UBC Faculty Research and Publications

Caregivers’ treatment-seeking behaviors and predictors of whether a child received an appropriate antimalarial… Kassam, Rosemin; Sekiwunga, Richard; Collins, John B; Tembe, Juliet; Liow, Eric Sep 6, 2016

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

Item Metadata


52383-12879_2016_Article_1815.pdf [ 538.07kB ]
JSON: 52383-1.0314181.json
JSON-LD: 52383-1.0314181-ld.json
RDF/XML (Pretty): 52383-1.0314181-rdf.xml
RDF/JSON: 52383-1.0314181-rdf.json
Turtle: 52383-1.0314181-turtle.txt
N-Triples: 52383-1.0314181-rdf-ntriples.txt
Original Record: 52383-1.0314181-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessCaregivers’ treatment-seeking behaviorsand predictors of whether a child receivedan appropriate antimalarial treatment:a household survey in rural UgandaRosemin Kassam1* , Richard Sekiwunga2, John B. Collins3, Juliet Tembe4 and Eric Liow1AbstractBackground: This study responds to a rural community’s concern that, despite national initiatives, malaria managementin young children falls short of national guidelines in their district. This study aimed to: (1) describe caregivers’ treatment-seeking behaviors in the rural district of Butaleja, (2) estimate the percentage of children who received an appropriateantimalarial, and (3) determine factors that maximized the likelihood of receiving an appropriate antimalarial. Appropriateantimalarial in this study is defined as having received only the Uganda’s age-specific first-line malaria treatment foruncomplicated and severe malaria during the course of the febrile illness.Methods: A household survey design was used in 2011 to interview 424 caregivers with a child aged five and underwho had fever within the two weeks preceding the survey. The survey evaluated factors that included: knowledge aboutmalaria and its treatment, management practices, decision-making, and access to artemisinin combination therapy (ACT)and information sources. Bivariate analysis, followed by logistic regression, was used to determine predictors of thelikelihood of receiving an appropriate antimalarial.Results: Home management was the most common first action, with most children requiring a subsequent action tomanage their fever. Overall, 20.9 % of children received a blood test, 68.4 % received an antimalarial, and 41.0 % receivedan ACT. But closer inspection showed that only 31.6 % received an appropriate antimalarial. These results confirm thatACT usage and receipt of an appropriate antimalarial in Butaleja remain well below the 2010/2015 target of 85 %. Whilenine survey items differentiated significantly whether a child had or had not received an appropriate antimalarial, ourlogistic regression model identified four items as independent predictors of likelihood that a child would receive anappropriate antimalarial: obtaining antimalarials from regulated outlets (OR = 14.99); keeping ACT in the home for futureuse (OR = 6.36); reporting they would select ACT given the choice (OR = 2.31); and child’s age older thanfour months (OR = 5.67).Conclusions: Few children in Butaleja received malaria treatment in accordance with national guidelines.This study highlighted the importance of engaging the full spectrum of stakeholders in the managementof malaria in young children - including licensed and unlicensed providers, caregivers, and family members.Keywords: Malaria, Treatment-seeking, Behavior, Children, Knowledge, Management, Access, Predictors, ACT,UgandaAbbreviations: ACT, Artemisinin combination therapy; AM, Antimalarial; AMFm, Affordable Medicines Facility-(Continued on next page)* Correspondence: rosemin.kassam@ubc.ca1School of Population and Public Health, Faculty of Medicine, University ofBritish Columbia, 2206 East Mall, Vancouver V6T 1Z3, BC, CanadaFull list of author information is available at the end of the article© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Kassam et al. BMC Infectious Diseases  (2016) 16:478 DOI 10.1186/s12879-016-1815-5(Continued from previous page)malaria; Apprp, Appropriate; CG, Caregiver; CI, Confidence Interval; CIHR, Canadian Institute of Health Research;Df, Degrees of freedom; Gov’t, Government; HP, trained health professional; iCCM, Integrated community casemanagement; MA, Malaria; Meds, Medicines; Mgmt, Management; OR, Odds Ratio; PHF, Public health facility;SSA, Sub-Saharan Africa; W/I, Within; WHO, World Health Organization; Y/N, Yes/no; η2, eta-squared; χ2,chi-square testBackgroundMalaria remains an important health challenge world-wide, with an estimated incidence of 219 million infec-tions and approximately 660,000 deaths reported in2010 [1]. Nearly 90 % of these deaths occurred in sub-Saharan Africa (SSA), of which 91 % were in childrenunder the age of five [1]. Children between three monthsand five years of age are most susceptible to adverseevents from malaria due to waning of their natural im-munity from maternal antibodies and an acquired im-munity that has not fully developed [2]. Among thosewho survive, many are left with persistent anemia, im-paired brain function, and/or paralysis, all of which ham-per physiological and cognitive development.Country-level data for 2010 indicate that 80 % of malariacases in the World Health Organization (WHO) Africanregion occur in just 17 countries, with Uganda beingamong the six most burdened countries [3]. Though adownward trend in the burden of malaria for childrenunder five has been observed in select regions of SSA, lim-ited regional data do not yet support this trend for Uganda[3–5]. In Uganda, where a large majority of the populationlives in high transmission areas, the social and economicburden of acute malaria and asymptomatic parasitemia onfamilies and governments is equally staggering [6]. Fam-ilies commonly incur out-of-pocket expenses when seek-ing antimalarial treatment for their children, as well as lossof income from missed work days [7].In 2004, artemisinin combination therapy (ACT) becamethe first-line malaria treatment for children older thanfour months, with artemether-lumafantrine (Coartem®) asthe first-line option and artesunate plus amodiaquine as analternative, and quinine as the second-line treatment [8, 9].The Uganda national case management guidelines wererevised in 2005 and 2006 to include the new treatmentpolicy, and provisions were made to distribute Coartem®to government and private-not-for-profit health facilitiesfor dispensing to communities free of charge [8]. Add-itionally, Uganda introduced a number of interventionsin an effort to bring diagnostics and treatment withACTs closer to the community [10]. At the time of thisstudy, the Uganda Ministry of Health approved twomajor national programs: the Integrated CommunityCase Management Program (iCCM) in mid-2010, andthe Affordable Medicines Facility – Malaria (AMFm) inSpring of 2011 [10].Despite such initiatives, national and regional data inUganda suggest that the use of first-line antimalarials inchildren under five years of age remains well below thenational 2010 and 2015 targets of 85 % [8, 11]. There is,however, limited district-level data, particularly in non-surveillance areas, to inform how closely practices withindifferent rural districts align with regional and nationaldata. Additionally, across Uganda, there is a paucity ofresearch investigating the extent to which childhoodmalaria is managed using only first-line treatments.This current study responds to a rural community’s con-cern with the under-use of effective antimalarials inchildren five and under during presumed malaria episodesin the district of Butaleja. As part of a large initiative todefine sustainable public health programs to improvemalaria case management in Butaleja, this study aimed to:(1) describe caregivers’ overall treatment-seeking behav-iors, (2) estimate the percentage of children who receivedan appropriate antimalarial, and (3) identify factors thatmaximized the likelihood of receiving an appropriate anti-malarial. Appropriate antimalarial in this study is definedas having received only the Uganda’s age-specific first-linemalaria treatment for uncomplicated and severe malariaduring the course of the febrile illness. These objectiveswere assessed across the district and within differentregions of the district.MethodsStudy designA cross-sectional household survey was interviewer-administered with caregivers of children five and underin Butaleja District, Uganda, following a recent fever epi-sode that was presumed to be malaria. The study’s field-work was conducted over the two-month period of Juneand July 2011. Ethics approval for the project had beenpreviously.SettingFormerly part of Tororo District, Butaleja was establishedas an independent district in 2005 and named after itsmain town of Butaleja where district headquarters are lo-cated [12]. Butaleja District is situated in eastern Uganda,bordered by Budaka District to the north, Mbale to theeast, Tororo to the south east, Bugiri to the south, andNamutumba to the west (Additional file 1: Figure S1)[12, 13]. Butaleja’s administrative structure consists ofKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 2 of 1510 sub-counties (mostly rural) and two town-councils(designated urban centers) [12]. At the time of this study,Butaleja had 66 parishes and 397 villages (K. Mweru, MD,written communication, April 2011). The major ethnicgroup is the Banyole tribe, representing approximately66 % − 85 % of the population, followed by the Bagweretribe at about 5 %, the Jopadhola tribe at 3 %, and otherethnicities (Basoga, Bagisu, Iteso, Baganda, Karimojong,Banyankole, Acholi, etc) making up less than 2 % [12, 13].The predominant language spoken across the district andby all tribes is Lunyole. The three largest religious groupsinclude Protestants (53 %), Muslims (30 %), and Catholics(17 %) (K. Mweru, MD, written communication, April2011). A large majority of the people are subsistencefarmers, with poverty being a district-wide phenomenon[14]. Life expectancy in 2008 was estimated at around47 years, and mortality rate for children under five in2006 was 152 per 1000 live births. Malaria was the singlehighest ranked cause of morbidity in the period 2007–2009, with about eight in every 10 persons experiencingmalaria/fever [12].The public health infrastructure in Uganda is strati-fied into four levels: by district, sub-county, parish, andvillage [15]. The following description is informed, inpart, by documents from the Regional District Office ofButaleja (K. Mweru, MD, written communication, April2011). At the highest level, there is one public hospitallocated in Busolwe Town-Council, providing outpatient,inpatient, surgical, and community health services. Atthe next lower level are 11 Health Centre IIIs located ineight sub-counties that are staffed by medical clinicalofficers, laboratory assistants, nurses, midwifes, andnursing assistants. These centres are regulated to ad-minister and dispense oral and injectable quinine, ACT,and sulfadoxine/pyrimethamine. At a still lower levelare the Health Centre IIs, which operate at the parishlevel. At the time of this study, there were 11 such cen-tres located in select parishes across seven sub-counties.Health Centers II are staffed by nurses and nursingassistants, and regulated to provide a more limitedrange of care, including administration and dispensingof oral and injectable quinine, artemisinin injection,ACT, and sulphadoxine-pyrimethamine. At the lowestlevel, Health Centres I exist as informal structures con-sisting of community volunteers (also known as villagehealth teams or community health workers - CHWs).Their role is to provide basic health services at thecommunity-level, such as the distribution of ACT, theprovision of health education advice, and referral tohigher level health centers. While it is advocated by theministry of health that every sub-county have at leastone Health Centre III and every parish have one HealthCentre II, at the time of this study the district fell shortof attaining this goal.In 2006, the Uganda National Malaria Control Pro-gram introduced a policy to provide the first-line anti-malarial ACT cost free from all levels of the healthsystem (levels I to hospitals), supplied as the Coartem®brand [16]. This formulation is composed of a fixedcombination of 20 mg of artemether and 120 mg oflumefantrine, supplied in pre-packed weight- and age-specific forms. However, at the time of this study, Coar-tem® had yet to be distributed to CHWs.Butaleja is also serviced by private outlets where otherACT brands can be purchased for a fee. The district,however, has no pharmacies, and the few licensed drugshops are mainly located in town centers and marketareas. Accordingly, unlicensed private vendors who donot have formal training in the management of malariaconstitute the large majority of private outlets in villagesacross Butaleja (R Kassam, E Liow, R Sekiwunga, Un-published work, August 2011).ParticipantsHousehold surveys were carried out with caregivers whomet the following inclusion criteria: (1) they had at leastone child five years or younger who had been febrilewithin the last two weeks, (2) they were the primary careprovider for the child (which included supervision, bath-ing, and feeding), (3) they resided within Butaleja Districtat the time of the study, (4) they spoke the common dis-trict dialect – Lunyole, (5) they agreed to participate, and(6) they willingly signed the consent form using thumb-print or written signature. Survey questions about currentpractice were asked in reference to the youngest child withfever (referred to as the “index child”). A household wasexcluded if its children had no fever within the two weeksprior to the survey, or if the child’s fever was confirmed bya qualified health professional to be unrelated to malaria.The presence of fever was determined by the caregiver,and for a large majority this was detected by touching theskin of the affected child.SamplingThe 2002 national census recorded the population ofButaleja District to be approximately 157,475, with anannual growth rate of 3.3 % [12]. Based on this census, itwas estimated that by 2010 the population would standat approximately 206,300, and be comprised of about41,240 households and 44,300 children under age five[12]. Assuming that every second household had a childof five years and under, the number of households withthe target population was estimated at 20,620. The Rao-soft sample size calculator showed that a representativesample of that number of households at a 5 % margin oferror with a 95 % confidence level required 380 house-holds to be interviewed [17].Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 3 of 15The sampling frame for the study covered all 10 sub-counties and two town-councils. Since the town-councilswere geographically situated centrally within the districtand shared similar urban characteristics, such as greateraccess to local amenities including licensed drug shopsand transportation, they were considered as a unit forsampling purposes. A purposive multi-step samplingprocess was used to ensure representation of all 10 sub-counties, varying household counts in different villages,different religious denominations, dominant tribes, andproximity to a government health center. In total, 35 dif-ferent villages were sampled from 27 of Butaleja’s 66 par-ishes. Sampling at the household level used a simplerandom process to avoid self-selection. With the help ofthe local leaders, a central point at each village level wasdetermined and each village was divided into four quad-rants using natural boundaries. Since no listing of house-holds with children five years and under was available, thefirst household in each of the quadrants was selected by adice throw, screened for eligibility, and recruited if the in-clusion and exclusion criteria were met. Subsequently,every second household was visited and assessed for eligi-bility. If no caregiver was available or if the household didnot meet the inclusion and exclusion criteria, then theimmediate next household was visited and assessed foreligibility. The process was repeated quadrant-by-quadrant until 12–13 households within each villagewere recruited. In total 424 eligible households wererecruited using this method.Research assistant training and deploymentThe survey was individually administered in the locallanguage of Lunyole by one of seven research assistantsrecruited from Butaleja District. The research assistantswere recent graduates from a Ugandan university withsome field experience and fluency in verbal and writtenEnglish and in the spoken local dialect. Prior to conduct-ing the survey, all seven research assistants participatedin a structured two-week training program consisting offace-to-face large and small group work and field-basedexercises. During the training, participants were briefedon the study’s goal and objectives, the household recruit-ment process, and the survey protocol. The purpose ofeach survey question along with its response format wasdiscussed in detail. The survey delivery process (first inEnglish, then in Lunyole) was practiced in small groupsand rehearsed in the field under the supervision of thestudy team, and each participant’s ability to elicit the ne-cessary information in an unbiased fashion was assessed.Because Lunyole is not a written language, questionsfrom the English version of the survey were translatedverbally during the training using group discussions andconsensus supported by local members of the researchteam who were fluent in both English and Lunyole.Instrument developmentA survey instrument was developed to assess caregivers’treatment-seeking behaviors for the management of malariain young children [see Additional file 2]. A literature review,existing survey instruments, measurement experts, andmalaria content experts informed the structure and contentof the survey. The final survey included 160 questionsconsisting of: (1) interviewer instructions, (2) researchassistants’ post-interview remarks, (3) household identi-fication, (4) index child’s disease presentation, (5) care-giver and household characteristics, and (6) interviewquestions. These questions addressed: (1) knowledgeabout malaria and its treatment, (2) access to ACTs, (3)management of the most recent fever episode in refer-ence to the index child, (4) malaria-related practices,(5) decision-making processes, and (6) informationsources accessed. Each question was carefully worded,its order of presentation within the survey was exam-ined, and its response format was keyed carefully to theintent of each question. In order to engage participantson the issue of malaria from the outset, personal anddemographic questions were deferred to the end of thesurvey. Responses were recorded free-form with answerchecklists keyed to the questions and a range of wordscales. For example, these included dichotomous scalessuch as: yes/no; true/false, and male/female; present/ab-sent scales such as: no problem/yes problem; and easy/dif-ficult scales for questions about various problems, ease ofaccomplishing various tasks, or ease of access to a publichealth facility. Additionally, caregivers identified variousmedications they obtained for the index child by pointingto photographs printed on laminated posters of all anti-malaria medicines, antibiotics, antihelmetics, analgesics,and antipyretics available within the district.Prior to administration, the survey was pre-tested withthe district key informants to confirm appropriateness ofthe questions and to ensure the use of culturally suitableterminology. For example, “malaria” and “omusuja” areterms frequently used interchangeably in the district torefer either to fever as a stand-alone symptom (referringto a hot body), or to a variety of illnesses with fever as acommon symptom. Similarly, “doctor” and “health profes-sional” are expressions commonly used to refer to anyonewho provides treatment. Thus, at the recommendation ofkey informants, the combined term of “omusuja gwesena”was used to ensure information collected would be aboutfever brought by mosquitos. For questions relating tohealth professionals, descriptions of premises where theyworked as well as their dress codes were also collected todistinguish between the roles of providers.Data collectionFollowing informed consent from primary caregivers,data were collected through household interviews.Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 4 of 15Quality control measures were introduced throughoutthe collection phase to ensure integrity of the data.These measures included a review of the final surveydocuments for completeness, spot checks by randomlyvisiting households and asking caregivers about theirexperiences with the survey, and cross-checks of thecompleted surveys with the audio recorded interviewsfor accuracy and completeness. On a day-to-day basis,this process was supported by the local research teamconsisting of the District Medical Health Officer - whoserved as the study’s site manager, a retired District HealthEducator - who served as the field coordinator, and a be-havioral scientist from Makerere University - who wasinvolved with the study design and oversight. ResearchAssistants were provided regular feedback and, when ne-cessary, caregivers were revisited to clarify information.Survey question data were transcribed from completedsurvey documents onto Excel spreadsheets for cleaningand verification, then subsequently transported into SPSS®21.0 for Windows for analysis.Measures and analysisThe study’s first objective was to provide an overview ofcaregivers’ treatment-seeking behavior when managing achild’s malaria episode. As such, frequencies of surveyquestion responses provided a first overview of results.The second objective of the study was to quantify thepercentage of children who received an appropriate anti-malarial vs. those who did not. Appropriate antimalarial inthis study is defined as having received only the Uganda’sage-specific first-line malaria treatment for uncomplicatedand severe malaria during the course of the febrile illness(K. Mweru, MD, written communication, April 2011) [9].For children younger than four months, this included re-ceiving quinine (oral or injectable) or artesunate (inject-able or rectal) therapy. For children four months andolder, an appropriate antimalarial was defined as receivingany ACT and/or artesunate (injectable or rectal) or quin-ine (injectable). For this study, all ACTs were consideredas legitimate first-line treatment. Simply asking a caregiverwhether the child had “received an antimalarial” allowedtoo much latitude for incorrect information, inappropriatecombinations, misidentified drugs, or simply lack ofknowledge. Therefore, an audit of all medications reportedto have been given to manage the current fever episode in-formed whether the appropriate antimalarial criteriahad been met for each individual child. In contrast,ACT usage in this study followed the definition of Finket al., with usage acknowledged if any of the medicinestaken during the febrile illness was an ACT [18]. Fre-quency of an appropriate antimalarial received was cal-culated for the whole of Butaleja District and for eachof its sub-counties and town-councils.The last objective of the study was to identify whetherthe likelihood of a child receiving an appropriate antimal-arial could be determined from caregivers’ answers to se-lect survey questions, including questions about caregivers,index child, and household demographic characteristics.Initially the focus was more on the content of the surveyquestions than their answers, with approximately 160different questions inspected for suitability for inclusion toassess whether the likelihood of a child receiving an appro-priate antimalarial could be determined from caregivers’answers to these questions. Bivariate analyses, using chi-square test (χ2) followed by eta-squared (η2), then examinedapproximately 125 questions to determine if they coulddistinguish whether or not a child received an appropriateantimalarial. During data cleaning and verification, certainfree-form or complex responses were recoded into simplerforms (Y/N; present/absent; mentioned/not mentioned;etc.) to allow more direct analysis of correlates of receivingan appropriate antimalarial or testing for differencesamong various sub-counties. Questions were retained if: χ2test demonstrated associations at or better than p < .05statistical significance, η2 of 2.0 % or more (hence account-ing for unexpectedly large fractions of variance), or therewas a strong conceptual reason for further examining spe-cific questions. Subsequent binary logistic regression testedwhether surviving questions-in-combination increasedpredictability. Examination of the data prior to analysis en-sured all predictor variables met the dichotomous level ofmeasurement and that multicollinearity would not presenta significant problem.ResultsRespondentsA total of 424 interviews were completed with primarycaregivers from households across Butaleja District,representing all 10 sub-counties and two town-councils.All caregivers who met the inclusion criteria agreed toparticipate. Complete information was provided for 399children, four surveys were partially completed becausecaregivers had to attend to other responsibilities, andone had missing demographic information. However, allsurveys were retained and analyzed for all possible in-formation that was present. Tables 1, 2 and 3 summarizecaregivers’, index children, and household characteristics.As shown in Table 1, the study sample adequatelyreflected the educational, occupational, tribal, and reli-gious diversities within the district.Caregivers’ practices and management strategiesFour themes relating to caregivers’ practice and malariamanagement strategies were evaluated, with three of thefour themes relating to the index child’s current episodeof fever: (1) first action taken by caregiver, (2) subse-quent action taken by caregiver, (3) care given to theKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 5 of 15index child over the course of the illness in the last twoweeks, and (4) households’ general practices. Additionalfile 3: Figure S2 and Additional file 4: Figure S3 illustrateselect first actions taken by caregivers, and Additionalfile 5: Figure S4 and Additional file 6: Figure S5 illustratethose taken during the course of the child’s illness.First action care for the index childDoing nothing (or only praying) was reported by veryfew caregivers (4.5 %). Almost three-quarters of indexchildren (73.6 %) received some form of action within6 h of their caregivers noticing the first symptoms and14.2 % between 6 and 24 h. Only 2.8 % waited morethan 24-h to receive any action. No information wasprovided for the remaining 5.0 % of the children. Intotal, 556 actions were provided by 424 caregivers, forTable 1 Demographic characteristics of caregiversCaregiver characteristics Mean SD Caregiver characteristics Number PercentAge (n = 406) 31 years 10 Currently Working for a Wage (n = 420)No 393 (93.6)n (%) Yes 27 (6.4)Gender (n = 420) Work Done in Last 12 Months (n = 420)Female 361 (86.0) Yes 347 (82.6)Male 59 (14.0) No 73 (17.4)Relationship to Head of Household (n = 418) Current Occupation (N = 419)Wife 310 (74.2) Peasant farmer 351 (83.8)Household head 72 (17.2) Housewife 35 (8.4)Daughter/son 15 (3.6) Petty trade/unskilled laborer 16 (3.8)Daughter-in-law 10 (2.4) Professional/shop keeper 7 (1.7)Parent 5 (1.2) Other 10 (2.4)Husband 3 (0.7) Tribe (n = 419)Sibling 3 (0.7) Banyole 312 (74.5)Relationship to Index Child (n = 424) Bagwere 55 (13.1)Mother 335 (79.0) Bagisu 18 (4.3)Father 46 (10.8) Basoga 16 (3.8)Grandparent 31 (7.3) Jopadhola 9 (2.1)Household head 9 (2.1) Itseso 4 (1.0)Aunt 3 (0.7) Other 5 (1.2)Part of the Household (n = 423) Religion (n = 424)Yes 422 (99.8) Protestant 211 (49.8)No 1 (0.2) Muslim 139 (32.8)Highest Level of Education (n = 419) Catholic 57 (13.4)None 73 (17.4) Born Again Christian 10 (2.4)Primary incomplete 228 (54.4) 7th Day Adventist 3 (0.7)Primary complete 48 (11.5) Other 4 (0.9)Secondary incomplete 54 (12.9)Secondary complete 7 (1.7)Post-secondary (technical/ University) 9 (2.1)Table 2 Demographic characteristics of index childrenIndex child characteristics Mean SDAge (n = 408) 22 months 16n (%)Gender (n = 424)Female 214 (50.5)Male 210 (49.5)Birth order (n = 423)Youngest 410 (96.9)2nd Youngest 12 (2.8)3rd Youngest 1 (0.2)Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 6 of 15an average 1.3 first actions per child. Home managementwas the most common first action reported by at least72.6 % of caregivers. Home management was defined asany care initiated from the home setting with resourceslocated within and around the household, such as homeremedy (home-stock medicines and physical supportivecare) and traditional herbs. Physical supportive care in-cluded non-pharmaceuticals, such as bathing, sponging,and giving tea or water for hydration. About three-fifths(63.9 %) of caregivers reported using only one type ofoption: medicines, supportive care, or traditional herbs;about one-third (29.5 %) used two of the three options;and a small proportion (3.1 %) used all three types ofoptions.Overall, for the first action, 26.4 % of caregivers reportedusing traditional herbs, 38.2 % used supportive care, andjust under two-thirds (62.3 %) gave some form of medi-cine. Almost half (48.8 %) of caregivers obtained medi-cines from an external source and one-third (29.9 %) usedmedicines from their home-stock. Approximately one-fifth (18.4 %) reported giving medicines from two sources(home-stock and an external source). The three mostcommon classes of medicines given in various combina-tions included antipyretics (54.7 %), antimalarials (24.3 %gave one antimalarial and 2.6 % gave two antimalarials),and antibiotics (12.7 %). Of the 114 children who receivedany antimalarial, about one-quarter (26.3 %) receivedACT, one-sixth (16.7 %) received quinine, and almostthree-fifths (58.8 %) received other antimalarials. For thefirst action, usage of a first-line antimalarial among the424 children was 15.1 %, and 7.1 % for ACT.Care over the course of the index child’s illness (within lasttwo weeks)Just over five-sixths (86.3 %) of index children required asubsequent action to manage their fever. About two-thirds (69.3 %) of caregivers reported seeking care froma trained health professional (a doctor, nurse or medicalattendant), with 72.1 % of these visits occurring at publichealth facilities. Caregivers were 2.3 times more likelyto seek care from an external source if they perceivedtheir child’s symptoms to be associated with severemalaria; although, no statistically significant relation-ship was found between caregivers’ distinction of mildTable 3 Demographic characteristics of householdsHousehold characteristics Number PercentCaregiver’s Perspective of Distance to Nearest Public Health Facility (n= 420)< ½ Mile 14 (3.3)½ to < 1 Mile 22 (5.2)1-5 Miles 308 (73.3)> 5 Miles 68 (16.2)Don’t know 8 (1.9)No. Rooms in this Household (n = 419)0 6 (1.4)1 225 (53.7)2 107 (25.5)3+ 81 (19.3)No. People Usually Sleep in this Household (n = 419)2-4 146 (34.8)5-7 202 (48.2)8+ 71 (16.9)House Structure (n = 421)Semi-permanent (Mud) 227 (53.9)Permanent (Brick) 193 (45.8)Uniport (Tin House) 1 (0.2)Type of Dwelling (n = 415)Independent 316 (76.1)Shared 85 (20.5)Muzigo 11 (2.7)Other 3 (0.7)Type of Toilet Facilities (n = 419)Traditional Pit/ Latrine 389 (92.8)Other 23 (5.5)Bush/filed/forest 7 (1.7)Type of Fuel Used (n = 420)Firewood 408 (97.1)Charcoal 11 (2.6)Electricity 1 (0.2)Transportation Owned (n = 420, Multiple Responses)Bicycle 258 (61.4)None 155 (36.9)Motorcycle 21 (5.0)Motor vehicle/Canoe/Boat 8 (1.0)Communication Owned (n = 419, Multiple Responses)Radio 265 (63.2)Mobile phone 200 (47.7)None 104 (24.8)Television (TV) 11 (2.6)Table 3 Demographic characteristics of households (Continued)Main Source of Information (n = 424, Multiple Responses)Word of mouth 312 (73.6)Radio 312 (73.6)Mobile phone 53 (12.5)None 13 (3.1)Other (TV, print media/mail/posters) 12 (2.8)Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 7 of 15versus severe malaria and whether a child received thefirst-line antimalarial.Fewer than one-quarter (20.8 %) of caregivers reportedthe child to have received a blood test when they pre-sented with the child at an external outlet, although thisstudy was unable to confirm the purpose of these tests.Of those children who were tested, about three-quarters(78.4 %) of them were carried out at public health fa-cilities (representing 32.6 % of children seen at publichealth facilities for assessment of their fever), 14.8 % at aprivate clinic, 9.1 % at a drug shop, and one was admin-istered by a CHW. Children who were reported to havereceived a blood test were no more likely to receive anappropriate antimalarial than those who had not re-ceived a test.Of the 424 children, nearly a third (31.6 %) receivedno antimalarial whatsoever, 31.6 % received an appro-priate antimalarial, and 36.8 % received subordinate an-timalarials. Whether children received an appropriateantimalarial or not, across all households surveyed,usage of ACT (41.0 %) or of an age-specific first-lineantimalarial (42.0 %) during the course of the illnessoccurred for fewer than half the children. Two categor-ies of subordinate antimalarial treatments were evident:(1) antimalarial treatments that did not include ACT(26.4 %), and (2) misuse of ACT itself (10.4 %). Exam-ples of the non-ACT antimalarial treatments (usedalone or in combination) included quinine (21.7 %),chloroquine (17.9 %), and other antimalarial such asmefloquine and sulphadoxine-pyrimethamine (1.9 %).Misuse of ACTs included administration of ACTs incombination with inferior antimalarials for children overfour months (7.3 %) or mis-prescription for infants underfour months (2.1 %).Three-quarters (75.1 %) of the 290 children who re-ceived an antimalarial did so within 24-h of presentingwith their first symptoms. While most caregivers (94.1 %)who were surveyed expressed that it was their preferenceto obtain an antimalarial from a public health facility, infact antimalarials were obtained from both public and pri-vate outlets: 61.3 % from public health facilities, 55.8 %from private sector outlets (18.5 % from private hospitals/clinics and 37.3 % from drug shops that were predomin-antly unlicensed), and 1.4 % from CHWs. Collectively, alarge majority (89.7 %) of caregivers who obtained anACT sourced it from a licensed outlet (public health fa-cility, CHW, or private hospital/clinic), with most (77.0 %)obtaining it from public health facilities. However, of allthose who visited a public health facility, only three-fifths(63.0 %) of children received an ACT and two-fifths(43.2 %) received an appropriate antimalarial. Nearly four-fifths (81.5 %) of caregivers used only one source forobtaining their antimalarial and fewer than one-fifth(18.5 %) reported using two sources.General household practiceOnly one-third (32.8 %) of caregivers believed ACT waseasy to find within their community and even fewer(16.5 %) believed ACT was affordable. It was commonpractice for caregivers to keep an antimalarial medicinein the home for future use in case their child got sick,with a little over half (55.5 %) of caregivers claiming tohave kept an antimalarial in the home over the past sixmonths. All ACTs given as part of the first action weregiven from households’ home-stock medicines. Whethercaregivers believed ACT was easy to find or was afford-able made no difference to whether the child actually re-ceived an appropriate antimalarial. However, children ofcaregivers who knew that ACT was the government’srecommended antimalarial were more likely to receivean appropriate antimalarial than those who knew noth-ing about government recommendations (41.6 % vs26.6 %, p < .002).Caregivers’ knowledgeFour themes related to knowledge were evaluated: (1)knowledge about malaria and its cause, (2) knowledgeabout the management of uncomplicated malaria, (3)knowledge about national policies regarding manage-ment of malaria, and (4) preparedness to respond ad-equately to an acute episode of malaria. Expanding on aclassification system by Bartolini and Zammarchi, allsymptoms mentioned by caregivers were individually ana-lyzed and clustered into the following five symptom-basedcategories: (1) subtle warnings signs of a child not beingwell, (2) early phase malaria symptoms, (3) classical symp-toms, (4) severe symptoms, and (5) non-malaria symp-toms [19].Knowledge about malaria – its symptoms and causeJust over half (55.7 %) of the caregivers expressed certaintyin their ability to recognize malaria. When asked whatsymptoms were associated with malaria, caregivers onaverage listed four different symptoms, and most were in-dicators of the classical and early phase symptoms associ-ated with malaria (94.8 % and 94.6 %, respectively). Amajority (80.2 %) of caregivers reported fever as a sign ofmalaria, with three-quarters (74.3 %) stating they werevery certain they would be able to diagnose fever.However, only 14.6 % of caregivers recognized chills to beassociated with malaria. While early phase symptoms col-lectively were mentioned most frequently, caregivers’knowledge about the individual symptoms was low, withonly 31.1 % mentioning diarrhea, 28.1 % malaise/weak-ness, 26.2 % vomiting, 5.0 % headache, and 3.6 % pain.Similarly, a small portion (20.3 %) mentioned symptomsassociated with severe malaria, with the three most com-mon being: inability to eat/drink/breastfeed (8.7 %), con-vulsions (5.4 %), and rapid breathing (2.8 %). There wereKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 8 of 15common misconceptions, even among those who had re-ported correct symptoms, with 60.8 % of caregivers link-ing runny nose and sneezing with malaria, and 36.1 %associating all types of cough with malaria. When asked tostate one main symptom of malaria, caregivers were gen-erally accurate in reporting an actual symptom of malaria,although these were not always aligned with what themedical community considers main (i.e., fever). Two-thirds (62.0 %) of caregivers reported legitimate indicatorsof malaria: 43.4 % listed one of the two classical symptoms(fever 40.3 %, chills 3.1 %), 13.0 % listed one of the earlyphase symptoms (diarrhea 4.2 %, malaise 4.2 % and vomit-ing 2.6 %), and 9.2 % listed subtle symptoms of malaria.However, almost one-third (38.0 %) of caregivers reportedsymptoms not clinically associated with malaria. Regard-ing the primary cause of malaria, only three-quarters(78.1 %) of the caregivers mentioned mosquito bites, 2.8 %mentioned a surrogate for mosquito bites such as environ-mental factors associated with proliferation of mosquitoes,and 19.1 % were misinformed, mentioning causes such aschange in season (6.5 %) and ingestion of certain foodsand fluids like drinking dirty water, or eating dirty, cold, orcertain types of foods (4.8 %). A small percentage (5.4 %)reported not knowing the cause.Knowledge about antimalarial medicinesMost caregivers (95.3 %) believed an antimalarial shouldbe started within 24-h of noticing fever. In terms ofmost effective medicines (which cured the best), aboutfour-sixth (68.2 %) of caregivers cited an actual antimalar-ial medicine as being the most effective, and one-quarter(23.1 %) acknowledged not knowing or citing a range ofother class of medications (8.7 %) including antibioticsand antipyretics/analgesics. Examples of antimalarialsreported included: ACT (35.1 %), quinine (17.7 %), andchloroquine (13.9 %). A similar fraction of caregivers saidthey would select for their children the very antimalarialsthey listed for the best cure if given the choice (ACT32.1 %, quinine 16.6 %, and chloroquine 15.2 %), but16.1 % mentioned they would select a non-antimalarialsuch as antipyretic/analgesic (10.4 %) and antibiotic(2.6 %), and 18.6 % stated they did not know what toselect.In terms of least effective antimalarials, 42.7 % re-ported not knowing, 24.5 % stated antipyretic/analgesic,16.3 % mentioned antimalarials other than ACT (chloro-quine 9.7 % and quinine 5.0 %), 5.2 % stated ACT, andthe rest stated a range of other drug classes includingantibiotics. As a last-choice antimalarial, only 8.1 %mentioned an ACT, one-fifth mentioned other antima-larials (chloroquine13.2 % and quinine 9.4 %), about aquarter (22.9 %) stated medicines not belonging to theantimalarial class such as antipyretics/analgesics, and therest (46.4 %) said they did not know.Knowledge about national policies and guidelinesOnly one-third (34.4 %) of caregivers reported knowingthat the recommended first-line treatment was ACT, butno one was aware that ACT was not recommended forchildren younger than four months of age. However, ahigher proportion (70.5 %) of caregivers were aware tostart an antimalarial within 24-h of the child experien-cing fever, with most (93.4 %) reporting that this expect-ation was attainable.Knowledge about where to access the best antimalarialsand advice about malariaAbout three-quarters (75.5 %) of caregivers believed thatthe best medicines were available from public health fa-cilities, pharmacies, private clinics, and CHWs. A similarportion (74.8 %) reported ACTs should be available forfree at public health facilities and from CHWs, but only32.8 % said they would be able to get an ACT from pub-lic sources if they needed it for their child. A comparablyhigh percentage (78.3 %) of caregivers identified trainedhealth providers such as doctors, nurses, and CHWs tobe the best individuals from whom to obtain adviceabout antimalarials.Assistance with critical decisionsThis section evaluated caregivers’ expressed preferencewith seeking assistance, when deciding about an anti-malarial, compared to what occurred for the index child.A large majority of caregivers (78.3 %) indicated that -given the choice - they would seek advice from a trainedhealth professional on how to treat malaria for childrenfive and under. However, what actually happened for theindex child fell slightly short of what caregivers indicatedas their preference.On average, the 290 index children received 1.3 anti-malarials over the course of their illness. Of these, 194children received only one antimalarial, 82 received twoantimalarials, and 14 received three antimalarials. Forthe first antimalarial given, fewer than three-quarters(66.2 %) of the 290 caregivers reported starting this atthe advice of trained health professionals, 21.2 % werestarted at the advice of drug shop vendors, and 7.2 % saidthey made the decision themselves. However, decisions togive an additional antimalarial commonly occurred at theadvice of health providers. A majority (85.4 %) of care-givers reported giving a second antimalarial at the adviceof trained health professionals, with few relying on drugshop vendors (6.1 %) or themselves (4.9 %). Similarly, ofthose caregivers who gave a third or fourth antimalarial, alarge portion indicated these decisions had been made bytrained health professionals (78.6 % and 100 %, respect-ively), with only 14.2 % of caregivers and 7.1 % of drugshop vendors deciding on the third antimalarial. On theday of the survey, only 15 of the 390 antimalarials wereKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 9 of 15still being given. Of the antimalarials stopped, two-fifths(44.8 %) had been stopped at the instruction of trainedhealth professionals, caregivers themselves made this deci-sion in the other two-fifth (39.4 %) of the cases, and onlyrarely were drug shop vendors (9.6 %), family members(3.7 %) or CHWs (0.3 %) involved in this decision.Accessing information sourcesInformation sources accessed by caregivers to learnabout approved malaria treatments were grouped intoone of the following three information source categories:(1) primary sources - defined as those sources that aremost likely to provide up-to-date factual informationconsistent with national guidelines and are amenable toregular updates. Examples include: public health providers(such as sensitization programs), informed communitymembers (such as CHWs and chairpersons), and newsmedia (such as newspaper, television, and radio), (2) sec-ondary sources - defined as those sources whose factualinformation can get outdated and are less amenable toregular updates, (3) hearsay – including either (a) infor-mation received from family or community members whoare not trained health providers, or (b) informationlearned on their own with no ability to name any sourcesexternal to themselves.On average, for the district, one-third (30.0 %) of care-givers reported they had learned that the “national policyrecommended ACT as the first-line treatment” from a pri-mary source, 69.1 % reported they learned this informationfrom hearsay (of which 67.7 % said “self”), and 0.9 % re-ported a secondary source. While almost all caregivers(93.4 %) reported knowing “when best to start an antimalar-ial after noticing fever”, only about half (55.2 %) mentionedlearning this from a primary source, with 44.8 % learningthis from hearsay (of which 41.0 % mentioned “self” or saidnothing). Similarly, two-thirds (59.4 %) of caregivers re-ported learning about “where ACT can be obtained fromwithin their community” from a primary source, with two-fifths (40.6 %) reporting hearing about this from hearsay (ofwhich 31.4 % mentioned “self” or said nothing).Predicting receipt of an appropriate antimalarial treatmentAs a first step, bivariate analyses identified nine surveyitems that differentiated significantly whether a childhad or had not received an appropriate antimalarial ator better than p < .05. Table 4 presents the nine predic-tors to receiving an appropriate antimalarial, plus a fewadditional items accounting for unusual fractions of vari-ance (η2 ≥ 2 %) or possessing other strong conceptualreasons for further testing. Children were most likely toreceive an appropriate antimalarial if the caregiver re-ported: being aware that ACT was the nationally recom-mended antimalarial (p = 0.002), that ACT cured thebest (p < .000), that they would choose ACT if given thechoice (p < .000), or they kept ACT in the home for fu-ture use (p < 0.000). Children were also more likely toreceive an appropriate antimalarial if the child was seenby a health professional (p < 0.000), if the child was seenat a public health facility (p < 0.000), or if antimalarialswere obtained from public health facilities, CHWs or pri-vate hospitals/clinics (p < 0.000). Other indicators includedcaregivers’ gender (p = 0.042) and child’s age (p = 0.015).Female caregivers were more likely to obtain an appropri-ate antimalarial for their child than were male caregivers,(33.1 % versus 20.3 %, p = .042), and infants four monthsand younger were distinctly disadvantaged compared toolder children who were more likely to receive an appro-priate antimalarial (7.1 % versus 34.6 %; X2 (7, 432) =13.10, p = .000). No other demographic characteristicssignificantly differentiated between whether a childhad received an appropriate antimalarial or not.Additionally, a binary logistic regression model deter-mined items that remained associated with the likeli-hood of receiving an appropriate antimalarial whenadjusted for other items. The analysis indicated the over-all model to be statistically significant (X2 (4) = 173.80,p < .000), with an 83.0 % predictive value. The results ofthe logistic model are shown in Table 4. Four surveyitems remained independent predictors of a child receiv-ing an appropriate antimalarial. Caregivers possessingthese predictors were characterized as knowledgeable andpro-active. In descending order of odds ratio (OR), theseitems included: (1) caregivers who obtained antimalarialsfrom public health facilities, CHWs, or private hospitals/clinics (OR = 14.99, p < .000), (2) caregivers who kept ACTin the home for future use (OR = 6.36, p < .000), (3) child’sage older than four months (OR = 5.67, p < .013), and (4)caregivers who reported that given the choice, they wouldselect ACT over other antimalarials (OR = 2.31, p < .000).Regional differences and receiving an appropriateantimalarial treatmentAcross the entire Butaleja District, 31.6 % of children re-ceived an appropriate antimalarial, but sharp differencesexisted among the 10 sub-counties and two town-councils – ranging from 8 % in Mazimasa in the ruralnorth-east to 52 % in Busolwe Town-Council in themore centralized sector of the district (Additional file 7:Figure S6). At either extreme, the rate of receiving anappropriate antimalarial falls far short of the govern-ment’s own policy of treatment availability, i.e., readilyavailable and free of charge in all communities.DiscussionThis study is the first to quantify caregivers’ treatment-seeking patterns for presumed malaria in children five andunder for Butaleja District. This study is also unique inquantifying the use of appropriate antimalarial treatment.Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 10 of 15Table 4 Predictors to whether a child received an appropriate antimalarial treatmentSurvey questionsa TotalrespondentsNo.Bivariate analysis Multivariate analysis (binary logistics step-wise regression)df χ2 p-value ɳ2 Step # Wald (X2) p-value OR 95 % CI % Accuracy ofpredictionReceipt of an appropriate antimalarial in Butaleja (N = 424): 32 %Community FactorsIs it easy to reach PHF 414 1 0.00 .969 0.000 – – NS – – –Child’s Personal FactorsAge (≤4 months vs Olderb) 423 7 17.34 .015 0.041 4 6.19 .013 5.67 1.44-22.23 83.0Gender (Male vs Female) 423 1 0.34 .559 0.001 – – NS – – –Birth order 423 2 2.75 .253 0.007 – – NS – – –CGs' Knowledge About AMsAwareness gov’t policy recommends (ACT vs Other) 423 1 9.59 .002 0.023 – – NS – – –Which AMs cure the best (ACT vs Other) 423 1 17.55 <.000 0.042 – – NS – – –Given the choice, which AM would you select first (ACT vs Other) 423 1 34.60 <.000 0.082 3 8.35 .004 2.31 1.31-4.07 81.4Knowing when to start an AM after noticing fever (Within 24 h vs Longerb) 423 2 0.36 .834 0.001 – – NS – – –Is CG knowledgeable about ACT being available in the community (Y/N) 416 1 0.06 .805 0.000 – – NS – – –Child’s Disease PresentationWas child seen by a health professional (Y/N) 421 1 12.82 <.000 0.030 – – NS – – –Where was the child seen by the health professional (PHF vs Other) 423 1 21.45 <.000 0.051 – – NS – – –Practice/Management FactorsWas medicine used for subsequent action (Y/N) 423 1 1.19 .275 0.003 – – NS – – –Where was AM (normally) obtained? (Regulated Outletsc vs Other) 423 1 24.25 <.000 0.057 1 71.94 <.000 14.99 8.02-28.02 71.8Were medicines kept in home over last 6 months for future use (Y/N) 416 1 2.74 .098 0.007 – – NS – – –Types of medicines kept as home remedy for future use on day of survey (ACT vs Other) 423 1 84.92 <.000 0.201 2 44.75 <.000 6.36 3.70-10.93 81.4CGs’ Personal & Demographic FactorsGender (Male vs Female) 419 1 4.12 .042 0.010 – – NS – – –Number of children ≤ 5 years (Only 1 vs Moreb) 423 5 2.81 .729 0.007 – – NS – – –aAbbreviations: antimalarial (AM), artemisinin combination therapy (ACT), caregiver (CG), government (Gov’t), public health facility (PHF), yes/no (Y/N)bFor logistic regression recoded into two categoriescRegulated outlets: community health worker, PHF, regulated private outletsKassametal.BMCInfectiousDiseases (2016) 16:478 Page11of15Overall, our study established that a substantial gap re-mains between Uganda’s national malaria policy and care-givers’ malaria management practices in Butaleja. Onlyone in five children with fever received any blood testwhatsoever, two in five received an ACT (alone or in com-bination with other antimalarials), and even fewer childrenreceived an appropriate antimalarial treatment.While a large majority of caregivers reported initiatingsome form of action within 24-h of noticing symptoms,for three out of four children this represented homemanagement. In Butaleja the use of home managementas initial care was more common than what has been re-ported nationally (72.6 % vs 42-61 %) or in other regionsof eastern Uganda (24 %) [15, 20, 21]. Similarly, the pro-portion of caregivers who used traditional herbs was alsohigher in Butaleja than elsewhere in eastern Uganda(26.4 % vs 19 %) [22]. Such practices are particularly dis-concerting given the findings by Rutebemberwa et al.,indicating the use of home management generally, andthe use of traditional herbs specifically, to be associatedwith delayed treatment-seeking at external sources [22].Over the course of the index child’s febrile illness,about half were reported to have been seen at a publichealth facility, a proportion larger than what has beenpreviously reported for eastern Uganda (range: 17–26 %)or nationally (21–24 %) [15, 22–25]. Despite this in-creased rate of visits to public facilities, ACT usageacross Butaleja was reported to be 41.0 % and receipt ofan appropriate antimalarial only 31.6 %. While treat-ment policies vary across SSA, according to the Ugandamalaria treatment guidelines, ACT constituted the first-line antimalarial for 90 % of index children in this study.ACT usage reported in our study was considerablyhigher than what has been estimated nationally prior to2011 (21–23 %), but comparable to 2011 and 2012 na-tional estimates (44 %) [15, 20, 26, 27]. None of theUgandan studies explicitly report on the prevalence ofchildren receiving only the first-line treatments, and itwas not possible to determine from these studies thepercentage of children for whom ACT was the only anti-malarial given. In this regard, our study provides valu-able insight by illustrating that an increase in ACT usagedoes not necessarily translate into receipt of an appro-priate antimalarial treatment.The emphasis on ACT usage in other studies is under-standable given that prompt treatment with ACT is cru-cial for preventing malaria deaths in young children.However, relying on ACT usage as the only indicator ofsuccess evades much needed policy discussions aroundthe use of multiple antimalarials to manage malaria inyoung children. Our study found the practice of poly-pharmacy to be common, with one in three caregivers inButaleja reporting using multiple antimalarials to man-age malaria in their children, often in combination withnon-antimalarials. This practice was confirmed in afollow-up study conducted in Butaleja, and has been re-ported in other regions of Uganda [15, 28–30]. The workby Rutebemberwa et al. found a large majority of care-givers in Iganga District preferred combination therapy,but caregivers’ views about which antimalarials to start achild on varied considerably [28]. Polypharmacy raisesseveral concerns, some of which include increased riskof adverse drug reactions, drug-drug/disease interac-tions, emergence of resistive strains, unnecessary thera-peutic intensity, and drug non-adherence [31].A further issue of concern with polypharmacy is the un-necessary expenditure incurred by households due to thesale of redundant drugs. In their study, Nabyonga-Oremet al. established that 82 % of caregivers living in mid-eastern districts, which included Butaleja, incurred out-of-pocket expenditures when seeking advice or treatment forthe management of malaria in young children, with 70 %incurring expenditures linked to acquisition of medicines[7]. For this region, expenditures related to purchase ofmedications was 16 % higher than the national average.While the cost due to drug mis-adventuring has not beenestimated for children in Uganda, studies in the westestimate such cost to be in the millions of dollars annually[32]. Future studies, therefore, need to take a broaderapproach when evaluating use of antimalarials by alsoconsidering the prevalence of appropriate antimalarialtreatment.This study examined caregivers’ responses to 160 differ-ent questions representing a variety of treatment-seekingbehaviors, to identify those that might distinguish whethera child received an appropriate antimalarial. While nine ofthese individual questions distinguished significantly usingbivariate analysis, our logistic regression model identifiedfour independent predictors of receiving an appropriateantimalarial: (1) caregivers who obtained antimalarialsfrom public health facilities, CHWs or private hospitals/clinics, (2) caregivers who kept ACT in the home for fu-ture use, (3) caregivers who reported that given the choice,they would select ACT over other antimalarials, and (4)child’s age older than four months. Collectively, the logis-tic model offered a high predictive value, with 83 % of thecases categorized correctly. While others have found anassociation between visits to public health facilities and re-ceipt of a first-line antimalarial treatment, our study foundvisits to a public health facility to be an independent pre-dictor only when an antimalarial was dispensed [23, 33].Additionally, ours is the only study to identify caregivers’preference for ACT and their practice of keeping home-stock supplies of ACT to be independent predictors ofreceiving an appropriate antimalarial. With respect todemographic characteristics, only the index child’s agewas found to be an independent predictor. The lack of as-sociation with other demographic characteristics was notKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 12 of 15surprising, considering that caregivers in Butaleja weremostly peasant farmers, thus their range of differences foreducational and socioeconomic variables may not havebeen large. National studies in Uganda have also foundtreatment for fever to vary with a child’s age, with fewerchildren under the age of one reported to have receivedan antimalarial or an antimalarial within the same or nextday compared to one to five year olds [15, 20].In rural communities such as Butaleja, where licensedprivate outlets are few and far between, and most house-holds are too poor to visit private clinics, public healthfacilities remain the only source for ACTs. Within thiscontext, it is not surprising that obtaining an antimalar-ial from a public health facility was found to be a signifi-cant predictor of receiving an appropriate antimalarial.What is noteworthy in this study is that fewer than halfof the caregivers reported visiting a public health facilityto manage their child’s febrile illness. Of those who didvisit a public facility, only 63.0 % of children received anACT. Given that the use of home-stock medicines andvisits to an unlicensed vendor remain common practice,only 43.2 % of those who visited a public health facilitycould be classified as having received an appropriateantimalarial. These findings confirm what has previouslybeen reported for management of childhood malaria inUganda: the use of public health facilities is limited,visits to a public facility does not ensure receipt of anACT, and self-management practices with sub-optimaltreatments at the household level remains a major obs-tacle to a child receiving an appropriate antimalarialtreatment [34]. Accordingly, this study proposes a needfor public health education to influence caregivers’ mal-aria management practices by educating them aboutACTS, the importance of using only the first-line treat-ments, and to know where to quickly obtain these treat-ments when a child experiences fever. However, in thecontext of a weak public health system and limited ac-cess to licensed private outlets, such initiatives alone willhave negligible impact. These findings suggest that limit-ing ACT access to only public health facilities and li-censed private outlets in rural settings such as Butalejais unlikely to improve malaria management to acceptablelevels. To improve access to ACTs across the entirepopulation spectrum, future public health initiatives inUganda will need to consider strengthening the healthsystem as a whole - including making training programsand subsidized ACTs more widely available at licensedand unlicensed private providers.LimitationsThe findings from this study need to be considered inthe context of potential limitations. First, the study re-lied on caregivers’ self-reported information. We, there-fore, recognize the potential for recall and reportingbiases to have influenced the quality of data collected.However, since caregivers were asked to focus on theiryoungest child who became sick in the previous twoweeks of the survey, and the child was present duringthe interview to help trigger details about the febrileepisode, we believe that recall bias was minimized. Simi-larly, reporting bias was minimized by assuring care-givers confidentiality and privacy during the interview.Furthermore, misclassification of treatments given toindex children was reduced by confirming all caregivers’verbal responses against laminated posters displayingphotographs of medicines commonly used for malaria inButaleja. Lastly, recruitment of the index child was basedon presumptive diagnosis rather than on confirmeddiagnosis of malaria, which could have resulted in selec-tion of children who may be suffering from diseasesother than malaria. Since fever commonly serves as aproxy for malaria in regions such as Butaleja where rapidor microscopic blood testing is rare, our selectionprocess followed real life practices. However, given thatthis study was conducted during a period of high malariatransmission, the sensitivity of fever being malaria waslikely high.ConclusionsUganda’s 2010 and 2015 targets of ensuring 85 % of chil-dren under five with fever receive malaria managementaccording to national guidelines are far from being met.In Butaleja, receipt of an appropriate antimalarial treat-ment in young children is undermined by limited accessto ACTs during acute episodes of malaria, over-relianceof households on home management, and a lack of care-giver knowledge of correct malaria treatment. Futurepublic health policies and programs, therefore, need toaddress the shortages of essential malaria commoditiesin recommended and frequently accessed treatment lo-cations. Particular emphasis should be placed on inte-grating the unlicensed private sector into standardizedcare, and on educating caregivers on the importance ofconfirmatory testing before proceeding to treatment,and on where free ACTs may be accessed within thecommunity.Additional filesAdditional file 1: Figure S1. Map of Butaleja District. (DOCX 93 kb)Additional file 2: Survey, Survey used to collect information aboutcaregivers’ malaria related treatment-seeking behaviors in Butaleja District.(PDF 1230 kb)Additional file 3: Figure S2. Types and frequency of homemanagement use: alone or in combination (n = 424). Abbreviations:management (Mgmt); medicines (Meds). (DOCX 24 kb)Additional file 4: Figure S3. Select first actions taken by caregivers(n = 424). Abbreviations: artemisinin combination therapy (ACT);Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 13 of 15antimalarial (AM); malaria (MA); management (Mgmt); trained healthprofessional (HP); within (W/I). (DOCX 64 kb)Additional file 5: Figure S4. Select actions taken by caregivers overthe course of the child’s fever episode (n = 424). Abbreviations:antimalarial (AM); public health facility (PHF); within (W/I). (DOCX 52 kb)Additional file 6: Figure S5. Antimalarial use over the course of thechild’s fever episode (n = 424). Abbreviations: appropriate (Apprp);artemisinin combination therapy (ACT); antimalarial (AM). (DOCX 24 kb)Additional file 7: Figure S6. Probability of receiving an appropriateantimalarial treatment across different regions of Butaleja District.(DOCX 69 kb)AcknowledgementsThis research was made possible by the contribution of caregivers whoshared their personal knowledge and experiences – we appreciate theirparticipation. We wish to recognize the Butaleja District officers and staff,Makerere University, and the entire Ugandan and Canadian team membersfor their support. We are specifically grateful to Dr. Kenneth Mweru (thenDistrict Health Officer for Butaleja District) for supervision of the field researchers,and to Mr. Daniel Hashasha for facilitating the initial contact with local villageleaders. We thank Ms. Cara Wirachowsky for her assistance with formatting, andDr. Gary Poole for proofreading and editorial comments.FundingThis study was part of a larger malaria exploratory study funded by theCanadian Institute of Health Research (CIHR) to define sustainablecommunity-based interventions for improving access and use of antimalarialdrugs in Uganda (CIHR Catalyst Grant). The funders had no role in study design,data collection and analysis, decision to publish, or preparation of themanuscript.Availability of data and materialsAt the time of obtaining consent, participants were assured that onlycollective descriptive data would be shared and published, no individualresponses (with or without identifiers) would be shared. Accordingly,as per our ethical responsibility, the dataset has not been made publicallyavailable. However, all descriptive data have been presented in the text,Figures, and Tables.Authors’ contributionsRK contributed to the conception and design of the study. RK, RS, and JTcontributed to acquisition of the data. RK analyzed and interpreted the datawith contribution from EL, and RS. RK drafted the manuscript article withinput from EL, JT, and RS. All authors have read and approved the finalmanuscript.Competing interestsThe authors declare that they have no competing interests.Consent for publicationNot applicable.Ethics approval and consent to participateEthics approval for the study was obtained from the Uganda NationalCouncil for Science and Technology (certificate number HS 906), SimonFraser University Office of Research Ethics (certificate number 2011 s0113),and the University of British Columbia’s Behavioral and Research Ethics Board(certificate number H10-02909). Only those caregivers who signed the consentform using thumbprint or written signature were recruited and interviewed.Author details1School of Population and Public Health, Faculty of Medicine, University ofBritish Columbia, 2206 East Mall, Vancouver V6T 1Z3, BC, Canada. 2ChildHealth and Development Centre, School of Medicine, Makerere University,P.O. Box 7062, Kampala, Uganda. 3Department of Educational Studies; Facultyof Education, University of British Columbia, Vancouver V6T 1Z4, BC, Canada.4The Islamic University, Mbale, Uganda.Received: 26 February 2016 Accepted: 1 September 2016References1. United Nations Children's Fund. Invest in the future: defeat malaria. Worldmalaria day 2013: focus on Africa. 2013. http://www.unicef.org/health/files/Malaria_brochure_18April2013.pdf. Accessed 30 Jan 2016.2. World Health Organization. Malaria: malaria in infants. 2013. http://www.who.int/malaria/areas/high_risk_groups/infants/en/. Accessed 30 Jan 2016.3. World Health Organization. World malaria report 2012: fact sheet. http://www.who.int/malaria/publications/world_malaria_report_2012/wmr2012_factsheet.pdf. Accessed 30 Jan 2016.4. Murray CJ, Rosenfeld LC, Lim SS, Andrews KG, Foreman KJ, Haring D, et al.Global malaria mortality between 1980 and 2010: a systematic analysis.Lancet. 2012;379(9814):413–31. doi:10.1016/S0140-6736(12)60034-8.5. Yeka A, Gasasira A, Mpimbaza A, Achan J, Nankabirwa J, Nsobya S, et al.Malaria in Uganda: challenges to control on the long road to elimination: I.Epidemiology and current control efforts. Acta Trop. 2012;121(3):184–95.doi:10.1016/j.actatropica.2011.03.004.6. Okello PE, Van Bortel W, Byaruhanga AM, Correwyn A, Roelants P, Talisuna A, etal. Variation in malaria transmission intensity in seven sites throughout Uganda.Am J Trop Med Hyg. 2006;75(2):219–25. http://www.ajtmh.org/content/75/2/219.full.pdf. Accessed 30 Jan 2016.7. Nabyonga-Orem J, Mugisha F, Okui AP, Musango L, Kirigia JM. Health careseeking patterns and determinants of out-of-pocket expenditure for malariafor the children under-five in Uganda. Malar J. 2013;12:175. doi:10.1186/1475-2875-12-175.8. Ministry of Health. Uganda malaria control strategic plan: 2005/06 - 2009/10.2005. http://archiverbm.rollbackmalaria.org/countryaction/nsp/uganda.pdf.Accessed 30 Jan 2016.9. Ministry of Health Uganda. Uganda Clinical Guideline 2010: NationalGuidelines for Management of Common Conditions. Kampala: The Republicof Uganda; 2010.10. Kassam R, Collins JB, Liow E, Rasool N. Narrative review of current context ofmalaria and management strategies in Uganda (Part I). Acta Trop. 2015;152:252–68. doi:10.1016/j.actatropica.2015.07.028.11. President's Malaria Initiative. President's Malaria Initiative Uganda malariaoperational plan for FY 2015. 2015. https://www.pmi.gov/docs/default-source/default-document-library/malaria-operational-plans/fy-15/fy-2015-uganda-malaria-operational-plan.pdf?sfvrsn=3. Accessed 03 Sept 2016.12. Uganda. Higher Local Government Statistical Abstract. Butaleja District.Kampala: The Republic of Uganda; 2009.13. Wikipedia. Butaleja District. http://en.wikipedia.org/wiki/Butaleja_District.Accessed 30 Jan 2016.14. Butaleja District Local Government NEMAN, United Nations DevelopmentProgramme (UNDP)/Poverty and Environment Initiatives (PEI). DistrictEnvironmental Policy - Butaleja. 2009. http://www.nemaug.org/district_policies/Butaleja_District_Environment_Policy.pdf. Accessed 30 Jan 2016.15. ACTwatch Group. Household survey, Uganda, 2012 survey report. 2013.http://www.actwatch.info/sites/default/files/content/publications/attachments/ACTwatch%20HH%20Report%20Uganda%202012.pdf. Accessed30 Jan 2016.16. Nanyunja M, Nabyonga-Orem J, Kato F, Kaggwa M, Katureebe C, Saweka J.Malaria treatment policy change and implementation: the case of Uganda.Malar Res Treat. 2011;2011:683167. doi:10.4061/2011/683167.17. Raosoft, Inc. Sample size calculator. http://www.raosoft.com/samplesize.html.Accessed 30 Jan 2016.18. Fink G, Dickens WT, Jordan M, Cohen JL. Access to subsidized ACT andmalaria treatment–evidence from the first year of the AMFm program in sixdistricts in Uganda. Health Policy Plan. 2014;29(4):517–27. doi:10.1093/heapol/czt041.19. Bartoloni A, Zammarchi L. Clinical aspects of uncomplicated and severemalaria. Mediterr J Hematol Infect Dis. 2012;4(1):e2012026. doi:10.4084/MJHID.2012.026.20. ACTwatch Group. Household survey report (baseline) Republic of Uganda03/09-04/09. 2009. http://www.actwatch.info/sites/default/files/content/publications/attachments/Uganda%20Household%20Baseline%2C%20ACTwatch%202009.pdf. Accessed 30 Jan 2016.21. Awor P, Wamani H, Bwire G, Jagoe G, Peterson S. Private sector drug shopsin integrated community case management of malaria, pneumonia, andKassam et al. BMC Infectious Diseases  (2016) 16:478 Page 14 of 15diarrhea in children in Uganda. Am J Trop Med Hyg. 2012;87 Suppl 5:92–6.doi:10.4269/ajtmh.2012.11-0791.22. Rutebemberwa E, Kadobera D, Katureebe S, Kalyango JN, Mworozi E, Pariyo G.Use of community health workers for management of malaria and pneumoniain urban and rural areas in eastern Uganda. Am J Trop Med Hyg. 2012;87Suppl 5:30–5. doi:10.4269/ajtmh.2012.11-0732.23. Littrell M, Gatakaa H, Evance I, Poyer S, Njogu J, Solomon T, et al. Monitoringfever treatment behaviour and equitable access to effective medicines in thecontext of initiatives to improve ACT access: baseline results and implicationsfor programming in six African countries. Malar J. 2011;10:327. doi:10.1186/1475-2875-10-327.24. Rutebemberwa E, Kallander K, Tomson G, Peterson S, Pariyo G. Determinantsof delay in care-seeking for febrile children in eastern Uganda. Trop Med IntHealth. 2009;14(4):472–9. doi:10.1111/j.1365-3156.2009.02237.x.25. Rutebemberwa E, Pariyo G, Peterson S, Tomson G, Kallander K. Utilizationof public or private health care providers by febrile children after user feeremoval in Uganda. Malar J. 2009;8:45. doi:10.1186/1475-2875-8-45.26. Uganda Bureau of Statistics (UBOS). Uganda demographic and healthsurvey 2011. 2012. http://www.ubos.org/onlinefiles/uploads/ubos/UDHS/UDHS2011.pdf. Accessed 30 Jan 2016.27. Uganda Bureau of Statistics (UBOS), Uganda Malaria Surveillance ProjectMolecular Laboratory, National Malaria Control Programme, ICF Macro.Uganda malaria indicator survey 2009. 2010;1-144. http://www.measuredhs.com/pubs/pdf/MIS6/MIS6.pdf. Accessed 30 Jan 2016.28. Rutebemberwa E, Nsabagasani X, Pariyo G, Tomson G, Peterson S, KallanderK. Use of drugs, perceived drug efficacy and preferred providers for febrilechildren: implications for home management of fever. Malar J. 2009;8:131.doi:10.1186/1475-2875-8-131.29. Tabuti JR. Herbal medicines used in the treatment of malaria in Budiopecounty. Uganda J Ethnopharmacol. 2008;116(1):33–42. doi:10.1016/j.jep.2007.10.036.30. Kassam R, Sekiwunga R, MacLeod D, Tembe J, Liow E. Patterns of treatment-seeking behaviors among caregivers of febrile young children: a Ugandanmultiple case study. BMC Public Health. 2016;116:160. doi:10.1186/s12889-016-2813-7.31. Manasse HR. Toward defining and applying a higher standard of qualityfor medication use in the United States. Am J Health Syst Pharm. 1995;52(4):374–9.32. Johnson JA, Bootman JL. Drug-related morbidity and mortality. A cost-of-illness model. Arch Intern Med. 1995;155(18):1949–56.33. Kemble SK, Davis JC, Nalugwa T, Njama-Meya D, Hopkins H, Dorsey G,Staedke SG. Prevention and treatment strategies used for the communitymanagement of childhood fever in Kampala. Uganda Am J Trop Med Hyg.2006;74(6):999–1007. http://www.ajtmh.org/content/74/6/999.full.pdf.Accessed 30 Jan 2016.34. Kassam R, Collins JB, Liow E, Rasool N. Caregivers’ treatment-seekingbehaviors and practices in Uganda - a systematic review (Part II).Acta Trop. 2015;152:269–81. doi:10.1016/j.actatropica.2015.07.029.•  We accept pre-submission inquiries •  Our selector tool helps you to find the most relevant journal•  We provide round the clock customer support •  Convenient online submission•  Thorough peer review•  Inclusion in PubMed and all major indexing services •  Maximum visibility for your researchSubmit your manuscript atwww.biomedcentral.com/submitSubmit your next manuscript to BioMed Central and we will help you at every step:Kassam et al. BMC Infectious Diseases  (2016) 16:478 Page 15 of 15


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items