should also be explored for the use in weight-estimation models. For example, humerus length could be an alternative measurement to ulna to tibia length. In a study exploring proxy measures for height in children with cerebral palsy, researchers found that humerus length was easier to measure compared to ulna length. (64) Moreover, using humerus length would require one less measurement when estimating weight because in order to correctly locate the mid-point of the upper arm for a MUAC measurement, the humerus length needs to be determined.  5.10. Conclusion This study suggests that a weight-estimation method based on mid-upper arm circumference (MUAC) and a surrogate measure for height, like ulna or tibia length, could potentially be used in developing country populations with high rates of undernutrition as a way to accurately estimate a child’s weight in the absence of a working scale. The MUAC-Ulna and MUAC-Tibia weight-prediction methods estimated children’s weight (18 months – 12 years) with the highest accuracy compared to other commonly used weight-prediction methods. Although these two weight-prediction models worked well in this study population, they need to be externally validated in other populations. If the two equations predict children’s weight in other populations to the same degree of accuracy seen in this study, they could potentially replace the use of other existing weight-prediction methods currently being used in developing country populations and improve the dosing of pediatric medications.  77  REFERENCES 1. World Health Organization, United Nations Children’s Fund. Low Birthweight: Country, regional and global estimates [Internet]. New York: UNICEF; 2004. Available from: http://www.unicef.org/publications/files/low_birthweight_from_EY.pdf 2. Lawn JE, Cousens S, Zupan J, Lancet Neonatal Survial Sterring Team. 4 million neonatal deaths: When? Where? Why? Lancet. 2005: 365(9462): 891-900. 3. World Health Organization. WHO Child Growth Standards: Methods and development: Length/height-for-age, weight-for-age, weight-for-length, weight-for-height and body mass index-for-age [Internet]. Geneva: World Health Organization; 2006. Available from: http://www.who.int/childgrowth/standards/Technical_report.pdf 4. de Onis M, Wijnhoven TMA, Onyango AW. Worldwide practices in child growth monitoring. J Pediatr. 2004: 144(4): 461-465. 5. World Health Organization, United Nations Children’s Fund. WHO child growth standards and the identification of severe acute malnutrition in infants and children. A joint statement by the World Health Organization and the United Nations Children's Fund [Internet]. Geneva: World Health Organization; 2009. Available from: http://www.who.int/nutrition/publications/severemalnutrition/9789241598163_eng.pdf 6. de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007: 85(9): 660-667. 7. Cogill B. Anthropometric Indicators Measurement Guide [Internet]. Washington, DC: Food and Nutrition Technical Assistance Project; 2003. Retrieved from: http://www.fantaproject.org/downloads/pdfs/anthro_2003.pdf  78  8. Flegal KM, Wei R, Ogden C. Weight-for-stature compared with body mass index-for-age growth charts for the United States from the Centers for Disease Control and Prevention. Am J Clin Nutr. 2002: 75(4): 761-766. 9. Ogden CL, Kuczmarski RJ, Flegal KM, Mei Z, Guo S, Grummer-Strawn LM, et al. Centers for Disease Control and Prevention 2000 growth charts for the United States: improvements to the 1977 National Center for Health Statistics version. Pediatrics. 2002: 109(1): 45-60. 10. Nandy S, Miranda JJ. Overlooking undernutrition? Using a composite index of anthropometric failure to assess how underweight misses and misleads the assessment of undernutrition in young children. Soc Sci Med. 2008: 66(9): 1963-1966. 11. World Health Organization. Physical status: the use and interpretation of anthropometry. World Health Organ Tech Rep Ser. 1995: 854: 1-452. 12. Geduld H, Hodkinson PW, Wallis LA. Validation of weight estimation by age and length based methods in the Western Cape, South Africa population. Emerg Med J. 2011: 28(10): 856-860. 13. World Health Organization. WHO expert committee on specifications for pharmaceutical preparations: Thirty-seventh edition. WHO tecnical report series 908. Geneva: World Health Organization; 2003. Available from: http://whqlibdoc.who.int/trs/who_trs_908.pdf 14. World Health Organization. Integrated Management of Childhood Illness Chart Booklet [Internet]. Geneva: World Health Organization; 2008. Available from: http://whqlibdoc.who.int/publications/2008/9789241597289_eng.pdf 15. Quick JD, Hogerzeil HV, Velasquez G, Rago L. Twenty-five years of essential medicines. Bull World Health Organ. 2002: 80(11): 913-914.  79  16. Oshikoya KA, Ojo OI. Medication errors in paediatric outpatient prescriptions of a teaching hospital in Nigeria. Nig Q J Hosp Med. 2007: 17(2):74-78. 17. Mesnson EN, Walker AS, Sharland M, Wells C, Tudor-Williams G, Riordan FA, et al. (2006). Underdosing of antiretrovirals in UK and Irish children with HIV as an example of problems in prescribing medicines to children, 1997-2005: cohort study. BMJ. 2006: 332: 1183. 18. Beggs SA, Cranswick NE, Reed MD. Improving drug use for children in the developing world. Arch Dis Child. 2005: 90(10): 1091-1093. 19. Johnson TN. The problems in scaling adult drug doses to children. Arch Dis Child. 2008: 93(3): 207-211. 20. World Health Organization. WHO Model Formulary for Children [Internet]. Geneva: World Health Organization. Available from: http://www.who.int/selection_medicines/list/WMFc_2010.pdf 21. Standing JF, Tuleu C. Paediatric formulations – Getting to the heart of the problem. Int J Pharm. 2005: 300(1-2): 56-66. 22. Ellis JC, L’homme RF, Ewings RM, Mulenga V, Bell F, Chileshe R, et al. Nevirapine concentrations in HIV-infected children treated with divided fixed-dose combination antiretroviral tablets in Malawi and Zambia. Antivir Ther. 2007: 12(2): 253-260. 23. Health Action International Africa. Practical Pharmacy for Developing Countries: Dispersible tablets: a novel dosage form for children. Nairobi: Health Action International Africa; 2010. Available from: http://apps.who.int/medicinedocs/documents/s17536en/s17536en.pdf  80  24. United Nations Children’s Fund. Supplies and Logistics: Dispersible Tablets [Internet]. New York: United Nations Children’s Fund; 2010. Available from: http://www.unicef.org/supply/index_53571.html 25. Larson C, Sauvé L, Senkungu JK, El Arifeen S, Brant R. Converting IMCI essential drugs from syrup to fixed dose dispersible tablet formulations: Development and validation of weight, height and age bands (unpublished). Submitted to JPPT, September 2011. 26. Ponnet M, Frederix K, Petdachai W, Wilson D, Eksaengsri A, Zachariah R. A drug dosage table is a useful tool to facilitate prescriptions of antiretroviral drugs for children in Thailand. Int J STD AIDS. 2005: 16(6): 420-426. 27. World Health Organization. Integrated Management of Childhood Illness Chart Booklet [Internet]. Geneva: World Health Organization; 2005. Available from: http://whqlibdoc.who.int/publications/2008/9789241597289_eng.pdf 28. Black K, Barnett P, Wolfe R, Young S. Are methods used to estimate weight in children accurate? Emerg Med. 2002: 14(2):160-165. 29. Argall JAW, Wright N, Mackway-Jones K, Jackson R. A comparison of two commonly used methods of weight estimation. Arch Dis Child. 2003: 88: 789-790. 30. Ramarajan N, Krishnamoorthi R, Strehlow M, Quinn J, Mahadevan SV. Internationalizing the Broselow tape: How reliable is weight estimation in Indian children. Acad Emerg Med. 2008: 15(5): 431-436. 31. So T, Farrington E, Absher RK. Evaluation of the accuracy of different methods used to estimate weights in the pediatric population. Pediatrics. 2009: 123(6): 1045-1051.  81  32. House DR, Ngetich E, Vreeman RC, Rusyniak DE. Estimating the weight of children in Kenya: Do the Broselow tape and aged-based formulas measure up? Ann Emerg Med. 2012: (Epub). 33. Cattermole GN, Leung PY, Mak PS, Graham CA, Rainer TH. Mid-arm circumference can be used to estimate children’s weights. Resuscitation. 2010; 81(9): 1105-1110. 34. Cattermole GN, Leung MP, So HK, Mak PS, Graham CA, Rainer TH. Age-based formulae to estimate children’s weight in the emergency department. Emerg Med J. 2011: 28(5): 390396. 35. Luscombe M, Owens B. Weight estimation in resuscitation: is the current formula still valid? Arch Dis Child. 2007: 92: 412-415. 36. Bryce J, Victora CG, Habicht JP, Vaughan JP, Black RE. The multi-country evaluation of the Integrated Management of Childhood Illness strategy: lessons for the evaluation of public health interventions. Am J Public Health. 2004: 94(3): 406-415. 37. Pariyo GW, Gouws E, Bryce J, Burnham G. Improving facility-based care for sick children in Uganda: training is not enough. Health Policy Plan. 2005: 20(Suppl 1): i58-i68. 38. Anglemyer BL, Hernandez C, Brice JH, Zou B. The accuracy of visual estimation of body weight in the ED. Am J Emerg Med. 2004: 22(7): 526-529. 39. Harris M, Patterson J, Morse J. Doctors, nurses, and parents are equally poor at estimating pediatric weights. Pediatr Emerg Care. 1999: 15(1): 17-18. 40. Grieg A, Ryan J, Glucksman E. How good are doctors at estimating children’s weight? J Accid Emerg Med. 1997: 14(2): 101-103. 41. Mackway-Jones K, Molyneux E, Phillips B, Wieteska S. Advanced paediatric life support – the practical approach. 4th edn. London: BMJ Books; 2001.  82  42. Theron L, Adams A, Jansen K, Robinson E. Emergency weight estimation in Pacific Island and Maori children who are large-for-age. Emerg Med Australas. 2005: 17(3): 238-243. 43. Needlman RD. The first year. In: Behrman RE, Kliegman RM, Jensen HB, eds. Nelson textbook of pediatrics. 17th edn. Philadelphia: Saunders; 2004: 31. 44. Australian Resuscitation Council. Guideline 12.3 – Flowchart for the sequential management of life-threatening arrhythmias in infants and children [Internet]. 2006. Available from: http://www.resus.org.au 45. Hamill PV, Drizd TA, Johnson CL, Reed RB, Roche AF. NCHS growth curves for children birth-18 years: United States. Vital Health Stat 11. 1977: 165: i-iv, 1-74. 46. Pollock L, Molyneux E, Jefferis O, Bunn J. Estimating children’s weights: a developing country perspective. Arch Dis Child. 2007: 92: 938-939. 47. World Health Organization. WHO management of the child with a serious infection or severe malnutrition guidelines for first-referral level in developing countries [Internet]. Geneva: World Health Organization; 2000. Available from: http://whqlibdoc.who.int/hq/2000/WHO_FCH_CAH_00.1.pdf 48. Tinning K, Acworth J. Make your best guess: an updated method for paediatric weight estimation in emergencies. Emerg Med Australas. 2007: 19(6): 528-534. 49. Thompson MT, Reading MJ, Acworth JP. Best Guess method for age-based weight estimation in paediatric emergencies: validation and comparison with current methods. Emerg Med Australas. 2007: 19(6): 535-542. 50. Krieser D, Nguyen K, Kerr D, Jolley D, Clooney M, Kelly A. Parental weight estimation of their child’s weight is more accurate than other weight estimation methods for determining children’s weight in an emergency department? Emerg Med J. 2007: 24(11): 756-759.  83  51. Rosenberg M, Greenberger S, Rawal A, Latimer-Pierson J, Thundiyil J. Comparison of Broselow tape measurements versus physician estimations of pediatric weights. Am J Emerg Med. 2011: 29(5): 482-488. 52. Ali K, Sammy I, Nunes P. Is the APLS formula used to calculate weight-for-age applicable to a Trinidadian population? BMC Emerg Med. 2012: 12: 9. 53. Varghese A, Vasudevan VK, Lewin S, Indumathi CK, Dinakar C, Rao SD. Do the lengthbased (Broselow) tape, APLS, Argall and Nelson’s formulae accurately estimate weight of Indian children? Indian Pediatr. 2006: 43(10): 889-894. 54. Garland JS, Kishaba RG, Nelson DB, Losek JD, Sobocinski KA. A rapid and accurate method of estimating body weight. Am J Emerg Med. 1986: 4(5): 390-393. 55. Traub SL, Johnson CE. Comparison of methods of estimating creatinine clearance in children. Am J Hosp Pharm. 1980: 37(2): 195-201. 56. Traub SL, Kichen L. Estimating ideal body mass in children. Am J Hosp Pharm. 1983: 40(1): 107-110. 57. Oakley PA. Inaccuracy and delay in decision making in paediatric resuscitation, and a proposed reference chart to reduce error. BMJ. 1988: 297(6652): 817-819. 58. Lubitz DS, Seidel JS, Chameides L, Luten RC, Zaritsky AL, Campbell FW. A rapid method for estimating weight and resuscitation drug dosages from length in the pediatric age group. Annals of Emerg Med. 1988: 17(6): 576-581. 59. Bourdeau S, Copeland J, Milne WK. Accuracy of the Broselow tape in estimating the weight of First Nations children. Can J Rural Med. 2011: 16(4): 121-125. 60. Milne WK, Yasin A, Knight J, Noel D, Lubell R, Filler G. Ontario children have outgrown the Broselow tape. CJEM. 2012: 14(1): 25-30.  84  61. Sinha M, Lezine MW, Frechette A, Foster KN. Weighing the pediatric patient during trauma resuscitation and its concordance with estimated weight using Broselow Luten emergency tape. Pediatr Emerg Care. 2012: 28(6): 544-547. 62. Gauld LM, Kappers J, Carlin JB, Robertson CF. Height prediction from ulna length. Dev Med Child Neurol. 2004: 46(7): 475-480. 63. Gauld LM, Kappers J, Carlin JB, Robertson CF. Prediction of childhood pulmonary function using ulna length. Am J Respir Crit Care Med. 2004: 168(7): 804-809. 64. Spender QW, Cronk CE, Charney EB, Stallings VA. Assessment of linear growth of children with cerebral palsy: use of alternative measure to height or length. Dev Med Child Neuro. 1989: 31(2): 206-214. 65. Bland MJ, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986: 1(8476): 307-310. 66. United Nations Children’s Fund. Statistics: Botswana [Internet]. New York: UNICEF; 2009. Available from: http://www.unicef.org/infobycountry/botswana_statistics.html 67. Kabue MM, Kekitiinwa A, Maganda A, Risser JM, Chan W, Kline MW. Growth in HIVinfected children receiving antiretroviral therapy at a pediatric infectious diseases clinic in Uganda. AIDS Patient Care STDS. 2008: 22(3): 245-251. 68. Berhane R, Bagenda D, Marum L, Aceng E, Ndugwa C, Bosch RJ, et al. Growth failure as a prognostic indicator of mortality in pediatric HIV infection. Pediatrics. 1997: 100(1): e7. 69. Henderson RA, Miotti PG, Saavedra JM, Dallabetta G, Chiphangwi J, Liomba G, et al. Longitudinal growth during the first 2 years of life in children born to HIV-infected mothers in Malawi, Africa. Pediatr AIDS HIV Infect. 1996: 7(2): 91-97.  85  70. Lepage P, Msellati P, Hitimana DG, Bazubagira A, Van Goethem C, Simonon A, et al. Growth of human immunodeficiency type 1-infected and uninfected children: A prospective cohort study in Kigali, Rwanda. Pediatr Infect Dis J. 1996: 15(6): 479-485. 71. Bobat R, Coovadia H, Moodley D, Coutsoudis A, Gouws E. Growth in early childhood in a cohort of children born to HIV-1-infected women from Durban, South Africa. Ann Trop Paediatr. 2001: 21(3): 203-210. 72. Weigel R, Phiri S, Chiputula F, Gumulira J, Brinkhof M, Gsponer T, et al. Growth response to antiretroviral treatment in HIV-infected children: a cohort study from Lilongwe, Malawi. Trop Med Int Health. 2010: 15(8): 934-944. 73. Miller TL, Evans SJ, Orav EJ, Morris V, McIntosh K, Winter HS. Growth and body composition in children infected with the human immunodeficiency virus-1. Am J Clin Nutr. 1993: 57(4): 588-592. 74. Arpadi SM. Growth failure in children with HIV infection. J Acquir Immune Defic Syndr. 2000: 25(Suppl 1): S37-S42. 75. Verweel G, van Roseeum AM, Hartwig NG, Wolfs TF, Scherpbier HJ, de Groot R. Treatment with highly active antiretroviral therapy in human immunodeficiency virus type 1infected children is associated with a sustained effect on growth. Pediatrics. 2002: 109(2): e25. 76. Newell ML, Borja MC, Peckham C, European Collaborative Study. Height, weight, and growth in children born to mothers with HIV-1 infection in Europe. Pediatrics. 2003: 111(1): e52-60. 77. Kline MW, Matusa RF, Copaciu L, Calles NR, Kline NE, Schwarzwald HL. Comprehensive pediatric human immunodeficiency virus care and treatment in Constanta, Romania:  86  Implementation of a program of highly active antiretroviral therapy in a resource-poor setting. Pediatr Infect Dis J. 2004: 23(8): 695-700. 78. Viani RM, Araneta MRG, Deville JG, Spector SA. (2004). Decrease in hospitalization and mortality rates among children with perinatally acquired HIV type 1 infection receiving highly active antiretroviral therapy. Clin Infect Dis. 2004: 39(5): 725-731. 79. Nachman SA, Lindsey JC, Moye J, Stanley KE, Johnson GM, Krogstad PA, et al. Pediatric AIDS clinical trials group 377 study team. Growth of human immunodeficiency virusinfected children receiving highly active antiretroviral therapy. Pediatr Infect Dis J. 2005: 24(4): 352-357. 80. Baleta A. Botswana reduces mother-to-child transmission of HIV. Lancet. 2010: 375(9730): 1954. 81. Chen JY, Ogwu AC, Svab P, Lockman S, Moffat HJ, Gaolathe T, et al. Antiretroviral treatment initiation among HIV-infected pregnant women with low CD4+ cell counts in Gaborone, Botswana. J Acquir Immune Defic Syndr. 2010: 54(1): 102-106. 82. United Nations Programme on HIV/AIDS. Progress report of the national response to the 2001 declaration of commitment on HIV and AIDS: Botswana country report 2010 [Internet]. Gaborone: UNAIDS. Available from: http://data.unaids.org/pub/Report/2010/botswana_2010_country_progress_report_en.pdf 83. Carrol W, Alexander J. Towards a better weight estimation in the seriously ill child – a comparison of methods. Arch Dis Child. 2001: 84 (Suppl 1): A12.  87  APPENDICES Appendix A. Participant Consent Form Information and Consent Agreement Form for the Evaluation of Potential Weight Proxies for Use in Drug Dosing Determination in Gaborone, Botswana Hello. My name is _______________________________________ and I am a (student, medical resident, research assistant) working with Botswana-Baylor Children's Clinical Center of Excellence and the University of British Columbia in Vancouver, Canada. We are asking your permission to allow us to measure your child’s weight, height, mid-upper arm circumference, and length of forearm or length of lower leg. The reason for taking these measurements is because we are determining simple and accurate substitute measures of weight for children, as weight scales are often unavailable. These measurements will tell us what dose of a drug to treat a child with. Your child’s participation is entirely voluntary and you may refuse to participate or may withdraw your child from this study at any time and without providing a reason for your decision. If you should withdraw, you have the choice not to permit the research team to use any information obtained from your child. By entering this study there is no known harm to your child. If you agree to participate, whatever information you provide will be kept strictly confidential, and will not be shared with anyone other than the survey team. Your participation is voluntary, and you can stop participating at any time and without providing a reason for your decision. If you decide not to participate it is your decision and we respect your decision. There are no consequences if you say no and your decision will not be shared with anyone, including government officials. Do you have any questions about the study? If you have any questions or concerns in the future you can contact us at (telephone number) or contact the study investigator, (name) at (e-mail address) or (mailing address). The measurements of your child will take a total of about 30 minutes to complete. In appreciation for the time this will take away from your other responsibilities we would like to remunerate you with 30 Pula (60 Pula if your child is over the age of 5). May I begin the measurements on your child now?  88  Consent: The caregiver agrees that the following has been explained to him/her: (completed by the research assistant)        The study has been explained to me. All my questions were answered. I understand that I have the right not to participate and the right to stop at any time. Yes I understand that I may refuse to participate without consequence. I am free now, and in the future, to ask any questions about the study. I have been told that my child’s personal information will be kept confidential.  No No No  Caregiver’s relation to child: Mother ____ Father ____ First degree relative most responsible for child’s care _____ The caregiver has provided verbal consent to participate in this study:  Ye  Name of Research Assistant obtaining verbal consent: ____________________________________  I hereby confirm that I have read this consent form to the caregiver, answered any questions, obtained verbal agreement for each of the above statements and have been given verbal consent for their child to participate in this study. Signed and dated: ______________________________  ________________  Signature of Research Assistant  Date  Signature of Co-Principal Investigator: ______________________________  ________________  Signature  Date  89  Appendix B. Survey Question  Module 1: CHILD AND CAREGIVER INFORMATION 1.  What is (NAME’S) birthdate?  2.  What is the (NAME’s) gender?  ________________ (DD/MM/YYYY)  0 = Male 1 = Female  3.  What is the HIV status of (NAME)?  0 = Negative for HIV (does not have HIV) 1 = Positive for HIV 2 = Unknown  4.  Is (NAME) currently taking ARVs?  0 = No 1 = Yes  5.  What is your age?  6.  How many people in total live in your household?  ______________ years  _____________ Adults _____________ Children _____________ Total (Adults + Children)  7.  What is the main reason for bringing (NAME) to the clinic today?  1 = Scheduled appointment/check-up 2 = Diarrhea 3 = Vomiting  ONLY CIRCLE THE MAIN REASON  4 = Cough 5 = Difficulty breathing 6 = Fever 7 = No appetite 8 = Other (Specify: ______________________________)  90  Appendix C. MUAC-Ulna Weight-Estimation Chart  MUAC (cm)  Ulna Length (cm) 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30  10 -2.4 -1.7 -1.0 -0.2 0.5 1.3 2.0 2.8 3.5 4.3 5.0 5.8 6.5 7.3 8.0 8.8 9.5 10.3 11.0 11.8 12.5 13.2 14.0 14.7 15.5 16.2 17.0 17.7 18.5 19.2 20.0 20.7 21.5 22.2 23.0 23.7 24.5 25.2 26.0 26.7 27.5  11 -0.9 -0.2 0.6 1.3 2.1 2.8 3.5 4.3 5.0 5.8 6.5 7.3 8.0 8.8 9.5 10.3 11.0 11.8 12.5 13.3 14.0 14.8 15.5 16.3 17.0 17.8 18.5 19.2 20.0 20.7 21.5 22.2 23.0 23.7 24.5 25.2 26.0 26.7 27.5 28.2 29.0  12 0.6 1.3 2.1 2.8 3.6 4.3 5.1 5.8 6.6 7.3 8.0 8.8 9.5 10.3 11.0 11.8 12.5 13.3 14.0 14.8 15.5 16.3 17.0 17.8 18.5 19.3 20.0 20.8 21.5 22.3 23.0 23.7 24.5 25.2 26.0 26.7 27.5 28.2 29.0 29.7 30.5  13 2.1 2.8 3.6 4.3 5.1 5.8 6.6 7.3 8.1 8.8 9.6 10.3 11.1 11.8 12.6 13.3 14.0 14.8 15.5 16.3 17.0 17.8 18.5 19.3 20.0 20.8 21.5 22.3 23.0 23.8 24.5 25.3 26.0 26.8 27.5 28.2 29.0 29.7 30.5 31.2 32.0  14 3.6 4.3 5.1 5.8 6.6 7.3 8.1 8.8 9.6 10.3 11.1 11.8 12.6 13.3 14.1 14.8 15.6 16.3 17.1 17.8 18.5 19.3 20.0 20.8 21.5 22.3 23.0 23.8 24.5 25.3 26.0 26.8 27.5 28.3 29.0 29.8 30.5 31.3 32.0 32.7 33.5  15 5.1 5.9 6.6 7.3 8.1 8.8 9.6 10.3 11.1 11.8 12.6 13.3 14.1 14.8 15.6 16.3 17.1 17.8 18.6 19.3 20.1 20.8 21.6 22.3 23.0 23.8 24.5 25.3 26.0 26.8 27.5 28.3 29.0 29.8 30.5 31.3 32.0 32.8 33.5 34.3 35.0  16 6.6 7.4 8.1 8.9 9.6 10.4 11.1 11.9 12.6 13.3 14.1 14.8 15.6 16.3 17.1 17.8 18.6 19.3 20.1 20.8 21.6 22.3 23.1 23.8 24.6 25.3 26.1 26.8 27.5 28.3 29.0 29.8 30.5 31.3 32.0 32.8 33.5 34.3 35.0 35.8 36.5  17 8.1 8.9 9.6 10.4 11.1 11.9 12.6 13.4 14.1 14.9 15.6 16.4 17.1 17.8 18.6 19.3 20.1 20.8 21.6 22.3 23.1 23.8 24.6 25.3 26.1 26.8 27.6 28.3 29.1 29.8 30.6 31.3 32.0 32.8 33.5 34.3 35.0 35.8 36.5 37.3 38.0  18 9.6 10.4 11.1 11.9 12.6 13.4 14.1 14.9 15.6 16.4 17.1 17.9 18.6 19.4 20.1 20.9 21.6 22.3 23.1 23.8 24.6 25.3 26.1 26.8 27.6 28.3 29.1 29.8 30.6 31.3 32.1 32.8 33.6 34.3 35.1 35.8 36.6 37.3 38.0 38.8 39.5  19 11.2 11.9 12.6 13.4 14.1 14.9 15.6 16.4 17.1 17.9 18.6 19.4 20.1 20.9 21.6 22.4 23.1 23.9 24.6 25.4 26.1 26.8 27.6 28.3 29.1 29.8 30.6 31.3 32.1 32.8 33.6 34.3 35.1 35.8 36.6 37.3 38.1 38.8 39.6 40.3 41.1  20 12.7 13.4 14.2 14.9 15.7 16.4 17.1 17.9 18.6 19.4 20.1 20.9 21.6 22.4 23.1 23.9 24.6 25.4 26.1 26.9 27.6 28.4 29.1 29.9 30.6 31.3 32.1 32.8 33.6 34.3 35.1 35.8 36.6 37.3 38.1 38.8 39.6 40.3 41.1 41.8 42.6  91  21 14.2 14.9 15.7 16.4 17.2 17.9 18.7 19.4 20.2 20.9 21.6 22.4 23.1 23.9 24.6 25.4 26.1 26.9 27.6 28.4 29.1 29.9 30.6 31.4 32.1 32.9 33.6 34.4 35.1 35.9 36.6 37.3 38.1 38.8 39.6 40.3 41.1 41.8 42.6 43.3 44.1  MUAC (cm)  Ulna Length (cm) 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30  21 14.2 14.9 15.7 16.4 17.2 17.9 18.7 19.4 20.2 20.9 21.6 22.4 23.1 23.9 24.6 25.4 26.1 26.9 27.6 28.4 29.1 29.9 30.6 31.4 32.1 32.9 33.6 34.4 35.1 35.9 36.6 37.3 38.1 38.8 39.6 40.3 41.1 41.8 42.6 43.3 44.1  22 15.7 16.4 17.2 17.9 18.7 19.4 20.2 20.9 21.7 22.4 23.2 23.9 24.7 25.4 26.1 26.9 27.6 28.4 29.1 29.9 30.6 31.4 32.1 32.9 33.6 34.4 35.1 35.9 36.6 37.4 38.1 38.9 39.6 40.4 41.1 41.8 42.6 43.3 44.1 44.8 45.6  23 17.2 17.9 18.7 19.4 20.2 20.9 21.7 22.4 23.2 23.9 24.7 25.4 26.2 26.9 27.7 28.4 29.2 29.9 30.7 31.4 32.1 32.9 33.6 34.4 35.1 35.9 36.6 37.4 38.1 38.9 39.6 40.4 41.1 41.9 42.6 43.4 44.1 44.9 45.6 46.3 47.1  24 18.7 19.5 20.2 20.9 21.7 22.4 23.2 23.9 24.7 25.4 26.2 26.9 27.7 28.4 29.2 29.9 30.7 31.4 32.2 32.9 33.7 34.4 35.2 35.9 36.6 37.4 38.1 38.9 39.6 40.4 41.1 41.9 42.6 43.4 44.1 44.9 45.6 46.4 47.1 47.9 48.6  25 20.2 21.0 21.7 22.5 23.2 24.0 24.7 25.4 26.2 26.9 27.7 28.4 29.2 29.9 30.7 31.4 32.2 32.9 33.7 34.4 35.2 35.9 36.7 37.4 38.2 38.9 39.7 40.4 41.1 41.9 42.6 43.4 44.1 44.9 45.6 46.4 47.1 47.9 48.6 49.4 50.1  26 21.7 22.5 23.2 24.0 24.7 25.5 26.2 27.0 27.7 28.5 29.2 30.0 30.7 31.4 32.2 32.9 33.7 34.4 35.2 35.9 36.7 37.4 38.2 38.9 39.7 40.4 41.2 41.9 42.7 43.4 44.2 44.9 45.6 46.4 47.1 47.9 48.6 49.4 50.1 50.9 51.6  27 23.2 24.0 24.7 25.5 26.2 27.0 27.7 28.5 29.2 30.0 30.7 31.5 32.2 33.0 33.7 34.5 35.2 35.9 36.7 37.4 38.2 38.9 39.7 40.4 41.2 41.9 42.7 43.4 44.2 44.9 45.7 46.4 47.2 47.9 48.7 49.4 50.1 50.9 51.6 52.4 53.1  28 24.8 25.5 26.2 27.0 27.7 28.5 29.2 30.0 30.7 31.5 32.2 33.0 33.7 34.5 35.2 36.0 36.7 37.5 38.2 39.0 39.7 40.4 41.2 41.9 42.7 43.4 44.2 44.9 45.7 46.4 47.2 47.9 48.7 49.4 50.2 50.9 51.7 52.4 53.2 53.9 54.7  29 26.3 27.0 27.8 28.5 29.3 30.0 30.7 31.5 32.2 33.0 33.7 34.5 35.2 36.0 36.7 37.5 38.2 39.0 39.7 40.5 41.2 42.0 42.7 43.5 44.2 44.9 45.7 46.4 47.2 47.9 48.7 49.4 50.2 50.9 51.7 52.4 53.2 53.9 54.7 55.4 56.2  30 27.8 28.5 29.3 30.0 30.8 31.5 32.3 33.0 33.8 34.5 35.2 36.0 36.7 37.5 38.2 39.0 39.7 40.5 41.2 42.0 42.7 43.5 44.2 45.0 45.7 46.5 47.2 48.0 48.7 49.4 50.2 50.9 51.7 52.4 53.2 53.9 54.7 55.4 56.2 56.9 57.7  92  Appendix D. MUAC-Tibia Weight-Estimation Chart  MUAC (cm)  Tibia Length (cm) 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30  13 -2.0 -1.1 -0.3 0.5 1.3 2.1 3.0 3.8 4.6 5.4 6.2 7.1 7.9 8.7 9.5 10.3 11.1 12.0 12.8 13.6 14.4 15.2 16.1 16.9 17.7 18.5 19.3 20.1 21.0 21.8 22.6 23.4 24.2 25.1 25.9 26.7 27.5 28.3 29.2 30.0 30.8  14 -1.1 -0.2 0.6 1.4 2.2 3.0 3.9 4.7 5.5 6.3 7.1 7.9 8.8 9.6 10.4 11.2 12.0 12.9 13.7 14.5 15.3 16.1 16.9 17.8 18.6 19.4 20.2 21.0 21.9 22.7 23.5 24.3 25.1 26.0 26.8 27.6 28.4 29.2 30.0 30.9 31.7  15 -0.2 0.7 1.5 2.3 3.1 3.9 4.7 5.6 6.4 7.2 8.0 8.8 9.7 10.5 11.3 12.1 12.9 13.7 14.6 15.4 16.2 17.0 17.8 18.7 19.5 20.3 21.1 21.9 22.8 23.6 24.4 25.2 26.0 26.8 27.7 28.5 29.3 30.1 30.9 31.8 32.6  16 0.7 1.5 2.4 3.2 4.0 4.8 5.6 6.5 7.3 8.1 8.9 9.7 10.5 11.4 12.2 13.0 13.8 14.6 15.5 16.3 17.1 17.9 18.7 19.5 20.4 21.2 22.0 22.8 23.6 24.5 25.3 26.1 26.9 27.7 28.6 29.4 30.2 31.0 31.8 32.6 33.5  17 1.6 2.4 3.3 4.1 4.9 5.7 6.5 7.3 8.2 9.0 9.8 10.6 11.4 12.3 13.1 13.9 14.7 15.5 16.3 17.2 18.0 18.8 19.6 20.4 21.3 22.1 22.9 23.7 24.5 25.4 26.2 27.0 27.8 28.6 29.4 30.3 31.1 31.9 32.7 33.5 34.4  18 2.5 3.3 4.1 5.0 5.8 6.6 7.4 8.2 9.1 9.9 10.7 11.5 12.3 13.1 14.0 14.8 15.6 16.4 17.2 18.1 18.9 19.7 20.5 21.3 22.2 23.0 23.8 24.6 25.4 26.2 27.1 27.9 28.7 29.5 30.3 31.2 32.0 32.8 33.6 34.4 35.2  19 3.4 4.2 5.0 5.9 6.7 7.5 8.3 9.1 9.9 10.8 11.6 12.4 13.2 14.0 14.9 15.7 16.5 17.3 18.1 19.0 19.8 20.6 21.4 22.2 23.0 23.9 24.7 25.5 26.3 27.1 28.0 28.8 29.6 30.4 31.2 32.0 32.9 33.7 34.5 35.3 36.1  20 4.3 5.1 5.9 6.7 7.6 8.4 9.2 10.0 10.8 11.7 12.5 13.3 14.1 14.9 15.8 16.6 17.4 18.2 19.0 19.8 20.7 21.5 22.3 23.1 23.9 24.8 25.6 26.4 27.2 28.0 28.8 29.7 30.5 31.3 32.1 32.9 33.8 34.6 35.4 36.2 37.0  21 5.2 6.0 6.8 7.6 8.5 9.3 10.1 10.9 11.7 12.6 13.4 14.2 15.0 15.8 16.6 17.5 18.3 19.1 19.9 20.7 21.6 22.4 23.2 24.0 24.8 25.6 26.5 27.3 28.1 28.9 29.7 30.6 31.4 32.2 33.0 33.8 34.7 35.5 36.3 37.1 37.9  22 6.1 6.9 7.7 8.5 9.4 10.2 11.0 11.8 12.6 13.4 14.3 15.1 15.9 16.7 17.5 18.4 19.2 20.0 20.8 21.6 22.4 23.3 24.1 24.9 25.7 26.5 27.4 28.2 29.0 29.8 30.6 31.4 32.3 33.1 33.9 34.7 35.5 36.4 37.2 38.0 38.8  23 7.0 7.8 8.6 9.4 10.2 11.1 11.9 12.7 13.5 14.3 15.2 16.0 16.8 17.6 18.4 19.2 20.1 20.9 21.7 22.5 23.3 24.2 25.0 25.8 26.6 27.4 28.2 29.1 29.9 30.7 31.5 32.3 33.2 34.0 34.8 35.6 36.4 37.3 38.1 38.9 39.7  24 7.9 8.7 9.5 10.3 11.1 12.0 12.8 13.6 14.4 15.2 16.0 16.9 17.7 18.5 19.3 20.1 21.0 21.8 22.6 23.4 24.2 25.0 25.9 26.7 27.5 28.3 29.1 30.0 30.8 31.6 32.4 33.2 34.1 34.9 35.7 36.5 37.3 38.1 39.0 39.8 40.6  25 8.8 9.6 10.4 11.2 12.0 12.8 13.7 14.5 15.3 16.1 16.9 17.8 18.6 19.4 20.2 21.0 21.8 22.7 23.5 24.3 25.1 25.9 26.8 27.6 28.4 29.2 30.0 30.9 31.7 32.5 33.3 34.1 34.9 35.8 36.6 37.4 38.2 39.0 39.9 40.7 41.5  93  26 9.6 10.5 11.3 12.1 12.9 13.7 14.6 15.4 16.2 17.0 17.8 18.6 19.5 20.3 21.1 21.9 22.7 23.6 24.4 25.2 26.0 26.8 27.7 28.5 29.3 30.1 30.9 31.7 32.6 33.4 34.2 35.0 35.8 36.7 37.5 38.3 39.1 39.9 40.7 41.6 42.4  MUAC (cm)  Tibia Length (cm) 10 10.5 11 11.5 12 12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 23 23.5 24 24.5 25 25.5 26 26.5 27 27.5 28 28.5 29 29.5 30  27 10.5 11.4 12.2 13.0 13.8 14.6 15.4 16.3 17.1 17.9 18.7 19.5 20.4 21.2 22.0 22.8 23.6 24.5 25.3 26.1 26.9 27.7 28.5 29.4 30.2 31.0 31.8 32.6 33.5 34.3 35.1 35.9 36.7 37.5 38.4 39.2 40.0 40.8 41.6 42.5 43.3  28 11.4 12.2 13.1 13.9 14.7 15.5 16.3 17.2 18.0 18.8 19.6 20.4 21.3 22.1 22.9 23.7 24.5 25.3 26.2 27.0 27.8 28.6 29.4 30.3 31.1 31.9 32.7 33.5 34.3 35.2 36.0 36.8 37.6 38.4 39.3 40.1 40.9 41.7 42.5 43.3 44.2  29 12.3 13.1 14.0 14.8 15.6 16.4 17.2 18.0 18.9 19.7 20.5 21.3 22.1 23.0 23.8 24.6 25.4 26.2 27.1 27.9 28.7 29.5 30.3 31.1 32.0 32.8 33.6 34.4 35.2 36.1 36.9 37.7 38.5 39.3 40.1 41.0 41.8 42.6 43.4 44.2 45.1  30 13.2 14.0 14.8 15.7 16.5 17.3 18.1 18.9 19.8 20.6 21.4 22.2 23.0 23.9 24.7 25.5 26.3 27.1 27.9 28.8 29.6 30.4 31.2 32.0 32.9 33.7 34.5 35.3 36.1 36.9 37.8 38.6 39.4 40.2 41.0 41.9 42.7 43.5 44.3 45.1 46.0  31 14.1 14.9 15.7 16.6 17.4 18.2 19.0 19.8 20.7 21.5 22.3 23.1 23.9 24.7 25.6 26.4 27.2 28.0 28.8 29.7 30.5 31.3 32.1 32.9 33.7 34.6 35.4 36.2 37.0 37.8 38.7 39.5 40.3 41.1 41.9 42.8 43.6 44.4 45.2 46.0 46.8  32 15.0 15.8 16.6 17.5 18.3 19.1 19.9 20.7 21.5 22.4 23.2 24.0 24.8 25.6 26.5 27.3 28.1 28.9 29.7 30.5 31.4 32.2 33.0 33.8 34.6 35.5 36.3 37.1 37.9 38.7 39.6 40.4 41.2 42.0 42.8 43.6 44.5 45.3 46.1 46.9 47.7  33 15.9 16.7 17.5 18.3 19.2 20.0 20.8 21.6 22.4 23.3 24.1 24.9 25.7 26.5 27.3 28.2 29.0 29.8 30.6 31.4 32.3 33.1 33.9 34.7 35.5 36.4 37.2 38.0 38.8 39.6 40.4 41.3 42.1 42.9 43.7 44.5 45.4 46.2 47.0 47.8 48.6  34 16.8 17.6 18.4 19.2 20.1 20.9 21.7 22.5 23.3 24.1 25.0 25.8 26.6 27.4 28.2 29.1 29.9 30.7 31.5 32.3 33.2 34.0 34.8 35.6 36.4 37.2 38.1 38.9 39.7 40.5 41.3 42.2 43.0 43.8 44.6 45.4 46.2 47.1 47.9 48.7 49.5  35 17.7 18.5 19.3 20.1 20.9 21.8 22.6 23.4 24.2 25.0 25.9 26.7 27.5 28.3 29.1 29.9 30.8 31.6 32.4 33.2 34.0 34.9 35.7 36.5 37.3 38.1 39.0 39.8 40.6 41.4 42.2 43.0 43.9 44.7 45.5 46.3 47.1 48.0 48.8 49.6 50.4  36 18.6 19.4 20.2 21.0 21.8 22.7 23.5 24.3 25.1 25.9 26.7 27.6 28.4 29.2 30.0 30.8 31.7 32.5 33.3 34.1 34.9 35.8 36.6 37.4 38.2 39.0 39.8 40.7 41.5 42.3 43.1 43.9 44.8 45.6 46.4 47.2 48.0 48.8 49.7 50.5 51.3  37 19.5 20.3 21.1 21.9 22.7 23.5 24.4 25.2 26.0 26.8 27.6 28.5 29.3 30.1 30.9 31.7 32.6 33.4 34.2 35.0 35.8 36.6 37.5 38.3 39.1 39.9 40.7 41.6 42.4 43.2 44.0 44.8 45.6 46.5 47.3 48.1 48.9 49.7 50.6 51.4 52.2  38 20.3 21.2 22.0 22.8 23.6 24.4 25.3 26.1 26.9 27.7 28.5 29.4 30.2 31.0 31.8 32.6 33.4 34.3 35.1 35.9 36.7 37.5 38.4 39.2 40.0 40.8 41.6 42.4 43.3 44.1 44.9 45.7 46.5 47.4 48.2 49.0 49.8 50.6 51.5 52.3 53.1  39 21.2 22.1 22.9 23.7 24.5 25.3 26.2 27.0 27.8 28.6 29.4 30.2 31.1 31.9 32.7 33.5 34.3 35.2 36.0 36.8 37.6 38.4 39.2 40.1 40.9 41.7 42.5 43.3 44.2 45.0 45.8 46.6 47.4 48.3 49.1 49.9 50.7 51.5 52.3 53.2 54.0  94  40 22.1 23.0 23.8 24.6 25.4 26.2 27.0 27.9 28.7 29.5 30.3 31.1 32.0 32.8 33.6 34.4 35.2 36.0 36.9 37.7 38.5 39.3 40.1 41.0 41.8 42.6 43.4 44.2 45.1 45.9 46.7 47.5 48.3 49.1 50.0 50.8 51.6 52.4 53.2 54.1 54.9  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0073523/manifest

Comment

Related Items