• View in gallery
    Figure 1.

    Study site. Map demonstrating locations of residence’s for both severe malaria cases and uncomplicated malaria cases within the catchment area of Jinja Hospital. This figure appears in color at www.ajtmh.org.

  • View in gallery
    Figure 2.

    Study profile.

  • View in gallery
    Figure 3.

    Venn diagram: Distribution of the three principal severe malaria syndromes among the cases. This figure appears in color at www.ajtmh.org.

  • 1.

    Noor AM, Kinyoki DK, Mundia CW, Kabaria CW, Mutua JW, Alegana VA, Fall IS, Snow RW, 2014. The changing risk of Plasmodium falciparum malaria infection in Africa: 2000–10: a spatial and temporal analysis of transmission intensity. Lancet 383: 17391747.

    • Search Google Scholar
    • Export Citation
  • 2.

    Bhatt S et al. 2015. The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015. Nature 526: 207211.

  • 3.

    White NJ, Pukrittayakamee S, Hien TT, Faiz MA, Mokuolu OA, Dondorp AM, 2014. Malaria. Lancet 383: 723735.

  • 4.

    World Health Organization, 2015. Malaria Treatment Guidelines. Geneva, Switzerland: World Health Organization.

  • 5.

    Chotivanich K, Udomsangpetch R, Simpson JA, Newton P, Pukrittayakamee S, Looareesuwan S, White NJ, 2000. Parasite multiplication potential and the severity of falciparum malaria. J Infect Dis 181: 12061209.

    • Search Google Scholar
    • Export Citation
  • 6.

    Zoungrana A, Chou YJ, Pu C, 2014. Socioeconomic and environment determinants as predictors of severe malaria in children under 5 years of age admitted in two hospitals in Koudougou district, Burkina Faso: a cross sectional study. Acta Trop 139: 109114.

    • Search Google Scholar
    • Export Citation
  • 7.

    Ilunga-Ilunga F, Leveque A, Ngongo LO, Laokri S, Dramaix M, 2015. Treatment-seeking paths in the management of severe malaria in children under 15 years of age treated in reference hospitals of Kinshasa, Democratic Republic of Congo. Trop Med Health 43: 1119.

    • Search Google Scholar
    • Export Citation
  • 8.

    Byakika-Kibwika P, Ndeezi G, Kamya MR, 2009. Health care related factors associated with severe malaria in children in Kampala, Uganda. Afr Health Sci 9: 206210.

    • Search Google Scholar
    • Export Citation
  • 9.

    Nacher M, Singhasivanon P, Vannaphan S, Treeprasertsuk S, Phanumaphorn M, Traore B, Looareesuwan S, Gay F, 2001. Socio-economic and environmental protective/risk factors for severe malaria in Thailand. Acta Trop 78: 139146.

    • Search Google Scholar
    • Export Citation
  • 10.

    Greenwood BM, Bradley AK, Greenwood AM, Byass P, Jammeh K, Marsh K, Tulloch S, Oldfield FS, Hayes R, 1987. Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Trans R Soc Trop Med Hyg 81: 478486.

    • Search Google Scholar
    • Export Citation
  • 11.

    Genton B, al-Yaman F, Alpers MP, Mokela D, 1997. Indicators of fatal outcome in paediatric cerebral malaria: a study of 134 comatose Papua New Guinean children. Int J Epidemiol 26: 670676.

    • Search Google Scholar
    • Export Citation
  • 12.

    Luckner D, Lell B, Greve B, Lehman LG, Schmidt-Ott RJ, Matousek P, Herbich K, Schmid D, Mba R, Kremsner PG, 1998. No influence of socioeconomic factors on severe malarial anaemia, hyperparasitaemia or reinfection. Trans R Soc Trop Med Hyg 92: 478481.

    • Search Google Scholar
    • Export Citation
  • 13.

    Al-Taiar A, Jaffar S, Assabri A, Al-Habori M, Azazy A, Al-Gabri A, Al-Ganadi M, Attal B, Whitty CJ, 2008. Who develops severe malaria? Impact of access to healthcare, socio-economic and environmental factors on children in Yemen: a case-control study. Trop Med Int Health 13: 762770.

    • Search Google Scholar
    • Export Citation
  • 14.

    Rees CP, Hawkesworth S, Moore SE, Dondeh BL, Unger SA, 2016. Factors affecting access to healthcare: an observational study of children under 5 years of age presenting to a Rural Gambian Primary Healthcare Centre. PLoS One 11: e0157790.

    • Search Google Scholar
    • Export Citation
  • 15.

    Romay-Barja M, Cano J, Ncogo P, Nseng G, Santana-Morales MA, Valladares B, Riloha M, Benito A, 2016. Determinants of delay in malaria care-seeking behaviour for children 15 years and under in Bata district, Equatorial Guinea. Malar J 15: 187.

    • Search Google Scholar
    • Export Citation
  • 16.

    Chuma J, Okungu V, Molyneux C, 2010. Barriers to prompt and effective malaria treatment among the poorest population in Kenya. Malar J 9: 144.

  • 17.

    Chukwuocha UM, Okpanma AC, Nwakwuo GC, Dozie IN, 2014. Determinants of delay in seeking malaria treatment for children under-five years in parts of south eastern Nigeria. J Community Health 39: 11711178.

    • Search Google Scholar
    • Export Citation
  • 18.

    Getahun A, Deribe K, Deribew A, 2010. Determinants of delay in malaria treatment-seeking behaviour for under-five children in south-west Ethiopia: a case control study. Malar J 9: 320.

    • Search Google Scholar
    • Export Citation
  • 19.

    Ahorlu CK, Koram KA, Ahorlu C, de Savigny D, Weiss MG, 2006. Socio-cultural determinants of treatment delay for childhood malaria in southern Ghana. Trop Med Int Health 11: 10221031.

    • Search Google Scholar
    • Export Citation
  • 20.

    Ansah EK, Gyapong M, Narh-Bana S, Bart-Plange C, Whitty CJ, 2016. Factors influencing choice of care-seeking for acute fever comparing private chemical shops with health centres and hospitals in Ghana: a study using case-control methodology. Malar J 15: 290.

    • Search Google Scholar
    • Export Citation
  • 21.

    Carme B, Plassart H, Senga P, Nzingoula S, 1994. Cerebral malaria in African children: socioeconomic risk factors in Brazzaville, Congo. Am J Trop Med Hyg 50: 131136.

    • Search Google Scholar
    • Export Citation
  • 22.

    Mulumba MP, Wery M, Ngimbi NN, Paluku K, Van der Stuyft P, De Muynck A, 1990. Childhood malaria in Kinshasa (Zaire). Influence of seasons, age, environment, and family social conditions. Med Trop (Mars) 50: 5364.

    • Search Google Scholar
    • Export Citation
  • 23.

    Safeukui-Noubissi I et al. 2004. Risk factors for severe malaria in Bamako, Mali: a matched case-control study. Microbes Infect 6: 572578.

  • 24.

    Koram KA, Bennett S, Adiamah JH, Greenwood BM, 1995. Socio-economic determinants are not major risk factors for severe malaria in Gambian children. Trans R Soc Trop Med Hyg 89: 151154.

    • Search Google Scholar
    • Export Citation
  • 25.

    Ossou-Nguiet PM, Okoko AR, Ekouya Bowassa G, Oko AP, Mabiala-Babela JR, Ndjobo Mamadoud IC, Moyen G, 2013. Determinants of cerebral malaria in Congolese children. Rev Neurol (Paris) 169: 510514.

    • Search Google Scholar
    • Export Citation
  • 26.

    Bennett S, Koram KA, Greenwood BM, 1999. Risk factors for severe malaria: importance of careful study design. Trans R Soc Trop Med Hyg 93: 679.

    • Search Google Scholar
    • Export Citation
  • 27.

    Carme B, Rogier C, Trape JF, 2000. Risk factors for severe malaria: importance of careful study design: a reply. Trans R Soc Trop Med Hyg 94: 89.

    • Search Google Scholar
    • Export Citation
  • 28.

    Nacher M, Singhasivanon P, Gay F, Silachamroon U, Looareesuwan S, 2001. Case-control studies on host factors in severe malaria. Trends Parasitol 17: 253254.

    • Search Google Scholar
    • Export Citation
  • 29.

    Driss A, Hibbert JM, Wilson NO, Iqbal SA, Adamkiewicz TV, Stiles JK, 2011. Genetic polymorphisms linked to susceptibility to malaria. Malar J 10: 271.

  • 30.

    Cunnington AJ, Walther M, Riley EM, 2013. Piecing together the puzzle of severe malaria. Sci Transl Med 5: 211ps18.

  • 31.

    Uganda Bureau of Statistics (UBOS) and ICF International, 2015. Uganda Malaria Indicator Survey 2014–15. Kampala, Uganda, and Rockville, MD: UBOS and ICF International.

  • 32.

    Uganda Bureau of Statistics (UBOS), 2014. Uganda National Household Survey 2012/2013. Kampala, Uganda: UBOS.

  • 33.

    Marsh K et al. 1995. Indicators of life-threatening malaria in African children. N Engl J Med 332: 13991404.

  • 34.

    WHO, 2014. Severe malaria. Trop Med Int Health 19 (Suppl 1): 7131.

  • 35.

    Talisuna AO, Noor AM, Mundia CW, Otieno V, Mitto B, Amratia P, Robert WS, 2013. An Epidemiological Profile of Malaria and its Control in Uganda. Analysis and report funded by Roll Back Malaria and Department for International Development. UK.

  • 36.

    Li H, Yuan Z, Su P, Wang T, Yu Y, Sun X, Xue F, 2016. A simulation study on matched case-control designs in the perspective of causal diagrams. BMC Med Res Methodol 16: 102.

    • Search Google Scholar
    • Export Citation
  • 37.

    Vyas S, Kumaranayake L, 2006. Constructing socio-economic status indices: how to use principal components analysis. Health Policy Plan 21: 459468.

    • Search Google Scholar
    • Export Citation
  • 38.

    Bursac Z, Gauss CH, Williams DK, Hosmer DW, 2008. Purposeful selection of variables in logistic regression. Source Code Biol Med 3: 17.

  • 39.

    World Health Organization, 2015. Country Profiles. Available at: http://www.who.int/malaria/publications/country-profile_uga.en.pdf. Accessed June 21, 2017.

  • 40.

    van der Laan MJ, 2008. Estimation based on case-control designs with known prevalence probability. Int J Biostat 4: 17.

  • 41.

    Jiang Y, Scott AJ, Wild CJ, 2006. Secondary analysis of case-control data. Stat Med 25: 13231339.

  • 42.

    Sommerfelt H, Steinsland H, van der Merwe L, Blackwelder WC, Nasrin D, Farag TH, Kotloff KL, Levine MM, Gjessing HK, 2012. Case/control studies with follow-up: Constructing the source population to estimate effects of risk factors on development, disease, and survival. Clin Infect Dis 55 (Suppl 4): S262S270.

    • Search Google Scholar
    • Export Citation
  • 43.

    Feikin DR, Nguyen LM, Adazu K, Ombok M, Audi A, Slutsker L, Lindblade KA, 2009. The impact of distance of residence from a peripheral health facility on pediatric health utilisation in rural western Kenya. Trop Med Int Health 14: 5461.

    • Search Google Scholar
    • Export Citation
  • 44.

    Rutebemberwa E, Pariyo G, Peterson S, Tomson G, Kallander K, 2009. Utilization of public or private health care providers by febrile children after user fee removal in Uganda. Malar J 8: 45.

    • Search Google Scholar
    • Export Citation
  • 45.

    Awor P, Wamani H, Tylleskar T, Peterson S, 2015. Drug seller adherence to clinical protocols with integrated management of malaria, pneumonia and diarrhoea at drug shops in Uganda. Malar J 14: 277.

    • Search Google Scholar
    • Export Citation
  • 46.

    Basu S, Andrews J, Kishore S, Panjabi R, Stuckler D, 2012. Comparative performance of private and public healthcare systems in low- and middle-income countries: a systematic review. PLoS Med 9: e1001244.

    • Search Google Scholar
    • Export Citation
  • 47.

    Rutebemberwa E, Kallander K, Tomson G, Peterson S, Pariyo G, 2009. Determinants of delay in care-seeking for febrile children in eastern Uganda. Trop Med Int Health 14: 472479.

    • Search Google Scholar
    • Export Citation
  • 48.

    Rutebemberwa E, Buregyeya E, Lal S, Clarke SE, Hansen KS, Magnussen P, LaRussa P, Mbonye AK, 2016. Assessing the potential of rural and urban private facilities in implementing child health interventions in Mukono district, central Uganda-a cross sectional study. BMC Health Serv Res 16: 268.

    • Search Google Scholar
    • Export Citation
  • 49.

    Liu J, Prach LM, Treleaven E, Hansen M, Anyanti J, Jagha T, Seaman V, Ajumobi O, Isiguzo C, 2016. The role of drug vendors in improving basic health-care services in Nigeria. Bull World Health Organ 94: 267275.

    • Search Google Scholar
    • Export Citation
  • 50.

    Awor P, Wamani H, Bwire G, Jagoe G, Peterson S, 2012. Private sector drug shops in integrated community case management of malaria, pneumonia, and diarrhea in children in Uganda. Am J Trop Med Hyg 87: 9296.

    • Search Google Scholar
    • Export Citation
  • 51.

    Hetzel MW, Dillip A, Lengeler C, Obrist B, Msechu JJ, Makemba AM, Mshana C, Schulze A, Mshinda H, 2008. Malaria treatment in the retail sector: knowledge and practices of drug sellers in rural Tanzania. BMC Public Health 8: 157.

    • Search Google Scholar
    • Export Citation
  • 52.

    Kioko U, Riley C, Dellicour S, Were V, Ouma P, Gutman J, Kariuki S, Omar A, Desai M, Buff AM, 2016. A cross-sectional study of the availability and price of anti-malarial medicines and malaria rapid diagnostic tests in private sector retail drug outlets in rural western Kenya, 2013. Malar J 15: 359.

    • Search Google Scholar
    • Export Citation
  • 53.

    Sudhinaraset M, Ingram M, Lofthouse HK, Montagu D, 2013. What is the role of informal healthcare providers in developing countries? A systematic review. PLoS One 8: e54978.

    • Search Google Scholar
    • Export Citation
  • 54.

    Opiyo N, Yamey G, Garner P, 2016. Subsidising artemisinin-based combination therapy in the private retail sector. Cochrane Database Syst Rev 3: CD009926.

    • Search Google Scholar
    • Export Citation
  • 55.

    Awor P, Wamani H, Tylleskar T, Jagoe G, Peterson S, 2014. Increased access to care and appropriateness of treatment at private sector drug shops with integrated management of malaria, pneumonia and diarrhoea: a quasi-experimental study in Uganda. PLoS One 9: e115440.

    • Search Google Scholar
    • Export Citation
  • 56.

    Rutta E, Liana J, Embrey M, Johnson K, Kimatta S, Valimba R, Lieber R, Shekalaghe E, Sillo H, 2015. Accrediting retail drug shops to strengthen Tanzania’s public health system: an ADDO case study. J Pharm Policy Pract 8: 23.

    • Search Google Scholar
    • Export Citation
  • 57.

    Rutta E et al. 2011. Increasing access to subsidized artemisinin-based combination therapy through accredited drug dispensing outlets in Tanzania. Health Res Policy Syst 9: 22.

    • Search Google Scholar
    • Export Citation
  • 58.

    Wafula FN, Miriti EM, Goodman CA, 2012. Examining characteristics, knowledge and regulatory practices of specialized drug shops in Sub-Saharan Africa: a systematic review of the literature. BMC Health Serv Res 12: 223.

    • Search Google Scholar
    • Export Citation
  • 59.

    Embrey M et al. 2016. Understanding the role of accredited drug dispensing outlets in Tanzania’s Health System. PLoS One 11: e0164332.

  • 60.

    Ansah EK, Whitty CJ, Bart-Plange C, Gyapong M, 2016. Changes in the availability and affordability of subsidised artemisinin combination therapy in the private drug retail sector in rural Ghana: before and after the introduction of the AMFm subsidy. Int Health 8: 427432.

    • Search Google Scholar
    • Export Citation
  • 61.

    Ewing VL, Lalloo DG, Phiri KS, Roca-Feltrer A, Mangham LJ, SanJoaquin MA, 2011. Seasonal and geographic differences in treatment-seeking and household cost of febrile illness among children in Malawi. Malar J 10: 32.

    • Search Google Scholar
    • Export Citation
  • 62.

    Uganda Bureau of Statistics, 2016. The National Population and Housing Census 2014 – Main Report. Kampala, Uganda: Uganda Bureau of Statistics.

  • 63.

    Uggla C, Mace R, 2016. Parental investment in child health in sub-Saharan Africa: a cross-national study of health-seeking behaviour. R Soc Open Sci 3: 150460.

    • Search Google Scholar
    • Export Citation
  • 64.

    Tusting LS et al. 2016. Why is malaria associated with poverty? Findings from a cohort study in rural Uganda. Infect Dis Poverty 5: 78.

  • 65.

    Ewing VL, Tolhurst R, Kapinda A, Richards E, Terlouw DJ, Lalloo DG, 2016. Increasing understanding of the relationship between geographic access and gendered decision-making power for treatment-seeking for febrile children in the Chikwawa district of Malawi. Malar J 15: 521.

    • Search Google Scholar
    • Export Citation
  • 66.

    UNICEF, Liverpool School of Tropical Medicine, 2011. Gender Influences on Child Survival, Health and Nutrition: A Narrative Review. New York, NY: UNICEF.

  • 67.

    Tolhurst R, Amekudzi YP, Nyonator FK, Bertel Squire S, Theobald S, 2008. “He will ask why the child gets sick so often”: the gendered dynamics of intra-household bargaining over healthcare for children with fever in the Volta Region of Ghana. Soc Sci Med 66: 11061117.

    • Search Google Scholar
    • Export Citation
  • 68.

    Maitra P, 2004. Parental bargaining, health inputs and child mortality in India. J Health Econ 23: 259291.

Past two years Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 565 251 28
PDF Downloads 312 121 19
 
 
 
 
 
 
 
 
 
 
 
 
 
 

 

 

 

Demographic, Socioeconomic, and Geographic Factors Leading to Severe Malaria and Delayed Care Seeking in Ugandan Children: A Case–Control Study

Arthur MpimbazaChild Health & Development Centre, Makerere University-College of Health Sciences, Kampala, Uganda;

Search for other papers by Arthur Mpimbaza in
Current site
Google Scholar
PubMed
Close
,
Grace NdeeziDepartment of Pediatrics & Child Health, Makerere University-College of Health Sciences, Kampala, Uganda;

Search for other papers by Grace Ndeezi in
Current site
Google Scholar
PubMed
Close
,
Anne KatahoireChild Health & Development Centre, Makerere University-College of Health Sciences, Kampala, Uganda;

Search for other papers by Anne Katahoire in
Current site
Google Scholar
PubMed
Close
,
Philip J. RosenthalDepartment of Medicine, University of California, San Francisco, California;

Search for other papers by Philip J. Rosenthal in
Current site
Google Scholar
PubMed
Close
, and
Charles KaramagiDepartment of Pediatrics & Child Health, Makerere University-College of Health Sciences, Kampala, Uganda;
Clinical Epidemiology Unit, Department of Medicine, Makerere University-College of Health Sciences, Kampala, Uganda

Search for other papers by Charles Karamagi in
Current site
Google Scholar
PubMed
Close

We studied associations between delayed care seeking, demographic, socioeconomic, and geographic factors and likelihood of severe malaria in Ugandan children. The study was based at Jinja Hospital, Uganda. We enrolled 325 severe malaria cases and 325 uncomplicated malaria controls matched by age and residence. Patient details, an itinerary of events in response to illness, household information, and location of participants’ residences were captured. Conditional logistic regression was used to determine risk factors for severe malaria and delayed care seeking. Delayed care seeking (≥ 24 hours after fever onset; odds ratio [OR] 5.50; 95% confidence interval [CI] 2.70, 11.1), seeking care at a drug shop as the initial response to illness (OR 3.62; 95% CI 1.86, 7.03), and increasing distance from place of residence to the nearest health center (OR 1.45; 95% CI 1.17, 1.79) were independent risk factors for severe malaria. On subgroup analysis, delayed care seeking was a significant risk factor in children with severe malaria attributable to severe anemia (OR 15.6; 95% CI 3.02, 80.6), but not unconsciousness (OR 1.13; 95% CI 0.30, 4.28). Seeking care at a drug shop (OR 2.84; 95% CI 1.12, 7.21) and increasing distance to the nearest health center (OR 1.18; 95% CI 1.01, 1.37) were independent risk factors for delayed care seeking. Delayed care seeking and seeking care at a drug shop were risk factors for severe malaria. Seeking care at a drug shop was also a predictor of delayed care seeking. The role of drug shops in contributing to delayed care and risk of severe malaria requires further study.

INTRODUCTION

Despite reported declines in malaria mortality, the burden of infection remains high, particularly in sub-Saharan Africa, where about 90% of worldwide morbidity and mortality occurs.1,2 Of five Plasmodium species that cause human infection, Plasmodium falciparum is the most virulent and is responsible for the large majority of infections in sub-Saharan Africa.3 Infection with P. falciparum results in one of three possible outcomes: asymptomatic parasitemia, defined as the presence of asexual parasites in the blood without symptoms; uncomplicated malaria, which entails febrile illness not associated with signs of severe disease; and severe malaria, characterized by various syndromes of organ dysfunction, which if not treated promptly may result in death.4 In children, severe malaria most commonly manifests as either severe anemia or cerebral malaria.3

Prompt access to appropriate antimalarial treatment, ideally within the first 24 hours after fever onset, is promoted by the World Health Organization (WHO) for prevention of progression from uncomplicated to severe malaria.4 This strategy is logical as progression to severe disease requires extensive parasite multiplication over time.5 Support for delayed care seeking as a risk factor for severe malaria is largely based on results from cross-sectional studies demonstrating association between longer duration of symptoms and severe malaria69 and demonstration of rapid progression of uncomplicated to severe malaria.10,11 However, some studies have not shown an association between duration of symptoms and risk of severe malaria.12,13

Demographic, socioeconomic, and behavioral factors and distance from residence to health facilities are known determinants of delayed seeking of appropriate care among children with malaria.1420 However, their contribution to severe malaria is unclear, with available studies yielding conflicting results.6,12,13,2125 Variations in findings may be attributable to differences in study design and analysis, varied impacts of complex host–parasite interactions that affect disease presentation and small sample sizes.2630

The association between delayed care seeking as specifically defined by the WHO (delay ≥ 24 hours after fever onset) and progression from uncomplicated to severe malaria remains undetermined. In addition, the predictors of delayed care seeking in children with malaria have not been well elucidated. The primary objective of this study was to determine risk factors of severe malaria, including delayed care seeking from public facilities. In addition, we assessed predictors of delayed care seeking in Ugandan children.

METHODS

Ethics statement.

Informed consent was obtained from the parents or guardians of all study participants. The study protocol was approved by the Uganda National Council of Science and Technology and the Institutional Review Boards of the School of Medicine, Makerere University-College of Health Sciences, and the University of California, San Francisco.

Study design.

A matched case–control design was used to compare demographic, socioeconomic, family-related, and geographic factors between cases and controls. Incident cases were children with severe malaria, and controls were children with uncomplicated malaria. Each case was matched to a control by age and subcounty of residence. After enrollment, information on events that took place in response to the child’s illness, from the onset of illness to the time when the child was enrolled, were systematically obtained and recorded, facilitating quantification of time to event data for two retrospective cohorts; case cohort and control cohort.

Site.

The study was based at the children’s ward of Jinja Hospital, Jinja District, Uganda. It is the largest hospital in eastern Uganda, serving as a regional referral hospital with a catchment area encompassing 12 districts located in the east central (ten) and central (two) regions of the country from where controls were enrolled (Figure 1). The catchment area is considered a region of moderate to high malaria transmission intensity.31 Nearly half of the population in the east central and central regions is in the lowest wealth bracket of the region and residents are engaged predominantly in subsistence agriculture.32

Figure 1.
Figure 1.

Study site. Map demonstrating locations of residence’s for both severe malaria cases and uncomplicated malaria cases within the catchment area of Jinja Hospital. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 5; 10.4269/ajtmh.17-0056

Matching.

Considering the possible confounding effect of age on associations between risk factors and severe malaria, controls were matched to cases by age. Targeted age groupings for matching were 6 to < 12 months, 1 to < 3 years, 3 to < 5 years, and ≥ 5 years. In addition, controls were matched to cases by location of residence (within a subcounty level) and calendar time (enrollment within one month of case.) We believe that the latter two matching criteria resulted in selection of controls with an exposure distribution identical to that of the population that gave rise to the cases, as evidenced by the similarity in characteristics of the uncomplicated malaria controls and neighborhood controls selected from within the same village as cases (Supplemental Table 1).

Severe malaria cases: definition and selection.

Cases were defined as hospitalized children aged 4 months to 10 years with a positive malaria blood smear and one of three severe malaria syndromes: 1) severe anemia (Hb < 5 g/dl; measured by HemoCue AB), 2) impaired consciousness (Blantyre coma score < 4),33 or 3) respiratory distress (intercostal or subcostal recession without crackles or other evidence of pneumonia on auscultation). When possible, comatose children (Blantyre coma score < 3) had a lumbar puncture performed to exclude children with bacterial or viral meningitis, and children with clinical evidence of a specific nonmalarial febrile illness or chronic disease were excluded. Cases were consecutively sampled as they presented to Jinja Hospital during the study period.

Uncomplicated malaria controls: definition and selection.

Uncomplicated malaria controls were children with fever and a blood smear positive for P. falciparum parasites, but without clinical evidence of severe malaria, danger signs,34 or other known causes of febrile illness. Children with history of a blood transfusion in the past 3 months were excluded. Eligible children were consecutively enrolled from a level III or IV health center located in the same geographical area (subcounty) as the matched case.

Data collection.

For both cases and controls information was collected using three questionnaires. The first questionnaire detailed patient demographics, history of current illness, physical examination findings, and initial laboratory findings. The second questionnaire detailed events during the course of illness before enrollment. An event consisted of reported symptoms, action taken, and intervention(s) given in response to illness. Date and time of occurrence of new symptoms, action taken, and intervention(s) were documented based on the respondent’s approximation. For each action, the decision-maker was documented. If an antimalarial was given, frequency and duration of treatment were recorded. Dosing was categorized as either appropriate or inappropriate, based on the WHO treatment guidelines.4 If the action was “sought care at a health center,” distance from home, time taken to reach the center, type of center, timing, and type of services offered was recorded in the order of events. Date and time of occurrence of two critical events were determined using probing interview techniques: first, when fever was first noted (onset of illness), and second, when the child arrived at the health center where the child was enrolled (time of seeking appropriate care). The third questionnaire was used to collect information concerning the child’s caregiver and head of household (relation to child, age, education level, religion, and employment status). House characteristics (building materials, roofing material, water source, toilet type, lighting, and cooking materials) and possessions were documented and used to construct a wealth index for each child.

Geographical coordinates.

Garmin Handheld Global Positioning System (GPS) Navigators (GARMIN eTrex Legend H) were used to record coordinates of participants’ homes and health centers where uncomplicated malaria cases were recruited. GPS coordinates of towns and public health facilities in Uganda were retrieved from external data sources.35 The GPS coordinates were measured using the NAVSTAR GPS Satellite Network System, and accuracy was enhanced using the Wide Area Augmentation System (http://www8.garmin.com/aboutGPS/waas.html). Waypoint data were imported from the GPS navigators and stored in Garmin Base Camp software as GPX files. Data were exported from Garmin Base Camp to QGIS (QGIS Development Team, 2009 QGIS Geographic Information System; http://www.qgis) for data viewing, editing, analysis, and projection of maps.

Sample size estimation.

Sample size was predetermined based on an existing study of genetic polymorphisms and malaria, powered to detect an association between α-thalassemia and severe malaria.

With the population prevalence of the α-thalassemia trait estimated at 44% and allowing for 5% of patients being excluded from analysis after enrollment because of false positive results, we required a total sample size of 650 children (325 children in each arm) to detect a minimum odds ratio (OR) of 0.5 for risk of severe malaria among children with α-thalassemia, assuming power of 80%, alpha of 0.05, and 35% prevalence of α-thalassemia in controls. With 325 cases and 325 controls, the study had 90% power to detect an OR of two in disease exposure, assuming 46% prevalence of delayed seeking of care among controls, with a 95% level of confidence.

Analysis.

All data were entered using Microsoft Access (Microsoft Corporation) and analyzed using STATA (version 14; STATA Corp., College Station, TX). Time taken to seek appropriate care was computed based on the difference between time when fever was first noted and time of arrival at the health center where the child was enrolled. Delayed care seeking was defined as care seeking ≥ 24 hours after fever onset at the health center where the child was enrolled and provided with appropriate antimalarial treatment. Descriptive statistics were reported as proportions and medians with interquartile ranges. Distributions of baseline characteristics in either study group were compared using the chi-squared test and the Wilcoxon signed rank test for categorical and continuous data, respectively. We used conditional logistic regression, a matched analytical approach, as our statistical method for identification of risk factors for severe malaria. Unmatched analysis on matched data (even when incomplete) compromises the validity of effect estimates; even though estimates may be more precise.36 For purposes of comparison, we also ran other models based on restricted conditional logistic regression (limited to successfully age-matched case–control pairs; equivalent to 74% of case–control pairs) and unconditional logistic regression, with adjustment for age. In the unadjusted analysis, univariable conditional logistic regression models and the Mantel–Haenszel matched pairs test were used to measure the association of continuous and categorical variables between cases and controls, respectively. Principal component analysis was used to create a wealth index from nine variables: ownership of mobile telephones, radios, clocks, cupboards, sofas, and tables; number of rooms in the house; number of meals eaten per day; access to an improved toilet; and roofing, wall, and floor materials. The first component accounted for 25.8% of the total variation, a value within the range of other studies.37 Suitability of the nine variables in performing factorial analysis was confirmed using Bartlett’s test of sphericity (P < 0.001) and the Kaiser–Meyer–Olkin Measure of Sampling Adequacy (0.787).

The purposeful selection approach recommended by Hosmer and Lemeshow was used to select variables for inclusion in the final logistic regression model.38 In brief, variables that demonstrated a significant (P < 0.25) association with severe malaria risk in the unadjusted analysis were included in a preliminary effects model. For the final effects model, variables were removed if they were nonsignificant (P > 0.15), did not contribute to collinearity, were not confounders, and did not impact on model fitness (see Supplemental Table 2). Age was included in the final model to account for residual confounding attributed to incomplete matching and possible selection bias introduced by matching on age. Household building materials were not included as explanatory variables in the final effects model because of “mullticollinearity” concerns with the wealth index, which was constructed based on building materials. Potential confounders were assessed using the Chest Command in Stata 14.0 and kept in the model if they changed the OR of predictor variables of interest by 15% or more. Model fitness was assessed using the likelihood ratio test. Considering that risk factors may differ depending on the severe malaria syndrome, we ran subgroup conditional logistic models for case–control pairs limited to cases with only severe anemia or unconsciousness, including variables that were only significant (P < 0.05) in the final model and that were clinically relevant confounders. In the subgroup analysis, the variables “danger symptoms on the first day of illness” and “having gametocytes” were associated with small numbers of discordant case–control pairs resulting in unintelligible effect estimates and confidence intervals (CI) and were thus excluded from the final models.

To factor in the contribution of distance to the nearest public health center to severe malaria occurrence, we ran a multivariable conditional logistic model limited to the study population with valid GPS data. The GPS coordinate data were used to compute distances between case and control homes and other locations, including visited health center and nearest public health center. Nearest public health centers were categorized as level II or higher, or level III or higher. Visited health centers represented the center where controls were enrolled and were subsequently listed as the center where cases would have initially sought formal care, even if they did not. In the final multivariable analysis, distance to health facilities was the distance between a home and the nearest public health center, regardless of whether cases or controls attended that center.

Determination of risk factors for delayed care seeking at the health center where a child was enrolled was based on logistic regression with a dichotomous outcome (care seeking within 24 hours after fever onset or not). To account for the biased representation of cases in our case–control sample, we performed a weighted logistic regression. The population incidence of cases was estimated to be 2,000 cases per 100,000 people per year (0.02) based on the WHO Uganda’s malaria country profile.39 This information was used to calculate sampling weights for cases and controls, as has been described by others.4042

RESULTS

Study participants.

We screened 1009 severe malaria cases and 528 uncomplicated malaria controls between March 2015 and March 2016 (Figure 2). Of the screened cases, lack of evidence of specified severe malaria criteria (427, 42%) was the most common reason for exclusion. Among controls, we excluded 196 (37%) children. Low levels of parasitemia, refusal to consent, clinical evidence of other febrile illness, and evidence of severe malaria were the common reasons for exclusion of controls. Of the 332 case–control pairs, seven were not included in the final analysis: three pairs had cases with negative malaria test results, one had a reenrolled case, and three had at least one missing questionnaire. Therefore, 325 matched case–control pairs were analyzed.

Figure 2.
Figure 2.

Study profile.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 5; 10.4269/ajtmh.17-0056

Baseline characteristics.

Severe malaria cases and uncomplicated malaria controls had similar characteristics, but cases were younger (median 1.9 versus 2.1 years; P < 0.001), more likely to be hyperparasitemic (10.4% versus 4.3%; P = 0.003), and more likely to have gametocytes identified (28.1% versus 15.1%; P < 0.001) at enrollment (Table 1). Compared with caregivers of controls, caregivers of cases were likely to be older (P = 0.01) and employed (P < 0.001). Household heads of cases had fewer years of education than those of controls (P = 0.03) and the median distance of case participants’ homes from the nearest public health center or town was greater than that of control homes (Table 1).

Table 1

Baseline characteristics of the study population

VariableUncomplicated malaria controlsSevere malaria casesP value
N = 325N = 325
Child characteristics
 Age, median (IQR)2.16 (1.34–3.15)1.99 (1.14–2.96)< 0.001
 Female, n (%)156 (48.1%)153 (47.0%)0.785
 Breast feed exclusively for 6 months210 (66.2%)230 (73.7%)0.041
 Sleeps in net, n (%)233 (71.6%)250 (76.9%)0.497
 Danger symptoms on day 1 of illness, n (%)20 (6.15%)38 (11.6%)0.013
 Parasite density (geometric mean)29,19826,1570.633
 Hyperparasitemia (> 250,000 parasites/µL)14 (4.3%)34 (10.4%)0.003
 Gametocytemia, n (%)50 (15.5%)91 (28.1%)< 0.001
 Hemoglobin9.8 (8.2–10.8)4.5 (3.5–6.3)< 0.001
Caregiver characteristics
 Age, median (IQR)26 (22–32)28 (23–35)0.015
 Mother care taker, n (%)284 (87.3%)261 (80.5%)0.018
 Years of education, median (IQR)7 (5–10)7 (5–10)0.279
 Employed, n (%)170 (52.6%)225 (70.7%)< 0.001
 Polygamous relationship, n (%)69 (27.2%)83 (32.4%)0.204
Head of household characteristics
 Age, median (IQR)32 (28–40)36 (30–45)0.139
 Years of education, median (IQR)8 (6–12)7 (5–11)0.033
 Employed, n (%)213 (79.4%)228 (84.7%)0.110
Home characteristics
 ≥ 3 children in home, n (%)44 (13.6%)72 (22.5%)0.003
 Residence near a water body, n (%)136 (42.3%)147 (46.0%)0.344
 Distance in km to nearest HC* (II and higher), median (IQR)1.37 (0.67–2.33)1.75 (0.89–2.80)0.008
 Distance in km to nearest HC* (III and higher), median (IQR)1.88 (0.95–3.38)3.33 (1.97–4.87)< 0.001
 Distance in km to listed HC (III and higher), median (IQR)3.46 (1.26–10.6)8.24 (4.10–15.9)< 0.001
 Distance in km to nearest town, median (IQR)2.18 (0.96–4.09)3.54 (1.78–5.42)< 0.001
 Socioeconomic position1 (lowest)86 (28.9%)66 (22.7%)0.159
269 (23.2%)75 (25.8%)
373 (24.5%)72 (24.8%)
4 (highest)69 (23.2%)77 (26.5%)

HC = health center; IQR = interquartile range.

Health center closet to the residence of the patient.

Health center where the control was enrolled.

Characteristics of children with severe malaria (cases).

Among the total of three severe malaria defining syndromes, the most common complication was severe anemia (Hb < 5 g/dl; 49.4%), followed by impaired consciousness (32.1%), and respiratory distress (18.5%; Figure 3). Approximately two-thirds of children with severe malaria presented with only one of the defining severe malaria syndromes, including severe anemia (64.8% of those with one defining syndrome) and impaired consciousness (26.5%). Respiratory distress was uncommon in isolation or in combination with other severe malaria defining syndromes.

Figure 3.
Figure 3.

Venn diagram: Distribution of the three principal severe malaria syndromes among the cases. This figure appears in color at www.ajtmh.org.

Citation: The American Journal of Tropical Medicine and Hygiene 97, 5; 10.4269/ajtmh.17-0056

Risk factors for severe malaria.

With unadjusted analysis, increasing age was protective against severe malaria (OR 0.56, P < 0.001; Table 2). Delayed care seeking at the health center (OR 3.77, P < 0.001) where a child was enrolled and seeking care at a drug shop as the first response to illness (OR 3.20, P < 0.001) were significantly associated with severe malaria (Table 2). If the child’s mother was the first decision-maker, compared with others being the decision-maker, the odds of severe malaria were reduced (OR 0.46, P < 0.001). Employment of the child’s caregiver (OR 2.14, P < 0.001) or the head of household (OR 2.05, P = 0.007) were associated with severe malaria (Table 2). With multivariable analysis, variables contributing to collinearity and of no significance to model fitness or interpretation were excluded (Supplemental Table 2: selection of variables). In the final model, delayed care seeking (OR 5.50, P < 0.001) and seeking care at a drug shop as an initial response to illness (OR 3.62, P < 0.001) were associated with severe malaria. Other factors associated with severe malaria were employment of the caregiver and the caregiver having more than three children aged less than 5 years. The child’s mother being the first decision-maker (OR 0.45, P = 0.014) was the only factor significantly protective against severe malaria. Increasing socioeconomic position was associated with increased likelihood of severe malaria. No significant interactions were noted in the final model. Other analytical approaches (restricted conditional logistic regression and unconditional logistic regression, with adjustment for age; Supplemental Table 3) yielded the same results as the overall conditional logistical regression model. Among patients with available GPS data, increasing distance to the nearest health center at level III or higher was associated with severe malaria (OR 1.45, P < 0.001; Table 2). Distance to the nearest health center at level II or higher was not associated with severe malaria. When the final regression model was restricted to a subgroup of cases and controls in which the case had severe malaria solely attributed to unconsciousness (N = 59 pairs) or severe anemia (N = 141 pairs), delayed seeking of care was a significant risk factor for severe malaria among children presenting with severe anemia (OR 15.6, P = 0.001), but not those presenting with unconsciousness (OR 1.13, P = 0.853; Table 3).

Table 2

Unadjusted and adjusted analysis of factors associated with severe malaria among 325 children with severe malaria (cases) and 325 children with uncomplicated malaria (controls) in Jinja District, Uganda

VariablesAll children (N = 325 pairs)Children with GPS data (N = 240 pairs)
Number of severe malaria casesUnadjusted analysisAdjusted analysisNumber of severe malaria casesAdjusted analysis
N = 325, n (%)*Odds ratioP valueOdds ratioP valueN = 240, n (%)*Odds ratioP value
Health care seeking characteristics
 Delayed to seek appropriate care287 (88.3%)3.77 (2.35, 6.03)< 0.0015.50 (2.70, 11.1)< 0.001213 (88.7%)5.76 (2.23–14.9)< 0.000
 Went to a drug shop as 1st response104 (32.0%)3.20 (2.11, 4.86)< 0.0013.62 (1.86, 7.03)< 0.00178 (32.5%)4.46 (1.93, 10.2)< 0.000
 Ant malarial given on day 1 of illness33 (10.1%)1.00 (0.60, 1.64)1.000§30 (12.5%)§
 Mother took decision on first day215 (67.4%)0.46 (0.32, 0.68)< 0.0010.45 (0.24, 0.78)0.010159 (67.9%)0.34 (0.15, 0.78)0.011
Child characteristics
 Age in yearNA0.56 (0.42, 0.76)< 0.0010.49 (0.32, 0.73)0.001NA0.73 (0.40, 1.33)0.309
 Female, n (%)153 (47.0%)0.94 (0.68, 1.30)0.744§117 (48.7%)§
 Breast feed exclusively for 6 months230 (73.7%)1.46 (1.00, 2.14)0.046§159 (69.1%)§
 Sleeps in net250 (76.9%)1.26 (0.80, 1.98)0.305§182 (81.6%)§
 Danger symptoms on day 1 of illness,38 (11.6%)2.12 (1.17, 3.84)0.0104.58 (1.73, 12.1)0.00232 (13.3%)4.62 (1.30, 16.3)0.018
 Hyperparasitemia (> 250,000 parasites/µL)34 (10.4%)2.42 (1.30, 4.52)0.003§22 (9.1%)§
 Gametocytemia91 (28.1%)2.10 (1.42, 3.11)< 0.0011.86 (1.05, 3.28)0.03266 (27.7%)1.98 (0.94, 4.19)0.072
Caregiver characteristics
 Age in yearsNA1.01 (0.99, 1.02)0.101§NA§
 Is the child’s mother261 (80.5%)0.60 (0.39, 0.92)0.020§196 (82.0%)§
 Years of educationNA0.99 (0.94, 1.03)0.747§§
 Employed225 (70.7%)2.14 (1.52, 3.00)< 0.0013.10 (1.77, 5.45)0.015163 (69.6%)2.89 (1.38, 6.03)0.005
 In a polygamous relationship83 (25.5%)1.25 (0.88, 1.79)0.205§67 (35.2%)§
Head of home characteristics
 Age in years, median (IQR)NA1.00 (0.99, 1.01)0.721§NA§
 Years of educationNA0.96 (0.92, 1.00)0.0640.94 (0.87, 1.00)0.078NA0.91 (0.83, 1.00)0.070
 Employed228 (84.7%)2.05 (1.20, 3.49)0.007§168 (83.1%)§
Home characteristics
 Distance in km to nearest HC (II and higher)§§§NA1.15 (0.91, 1.47)0.231
 Distance in km to nearest HC (III and higher)§§§NA1.45 (1.17, 1.79)< 0.001
 Distance in km to nearest town§§§NA§
 ≥ 3 children in home, n (%)72 (22.5%)1.85 (1.22, 2.81)0.0032.46 (1.20, 5.04)0.01350 (21.2%)2.78 (1.02, 7.60)0.045
 Residence near a water body147 (46.0%)1.18 (0.85, 1.64)0.317§105 (44.8%)§
 Cement floor121 (37.2%)0.78 (0.55, 1.09)0.149§91 (38.4%)§
 Dung floor107 (32.9%)2.51 (1.68, 3.76)< 0.001§77 (32.0%)§
 Iron roof283 (87.0%)0.75 (0.45, 1.23)0.256§210 (88.6%)§
 Grass roof30 (9.2%)1.04 (0.60, 1.80)0.888§23 (9.5%)§
 Mud wall68 (20.9%)1.27 (0.85, 1.90)0.225§50 (20.8%)§
 Cement wall211 (65.7%)1.01 (0.72, 1.43)0.933§158 (66.6%)§
 Socioeconomic position1 (lowest)66 (22.7%)ReferentReferent46 (21.6%)Referent
275 (25.8%)1.61 (0.97, 2.69)0.0642.14 (1.31, 7.30)0.05460 (28.1%)3.26 (1.18, 8.97)0.022
372 (24.8%)1.38 (0.82, 2.31)0.2181.21 (0.56, 2.65)0.61652 (24.4%)1.91 (0.63, 5.82)0.251
4 (highest)77 (26.5%)1.71 (1.00, 2.92)0.0462.33 (1.09, 4.98)0.02855 (25.8%)2.80 (1.05, 7.47)0.039

GPS = global positioning system; IQR = interquartile range; NA = not applicable.

Column frequency.

Analysis applied to all children; distance data not included in the analysis.

Analysis restricted to children with valid GPS data; distance data included in the analysis.

Not included in final model.

Table 3

Adjusted analysis of factors associated for specified sub-groups of children with severe malaria in Jinja District, Uganda

VariablesSevere malaria anemia (N = 141 pairs)Impaired consciousness (N = 59 pairs)
Odds ratioP valueOdds ratioP value
Health care seeking characteristics
 Delayed to seek appropriate care15.6 (3.02, 80.6)0.0011.13 (0.30, 4.28)0.853
 Went to a drug shop as 1st response7.55 (1.94, 29.2)0.0035.41 (1.25, 23.2)0.023
 Mother took decision on first day0.40 (0.15, 1.08)0.0710.43 (0.09, 2.04)0.294
Child characteristics
 Age in year0.38 (0.17, 0.82)0.0150.46 (0.21, 0.99)0.048
 Danger symptoms on day 1 of illness****
 Gametocytemia4.14 (1.53, 11.1)0.005*
Caregiver characteristics
 Employed3.81 (1.44, 10.0)0.0074.29 (1.13, 16.2)0.032
Head of home characteristics
 Years of education0.92 (0.82, 1.05)0.2551.08 (0.91, 1.29)0.359
Home characteristics
 ≥ 3 children in home, n (%)2.14 (0.60, 7.62)0.2407.75 (0.96, 62.4)0.054
 Dung floor
 Mud wall
 Socioeconomic position1 (lowest)ReferentReferent
21.81 (0.53, 6.21)0.3420.65 (0.08, 5.02)0.681
30.64 (0.16, 2.60)0.5410.42 (0.05, 2.97)0.386
4 (highest)2.60 (0.77, 8.70)0.1200.77 (0.96, 6.20)0.808

Not included in final model.

Risk factors for delayed care seeking.

As delayed care seeking from health centers where study participants were enrolled and provided with appropriate care was associated with increased risk of severe malaria, we studied factors associated with this delay. With unadjusted analysis, seeking care at a drug shop as the initial response to illness and caregivers in a polygamous relationship were significantly associated with delayed (≥ 24 hours after fever onset) care seeking (Table 4). With adjusted analysis, seeking care at a drug shop was significantly associated with delayed care seeking (OR 2.84, P = 0.028). The caregiver (OR 2.35, P = 0.018) being in a polygamous relationship was significantly associated with delayed care seeking. Considering children with valid GPS data, distance to the nearest level III and higher health center was the only factor significantly associated with delay in care seeking (OR 1.18, P = 0.031). In the model including distance data, seeking care at a drug shop did not demonstrate a significant association with delay in care seeking.

Table 4

Weighted unadjusted and adjusted analysis for factors associated with delay in care seeking (≥ 24 hours after fever onset) at heath centers where children were enrolled

VariablesAll children (N = 650)Children with GPS data (N = 480)
Children who delayed to seek careUnadjustedAdjustedChildren who delayed to seek careAdjusted
N = 513, n (%)*Odds ratioP valueN = 383, n (%)*P valueN = 383, n (%)*Odds ratioP value
Health care seeking characteristics
 Went to a drug shop as 1st response127 (24.7%)2.72 (1.14, 6.50)0.0242.84 (1.12, 7.21)0.02896 (25.0%)1.73 (0.80, 3.70)0.158
 Ant malarial given on day 1 of illness55 (10.7%)1.41 (0.61, 3.22)0.411§47 (12.2%)§
 Mother took decision on first day366 (71.3%)0.49 (0.25, 0.95)0.0370.63 (0.28, 1.41)0.266277 (72.3%)0.81 (0.36, 1.82)0.614
Child characteristics
 Age in yearsNA0.94 (0.81, 1.10)0.483§NA§
 Female242 (47.1%)0.92 (0.58, 1.47)0.745§191 (49.8%)§
 Breast feed exclusively for 6 months337 (65.9%)0.57 (0.34, 0.97)0.042§243 (63.4%)§
 Sleeps in net373 (72.1%)0.88 (0.47, 1.64)0.695§272 (71.0%)§
 Danger symptoms on day 1 of illness46 (8.9%)2.49 (0.76, 8.16)0.132§37 (9.6%)§
 Hyperparasitemia (> 250,000 parasites/µL)41 (7.9%)2.66 (0.62, 11.3)0.183§26 (6.7%)§
 Gametocytemia119 (23.2%)1.17 (0.61, 2.27)0.624§91(23.7%)§
Caregiver characteristics
 Age in yearsNA1.01 (0.99, 1.03)0.156§NA§
 Child’s mother421 (82.0%)0.42 (0.18, 0.99)0.048§316 (82.5%)§
 Years of educationNA0.93 (0.86, 0.99)0.0380.96 (0.88, 1.05)(0.389)NA0.91 (0.83, 0.99)0.035
 Employed314 (61.2%)0.87 (0.54, 1.40)0.591§224 (58.4%)
 In a polygamous relationship128 (24.9%)1.94 (1.03, 3.65)0.0402.35 (1.15, 4.80)0.018104 (27.1%)2.79 (1.18, 6.57)0.019
Head of household characteristics
 Age in yearsNA1.00 (0.99, 1.01)0.441§NA§
 Years of educationNA0.96 (0.91, 1.02)0.253§NA§
 Employed341 (80.2%)0.44 (0.21, 0.93)0.0320.45 (0.19, 1.06)0.071252 (78.0%)0.52 (0.22, 1.26)0.152
Home characteristics
 Distance in km to nearest HC (> HC II)NA§§NA1.11 (0.89, 1.38)0.349
 Distance in km to nearest HC III (> HC III)NA§§NA1.18 (1.01, 1.37)0.031
 Distance in km to nearest townNA§§NA§
 ≥ 3 children in home, n (%)97 (18.9%)1.22 (0.60, 2.46)0.569§68 (17.7%)§
 Residence near a water body222 (43.2%)1.11 (0.69, 1.80)0.641§168 (43.8%)§
 Cement floor191 (37.2%)0.55 (0.34, 0.89)0.015§138 (36.0%)§
 Dung floor132 (25.7%)1.05 (0.57, 1.93)0.875§95 (24.8%)§
 Iron roof451 (87.9%)0.98 (0.45, 2.12)0.960§335 (87.4%)§
 Grass roof49 (9.5%)1.17 (0.50, 2.71)0.714§40 (10.4%)§
 Mud wall99 (19.3%)1.23 (0.65, 2.33)0.511§75 (19.5%)§
 Cement wall331 (64.5%)0.73 (0.44, 1.21)0.223§247 (64.4%)§
 Socioeconomic position1 (lowest)124 (26.9%)ReferentReferent96 (27.6%)Referent
2119 (25.8%)1.32 (0.50, 3.47)0.5611.05 (0.44, 2.51)0.90593 (26.8%)1.00 (0.45, 2.36)0.986
3106 (22.9%)1.11 (0.44, 2.77)0.8150.48 (0.20, 1.17)0.10776 (21.9%)0.90 (0.19, 1.75)0.823
4 (highest)112 (24.3%)1.19 (0.42, 3.41)0.7360.55 (0.23, 1.33)0.19382 (23.6%)1.02 (0.26, 2.56)0.954

NA = not applicable.

Column frequency.

Analysis applied to all children; distance data not included in the analysis.

Analysis restricted to children with valid GPS data; distance data included in the analysis.

Not included in final model.

DISCUSSION

We used a case–control design to assess factors associated with severe malaria in Uganda. Delayed care seeking at the health center where the child was enrolled was a potent risk factor for severe malaria. Seeking care at a drug shop as the initial response to illness and increasing distance from place of residence to the nearest health center were associated with delayed care seeking and with severe malaria. Associations between distance to a health center and delayed care seeking,43 and between delayed care seeking and severe malaria have been described previously.6,8 We add to our understanding of the pathway toward severe malaria the important influence of drug shops; seeking care at a drug shop was strongly associated with both delayed care seeking and likelihood of severe malaria.

Prior studies have shown inconsistent results regarding the impact of delayed care seeking on severe malaria risk.69 These studies demonstrated association between longer duration of symptoms or arbitrarily set cut-offs defining duration of symptoms before admission and severe malaria, but they did not consider the standard WHO definition of delay and severe malaria risk. One study conducted in the Gambia showed lack of association between duration of symptoms and severe malaria, a finding attributed to difficulties in recollection of accurate historical information by distressed mothers.12 Our study design enabled accurate history taking, and we found that delayed care seeking at health centers was a risk factor for severe malaria, even when delay was defined based on a narrow time frame of 24 hours after fever onset. Interestingly, delayed care seeking was associated with severe malaria due to anemia, but not that due to unconsciousness. Lack of association between delayed care seeking within 24 hours and severe malaria attributed to unconsciousness, a surrogate for cerebral malaria, suggests that in these children illness evolution was too rapid to be impacted by improved care seeking. Considering these results, whereas it remains imperative to urge prompt care for all febrile children, only efforts to prevent malaria may have significant impacts on the risk of cerebral malaria.

Drug shops are a popular choice of care in Uganda.44,45 Drug shops are more accessible and approachable than health facilities, even in rural settings. This improved accessibility has led some experts to advocate a role of drug shops in improving access to health services by filling in gaps in the health sector.4649 However, we found a strong association between seeking care at a drug shop as the initial response to illness and delayed care seeking at public health centers. Seeking care at drug shops was also independently associated with an increased risk of severe malaria. Our findings implicate drug shops as barriers to access to appropriate antimalarial treatment, presumably by offering false assurance of adequate care and thereby allowing for progression from uncomplicated to severe malaria. In a survey conducted in Eastern Uganda, only 25% of children with fever seen at a drug shop received an artemisinin-based combination therapy (ACT), the first-line therapy for malaria in Africa. Of those who received an ACT, only 18% completed a full course of therapy.50 Similar findings were demonstrated in Tanzania, with prohibitive costs charged for appropriate doses.51 Among sampled drug shops in western Kenya, nonrecommended and cheaper monotherapies were available, potentially encouraging inadequate therapy.52 Subtherapeutic dosing has obvious potential bad consequences, including failure to cure infections, prolongation of illness, selection of drug resistance, and progression to severe disease.51 In our study, initially seeking care at a drug shop more than tripled the likelihood of severe malaria compared with children who did not access drug shops.

Our results suggest that it may be appropriate to rethink policies that use informal drug shops as tools for malaria treatment and control. These shops suffer from lack of medical knowledge, stocking of unregistered poor quality drugs, and the potential for profit-driven suboptimal practices.53 Interventions including subsidizing costs of ACTs sold at drug shops, training of drug vendors, and registration of drug shops have been shown to improve standards of malaria case management.54,55 In Tanzania, the National Malaria Control Program adopted the accrediting drug dispensing outlet (ADDO) program as a private sector mechanism to supplement the distribution of subsidized ACTs.56 This approach resulted in increased access to affordable ACTs for underserved populations.57 However, this success was not without limitations, and sustaining positive changes remains a concern.56,58 Furthermore, a recent assessment of the role of ADDOs in the health care system showed dispensing practices for pneumonia deviating from the norm, a practice attributed to patient demand for antibiotics and profit motivated practices.59 Similar trends might be seen with ACTs for malaria. In Ghana, the Affordable Medicines Facility–malaria initiative resulted in significant increases in availability of ACTs in drug shops, with less than half of the shops having ACTs soon after the intervention and over two years after initiation, availability of ACTs remained unchanged. Also of concern, the availability of nonstandardized herbal based antimalarials remained high during and after the intervention, sometimes surpassing availability of ACTs.60 Considering this background and our results, more study is needed to determine if optimal malaria control strategies should include drug shops as key providers of care or if emphasis should be directed primarily towards improving care at public health centers.

Longer distance from place of residence to public health centers was a notable risk factor for both delayed care seeking and severe malaria, consistent with findings from other studies showing that distance to a health center is a deterrent to prompt use of the more formal health services offered at public health centers.15,18,20,43,61 This finding is enlightening, considering that for both cases and controls average distances from patients’ homes to the nearest health center were well within a 5 km radius, the acceptable distance between a population and a health center recommended by the Ministry of Health, Uganda.62 Thus, factors beyond geographical access may have deterred caregivers from seeking care at public health centers. Associations between distance to health centers and both delayed seeking of care and severe malaria were only significant when considering centers at level III or higher. These results suggest that lower-level health centers offered less efficacious care, highlighting the need to improve services offered at these centers.

Other risk factors associated with increased likelihood of severe malaria included the following. First, gametocytemia was associated with severe malaria, likely because gametocytemia serves as a marker for prolonged infection, with gametocytogenesis peaking 7–10 days after asexual parasitaemia.3 Second, families with more than three children less than five years of age in the household were associated with severe malaria, suggesting that social dynamics linked to caring for multiple children compromised good family support.63 Third, surprisingly, having an employed caregiver or head of household was associated with severe malaria although there was no association between employment and delayed care seeking. Fourth, higher socioeconomic status of the household was associated with severe malaria, contrasting with findings from other studies.6 This finding is unexplained, but an association between higher socioeconomic status and decreased exposure to mosquitoes,64 leading to decreased malaria specific immunity and predisposition to severe disease might play a role. The mother being the first decision-maker was the only protective factor against severe malaria, highlighting the importance of women’s autonomy in a home, a characteristic associated with appropriate care seeking.6568 Increasing age was protective against severe malaria, a paradoxical finding considering that we matched on age. However, this finding could be explained by incomplete matching, and wide age ranges within matched case-control pairs.

This study had limitations. First, it was subject to inherent limitations of case–control studies, particularly selection bias attributable to enrollment of nonrepresentative controls although we enrolled geographically and time matched controls to minimize this bias. Second, recall bias may have inadvertently affected the validity of the historical information collected. Caregivers of sicker children may have exaggerated or underrated their responses in anticipation of benefit, and their responses may have been incomplete as a result of anxiety attributable to their child’s state of health. Third, inability to certainly exclude comorbidities in children with parasitemia may have resulted in misclassification of either cases or controls. Lastly, attaining balanced numbers of the three studied manifestations of severe malaria was a challenge, limiting sample size for some syndromes, and thus minimizing our ability to draw comparisons across the different syndromes.

Notwithstanding, our study provides useful insights into important modifiable risk factors for delayed care seeking and for severe malaria. We show that delayed care seeking and seeking care at a drug shop as the first response to illness were significantly associated with risk of severe disease. In addition, we demonstrate that prompt care seeking at public health centers remains a challenge, with distance from residence to health centers a hindrance in accessing these facilities. Considering their potential to cause harm, caution is recommended in incorporating drug shops as service providers. Above all, enhancement of formal public health systems in sub-Saharan Africa should be a high priority.

Supplementary Material

Acknowledgments:

We thank the clinical study team of Yasin Kisambira, Jessica Tagobera, Annet Nabweteme, Prossy Ewinyo, Azizz Kiwanuka, Rose Nabirye, Peter Wambi, Ronald Bayisuka, William Ambayo, Juliet Nabunjje, Fiona Kassana, David Katikati, Sally Opus, and Benjamin Buyi. We also thank health workers in Jinja Hospital and other participating health facilities for supporting the study team. We thank the staff of the Infectious Diseases Research Collaboration for providing administrative support and Kelly Wilson and Grant Guyen from the University of Washington for providing guidance in the mapping analysis. We would also like to thank Sarah Staedke for providing methodological advice to the study team. Finally, we are grateful to the parents, guardians, and caretakers who agreed to take part in this study.

REFERENCES

  • 1.

    Noor AM, Kinyoki DK, Mundia CW, Kabaria CW, Mutua JW, Alegana VA, Fall IS, Snow RW, 2014. The changing risk of Plasmodium falciparum malaria infection in Africa: 2000–10: a spatial and temporal analysis of transmission intensity. Lancet 383: 17391747.

    • Search Google Scholar
    • Export Citation
  • 2.

    Bhatt S et al. 2015. The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015. Nature 526: 207211.

  • 3.

    White NJ, Pukrittayakamee S, Hien TT, Faiz MA, Mokuolu OA, Dondorp AM, 2014. Malaria. Lancet 383: 723735.

  • 4.

    World Health Organization, 2015. Malaria Treatment Guidelines. Geneva, Switzerland: World Health Organization.

  • 5.

    Chotivanich K, Udomsangpetch R, Simpson JA, Newton P, Pukrittayakamee S, Looareesuwan S, White NJ, 2000. Parasite multiplication potential and the severity of falciparum malaria. J Infect Dis 181: 12061209.

    • Search Google Scholar
    • Export Citation
  • 6.

    Zoungrana A, Chou YJ, Pu C, 2014. Socioeconomic and environment determinants as predictors of severe malaria in children under 5 years of age admitted in two hospitals in Koudougou district, Burkina Faso: a cross sectional study. Acta Trop 139: 109114.

    • Search Google Scholar
    • Export Citation
  • 7.

    Ilunga-Ilunga F, Leveque A, Ngongo LO, Laokri S, Dramaix M, 2015. Treatment-seeking paths in the management of severe malaria in children under 15 years of age treated in reference hospitals of Kinshasa, Democratic Republic of Congo. Trop Med Health 43: 1119.

    • Search Google Scholar
    • Export Citation
  • 8.

    Byakika-Kibwika P, Ndeezi G, Kamya MR, 2009. Health care related factors associated with severe malaria in children in Kampala, Uganda. Afr Health Sci 9: 206210.

    • Search Google Scholar
    • Export Citation
  • 9.

    Nacher M, Singhasivanon P, Vannaphan S, Treeprasertsuk S, Phanumaphorn M, Traore B, Looareesuwan S, Gay F, 2001. Socio-economic and environmental protective/risk factors for severe malaria in Thailand. Acta Trop 78: 139146.

    • Search Google Scholar
    • Export Citation
  • 10.

    Greenwood BM, Bradley AK, Greenwood AM, Byass P, Jammeh K, Marsh K, Tulloch S, Oldfield FS, Hayes R, 1987. Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Trans R Soc Trop Med Hyg 81: 478486.

    • Search Google Scholar
    • Export Citation
  • 11.

    Genton B, al-Yaman F, Alpers MP, Mokela D, 1997. Indicators of fatal outcome in paediatric cerebral malaria: a study of 134 comatose Papua New Guinean children. Int J Epidemiol 26: 670676.

    • Search Google Scholar
    • Export Citation
  • 12.

    Luckner D, Lell B, Greve B, Lehman LG, Schmidt-Ott RJ, Matousek P, Herbich K, Schmid D, Mba R, Kremsner PG, 1998. No influence of socioeconomic factors on severe malarial anaemia, hyperparasitaemia or reinfection. Trans R Soc Trop Med Hyg 92: 478481.

    • Search Google Scholar
    • Export Citation
  • 13.

    Al-Taiar A, Jaffar S, Assabri A, Al-Habori M, Azazy A, Al-Gabri A, Al-Ganadi M, Attal B, Whitty CJ, 2008. Who develops severe malaria? Impact of access to healthcare, socio-economic and environmental factors on children in Yemen: a case-control study. Trop Med Int Health 13: 762770.

    • Search Google Scholar
    • Export Citation
  • 14.

    Rees CP, Hawkesworth S, Moore SE, Dondeh BL, Unger SA, 2016. Factors affecting access to healthcare: an observational study of children under 5 years of age presenting to a Rural Gambian Primary Healthcare Centre. PLoS One 11: e0157790.

    • Search Google Scholar
    • Export Citation
  • 15.

    Romay-Barja M, Cano J, Ncogo P, Nseng G, Santana-Morales MA, Valladares B, Riloha M, Benito A, 2016. Determinants of delay in malaria care-seeking behaviour for children 15 years and under in Bata district, Equatorial Guinea. Malar J 15: 187.

    • Search Google Scholar
    • Export Citation
  • 16.

    Chuma J, Okungu V, Molyneux C, 2010. Barriers to prompt and effective malaria treatment among the poorest population in Kenya. Malar J 9: 144.

  • 17.

    Chukwuocha UM, Okpanma AC, Nwakwuo GC, Dozie IN, 2014. Determinants of delay in seeking malaria treatment for children under-five years in parts of south eastern Nigeria. J Community Health 39: 11711178.

    • Search Google Scholar
    • Export Citation
  • 18.

    Getahun A, Deribe K, Deribew A, 2010. Determinants of delay in malaria treatment-seeking behaviour for under-five children in south-west Ethiopia: a case control study. Malar J 9: 320.

    • Search Google Scholar
    • Export Citation
  • 19.

    Ahorlu CK, Koram KA, Ahorlu C, de Savigny D, Weiss MG, 2006. Socio-cultural determinants of treatment delay for childhood malaria in southern Ghana. Trop Med Int Health 11: 10221031.

    • Search Google Scholar
    • Export Citation
  • 20.

    Ansah EK, Gyapong M, Narh-Bana S, Bart-Plange C, Whitty CJ, 2016. Factors influencing choice of care-seeking for acute fever comparing private chemical shops with health centres and hospitals in Ghana: a study using case-control methodology. Malar J 15: 290.

    • Search Google Scholar
    • Export Citation
  • 21.

    Carme B, Plassart H, Senga P, Nzingoula S, 1994. Cerebral malaria in African children: socioeconomic risk factors in Brazzaville, Congo. Am J Trop Med Hyg 50: 131136.

    • Search Google Scholar
    • Export Citation
  • 22.

    Mulumba MP, Wery M, Ngimbi NN, Paluku K, Van der Stuyft P, De Muynck A, 1990. Childhood malaria in Kinshasa (Zaire). Influence of seasons, age, environment, and family social conditions. Med Trop (Mars) 50: 5364.

    • Search Google Scholar
    • Export Citation
  • 23.

    Safeukui-Noubissi I et al. 2004. Risk factors for severe malaria in Bamako, Mali: a matched case-control study. Microbes Infect 6: 572578.

  • 24.

    Koram KA, Bennett S, Adiamah JH, Greenwood BM, 1995. Socio-economic determinants are not major risk factors for severe malaria in Gambian children. Trans R Soc Trop Med Hyg 89: 151154.

    • Search Google Scholar
    • Export Citation
  • 25.

    Ossou-Nguiet PM, Okoko AR, Ekouya Bowassa G, Oko AP, Mabiala-Babela JR, Ndjobo Mamadoud IC, Moyen G, 2013. Determinants of cerebral malaria in Congolese children. Rev Neurol (Paris) 169: 510514.

    • Search Google Scholar
    • Export Citation
  • 26.

    Bennett S, Koram KA, Greenwood BM, 1999. Risk factors for severe malaria: importance of careful study design. Trans R Soc Trop Med Hyg 93: 679.

    • Search Google Scholar
    • Export Citation
  • 27.

    Carme B, Rogier C, Trape JF, 2000. Risk factors for severe malaria: importance of careful study design: a reply. Trans R Soc Trop Med Hyg 94: 89.

    • Search Google Scholar
    • Export Citation
  • 28.

    Nacher M, Singhasivanon P, Gay F, Silachamroon U, Looareesuwan S, 2001. Case-control studies on host factors in severe malaria. Trends Parasitol 17: 253254.

    • Search Google Scholar
    • Export Citation
  • 29.

    Driss A, Hibbert JM, Wilson NO, Iqbal SA, Adamkiewicz TV, Stiles JK, 2011. Genetic polymorphisms linked to susceptibility to malaria. Malar J 10: 271.

  • 30.

    Cunnington AJ, Walther M, Riley EM, 2013. Piecing together the puzzle of severe malaria. Sci Transl Med 5: 211ps18.

  • 31.

    Uganda Bureau of Statistics (UBOS) and ICF International, 2015. Uganda Malaria Indicator Survey 2014–15. Kampala, Uganda, and Rockville, MD: UBOS and ICF International.

  • 32.

    Uganda Bureau of Statistics (UBOS), 2014. Uganda National Household Survey 2012/2013. Kampala, Uganda: UBOS.

  • 33.

    Marsh K et al. 1995. Indicators of life-threatening malaria in African children. N Engl J Med 332: 13991404.

  • 34.

    WHO, 2014. Severe malaria. Trop Med Int Health 19 (Suppl 1): 7131.

  • 35.

    Talisuna AO, Noor AM, Mundia CW, Otieno V, Mitto B, Amratia P, Robert WS, 2013. An Epidemiological Profile of Malaria and its Control in Uganda. Analysis and report funded by Roll Back Malaria and Department for International Development. UK.

  • 36.

    Li H, Yuan Z, Su P, Wang T, Yu Y, Sun X, Xue F, 2016. A simulation study on matched case-control designs in the perspective of causal diagrams. BMC Med Res Methodol 16: 102.

    • Search Google Scholar
    • Export Citation
  • 37.

    Vyas S, Kumaranayake L, 2006. Constructing socio-economic status indices: how to use principal components analysis. Health Policy Plan 21: 459468.

    • Search Google Scholar
    • Export Citation
  • 38.

    Bursac Z, Gauss CH, Williams DK, Hosmer DW, 2008. Purposeful selection of variables in logistic regression. Source Code Biol Med 3: 17.

  • 39.

    World Health Organization, 2015. Country Profiles. Available at: http://www.who.int/malaria/publications/country-profile_uga.en.pdf. Accessed June 21, 2017.

  • 40.

    van der Laan MJ, 2008. Estimation based on case-control designs with known prevalence probability. Int J Biostat 4: 17.

  • 41.

    Jiang Y, Scott AJ, Wild CJ, 2006. Secondary analysis of case-control data. Stat Med 25: 13231339.

  • 42.

    Sommerfelt H, Steinsland H, van der Merwe L, Blackwelder WC, Nasrin D, Farag TH, Kotloff KL, Levine MM, Gjessing HK, 2012. Case/control studies with follow-up: Constructing the source population to estimate effects of risk factors on development, disease, and survival. Clin Infect Dis 55 (Suppl 4): S262S270.

    • Search Google Scholar
    • Export Citation
  • 43.

    Feikin DR, Nguyen LM, Adazu K, Ombok M, Audi A, Slutsker L, Lindblade KA, 2009. The impact of distance of residence from a peripheral health facility on pediatric health utilisation in rural western Kenya. Trop Med Int Health 14: 5461.

    • Search Google Scholar
    • Export Citation
  • 44.

    Rutebemberwa E, Pariyo G, Peterson S, Tomson G, Kallander K, 2009. Utilization of public or private health care providers by febrile children after user fee removal in Uganda. Malar J 8: 45.

    • Search Google Scholar
    • Export Citation
  • 45.

    Awor P, Wamani H, Tylleskar T, Peterson S, 2015. Drug seller adherence to clinical protocols with integrated management of malaria, pneumonia and diarrhoea at drug shops in Uganda. Malar J 14: 277.

    • Search Google Scholar
    • Export Citation
  • 46.

    Basu S, Andrews J, Kishore S, Panjabi R, Stuckler D, 2012. Comparative performance of private and public healthcare systems in low- and middle-income countries: a systematic review. PLoS Med 9: e1001244.

    • Search Google Scholar
    • Export Citation
  • 47.

    Rutebemberwa E, Kallander K, Tomson G, Peterson S, Pariyo G, 2009. Determinants of delay in care-seeking for febrile children in eastern Uganda. Trop Med Int Health 14: 472479.

    • Search Google Scholar
    • Export Citation
  • 48.

    Rutebemberwa E, Buregyeya E, Lal S, Clarke SE, Hansen KS, Magnussen P, LaRussa P, Mbonye AK, 2016. Assessing the potential of rural and urban private facilities in implementing child health interventions in Mukono district, central Uganda-a cross sectional study. BMC Health Serv Res 16: 268.

    • Search Google Scholar
    • Export Citation
  • 49.

    Liu J, Prach LM, Treleaven E, Hansen M, Anyanti J, Jagha T, Seaman V, Ajumobi O, Isiguzo C, 2016. The role of drug vendors in improving basic health-care services in Nigeria. Bull World Health Organ 94: 267275.

    • Search Google Scholar
    • Export Citation
  • 50.

    Awor P, Wamani H, Bwire G, Jagoe G, Peterson S, 2012. Private sector drug shops in integrated community case management of malaria, pneumonia, and diarrhea in children in Uganda. Am J Trop Med Hyg 87: 9296.

    • Search Google Scholar
    • Export Citation
  • 51.

    Hetzel MW, Dillip A, Lengeler C, Obrist B, Msechu JJ, Makemba AM, Mshana C, Schulze A, Mshinda H, 2008. Malaria treatment in the retail sector: knowledge and practices of drug sellers in rural Tanzania. BMC Public Health 8: 157.

    • Search Google Scholar
    • Export Citation
  • 52.

    Kioko U, Riley C, Dellicour S, Were V, Ouma P, Gutman J, Kariuki S, Omar A, Desai M, Buff AM, 2016. A cross-sectional study of the availability and price of anti-malarial medicines and malaria rapid diagnostic tests in private sector retail drug outlets in rural western Kenya, 2013. Malar J 15: 359.

    • Search Google Scholar
    • Export Citation
  • 53.

    Sudhinaraset M, Ingram M, Lofthouse HK, Montagu D, 2013. What is the role of informal healthcare providers in developing countries? A systematic review. PLoS One 8: e54978.

    • Search Google Scholar
    • Export Citation
  • 54.

    Opiyo N, Yamey G, Garner P, 2016. Subsidising artemisinin-based combination therapy in the private retail sector. Cochrane Database Syst Rev 3: CD009926.

    • Search Google Scholar
    • Export Citation
  • 55.

    Awor P, Wamani H, Tylleskar T, Jagoe G, Peterson S, 2014. Increased access to care and appropriateness of treatment at private sector drug shops with integrated management of malaria, pneumonia and diarrhoea: a quasi-experimental study in Uganda. PLoS One 9: e115440.

    • Search Google Scholar
    • Export Citation
  • 56.

    Rutta E, Liana J, Embrey M, Johnson K, Kimatta S, Valimba R, Lieber R, Shekalaghe E, Sillo H, 2015. Accrediting retail drug shops to strengthen Tanzania’s public health system: an ADDO case study. J Pharm Policy Pract 8: 23.

    • Search Google Scholar
    • Export Citation
  • 57.

    Rutta E et al. 2011. Increasing access to subsidized artemisinin-based combination therapy through accredited drug dispensing outlets in Tanzania. Health Res Policy Syst 9: 22.

    • Search Google Scholar
    • Export Citation
  • 58.

    Wafula FN, Miriti EM, Goodman CA, 2012. Examining characteristics, knowledge and regulatory practices of specialized drug shops in Sub-Saharan Africa: a systematic review of the literature. BMC Health Serv Res 12: 223.

    • Search Google Scholar
    • Export Citation
  • 59.

    Embrey M et al. 2016. Understanding the role of accredited drug dispensing outlets in Tanzania’s Health System. PLoS One 11: e0164332.

  • 60.

    Ansah EK, Whitty CJ, Bart-Plange C, Gyapong M, 2016. Changes in the availability and affordability of subsidised artemisinin combination therapy in the private drug retail sector in rural Ghana: before and after the introduction of the AMFm subsidy. Int Health 8: 427432.

    • Search Google Scholar
    • Export Citation
  • 61.

    Ewing VL, Lalloo DG, Phiri KS, Roca-Feltrer A, Mangham LJ, SanJoaquin MA, 2011. Seasonal and geographic differences in treatment-seeking and household cost of febrile illness among children in Malawi. Malar J 10: 32.

    • Search Google Scholar
    • Export Citation
  • 62.

    Uganda Bureau of Statistics, 2016. The National Population and Housing Census 2014 – Main Report. Kampala, Uganda: Uganda Bureau of Statistics.

  • 63.

    Uggla C, Mace R, 2016. Parental investment in child health in sub-Saharan Africa: a cross-national study of health-seeking behaviour. R Soc Open Sci 3: 150460.

    • Search Google Scholar
    • Export Citation
  • 64.

    Tusting LS et al. 2016. Why is malaria associated with poverty? Findings from a cohort study in rural Uganda. Infect Dis Poverty 5: 78.

  • 65.

    Ewing VL, Tolhurst R, Kapinda A, Richards E, Terlouw DJ, Lalloo DG, 2016. Increasing understanding of the relationship between geographic access and gendered decision-making power for treatment-seeking for febrile children in the Chikwawa district of Malawi. Malar J 15: 521.

    • Search Google Scholar
    • Export Citation
  • 66.

    UNICEF, Liverpool School of Tropical Medicine, 2011. Gender Influences on Child Survival, Health and Nutrition: A Narrative Review. New York, NY: UNICEF.

  • 67.

    Tolhurst R, Amekudzi YP, Nyonator FK, Bertel Squire S, Theobald S, 2008. “He will ask why the child gets sick so often”: the gendered dynamics of intra-household bargaining over healthcare for children with fever in the Volta Region of Ghana. Soc Sci Med 66: 11061117.

    • Search Google Scholar
    • Export Citation
  • 68.

    Maitra P, 2004. Parental bargaining, health inputs and child mortality in India. J Health Econ 23: 259291.

Author Notes

Address correspondence to Arthur Mpimbaza, Child Health & Development Centre, Makerere University-College of Health Sciences, P.O. Box 6717, Kampala, Uganda. E-mail: arthurwakg@yahoo.com

Financial support: This research was supported by two training awards from the NIH Fogarty International Center, the University of California Global Health Institute GloCal Health Fellowship (TW009343) and the Training in Malaria Research in Uganda program (TW007375). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Authors’ addresses: Arthur Mpimbaza, Grace Ndeezi, Anne Katahoire, and Charles Karamagi, College of Health Sciences, Makerere University, Kampala, Uganda, E-mails: arthurwakg@yahoo.com, gndeezi@gmail.com, annekatahoire@yahoo.co.uk, and ckaramagi2000@yahoo.com. Philip J. Rosenthal, University of California, San Francisco, CA, Email: philip.rosenthal@ucsf.edu.

Save