Access to health care and mortality of children under 5 years of age in the Gambia: a case–control study

216 Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Objective To assess whether traditional measures of access to health care (distance and travel time to a facility) and non-traditional measures (social and financial support indicators) are associated with mortality among children under 5 years of age in the Gambia. Methods We conducted a case–control study in a population under demographic surveillance. Cases (n = 140) were children under 5 years of age who died between 31 December 2003 and 30 April 2006. Each case was matched in age and sex to five controls (n = 700). Information was gathered by interviewing primary caregivers. The data were analysed using conditional logistic regression. Findings Of traditional measures of access, only rural versus urban/periurban residence was important: children from rural areas were more likely to die (OR: 4.9; 95% confidence interval, CI: 1.2–20.2). For non-traditional measures, children were more likely to die if their primary caregivers lacked help with meal preparation (OR: 2.3; 95% CI: 1.2–4.1), had no one to relax with (OR: 1.8; 95% CI: 1.1–2.9), had no one who could offer good advice (OR: 23.1; 95% CI: 4.3–123.4), had little say over how earned money was spent (OR: 12.7; 95% CI: 1.3–127.6), were unable to cut spending for health care (OR: 2.5; 95% CI: 1.5–4.2) or had to carry out odd jobs to pay for the care (OR: 3.4; 95% CI: 2.1–5.5). A protective effect was observed when the caregiver had other children to care for (OR: 0.2; 95% CI: 0.1–0.5). Conclusion Improving access to health-care for children in the Gambia and similar settings is not simply a matter of reducing travel time and distance to a health facility, but requires improvements in caregivers’ support networks and their access to the financial resources they need. Une traduction en français de ce résumé figure à la fin de l’article. Al final del artículo se facilita una traducción al español. Access to health care and mortality of children under 5 years of age in the Gambia: a case–control study Merrin E Rutherford,a John D Dockerty,a Momodou Jasseh,b Stephen RC Howie,b Peter Herbison,a David J Jeffries,b Melissa Leach,c Warren Stevens,b Kim Mulholland,d Richard A Adegbola b & Philip C Hill b .ةلاقلما هذهل لماكلا صنلا ةياهن في ةصلاخلا هذهل ةيبرعلا ةمجترلا a Centre for International Health, University of Otago, Dunedin, New Zealand. b Bacterial Diseases Programme, Medical Research Council Laboratories, Banjul, the Gambia. c Institute of Development Studies, University of Sussex, Brighton, England. d Infectious Diseases Epidemiology Unit, London School of Hygiene and Tropical Medicine, London, England. Correspondence to Philip C Hill (e-mail: philip.hill@otago.ac.nz). (Submitted: 15 February 2008 – Revised version received: 7 July 2008 – Accepted: 11 July 2008 – Published online: 29 January 2009 ) Introduction Every year approximately 10 million children under 5 years of age die throughout the world, mostly in developing coun- tries.1 Of every 1000 children born in sub-Saharan Africa, approximately 170 die, compared with less than 10 of those who are born in developed countries.2 The fourth Millennium Development Goal is to reduce child and infant mortality in the world by two-thirds by 2015,2 yet current estimates suggest that at least 44 developing countries have less than a 20% chance of achieving the goal.1 An understanding of the risk factors associated with child mortality and the design of appropriate interventions are urgently required. Population-based studies suggest that poor access to health care, which results in delayed attendance at a health facility or none at all, may be a key determinant of mortality in children under 5 years of age in developing countries.3,4 To assess this relationship, researchers have adopted simple traditional measures, such as distance from the household to a health facility,5–7 availability of transportation8 and health- care costs.9 Their findings have been inconclusive. Recently, qualitative studies10–12 have suggested the need to consider additional barriers, such as lack of social support for primary caregivers, limited caregiver autonomy in decision-making and financial matters, and social exclusion at health clinics. Studies that assess the relationship between these factors and child death are needed. Therefore, we evaluated traditional measures of access to health care in addition to non-traditional measures to study mortality in children under 5 years of age in the Gambia. Methodology Population and study design We conducted a case-control study within the Farafenni De- mographic Surveillance System (FDSS), on the North Bank of the River Gambia. The FDSS was established in 1981 as a platform for research, its location being considered particu- larly ideal for malaria studies. The FDSS covers approximately 17 000 rural and 26 000 urban dwellers belonging to three main ethnic groups (Wolloff, Mandinka and Fula) among which young age, high fertility, low educational levels, fre- quent polygamy and widespread poverty prevail.13 Water is gathered primarily from communal wells with hand pumps, and 12% of the population utilizes uncovered wells. Pit latrines predominate. Villages have no electricity, and trans- portation routes are dirt roads.13 217Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the GambiaMerrin E Rutherford et al. The cases were all children under 5 years of age under coverage by the FDSS who had died between 31 Decem- ber 2003 and 30 April 2006. Children were excluded from the study if their primary caregiver could not be located or had since died or if their cause of death was unrelated to access to health care (e.g. congenital malformations, in-hospital neonatal deaths, immedi- ate death after an injury, etc.). Verbal autopsies, available for 92% of the cases, were conducted within 3 months of death by trained field workers using the standardized INDEPTH form, as part of routine FDSS practice.14 They were assessed by consensus between two physicians. Controls were children under 5 years of age who were registered in the FDSS and who were alive at the time their respective cases died. They were excluded from the study if the primary caregiver could not be located or had since died. From the FDSS population, 10 controls were randomly identified for and matched with each case by age (within the same birth month) and sex. Only the first five random choices were selected, followed by the next child on the list of random choices in the event of refusal to participate. On the assumption that 70% of the controls would take at least 30 min- utes or more to travel to a health-care facility, we estimated that at least 120 cases and 600 controls were required to detect an odds ratio (OR) of 2.0 for less than or more than 30 minutes of travel time as a binary variable at a power of 80% and an alpha level of 0.05. Data collection After written informed consent was obtained from primary caregivers, data were collected from them in their na- tive language between 11 May 2006 and 25 October 2006. Interviews were conducted, whenever possible in private locations, by six trained and experienced Gambian fieldworkers who were famil- iar with the area and local languages. The questionnaire had seven sec- tions: (i) demographic characteristics; (ii) child characteristics (such as ethnic- ity); (iii) child health; (iv) household features, including amenities and wealth; (v) social support; (vi) health- seeking behaviour and (vii) traditional measures of access to health care. A household crowding index was adopted from a previous Gambian study.15 Social and financial questions relevant to ac- cessing health care were identified and refined through a literature review that yielded 11 social support questions from a medical outcomes study16 plus social17–20 and economic5,21–24 questions from previous studies in the Gambia and elsewhere. These were refined fol- lowing focus group discussions, two rounds of piloting and expert review. The focus groups were composed of mothers with children from three vil- lages. The mothers of identified cases were excluded. The reference date for each control interview was the day on which the control was the same age as the case child upon death. Personal prompts in relation to the reference dates were developed using an Islamic calendar and significant village-specific events. To determine which health centres were appropriate and should be used to calculate distance and time from the households, we ascertained when each centre was established, whether it had the capacity to treat a seriously ill child, and what vaccinations and essential medications it had in stock. Respondents were allocated to the nearest centre considered appropriate. Their choice took precedence if it dif- fered from the facility assigned. Using a hand-held global positioning system (Garmin GPS 12 Personal Navigator®), we calculated the distance to each health centre from the centre of each village and the 45 periurban suburbs. Travel time was calculated by replicating the mode of travel each respondent used: walking, horse or donkey cart, bush taxi (van to transport 20 to 30 people), private taxi or car. Transportation costs were obtained from taxi station atten- dants and respondents. Data management and analysis Completed questionnaires were checked immediately and queries were resolved. Available demographic data were ex- tracted from the FDSS database and cross-checked with hard copy question- naires. Other data from the question- naires were double entered into a data- base, verified and analysed using Stata software version 8.0 (Stata Corp, College Station, TX, United States of America). ORs were derived using conditional logistic regression. Confounding factors, in addition to variables felt to be related to access to health care, were incorpo- rated into the questionnaire based on a review of relevant literature and on con- siderations of plausibility. Such factors included demographic characteristics, breastfeeding history, vaccination status and vitamin A supplementation status. These were considered for inclusion in multivariate models using the change in estimate methodology.25 For each variable, possible confounders were en- tered singly into a multivariable model. Any variable that caused the crude OR to change by more than 10%, or that made a non-significant result become significant, were selected for the final multivariate model. We applied the likelihood ratio test to detect interac- tions between variables. Correlation coefficients were calculated separately for social and financial variables. Fig. 1. Case and control recruitment process, May–October 2006, for study conducted in the FDSS, the Gambia Eligible cases 163 Approached controls 776 Refused 8 Untraceable 1 Migrated 9 Other 4 Interviewed cases 141 Retrospectively excluded 1 Included in final analysis 140 Refused 9 Untraceable 9 Migrated 42 Other 16 Interviewed controls 700 Included in final analysis 700 FDSS, Farafenni demographic surveillance system. 218 Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the Gambia Merrin E Rutherford et al. The Joint Government of the Gambia/Medical Research Council Eth- ics Committee approved the study. Results We recruited 141 of 163 possible cases (Fig. 1). One case was later retrospec- tively excluded because the death bore no relation to health-care access. We ap- proached 776 randomly selected con- trols to yield 700 for interview (Fig. 1); 76 were the second choice, and 4 were the third choice. Refusal rates were less than 5%. The child’s mother was the primary caregiver for 135 (96.4%) cases and 690 (98.6%) controls. The mean age of mothers of case children (28 years; range: 16–50) did not differ significantly from that of control children’s mothers. Cases and controls did not differ sig- nificantly in either ethnicity or primary caregiver’s educational level. Sixty nine (49.3%) cases died before the age of 1 year (Table 1), and more than half the deaths occurred at home (52.9%). The commonest causes of death were fever of unknown origin (23.3%) and diarrhoea (20.2%). The terminal illness lasted less than a week in the majority of cases (84.3%), and 89.3% of the children were seen at a government health facility. A longer distance to travel (more than 3 km) and higher costs to reach the health facility (more than 5 dalasi; 1 US dollar = 30 dalasi) were significantly associated with child death in univari- ate analyses (Table 2). However, after controlling for possible confounding factors (referenced at the bottom of each table), only place of residence re- tained significance: rural dwellers had a significantly greater likelihood of death than urban dwellers (OR: 4.9; 95% confidence interval, CI: 1.2–20.2). In contrast, 10 non-traditional mea- sures of access to health care were signifi- cantly associated with child death after controlling for confounders (Table 3 and Table 4). These included measures of primary caregiver social support: not having someone to relax with (OR: 1.8; 95% CI: 1.1–2.9); not having someone else to prepare meals (OR: 2.3; 95% CI: 1.2–4.1); not participating in social groups (OR: 3.6; 95% CI: 1.9–6.9); and not knowing anyone who could provide good advice in a crisis situation (OR: 23.1; 95% CI: 4.3–123.4). Surprisingly, not having someone who showed car- ing and kindness to the caregiver was Table 1. Characteristics of children under 5 years of age who died in the FDSS, the Gambia, between 31 December 2003 and 30 April 2006 Characteristic No. (%) Age of child at death, in months 0–6 40 (28.6) 7–12 29 (20.7) 13–24 31 (22.1) 25–60 40 (28.6) Cause of death (n = 129) Fever of unknown origin 30 (23.3) Diarrhoea 26 (20.2) Pneumonia 15 (11.6) Meningitis or cerebral malaria 17 (13.2) Neonatal sepsis 6 (4.7) Vomiting 3 (2.3) Non-specific/other 32 (24.8) Place of death Hospital 45 (32.1) Health centre 13 (9.3) Home 74 (52.9) Other 8 (5.7) Duration of illness, in days 1–2 76 (54.3) 3–7 42 (30.0) 8–14 10 (7.1) 15–30 6 (4.3) > 31 6 (4.3) Highest level of treatment received for illness Health centre 60 (42.9) Hospital 65 (46.4) Village health worker 5 (3.6) Traditional healer/marabout 5 (3.6) Private physician/pharmacist/drug shop keeper 4 (2.9) Unknown 1 (0.7) FDSS, Farafenni demographic surveillance system. protective against child death (OR: 0.4; 95% CI: 0.2–0.9). Moderately strong correlations (r = 0.60–0.69) were found between each of the following: having someone to talk with, having someone to relax with, having someone who understood your problems and having someone who showed caring and kind- ness. Also, having someone to help with domestic duties and having someone to prepare meals were moderately cor- related with each other (r = 0.6). Children whose primary caregivers had little say over how money was spent (Table 4) were at increased risk of dying compared with those whose caregivers could always decide (OR: 12.7; 95% CI: 1.3–127.6). Children whose primary caregivers could not cut back on spend- ing for health care were at greater risk of dying (OR: 2.5; 95% CI: 1.5–4.2). The same was true of children whose primary caregivers had to do odd jobs to generate money for health care (OR: 3.4; 95% CI: 2.1–5.5). Measures of correlation be- tween financial variables were not greater than r = 0.3 for any comparison. Two significant interactions were identified: for children whose caregivers were unable to cut back on health-care spending but could use their savings to pay for health care, the OR for child death was 0.2 (95% CI: 0.1–0.9) as opposed to 5.3 (95% CI: 2.4–11.5) for children whose caregivers were unable to cut back on health-care spending or to use their savings for this purpose; rural children whose caregivers did not have an outside source of money to pay for health care had an OR for child death of 2.1 (95% CI: 1.2–2.7) when compared with urban children (OR: 1.0; 95% CI: 0.5–1.8). 219Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the GambiaMerrin E Rutherford et al. Table 2. Association between traditional indicators of access to health care and mortality in children under 5 years of age, FDSS, the Gambia Variable Cases No. (%) Controls No. (%) Crude OR (95% CI) Adjusted OR (95% CI) Location of residence Urban/peri-urban 57 (40.7) 386 (55.1) 1.0 1.0 Rural 83 (59.3) 314 (44.9) 1.8 (1.2–2.6) 4.9 (1.2–20.2)a Distance to health centre, in km 0–3 62 (44.3) 400 (57.1) 1.0 1.0 > 3 to 6 57 (40.7) 214 (30.6) 1.7 (1.2–2.6) 1.2 (0.5–2.6) > 6 21 (15.0) 86 (12.3) 1.6 (0.9–2.8) 1.8 (0.6–5.6)b Distance, in km, and travel time, in minutes, index < 3 km and < 30 min 51 (36.4) 328 (46.9) 1.0 1.0 > 3 km and < 30 min or < 3 km and > 30 min 33 (23.6) 142 (20.3) 1.5 (0.9–2.4) 1.2 (0.7–2.1) > 3 km and > 30 min 56 (40.0) 230 (32.9) 1.6 (1.0–2.4) 1.1 (0.7–1.9)c Cost of reaching health centre, in dalasi 0–4 113 (80.7) 600 (85.7) 1.0 1.0 5–9 20 (14.3) 58 (8.3) 1.8 (1.1–3.2) 0.9 (0.3–3.2) ³ 10 7 (5.0) 42 (6.0) 0.9 (0.4–2.1) 0.7 (0.2–2.4)d FDSS, Farafenni demographic surveillance system; OR, odds ratio. Factors included in multivariate analysis: a Allocated health centre, distance to a health centre, readily available money, child vitamin A status, child vaccination status; b Allocated health centre, travel time to a health centre, readily available money, child vitamin A status, child vaccination status; c Rural residence; d Travel time to a health centre, distance to a health centre, readily available money, child vitamin A status, vaccination status, rural residence. With respect to time availability, children whose primary caregivers cared for other children were less likely to die than those whose primary caregivers cared for no other children, regardless of the number of children cared for: (1–3 other children: OR: 0.2; 95% CI: 0.1–0.5; 4 or more other children: OR: 0.2; 95% CI: 0.1−0.4). Discussion Major findings After controlling for possible confound- ers, a rural place of residence was sig- nificantly associated with an increased risk of death before the age of 5 years, but no other traditional measure of access to health care retained statistical significance. However, 10 additional, non-traditional variables, including indicators of social support for the primary caregiver, his/her degree of fi- nancial autonomy and his/her source of revenue for health-care expenses, were significantly associated with child death. These findings have important implica- tions for the design of new interventions against child death in the Gambia and other developing countries and present new avenues for further research. Comparison with other studies The increased risk of death in rural dwellers was not explained by a longer physical distance to a health care facil- ity. It may be that rural dwellers simply take longer to seek appropriate help. In Malawi, rural residence was significantly associated with an increased risk of late medical treatment,26 while in a multi- country analysis of sub-Saharan African countries, sick urban children were more likely to visit a health facility than rural children.27 The lack of a significant association in our study between dis- tance to a health care facility and child death is consistent with the results of studies performed elsewhere in Africa: non-significant increases in the risk of child death in association with increas- ing distance to health care were reported from Burkina Faso6 and the northern Democratic Republic of the Congo.7 This suggests that in such settings, physical distance to a health facility has little effect on access to health care. The importance of social support in relation to access to health care has been proposed through qualitative research in Chad,11 the Gambia,18 Guinea12 and India.19 Social support is thought to facilitate health care access through increasing help with domestic tasks, peer encouragement and knowledge-sharing. In Ethiopia, a significant protective association be- tween increased social capital (defined in terms of membership in community groups, evidence of leadership, ability to borrow money, trustworthiness and interpersonal indicators) and child death was identified.20 Some caregiv- ers are also socially excluded at health centres and treated in a discriminatory fashion.18 Thus, impaired access affects even those who attend health facilities. Financial autonomy is likely to increase the primary caregiver’s abil- ity to access health care and pay for it. Qualitative research in Ghana has drawn attention to the importance of female financial autonomy for house- hold expenditure in terms of the ability to obtain health care,28 while children of Tanzanian women who were the sole decision-makers regarding health care were significantly less likely to die.29 Our findings show that children of caregivers who did not have to do odd jobs to pay for health care or who were able to generate money for health care by cutting back on spending were less likely to die. In Viet Nam, poor people who had to borrow money delayed seeking treatment at health facilities.30 Together, these findings indicate that lack of an available monetary source to cover health care costs delays health care seeking behaviour and increases the risk of death. When investigating the caregiver’s time availability, having to care for other children besides the one at risk of death showed a protective effect clearly not explained by high maternal parity or overcrowding. In contrast, Van den Broeck et al. showed that the risk of child death in the northern Democratic Republic of the Congo increased with a parity of five or more as a result of having limited time for child care.7 It is possible that primary caregivers that 220 Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the Gambia Merrin E Rutherford et al. Table 3. Association between primary caregiver social support indicators and mortality in children under 5 years of age, FDSS, the Gambia Variable name Cases No. (%) Controls No. (%) Crude OR (95% CI) Adjusted OR (95% CI) Had someone who understood your problems Most or all of the time 97 (69.3) 490 (70.0) 1.0 1.0 Some of the time 35 (25.0) 166 (23.7) 1.1 (0.7–1.6) 1.1 (0.7–1.6) A little or none of the time 8 (5.7) 44 (6.3) 0.9 (0.4–2.0) 0.9 (0.4–2.0)* Had someone who showed you kindness and caring (n = 699 due to missing data) Most or all of the time 91 (65.0) 428 (61.2) 1.0 1.0 Some of the time 41 (29.3) 195 (27.9) 1.0 (0.7–1.5) 0.8 (0.5–1.3)a A little or none of the time 8 (5.7) 76 (10.9) 0.5 (0.2–1.1) 0.4 (0.2–0.9)a Had someone to relax with Most or all of the time 70 (50.0) 340 (48.6) 1.0 1.0 Some of the time 30 (21.4) 248 (35.4) 0.6 (0.4–0.9) 0.5 (0.3–0.9)b A little or none of the time 40 (28.6) 112 (16.0) 1.7 (1.1–2.7) 1.8 (1.1–2.9)b Had someone to take you to a doctor (n = 699 due to missing data) Most or all of the time 130 (92.9) 665 (95.1) 1.0 1.0 Some of the time 7 (5.0) 31 (4.4) 1.1 (0.5–2.6) 1.1 (0.5–2.7)c A little or none of the time 3 (2.1) 3 (0.4) 5.0 (1.0–24.9) 4.5 (0.9–22.9)c Had someone to prepare meals if you were unable to Most or all of the time 103 (73.6) 541 (77.3) 1.0 1.0 Some of the time 18 (12.9) 108 (15.4) 0.9 (0.5–1.5) 0.9 (0.5–1.6)d A little or none of the time 19 (13.6) 51 (7.3) 2.0 (1.1–3.5) 2.3 (1.2–4.1)d Had someone to help with domestic duties Most or all of the time 101 (72.1) 480 (68.6) 1.0 1.0 Some of the time 18 (12.9) 153 (21.9) 0.5 (0.3–0.9) 0.5 (0.3–0.9)e A little or none of the time 21 (15.0) 67 (9.6) 1.5 (0.9–2.6) 1.8 (1.0–2.9)e Participation in social groups Yes 119 (85.0) 639 (91.3) 1.0 1.0 No 21 (15.0) 61 (8.7) 1.8 (1.1–3.1) 3.6 (1.9–6.9) f Had someone to talk with Most or all of the time 80 (57.1) 421 (60.1) 1.0 1.0 Some of the time 43 (30.7) 228 (32.6) 1.0 (0.7–1.5) 1.0 (0.7–1.5) A little or none of the time 17 (12.1) 51 (7.3) 1.7 (1.0–3.2) 1.7 (1.0–3.2)* Had someone to give good advice in a crisis Most or all of the time 119 (85.0) 586 (83.7) 1.0 1.0 Some of the time 15 (10.7) 111 (15.9) 0.6 (0.4–1.2) 0.8 (0.4–1.4)g A little or none of the time 6 (4.3) 3 (0.4) 13.3 (2.7–66.5) 23.1 (4.3–123.4)g FDSS, Farafenni demographic surveillance system; OR, odds ratio. * Association not altered more than 10% following multivariate analysis. Factors included in multivariate analysis: a Primary caregiver ethnicity, allocated health centre, number of members in household; b Primary caregiver ethnicity, allocated health centre, rural residence; c Allocated health centre; d Allocated health centre; e Rural residence; f Rural residence, primary caregiver ethnicity, allocated health centre, number of members in household; g Primary caregiver ethnicity, primary caregiver household position, rural residence, allocated health centre. look after more children have greater experience with serious illness and seek help earlier. Previous research has shown that poor symptom recognition delays health care seeking and that caregiver knowledge about health protects against child death.28,31,32 The majority of deaths in this study were caused by infectious diseases. In over 80% of the cases the final illness lasted less than 1 week, and over half the children died at home (53%). Ac- cording to previous research using verbal autopsies in the Gambia33 and Guinea- Bissau,34 acute diarrhoea and fever are the leading causes of death in children under 5 years of age. Studies in the Gam- bia,33 Guinea-Bissau34 and the United Republic of Tanzania21 have shown that children most commonly die at home. In this study the number of children who received modern health care (as opposed to traditional treatments) for their final illness is consistent with the findings of recent studies in Guinea-Bissau34 and the United Republic of Tanzania,21 where 93% and 75% of children, respectively, sought modern health care. Study strengths and weaknesses Selection from the community at large allowed inclusion of cases and controls who did not utilize health care services. We chose death as the outcome because it represents severe illness, it is clearly defined and measurable, and its risk factors in children under 5 years of age in developing countries are poorly known. Future studies might include other outcome variables. The use of FDSS data for selection ensured that cases and controls were from the same population and that the majority of cases were included, which minimized selection bias. We considered selecting controls who were ill, but frequent illness is extremely common, if not universal, among Gambian children, and both cause-specific and non- cause-specific definitions of illness are problematic. The FDSS population may not be fully representative, as indi- viduals need to have resided in the area for 3 months before being registered, so recent migrants and the transient population may have been excluded. However, because these people are probably the most susceptible to social exclusion, including them would prob- ably have strengthened our findings. 221Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the GambiaMerrin E Rutherford et al. Table 4. Association between primary caregiver financial indicators of access to health care and child death, FDSS, the Gambia Variable Cases No. (%) Controls No. (%) Crude OR (95% CI) Adjusted OR (95% CI) Primary caregiver wealth Amount of income generated by primary caregiver, in dalasi 0 95 (67.9) 383 (54.7) 1.0 1.0 1–500 23 (16.4) 115 (16.4) 0.8 (0.5–1.3) 1.0 (0.5–2.0)a > 500 22 (15.7) 202 (28.9) 0.4 (0.3–0.7) 0.6 (0.3–1.2)a Received money from an outside source Yes 53 (37.9) 286 (40.9) 1.0 1.0 No 87 (62.1) 414 (59.1) 1.1 (0.8–1.6) 1.1 (0.8–1.6)* Financial autonomy (95 cases and 383 controls had no income) Could decide how earned money was spent All 41 (91.1) 306 (96.5) 1.0 1.0 Some, a little, or none 4 (8.9) 11 (3.5) 9.7 (1.0–92.5) 12.7b (1.3–127.6)b Used own money to pay for health care Yes 12 (8.6) 36 (5.1) 1.0 1.0 No 128 (91.4) 664 (94.9) 0.6 (0.3–1.2) 0.5 (0.2–1.0)c Monetary sources used to cover health care costs (indicators of availability) Savings Yes 72 (51.4) 393 (56.1) 1.0 1.0 No 68 (48.6) 307 (43.9) 1.2 (0.8–1.7) 1.2 (0.8–1.7)* Cutting back on spending Yes 18 (12.9) 186 (26.6) 1.0 1.0 No 122 (87.1) 514 (73.4) 2.5 (1.5–4.2) 2.5 (1.5–4.2)* Selling goods of livestock Yes 59 (42.1) 233 (33.3) 1.0 1.0 No 81 (57.9) 467 (66.7) 0.7 (0.5–1.0) 0.9 (0.6–1.3)d Monetary sources used to cover child death and health care costs (indicators of unavailability) Doing odd jobs No 103 (73.6) 640 (91.4) 1.0 1.0 Yes 37 (26.4) 60 (8.6) 3.8 (2.4–6.2) 3.4 (2.1–5.5)e Borrowing from friends No 133 (95.0) 664 (94.9) 1.0 1.0 Yes 7 (5.0) 36 (5.1) 1.0 (0.4–2.2) 0.7 (0.3–1.7) f Borrowing from lenders No 137 (97.9) 682 (97.4) 1.0 1.0 Yes 3 (2.1) 18 (2.6) 0.8 (0.2–2.8) 0.6 (0.2–2.2)g Time availability Average amount of time spent working at any occupation, in hours 0–5 9 (6.4) 109 (15.6) 1.0 1.0 6–10 108 (77.1) 511 (73.0) 2.6 (1.3–5.3) 2.9 (0.8–10.2)h 11–15 23 (16.4) 80 (11.4) 3.6 (1.6–8.4) 2.5 (0.6–10.1)h Number of other children cared for by primary caregiver 0 14 (10.0) 28 (4.0) 1.0 1.0 1–3 90 (64.3) 462 (66.0) 0.4 (0.2–0.8) 0.2 (0.1–0.5) i ³ 4 36 (25.7) 210 (30.0) 0.3 (0.2–0.7) 0.2 (0.1–0.4) i FDSS, Farafenni demographic surveillance system; OR, odds ratio. * Association not altered beyond 10% following multivariate analysis. Factors controlled for in multivariate analysis: a Allocated health centre, readily available money, vitamin A status, vaccination status, time spent working, rural residence; b Primary caregiver educational level; c Primary caregiver ethnicity, primary caregiver age, rural residence, time spent working, primary caregiver education level; d Rural residence, allocated health centre; e Rural residence; f Rural residence, allocated health centre; g Allocated health centre; h Rural residence, allocated health centre, vaccination status, vitamin A status; i Number of pregnancies of primary caregiver. 222 Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the Gambia Merrin E Rutherford et al. Recall bias was minimized by means of personal prompts, and responder bias was reduced by conducting interviews in private locations and not provid- ing response options. Low refusal and drop-out rates made non-participation bias unlikely. We controlled for a wide range of confounding factors by means of multivariate analysis, although we could not assess birth weight. Conclusion A simplistic approach to health care provision in countries such as the Gam- bia, based on distance and travel time to a health facility, may be inadequate. In light of the priority to achieve the fourth Millennium Development Goal by 2015, reductions in child mortality through improved health care access will benefit from a focus on additional mea- sures. The strengthening of social sup- port networks for children’s caregivers might be achieved through community group development and implementa- tion of community systems that can provide them with the support they need when seeking health care.20 It is important to also improve their access to financial resources through tools such as microfinancing 35. Further research in other settings is required to confirm these findings. ■ Acknowledgements We thank the population covered by the FDSS for its time and cooperation; the Gambian Government; Pierre Gomez and Simon Donkor for data manage- ment; our fieldworkers and village reporters; and Prof. Charlotte Paul for her useful comments. Funding: This study was funded by The Medical Research Council of the United Kingdom and the University of Otago, New Zealand. Competing interests: None declared. Résumé Accès aux soins de santé et mortalité des enfants de moins de 5 ans en Gambie : étude cas-témoins Objectif Evaluer les éventuelles associations entre les mesures classiques de l’accès aux soins (distance et durée du trajet jusqu’à l’établissement) ou les mesures non conventionnelles (indicateurs de soutien social et financier) et la mortalité des enfants de moins de 5 ans en Gambie. Méthodes Nous avons mené une étude cas-témoins dans une population sous surveillance démographique. Les cas (n = 140) étaient des enfants de moins de 5 ans, décédés entre le 31 décembre 2003 et le 30 avril 2006. Chaque cas était apparié selon l’âge et le sexe à 5 témoins (n = 700). Des informations ont été recueillies en interrogeant les aidants principaux. Les données ont été analysées par régression logistique classique. Résultats Parmi les mesures classiques de l’accès aux soins, seule la distinction entre résidence en milieu rural et résidence en milieu urbain ou périurbain était importante : les enfants provenant de zones rurales avaient une plus grande probabilité de mourir (OR : 4,9 ; IC à 95 % 1,2-20,2). S’agissant des mesures non conventionnelles, la probabilité de décès des enfants était plus forte si leur aidant principal manquait d’aide pour la préparation des repas (OR : 2,3 ; IC à 95 % : 1,2-4,1), n’avait personne avec qui se détendre (OR : 1,8 ; IC à 95 % : 1,1-2,9), n’avait personne pour lui prodiguer de bons conseils (OR : 23,1 ; IC à 95 % : 4,3-123,4), n’avait pas son mot à dire sur la façon dont l’argent gagné était dépensé (OR : 12,7 ; IC à 95 % : 1,3-127,6), était dans l’incapacité de réduire les dépenses de santé (OR : 2,5 ; IC à 95 % : 1,5-4,2) ou devait effectuer des «petits boulots» pour payer ces dépenses (OR : 3,4 ; IC à 95 % : 2,1-5,5). Un effet protecteur a été observé lorsque l’aidant principal devait également s’occuper d’autres enfants (OR : 0,2 ; IC à 95 % : 0,1-0,5). Conclusion Améliorer l’accès aux soins des enfants en Gambie ou dans d’autres contextes similaires ne signifie pas seulement réduire le temps de trajet et la distance jusqu’à l’établissement de soins, mais suppose un renforcement des réseaux de soutien aux aidants et un meilleur accès pour eux aux ressources financières dont ils ont besoin. Resumen Acceso a atención de salud y mortalidad de menores de 5 años en Gambia: estudio de casos y controles Objetivo Determinar si las medidas tradicionales (distancia y tiempo de viaje hasta un establecimiento) y no tradicionales (indicadores de apoyo social y financiero) de acceso a la atención de salud guardan relación con la mortalidad entre los menores de 5 años en Gambia. Métodos Llevamos a cabo un estudio de casos y controles en una población sometida a vigilancia demográfica. Los casos (n = 140) fueron niños menores de 5 años fallecidos entre el 31 de diciembre de 2003 y el 30 de abril de 2006. Cada caso se emparejó por edad y sexo con cinco controles (n = 700). Se entrevistó a los cuidadores principales para reunir la información pertinente, analizándose luego los datos mediante regresión logística condicional. Resultados De las medidas tradicionales de acceso, sólo la residencia rural frente a la urbana/periurbana tenía un efecto importante: los niños de las zonas rurales tenían más probabilidades de morir (OR: 4,9; intervalo de confianza (IC) del 95%: 1,2–20,2). En cuanto a las medidas no tradicionales, los niños tenían más probabilidades de morir cuando sus cuidadores principales carecían de ayuda para preparar las comidas (OR: 2,3; IC95%: 1,2–4,1), no disponían de nadie con quien distenderse (OR: 1,8; IC95%: 1,1–2,9), no tenían a nadie que pudiera ofrecerles buenos consejos (OR: 23,1; IC95%: 4,3–123,4), no podían influir apenas en las decisiones de gasto del dinero ganado (OR: 12,7; IC95%: 1,3–127,6), no podían reducir el gasto en atención de salud (OR: 2,5; IC95%: 1,5–4,2) o tenían que desempeñar trabajos poco corrientes para pagar la atención (OR: 3,4; IC95%: 2,1–5,5). Un efecto protector fue que el cuidador tuviera que ocuparse también de otros niños (OR: 0,2; IC95%: 0,1–0,5). Conclusión La mejora del acceso a la atención sanitaria para los niños de Gambia y entornos similares no estriba simplemente en reducir el tiempo de viaje y la distancia a los centros de salud. Es preciso mejorar las redes de apoyo a los cuidadores y su acceso a los recursos financieros que necesitan. 223Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the GambiaMerrin E Rutherford et al. صخلم دهاوشلاو تلااحلل ةسارد :ايبماغ في رمعلا نم ةسماخلا نود لافطلأا تايفوو ةيحصلا ةياعرلا لىإ لوصولا ةيحصلا ةياعرلا لىإ لوصولل ةيديلقتلا سيياقملل ناك اذإ ام ميـيقت :فدهلا تاشرؤم( ةيديلقتلا يرغ سيياقمللو )هيلإ رفسلا نمزو قفرلما نع دعُبلا( ةسماخلا نود لافطلأا ينب تايفولا تلادعبم طابترا )ليالماو يعماتجلاا معدلا .ايبماغ في رمعلا نم ةيناكس ةعومجم في دهاوشلاو تلااحلل ةسارد نوثحابلا ىرجأ :ةقيرطلا نود لافطلأا نم 140 تلااحلا ددع غلبو .فيارغويمدلا دصترلل ةعضاخ برمسيد/لولأا نوناك 31 ينب ةرـتفلا في اوتام نيذلا رمعلا نم ةسماخلا دهاوش ةسمخ ةلاح لك لباقم عضوو .2006 ليربأ/ناسين 30و 2003 ينئماقلا عم تلاباقلما قيرط نع تامولعلما اوعمجو .)700 دهاوشلا ددع( يتسجوللا ف ُّوحتلا مادختساب تايطعلما اوللحو .ةيحصلا ةياعرلا ءاتيإ لىع .يطشرلا ةياعرلل لوصولل ةيديلقتلا سيياقلما نم يأ ةيمهأ ينبتت لم :تادوجولما في لافطلأاف ؛اهيحاوض في وأ ندلما في وأ فيرلا في ةماقلإا ءانثتساب ةيحصلا ةقث ةلصافب 4.9 ةيحجرلأا ةبسن( ؛تولما لماتحلا ًاضرعت ثركأ اوناك فيرلا يرغ سيياقملل ةبسنلاب امأ .)20.2 – 1.2 ةيحجرلأا ةبسن تغلب ذإ %95 ءاتيإ لىع نوئماقلا رقتفا مالك دادزي لافطلأا توم لماتحا ناكف ةيديلقتلا ةقث ةلصافب ،2.3 ةيحجرلأا ةبسن( تابجولا دادعإ لىع ةردقلل ةيلولأا ةياعرلا ًاصخش نودقتفي اوناك وأ ؛)4.1و 1.2 ينب ةيحجرلأا ةبسن تحوارت ذإ %95 تحوارت ذإ %95 ةقث ةلصافب ،1.8 ةيحجرلأا ةبسن( هعم ءاخترسلاا مهنكيم مدقي يذلا صخشلا نودقتفي اوناك وأ ،)2.9 و 1.1 ينب ةيحجرلأا ةبسن تحوارت ذإ %95 ةقث ةلصافب 23.1 ةيحجرلأا ةبسن( ةديجلا ةحيصنلا مهل قافنإ ةيفيك لوح يأر مهل نكي لم وأ )123.4و 4.3 ينب ةيحجرلأا ةبسن تحوارت ذإ %95 ةقث ةلصافب 12.7 ةيحجرلأا ةبسن( اهنوبسكي يتلا لاوملأا لىع قافنلإا ليلقت مهرودقبم نكي لم وأ ؛)127.6و 1.3 ينب ةيحجرلأا ةبسن ةبسن تحوارت ذإ %95 ةقث ةلصافب 2.5 ةيحجرلأا ةبسن( ةيحصلا ةياعرلا ىتح نأشلا ةليئض فئاظوب مايقلا مهيلع ناك وأ )4.2و 1.5 ينب ةيحجرلأا ةقث ةلصافب 3.4 ةيحجرلأا ةبسن( ةياعرلا تاقفن فيلاكت عفد نم اونكمتي يرثأت دوجو ظحول دقو .)5.5و 2.1 ينب ةيحجرلأا ةبسن تحوارت ذإ %95 مهتياعر لىوتي نورخآ لافطأ ةياعرلا ءاتيإ لىع مئاقلا ىدل نوكي امدنع ئيماح 0.1 ينب ةيحجرلأا ةبسن تحوارت ذإ %95 ةقث ةلصافب 0.2 ةيحجرلأا ةبسن( .)0.5و لافطلأل ةبسنلاب ةيحصلا ةياعرلا لىإ لوصولا لُبُس ينسحت صرتقي لا :جاتنتسلاا وأ رفسلل مزلالا تقولا ليلقت درجم لىع اههباشت يتلا قطانلماو ايبماغ في ينئماقلل ةمعادلا تاكبشلا ينسحت اضيأ بلطتي لب ،يحصلا قفرلما لىإ ةفاسلما .اهنوجاتحي يتلا ةيلالما دراولما يرفوتو ةيحصلا ةياعرلا ءاتيإ لىع References Murray CJ, Laakso T, Shibuya K, Hill K, Lopez AD. Can we achieve Millennium 1. Development Goal 4? New analysis of country trends and forecasts of under-5 mortality to 2015. Lancet 2007;370:1040-54. PMID:17889243 doi:10.1016/S0140-6736(07)61478-0 Black2. RE, Morris SS, Bryce J. Where and why are 10 million children dying every year? [see comment]. Lancet 2003;361:2226-34. PMID:12842379 doi:10.1016/S0140-6736(03)13779-8 Pison3. G, Trape JF, Lefebvre M, Enel C. Rapid decline in child mortality in a rural area of Senegal. Int J Epidemiol 1993;22:72-80. PMID:8449650 doi:10.1093/ije/22.1.72 Greenwood4. BM, Bradley AK, Byass P, Greenwood AM, Menon A, Snow RW, et al. Evaluation of a primary health care programme in the Gambia. II. Its impact on mortality and morbidity in young children. J Trop Med Hyg 1990;93:87-97. PMID:2325198 Magnani5. RJ, Rice JC, Mock NB, Abdoh AA, Mercer DM, Tankari K. The impact of primary health care services on under-five mortality in rural Niger. Int J Epidemiol 1996;25:568-77. PMID:8671558 doi:10.1093/ije/25.3.568 Becher6. H, Muller O, Jahn A, Gbangou A, Kynast-Wolf G, Kouyate B. Risk factors of infant and child mortality in rural Burkina Faso. Bull World Health Organ 2004;82:265-73. PMID:15259255 Van7. den Broeck J, Eeckels R, Massa G. Maternal determinants of child survival in a rural African community. Int J Epidemiol 1996;25:998-1004. PMID:8921486 doi:10.1093/ije/25.5.998 Tsay8. C. Possible effects of transportation on mortality differentials in central Taiwan. Jing Ji Lun Wen 1985;13:145-66. PMID:12178379 James9. C, Morris SS, Keith R, Taylor A. Impact on child mortality of removing user fees: simulation model. BMJ 2005;331:747-9. PMID:16195292 doi:10.1136/bmj.331.7519.747 Baume10. C, Helitzer D, Kachur SP. Patterns of care for childhood malaria in Zambia. Soc Sci Med 2000;51:1491-503. PMID:11077952 doi:10.1016/ S0277-9536(00)00049-6 Hampshire11. K. Networks of nomads: negotiating access to health resources among pastoralist women in Chad. Soc Sci Med 2002;54:1025-37. PMID:11999500 doi:10.1016/S0277-9536(01)00078-8 Gordon12. AJ, Yoder S, Camara M. Responding to childhood illness in upper Guinea: Ecological approaches, prospective ethnography, and health promotion with IMCI (intergrated management of childhood illness). Paper presented at the conference: Locating the field: the ethnography of medical research in Africa, Kilifi, Kenya, 4–9 December 2005. Farafenni13. Demographic Surveillance System Report: profile of the Farafenni DSS, the Gambia. Farafenni: Medical Research Council; 2004. Standardized14. VA questionnaire (revised August 2003). Available from: http:// www.indepth-network.org/index.php?option=com_content&task=view&id=9 8&Itemid=186 [accessed on 20 May 2008]. Hill PC, Stevens W, Hill S, Bah J, Donkor SA, Jallow A, et al. Risk factors 15. for defaulting from tuberculosis treatment: a prospective cohort study of 301 cases in the Gambia. Int J Tuberc Lung Dis 2005;9:1349-54. PMID:16466057 Sherbourne16. CD, Stewart AL. The MOS social support survey. Soc Sci Med 1991;32:705-14. PMID:2035047 doi:10.1016/0277-9536(91)90150-B Begin17. F, Frongillo EA Jr, Delisle H. Caregiver behaviors and resources influence child height-for-age in rural Chad. J Nutr 1999;129:680-6. PMID:10082774 Cassell18. JA, Leach M, Fairhead JR, Small M, Mercer CH. The social shaping of childhood vaccination practice in rural and urban Gambia. Health Policy Plan 2006;21:373-91. PMID:16940303 doi:10.1093/heapol/czl020 Households19. U-KM. Kinship and access to reproductive health care among rural Muslim women in Jaipur. Econ Pol Wkly (Mumbai, India) 1999;34:10. Fantahun M, Berhane Y, Wall S, Byass P, Hogberg U. Women’s involvement in 20. household decision-making and strengthening social capital-crucial factors for child survival in Ethiopia. Acta Paediatr 2007;96:582-9. PMID:17306012 doi:10.1111/j.1651-2227.2007.00147.x Armstrong Schellenberg JR, Nathan R, Abdulla S, Mukasa O, Marchant TJ, 21. Tanner M, et al. Risk factors for child mortality in rural Tanzania. Trop Med Int Health 2002;7:506-11. PMID:12031072 doi:10.1046/j.1365- 3156.2002.00888.x Binka FN, Maude GH, Gyapong M, Ross DA, Smith PG. Risk factors for 22. child mortality in northern Ghana: a case-control study. Int J Epidemiol 1995;24:127-35. PMID:7797334 doi:10.1093/ije/24.1.127 Pickering H, Hayes RJ, Ng’andu N, Smith PG. Social and environmental 23. factors associated with the risk of child mortality in a periurban community in The Gambia. Trans R Soc Trop Med Hyg 1986;80:311-6. PMID:3787693 doi:10.1016/0035-9203(86)90045-3 Shier RP, Dollimore N, Ross DA, Binka FN, Quigley M, Smith PG. Drinking 24. water sources, mortality and diarrhoea morbidity among young children in northern Ghana. Trop Med Int Health 1996;1:334-41. PMID:8673836 doi:10.1046/j.1365-3156.1996.d01-55.x 224 Bull World Health Organ 2009;87:216–224 | doi:10.2471/BLT.08.052175 Research Health-care access and child mortality in the Gambia Merrin E Rutherford et al. Maldonado G, Greenland S. Simulation study of confounder-selection 25. strategies. Am J Epidemiol 1993;138:923-36. PMID:8256780 Holtz TH, Kachur SP, Marum LH, Mkandala C, Chizani N, Roberts JM, et al. 26. Care seeking behaviour and treatment of febrile illness in children aged less than five years: a household survey in Blantyre District, Malawi. Trans R Soc Trop Med Hyg 2003;97:491-7. PMID:15307408 doi:10.1016/S0035- 9203(03)80003-2 Fosu GB. Childhood morbidity and health services utilization: cross-27. national comparisons of user-related factors from DHS data. Soc Sci Med 1994;38:1209-20. PMID:8016686 doi:10.1016/0277-9536(94)90186-4 Hill Z, Kendall C, Arthur P, Kirkwood B, Adjei E. Recognizing childhood 28. illnesses and their traditional explanations: exploring options for care-seeking interventions in the context of the IMCI strategy in rural Ghana. Trop Med Int Health 2003;8:668-76. PMID:12828551 doi:10.1046/j.1365- 3156.2003.01058.x Mtango FD, Neuvians D, Broome CV, Hightower AW, Pio A. Risk factors for 29. deaths in children under 5 years old in Bagamoyo district, Tanzania. Trop Med Parasitol 1992;43:229-33. PMID:1293726 Ensor T, San PB. Access and payment for health care: the poor of Northern 30. Vietnam. Int J Health Plann Manage 1996;11:69-83. PMID:10157066 doi:10.1002/(SICI)1099-1751(199601)11:1<69::AID-HPM414>3.0.CO;2-P Reyes H, Perez-Cuevas R, Salmeron J, Tome P, Guiscafre H, Gutierrez G. 31. Infant mortality due to acute respiratory infections: the influence of primary care processes. Health Policy Plan 1997;12:214-23. PMID:10173402 doi:10.1093/heapol/12.3.214 Velema JP, Alihonou EM, Gandaho T, Hounye FH. Childhood mortality 32. among users and non-users of primary health care in a rural west African community. Int J Epidemiol 1991;20:474-9. PMID:1917252 doi:10.1093/ ije/20.2.474 Greenwood BM, Greenwood AM, Bradley AK, Tulloch S, Hayes R, Oldfield FS. 33. Deaths in infancy and early childhood in a well-vaccinated, rural, West African population. Ann Trop Paediatr 1987;7:91-9. PMID:2441658 Sodemann M, Jakobsen MS, Molbak K, Alvarenga IC Jr, Aaby P. High mortality 34. despite good care-seeking behaviour: a community study of childhood deaths in Guinea-Bissau. Bull World Health Organ 1997;75:205-12. PMID:9277007 Pronyk PM, Hargreaves JR, Morduch J. Microfinance programs and better 35. health: prospects for sub-Saharan Africa. JAMA 2007;298:1925-7. PMID:17954543 doi:10.1001/jama.298.16.1925