Payment for deliveries in Sierra Leone.

Payment for deliveries in Sierra Leone N.C. Edwards,' N.J. Birkett,2 & P.A. Sengeh3 The type and amount of payment for deliveries were investigated in 1982 during a survey on health status in two districts. Data on the payments made for 83S% of the 2591 deliveries in 535 randomly selected study villages showed that the most common method of payment was in cash only. Payments in kind were mostly given to trained traditional birth attendants (TBAs) (for 38.1% of their deliveries) and rare for professional staff (2.9% of deliveries). The total amount paid for a delivery differed significantly with the type of birth attendant (P < 0.00001) and the place of delivery (hospital, peripheral health unit or home) (P < 0.00001). The total average payment for a delivery was highest for professional birth attendants (Le 16.60) and lowest for untrained TBAs (Le 4.85) (Le 2 = approx. US$ 1 at the time of the study). The outcome of a delivery had a significant effect on the amount paid. Payments were significantly higher for stillbirths than for live births among professional and auxiliary birth attendants (P < 0.0001). However, the trained and untrained TBAs received less payment for stillbirths (Le 2.25) than for live births (Le 4.89) (P = 0.0146). The results show that there are several levels of financial disincentives for pregnant women requiring the services of trained auxiliary or professional health workers at the time of delivery. Introduction In developing countries, untrained personnel con- duct over two thirds of all deliveries (1).a Although many of these are normal births, high rates of mater- nal and perinatal mortality (2-4) attest to the fact that health care services are inaccessible or inappro- priately delivered to women who are at risk during pregnancy or childbirth and in the postnatal period. Several key factors contribute to this situation. First, the number of trained health personnel in the rural areas of many developing countries is inade- quate, the urban centres attracting a larger propor- tion of health care providers (5-7). Secondly, women may prefer traditional birth attendants (TBAs) whose services include, for example, the observation of cultural rites during the birth process (1, 8)a, b Thirdly, in countries such as Sierra Leone, TBAs are respected citizens chosen by the community for training and they have a strong rapport with com- munity members (8)," in sharp contrast to the young ' School of Nursing, McMaster University, Hamilton, Ontario, Canada. Requests for reprints should be sent to this author at P.O. Box 327, Metcalfe, Ontario KOA 2PO, Canada. 2 Department of Epidemiology and Community Health, University of Ottawa, Ottawa, Ontario, Canada. 3 Monitoring and Evaluation Officer, Bo Pujehun Rural Develop- ment Project, Sierra Leone. ' Traditional birth attendants: activities, problems and prospects for the future. Unpublished WHO document prepared for the Inter- regional Consultation on Traditional Birth Attendants, 10-14 December 1979, Mexico City, Mexico. b Wliams, B. 7The traditional birth attendant: training and uti- lization. Sierra Leone Ministry of Health, 1979. secondary school graduates who are self-selected for the midwifery role and are often without personal experience of childbirth. Finance has also been identified as a factor which reduces the accessibility of adequate health care for village residents (8-12). However, there have been few reports of studies documenting the impact of health care costs on the accessibility of services to pregnant women in developing countries. The present survey provided an opportunity to determine the type and amount of payment made for deliveries by trained versus untrained personnel. It also allowed us to investigate the relationship between payment and perinatal health outcomes. Background Sierra Leone is located in the equatorial lowlands of West Africa; the majority (85%) of the 3.7 million inhabitantsc live in the rural areas (13) where trans- port and communication services are limited and literacy rates are often below 25% (14). The cross- sectional survey was carried out in two districts of the Southern Province where an integrated rural development programme had been initiated by the government of Sierra Leone and the German Agency for Technical Cooperation (GTZ) in 1980. The popu- lation of approximately 418 000 in the two study dis- tricts reside in over 2100 villages (13). Mortality rates are particularly high among infants and young children, estimates of rates for the c Central Statlstics Office. 1986 population census of Sierra Leone (unofficial results, 1986). Bulletin of the World Health Organization, 67 (2): 163-169 (1989) © World Health Organization 1989 163 N. C. Edwards et al. study area showing considerable variation depend- ing on the data source, the type of data collection, and the demographic techniques applied (15)." Data from the 1970s provide estimates of infant mortality rates for districts in the Southern Province which range from 137 (16) to 308 per 1000 live births."f Approximately 25% of all female deaths in the 15-44-year age group are due to complications of pregnancy and childbirth (17). Heafth care services. Health care in the rural areas is provided by both traditional practitioners and western health care facilities. The traditional sector constitutes a variety of health workers, many of whom have learned their skills through an informal apprenticeship method of training (e.g., TBAs and bone setters). TBAs provide a large proportion of the antenatal, intrapartum and postnatal care in the rural areas of Sierra Leone. It is estimated that they attend over 85% of deliveries in the rural areas (15, 17). The western health care sector includes govern- ment peripheral health units (PHUs), district and provincial hospitals, and mission clinics and hospi- tals. PHUs are the most accessible rural health care facilities, each unit being staffed by at least one maternal and child health aide (MCH aide). These auxiliary nurses are primarily responsible for ante- natal care, deliveries and child care. In an effort to increase the accessibility of health services to the populace, the Ministry of Health has attempted to provide health care at minimal cost. However, low wages for government health workers, the high cost of imported drugs, and lack of incen- tives for rural health workers are factors which may lead to a charge for health care. The magnitude of these costs cannot be determined from government health records. Thus it is important for health plan- ners to obtain data on the direct costs of health ser- vices for patients. This type of information reflects the true financial disincentives which may discourage patients from using the available services. d Williams, B. Ministry of health policy on non-governmental organizations in primary health care in Sierra Leone. Presented at Primary Health Care Seminar, Nixon Memorial Methodist Hos- pital, Segbwema, Sierra Leone, 1980. * Kandeh, H.B.S. Infant and childhood mortality differentials in Sierra Leone. Presented at Expert Group Meeting on Fertility and Mortality Levels, Patterns and Trends in Africa and their Policy Implications, held in Liberia, 1977. ' Kand.h, H.B.S. & Dow, T.E. Correlates of infant mortality in the chiefdoms of Sierra Leone. Presented at national seminar on population, employment and development planning in Sierra Leone, Freetown, 1980. Methodology The target population for the cross-sectional survey included all the people in the two districts (Bo and Pujehun). The sampling frame was villages with a population of under 2000 as listed in the 1974 census (the most recent census data available at the time of the study). The target area was divided into eight strata on the basis of the following criteria: district (Bo versus Pujehun); village population (<200 versus 200-2000); distance to the PHU (<5 km (3 miles) versus >5 km). Within each stratum, a simple cluster sample of villages was selected with each village having an equal probability of selection. The number of villages selected from each stratum was chosen to ensure approximately the same number of residents in each stratum-specific sample. After selec- tion, local village chiefs were approached for per- mission to interview all households in the village. There was no replacement for villages where ihe chief refused access. Only data (collected during the interview) rela- ting to vital events and payment for deliveries were considered. Retrospective questioning of all house- holds was used to obtain information on all women of child-bearing age who experienced one or more of the following vital events during the calendar year 1982: (a) delivered a child (live birth or stillbirth); (b) lost a child under one year of age; (c) died during pregnancy, childbirth or the postpartum period. When a mother was not available for interview, the husband or another knowledgeable informant was asked to respond. In addition to details of the vital event (date, cause of death, data source used to verify information), respondents were questioned about features of the delivery (type of birth attendant, amount and type of payment, place of delivery, and transportation costs to the health facilities), charac- teristics of the mother (age, education), and prenatal care (antenatal clinic attendance, tetanus toxoid immunization). Respondents were asked whether or not they had given the birth attendant any payment for the delivery. The total amount of cash and the nature and quantity of the payment in kind were recorded; with regard to the latter, the current market prices in the main provincial town (population, 40000) were used to assign a cash value to the payments given in kind. To obtain data on transportation costs, know- ledgeable village informants such as the village chief were asked to describe: how they would travel from their village to the nearest PHU and closest hospital and the cost of this transportation. Each mode of transportation for the trip was identified and the 164 Payment for deliveries In Sierra Leone Table 1: Cost of transport to peripheral health units (PHUs) and hospitals In Bo and Pulehun districts Cost of transport (in leones)b Significancec District and stratum" To PHU To hospital (t-test) Bo district:d Small, <3 miles 0.33 ± 0.12 (117)° 4.33 ± 0.54 (117) -14.2** Large, <3 miles 0 35 + 0.21 (22) 5.51 ± 1.41 (21) -7-3** Small, >3 miles 1.86 + 0.36 (126) 6.21 ± 0.59 (127) -12.4** Large, >3 miles 1.61 ± 0.90 (25) 5.53 ± 1.13 (24) -5.4** Total 1.11 + 0.16 (290) 5.36 ± 0.32 (289) -23.15*** Puiehun district:' Small, <3 miles 0.81 ± 0.37 (120) 7.82 ± 2.03 (118) -6.7** Large, <3 miles 0.50 ± 0.35 (22) 11.59 ± 7.21 (22) -3.0* Small, >3 miles 2.35 + 0.54 (70) 13.42 ± 3.65 (71) -5.8** Large, >3 miles 2.02 + 1.79 (17) 17.79 ± 1.05 (17) -2.8* Total 1.34 + 0.24 (229) 10.67 ± 1.56 (228) -11 .57** Overall total 1.19 + 0.14 (519) 7.33 ± 0.61 (517) 19.16** Small and large refer to village populations of <200 and 200-2000, respectively. Distances are from the village to the PHU or hospital (3 miles = 5 km approx.). b Costs are given as the mean ±95% C.l. At the time of the study, Leone 2 = USS 1 approximately. c Unpaired t-test value: *P < 0.01, **P < 0.001, and ***P < 0.0001 (two-tailed). d Costs not available for 10 cases to the PHU and 11 cases to hospital. " Figures in parentheses are the number of cases in each category. ' Costs not available for 6 cases to the PHU and 7 cases to hospital. cost of each stated. All costs were recorded in the local currency (leone, Le). At the time of this study, Le 2 = US$ 1 approximately. Data collection The survey team, consisting of nineteen males employed by the Ministry of Health (Endemic Disease Control Unit), underwent a nine-day train- ing programme which included mock interviews in the classroom and actual interviews in three villages outside the study sample. Training continued until levels of inter- and intra-observer agreement of over 95 percent were obtained on all questions. A detailed description of the development and pretesting of the interviewer-administered question- naire is provided elsewhere (18). The validity and reliability of the responses to questions were checked using written records as the standard. Supervision in the field was provided by a team of fourteen supervisors who met the interviewers at least once a week during the data collection phase and accompanied them on 183 of the household interviews to confirm that the responses were record- ed correctly. The supervisors also re-interviewed a sample of households to ascertain the validity of the responses recorded by the interviewer when the supervisor was not present. Data collection was completed over a four-month period. Data were computerized by project staff in Sierra Leone using MICROSTAT on an Apple II. The data were transferred to a database developed under SIR on a VAX-750 in Canada for analysis. Adjustment for sampling design and clustring. Esti- mates of variance and confidence intervals were obtained after allowance for the sampling design. Formulae from sampling theory for a stratified single-stage cluster sample were used (19). All calcu- lations were done with the special purpose computer programme SUPERCARP (20). Comparisons of the cost of delivery in different groups were done using a one-way ANOVA after log-transforming the data to reduce variance hetero- scedasticity. Findings Response rate A village-level response rate was calculated for each of the study strata in the two districts. In Bo district, data were collected in 300 of the 338 villages (88.8%); 33 villages were deserted, and in five villages the enumerators were refused permission to work (total population of these 38 villages was 1678 or 4.4% of the estimated total population in the 338 study villages). In Pujehun district, the response rate was lower. The survey was completed in 235 of 311 villages (75.6%); 15 villages were deserted, and in 61 villages the enumerators were unable to complete the work primarily because local authorities deemed these areas unsafe due to political unrest (total population of these 76 villages was 19.8% of the total popu- lation of the original 311 study villages). 165 N. C. Edwards et al. Table 2: Average total payment for deliveries In cash and kInd In Bo and Pujehun districts by stratum, 1982 Total payment for deliveries District and stratum' (in leones)b c Bo district: Small, <3 miles 5.47 ± 0.49 (278)d Large, <3 miles 5.37 ± 0.83 (293) Small, >3 miles 6.37 ± 1.01 (316) Large, >3 miles 4.49 ± 0.57 (242) Total 5.90 ± 0.56 (1129) Pujehun district: Small, <3 miles 5.15 ± 0.81 (365) Large, <3 miles 5.38 ± 0.88 (292) Small, >3 miles 4.00 ± 0.51 (193) Large, >3 miles 4.20 ± 0.80 (185) Total 4.75 ± 0.53 (1035) Overall total 5.50 ± 0.41 (2164) Small and large refer to village populations of <200 and 200- 2000, respectively. Distances are from the village to the PHU or hospital (3 miles = 5 km approx.). b Payments are given as the mean ±95% C.l (Leone 2 = USS 1 approx.). c ANOVA: Df = 7, 2092; F = 6.3074, P < 0.00001 (unweighted, log-transformed data). d Figures in parentheses are the number of cases in each cate- gory. Data were collected on a total of 1378 and 1213 births in Bo and Pujehun districts, respectively. A total of 54 stillbirths and 2537 live births were reported in the two districts, including 18 multiple births; 76.8% of all these births were attended by untrained TBAs (the type of birth attendant could not be categorized for 4.4% of births, n = 114, owing to insufficient information). Ninety percent (n = 2327) of all deliveries took place in a village setting without formal health care facilities: e.g., in the home of the mother or the TBA, 67.4% (n = 1746); in a bondu (women's secret society) hut, 13.7% (n = 356); in a kitchen, 5.5% (n = 140). Only 0.13% of all deliveries were caesarian sections. The maternal mortality rates were 444 per 100000 live births in Bo district and 422 per 100000 in Pujehun district. Causes of the 12 reported mater- nal deaths included postpartum haemorrhage (1), eclampsia (1), obstructed labour (2), anaemia (2), and other causes (6). The fetal mortality rates in Bo and Pujehun dis- tricts were 18.9 and 23.1 per 1000, respectively. Transportation costs Only 37.4% of villages in Bo district and 49.4% of study villages in Pujehun district reportedly had a road leading into the village which was motorable using vehicles with four wheels during both dry and rainy seasons. The average cost of travel (Table 1) to a peripheral health unit (PHU) ranged from Le 0.33 to Le 1.86 in Bo district and from Le 0.50 to Le 2.35 in Pujehun district. The average cost of transporta- tion to a hospital as compared with transport to a PHU was significantly higher for all strata (P < 0.1). Payment for deliveries Data on the payment given for deliveries was pro- vided for 83.5% (n = 2164) of all reported births. The total payment for deliveries in the four strata of each district is summarized in Table 2, the most common method across all strata being payment in cash only. The average total payment for deliveries in the four strata in Bo district ranged from Le 4.49 to Le 6.37, and was slightly lower (ranging from Le 4.00 to Le 5.38) in Pujehun district. Payment in kind (with or without supplemen- tary payment in cash) was more frequent for trained TBAs (38.1%) and untrained TBAs (27.2%) than for professional staff like physicians and nurses (2.9%). Table 3: Average payments In cash and kind by category of birth attendant In Bo and Pujehun districts, 1982 Type of payment" Category of birth attendant In cashb In kindc Totald Professional 15.64 ± 6.85 (68)' 8.93 + 4.14 (2) 16.60 ± 7.06 (68) Auxiliary nurseg 7.39 ± 0.95 (222) 4.15 ± 1.41 (17) 7.62 ± 0.92 (225) Trained TBA 5.00 ± 0.87 (97) 3.70 + 1.00 (43) 6.05 ± 1.43 (113) Untrained TBA 3.08 ± 0.25 (1627) 4.96 + 0.40 (460) 4.85 ± 0.35 (1690) All attendants 3.92 ± 0.39 (2014) 4.92 ± 0.37 (522) 5.47 ± 0.42 (2096)h ' Payments are given in leones as the mean ±95% C.I. (Leone 2 = US$ 1 approx.). b ANOVA (payment in cash): Df = 3, 2010; F = 168.78; P < 0.00001 (unweighted, log-transformed data). c ANOVA (payment in kind): Df = 3, 518; F = 1.56; P = 0.199 (unweighted, log-transformed data). d ANOVA (total payment in cash and kind): Df = 3, 2092; F = 86.34; P < 0.00001 (unweighted, log-transformed data). ' Professional birth attendants include physicians, state-registered nurse midwives, and state-enrolled community health nurses. 'Auxiliary nurses include maternal and child health aides and village maternity assistants. Figures in parentheses are the number of cases in each category. h The category of birth attendant could not be determined for 68 births; these births are not included in this Table. 166 Payment for deliveries In Sierra Leone Table 4: Average payments In cash and kind by the place of delivery In Bo and Pulehun districht, 1982 Type of payment" Location In cashb In kindc Totald Hospital 16.87 ± 9.03 (46)' 8.93 ± 4.14 (2) 18.59 ± 9.69 (46) Peripheral health unit 8.49 ± 0.81 (159) 3.85 ± 2.53 (10) 8.74 ± 0.78 (161) Village 3.36 ± 0.25 (1877) 4.86 ± 0.38 (527) 5.03 ± 0.32 (1957) All locations 3.94 ± 0.38 (2082) 4.89 ± 0.37 (539) 5.48 ± 0.41 (2164) Payments are given in leones as the mean ±95% C.l. (Leone 2 = USS 1 approx.). b ANOVA (payment in cash): Df = 2, 2079; F = 185.22; P < 0.00001 (unweighted, log-transformed data). c ANOVA (payment in kind): Df = 2, 536; F = 4.119; P = 0.168 (unweighted, log-transformed data). d ANOVA (total payment in cash and kind): Df = 2, 2161; F = 100.219; P < 0.00001 (unweighted, log-transformed data). ° Figures in parentheses are the number of cases in each category. The average total payment for deliveries (in cash and kind) was highest for professional staff (Le 16.60) and lowest for untrained TBAs (Le 4.85). Dif- ferences in payment in cash and total payment (in cash and kind) for a delivery, by birth attendant category, are statistically highly significant (Table 3). However, no significant differences were found in the amount of payment in kind by type of birth atten- dant. The average payment for deliveries attended by trained TBAs was significantly greater than the average payment to untrained TBAs (unpaired t-test: Df = 1,802; t = 2.47; P = 0.014; log-transformed, weighted data). Similarly, the average total amount paid was significantly more for deliveries in a hospital (Le 18.59) than for deliveries in a PHU (Le 8.74) or in a village (Le 5.03) (F = 100.2; Df = 2, 2161; P < 0.00001; log-transformed, unweighted data). A multivariate model indicates no effect of the mother's educational status (a socioeconomic status surrogate) (F = 1.69; Df = 2, 2078; P = 0.19), but a strong effect on the cost depending on the type of birth attendant (F = 80.22; Df = 3, 2078; P < 0.0005) after adjustment for education. Payment for stillbirths. The relationship between payment for deliveries and fetal mortality was explored by comparing the payment reported for stillbirths with that stated for live births. Data on total cost of delivery and category of birth attendant was provided for 50% of stillbirths (27 out of 54 stillbirths) and 81.6% of live births (2069 out of 2537 live births). A two-way analysis of variance was done com- paring payment for deliveries by type of birth atten- dant and delivery outcome (i.e., live birth or stillbirth). We found a highly significant interaction effect (F = 5.90; Df= 3, 2088; P < 0.001; log- transformed, unweighted data), and we therefore examined the impact of delivery outcome on delivery payment with groups of birth attendant categories. As shown in Table 5, the amount paid for stillbirths in comparison with live births is significantly more for professional and auxiliary birth attendants (F = 7.84; Df = 1, 291; P < 0.0001; unweighted, log- transformed data), and significantly less for tradi- tional birth attendants (F = 5.97; Df = 1, 1801; P = 0.0146; unweighted, log-transformed data). Discussion In our study, we found that TBAs receive less payment for attending a delivery and are more likely to receive at least some payment in kind rather than cash alone, compared with other categories of birth attendants. This holds irrespective of the mother's educational status. Traditionally, TBAs do not request payment for their services. However, a frequent complaint made by TBAs who have completed the training prog- ramme offered by the government is that they are not paid enough by the women they deliver (21). A Table 5: Total payments In cash and kind for delivery by delivery outcome and category of birth attendant Delivery outcome Total payment (in leones)" Professional and auxiliary birth attendants:b Stillbirth 46.52 + 9.30 (5)C Live birth 9.60 1.42 (288) Total 9.96 + 1.60 (293) Trained and untrained traditional birth aftendants:d Stillbirth 2.25 ± 0.70 (22) Live birth 4.89 ± 0.33 (1781) Total 4.85 ± 0.32 (1803) Payments are given as the mean +95% C.l. (Leone 2 = US$ 1 approx.). b ANOVA: Df = 1, 291; F = 7.835; P < 0.0001 (unweighted, log- transformed data). c Figures in parentheses are the number of cases in each cate- gory. d ANOVA: Df = 1, 1801; F = 5.97; P = 0.0146 (unweighted, log- transformed data). 167 N. C. Edwards et al. reason sometimes given for failure to pay these trained TBAs is that the village women view them as government employees (salaried workers), trained by the government and thus no longer requiring payment by village clientele. On the other hand, it is interesting to note that the average payment by mothers for trained TBAs is significantly higher than that for untrained TBAs. Whether this is because the mothers perceive them to have increased status, or whether it reflects a selection bias operating with respect to who gets trained (the Chief's wife versus village midwife), or whether it represents higher payment demanded by trained TBAs cannot be determined from our results. However, it does suggest that training a TBA might make her less accessible to mothers who deliver in the village. Pro- spective data comparing the delivery fee paid to a TBA prior to and following training would help to answer this question. Theoretically, deliveries in government PHUs and in district and provincial hospitals are provided free or at minimal cost (i.e., Le 0.10 to Le 0.20 daily). Our results indicate that this is not the case. The average fee paid directly by the mother for these deliveries is significantly higher than that paid to a TBA. Mortality rates for deliveries by trained per- sonnel are also higher than those by TBAs.9 This suggests that the perceived fee schedule might be a deterrent to women requiring the services of trained personnel. Our study data come from PHUs and hospitals run by the government or by missions and we have not attempted to differentiate between the cost of delivering in one or the other. However, one particular category of birth attendant, the maternal and child health aide, works almost exclusively in government-operated PHUs. Thus, it can be seen from our results that payment is made to govern- ment health workers in PHUs where the services are supposed to be provided at minimal or no cost. Transportation costs tend to be low, assuming that vehicles which are on their regular routes are used. However, some villages have only one vehicle leaving the village early in the morning and returning the same evening. If a medical emergency arises such as eclampsia or a prolapsed cord, for which a vehicle must make a special trip, then the transportation costs are significantly greater. The overall average payment for a delivery in the two districts is Le 5.50. This amount is only a portion of the costs incurred by a woman who deliv- ers in hospital. The average cost of one-way trans- " Edwards, N. & Komba Kono, G. Deliveries conducted by trained and untrained birth attendants in Sierra Leone: a comparison of fetal and neonatal mortality rates and an investigation of related factors. Presented at the International Epidemiological Associ- ation Xth Scientific Meeting, Vancouver, BC, August 1984. portation to a hospital is approximately equal to the delivery cost in Bo district and almost double the actual delivery cost in Pujehun district. Given that rural women in Sierra Leone would rarely go to a PHU or hospital for delivery without being accom- panied by at least one other person, the actual trans- portation costs for a delivery are at least twice that indicated by our data. The high fetal and maternal mortality rates provide some evidence of the diffi- culties in making referrals which impede timely and appropriate interventions, e.g., caesarian sections. Our finding of an increased cost for stillbirths born in hospital must be interpreted with caution as data were collected on just five stillbirths delivered by professional or auxiliary staff. It may be that the women who delivered a stillbirth in hospital had complications and a higher fee was charged for the drugs and laboratory tests that were required. In contrast, the TBAs were paid less for a stillbirth than for a live birth. This may reflect underlying tradi- tional values and customs concerning the outcome of a delivery. This matter warrants further investiga- tion. Acknowledgements Nancy Edwards held a scholarship from the Canadian International Development Agency and a fellowship from The Commonwealth Foundation at the time of this study. Resume Prix de l'accouchement en Sierra Leone Dans le cadre d'une enquete transversale sur la situation sanitaire dans deux districts (Bo et Pujehun), on a analyse les formes de paiement et les montants verses pour les accouchements. Des donnees retrospectives sur ces paiements ont ete recueillies en 1982 au sujet de 83,5% des 2591 accouchements ayant eu lieu dans 535 villages choisis au hasard pour l'etude. On a egalement analyse le cout du moyen de transport necessaire pour se rendre au poste de sante peripherique ou 'a l'h6pital. La forme la plus courante etait un paiement uniquement en especes. Les paiements en nature concernaient surtout les accoucheuses tradition- nelles qualifiees (38,1% des accouchements qu'elles avaient pratiqu6s) et ils etaient rares pour le personnel de niveau professionnel (2,9% des accouchements). Le montant total verse pour un accouchement differait beaucoup en fonction du niveau de la personne presente (ANOVA P < 0,00001) et du lieu de l'accouchement (ANOVA 168 Payment for deliveries In Sierra Leone P < 0,00001). En moyenne, la somme versee pour un accouchement etait maximale pour un agent professionnel (par exemple medecin ou infirmiere qualifiee), soit 16,60 leones, et minimale pour une accoucheuse non qualifiee, soit 4,85 Le (2 Le = environ US$ 1). C'etait a l'h6pital que le prix moyen d'un accouchement 6tait le plus eleve (18,59 Le); venaient ensuite le poste de sante p6ripherique (8,74 Le) puis l'accouchement au village (5,03 Le). D'autre part, I'issue de l'accouchement a eu un effet sur la somme versee. Celle-ci etait net- tement plus elevee pour un enfant mort-ne que pour une naissance vivante en presence d'un professionnel ou d'une accoucheuse auxiliaire (P <0,0001). Toutefois, les accoucheuses tradition- nelles, qualifiees ou non, etaient moins payees pour un enfant mort-ne (2,25 Le) que pour une naissance vivante (4,89 Le) (P = 0,0146). D'une facon gen6rale, les accoucheuses traditionnelles etaient moins bien payees et avaient plus de chance de recevoir au moins une partie du paie- ment en nature que les autres categories d'agents de sante. Cela etait vrai independamment du niveau d'instruction de la m6re. Le cout du transport (aller simple) pour se rendre dans un hopital du district de Bo etait a peu pres equivalent au prix de l'accouchement tandis que, dans le district de Pujehun, il atteignait presque le double. Le fait que la plus grande partie des accouchements dans les petits villages et les villages recules aient ete pratiques par un personnel non forme donne a penser que l'absence de moyens de transport et le cout sont peut-etre des facteurs qui emp6chent d'utiliser les services offerts par le poste de sant6 peripherique et lth6pital. On voit donc qu'il existe plusieurs obstacles financiers auxquels se heurtent les femmes enceintes qui ont besoin des services d'agents de sante auxiliaires ou professionnels qualifies au moment de l'accouchement. II s'agit du prix du transport jusqu'au centre de sante, du paiement de l'accouchement en esp&ces plutot qu'en nature et du prix de l'accouchement proprement dit. References 1. Traditional midwives and family planning. Population reports, 8: 437-488 (1980). 2. Ross, D.A. The trained traditional birth attendant and neonatal tetanus. In: Mangay Maglacas, A. & Simons, J., ed. The potential of the traditional birth attendant. Geneva, World Health Organization, 1986 (WHO Offset Publication No. 95), pp. 8-21. 3. Chen, P.C.Y. An assessment of the training of the tra- ditional birth attendant of rural Malaysia. Medical journal of Malaysia, 31: 93-99 (1976). 4. Mahler, H. The safe motherhood initiative: a call to action. Lancet, 1: 668-670 (1987). 5. Evans, J.R. et al. Shattuck lecture - Health care in the developing world: problems of scarcity and choice. New England journal of medicine, 305: 1117- 1127 (1981). 6. Navarro, V. The underdevelopment of health or the health of underdevelopment: an analysis of the dis- tribution of human health resources in Latin America. International journal of health services, 4: 5-27 (1974). 7. Stock, R. Health care for some: a Nigerian study of who gets what, where and why? International journal of health services, 15: 469-484 (1985). 8. Chen, P.C.Y. Traditional and modern medicine in Malaysia. Social science and medicine, 15A: 127-136 (1981). 9. Lasker, J.N. Choosing among therapies: illness behaviour in the Ivory Coast. Social science and medicine, ISA: 157-168 (1981). 10. IDS Health Group. Health needs and health services in rural Ghana. Health needs and health services in the districts studied-Part A. Social science and medicine, ISA: 417-426 (1981). 11. IDS Health Group. Health needs and health services in rural Ghana. The assessment of resources and organization-Part A. Social science and medicine, 15A: 453-458 (1981). 12. Kroeger, A. South American Indians between tradi- tional and modern health services in rural Ecuador. Bulletin of the Pan American Health Organization, 16: 242-252 (1982). 13. Central StatIstIcs Office. 1974 population census of Sierra Leone. Freetown, 1976. 14. Kaplan, J. Area handbook for Sierra Leone. Wash- ington, DC, American University, 1976. 15. Williams, B. et al. Maternal and child health care booklet. Freetown, Ministry of Health, 1977. 16. Ministry of Health. Medical statistics reports for 1976/ 1977, Sierra Leone. Freetown, 1978. 17. Hardiman, M.G.W. & Mldgley, J.O. Planning and the health of mothers and children in the rural areas of Sierra Leone. Journal of tropical pediatrics, 7: 83-87 (1981). 18. Edwards, N. Traditional birth attendants: key provi- ders of maternal and infant care in Sierra Leone. Western journal of nursing research, 9: 335-347 (1987). 19. Cochrane, W.G. Sampling techniques, 3rd ed. Toronto, John Wiley & Sons, 1977, pp. 249-270. 20. Hldiroglou, M.A. et al. Super Carp. Survey Section Statistical Laboratory, Iowa State University, 1980. 21. Williams, B. & Yumkella, F. An evaluation of the training of traditional birth attendants in Sierra Leone and their performance after training. In: Mangay Maglacas, A. & Simons, J. ed. The potential of the traditional birth attendant. Geneva, World Health Organization, 1986 (WHO Offset Publication No. 95), pp. 35-50. Reprint No. 4963 169