Willingness to pay for child survival: Results of a national survey in Central African Republic

Social Science and Medicine

Volumn 43, Issue 6, 1996, Pages 985-998

  Weaver, Marcia ^a   Ndamobissi, Robert ^b   Kornfield, Ruth ^d   Blewane, Cesaire ^c   Sathe, Antoine ^c   Chapko, Michael ^d   Bendje, Nicholas ^d   Nguembi, Emmanuel ^d   Senwara Defiobona, Jacques ^c  

^a UNIVERSITY OF WASHINGTON   (United States)

^b Min of Economics Plan S   (Central African Republic)

^c Directorate of Preventive Medicine   (Central African Republic)

^d NONE

Author keywords

Child survival; Contingent valuation; Health care finance; Quality of care; User fees

Indexed keywords

AFRICA SOUTH OF THE SAHARA; ARTICLE; CHILD HEALTH; DIARRHEA; HEALTH CARE COST; HEALTH CARE QUALITY; HOUSEHOLD; HUMAN; INTESTINE PARASITE; MALARIA; MEDICAL FEE; RESPIRATORY TRACT INFECTION; RURAL AREA; SEXUALLY TRANSMITTED DISEASE; URBAN AREA;

CHILD SURVIVAL; HEALTH CARE FINANCING; QUALITY OF CARE; USER FEES; WILLINGNESS TO PAY;

CENTRAL AFRICAN REPUBLIC;

EID: 0030222815     PISSN: 02779536     EISSN: None     Source Type: Journal    
DOI: 10.1016/0277-9536(96)00015-9     Document Type: Article

References (35)

1. 1. For a synthesis of cost recovery schemes in Africa, most of which involve user fees for drugs, see World Bank, Africa Technical Department, Population, Health and Nutrition Department. Pharmaceutical Expenditures and Cost Recovery Schemes in Sub-Saharan Africa. Technical Working Paper No. 4, 1992.
2. 2. Collins D. H. and Hussein I. M. Financial Sustainability of Health Programs in Kenya: The Role of Cost Sharing. Oral Presentation at the 121st Annual Meeting of the American Public Health Association, 1993.
3. 3. Litvak J. I. and Bodart C. User fees plus quality equals improved access to health care: results of a field experiment in Cameroon. Soc. Sci. Med. 37(3), 369, 1993.
4. 4. The Ministry of Public Health and Social Affairs in Central African Republic received a National Council on International Health award in June 1990 for their child survival program.
5. 5. Waddington C. J. and Enyimayew K. A. A Price to pay: the impact of user charges in Ashanti-Akim District, Ghana. Int. J. Hlth Planning Management 4, 25, 43, 1989.
6. 6. Yoder R. A. Are people willing and able to pay for health services? Soc. Sci. Med. 29, 36, 1989.
7. 7. Frankish J. G. Day hospital fees and accessibility of essential health services. South African Medical J. 70, 286, 1986.
8. 8. Ministere de la Sante Publique du Niger et la Cooperation Medicale Belge. Financement des Medicaments par la Population: Experience de Tibiri (Dosso). Bilan Final. Projet RESSFOP, 1990.
9. 9. Ministere de la Sante Publique et des Affaires Sociales du Mali. Projet Magasins-Sante. Rapport Annuel 1988. Projet FED et Medecins Sans Frontieres-Belgique, 1988.
10. 10. Mitchell R. C. and Carson R. T. Using Surveys to Value Public Goods: the Contingent Valuation Method, pp. 235-259. Resources for the Future. Washington, DC, 1989.
11. 11. United States Department of Commerce. National Oceanic and Atmospheric Administration. Natural resource damage assessments under the Oil Pollution Act of 1990. Federal Register 58(10), 4601, January 15, 1993.
12. 12. Johannesson M., Jonsson B. and Borgquist L. Willingness to pay for antihypertensive therapy - results of a Swedish pilot study. J. Hlth Economics 10, 461, 1991.
13. 13. Johannesson M., Johansson P.-O., Kristrom B. and Gerdtham U.-G. Willingness to pay for antihypertensive therapy - further results. J. Hlth Economics 12, 95, 1993.
14. 14. Neumann P. and Johannesson M. The willingness to pay for in vitro fertilization: a pilot study using contingent valuation. Medical Care 32, 686, 1994.
15. 15. Whittington D., Briscoe J., Mu X. and Barron W. Estimating the willingness to pay for water services in developing countries: a case study in the use of contingent valuation surveys in Southern Haiti. Economic Development Cultural Change 293, 1990.
16. 16. Whittington D., Smith V. K., Okorafor A., Okore A. and Liu J. L. Giving respondents time to think in contingent valuation studies: a developing country application. J. Environ. Economics Management 22, 205, 1992.
17. 17. For more information on the survey design process, see Weaver M., Bendji N., Blewane C., Kornfield R. and Sathe A. Survey of Expenditures and Willingness to Pay for Health Care: Bangui, Central African Republic. Health Finance and Sustainability Project Technical Note, 1992.
18. 18. Barker B. Central African Republic. Health Finance and Sustainability Project Trip Report, 1992.
19. 19. Cameron T. A. A new paradigm for valuing non-market goods using Referendum Data: maximum likelihood estimation by censored logistic regression. J. Environ. Economics Management 15, 355, 1988.
20. 20. Weaver M., Ndamobissi R., Kornfield R., Chapko M., Blewane C., Sathe A. and Ngueretia L. P. Willingness to Pay for Health Care: A Comparison of Contingent Valuation and Traditional Economic Methods. Health Finance and Sustainability Project Technical Report, 1993.
21. 21. Marija J. Norusis/SPSS Inc. SPSS/PC + V2.0 Base Manual. p. B-184. SPSS Inc., Chicago, IL, 1988.
22. 22. Friedman M. The Theory of the Consumption Function. Princeton University Press. Princeton, NJ, 1957.
23. 23. Gertler P. and van der Gaag J. The Willingness to Pay for Medical Care: Evidence from Two Developing Countries, p. 80. Johns Hopkins University Press for the World Bank, Baltimore, 1990.
24. 24. Abel-Smith B. and Rawal P. Can the poor afford 'free' health services? A case study of Tanzania. Hlth Policy Planning 7, 331, 1993.
25. 25. Weaver M. User fees and patient behavior: evidence from Niamey National Hospital in Niger. Hth Policy Planning 10, 350, 1995.
26. 26. The survey design included 1280 households from 80 census tracts, but a team of interviewers was unable to conduct the survey in one census tract in Health Region II.
27. 27. There were separate listings for urban (towns with more than 5000 residents including Bangui) and rural (villages with less than 5000 residents) census tracts. For urban census tracts, a sampling interval was obtained by dividing the total number of individuals in urban census tracts by 29 (36% of 80 census tracts). For urban census tracts except Bangui, the sampling interval was reduced by one-half, because the sample was stratified to overrepresent these households. A starting point for the selection of census tracts was selected as a random number between one and the sampling interval. The first census tract was the one in which the individual who corresponded to the starting point lived. The sampling interval was added to the starting point, and the second census tract was the one in which the individual who corresponded to that sum lived. The procedure for rural households was similar, with the exception that the sampling interval was obtained by dividing the total number of individuals in rural census tracts by 38.
28. 28. The supervisor of the team of interviewers had the number of households for each of the census tracts where his/her team conducted interviews. The sampling interval within census tracts was obtained by dividing the number of households per census tract by 16 (the number of households interviewed in each census tract). A starting point for the selection of households was selected at random from pieces of folded paper numbered between one and the sampling interval. In rural census tracts where there was more than one village per census tract, the supervisor of the team of interviewers also had information on the number of households per village. The number of households selected from each village was proportional to its share of the number of households in the census tract. The sampling interval within the village was obtained by dividing the total number of households in the village by the number of households to be interviewed in that village. A starting point was selected as described above.
29. 29. There were 516 outpatient visits and hospitalizations in one month or an estimated 6192 visits and hospitaliz* ations in one year or 0.71 per person for the sample population of the 8715 people (1268 households multiplied by 6.9, the mean number of persons per household). In comparison, the official statistics are that there were 1,521,759 visits and hospitalizations in 1988 or 0.71 per person for a population of 2,463,616. For the health statistics see Republique Centrafricaine, Ministere de la Sante Publique et des Affaires Sociales. Bulletin Annuel d'Information Sanitaire Annee 1989, pp. 40-42. The official statistics may over-report utilization of public facilities and under-report utilization of private facilities.
30. 30. It is somewhat unorthodox to include zeros in health expenditure statistics. Standard practice in the United States is to report the percentage of people who were ill that used a facility and the average expenditures among users. This practice is based on the premise that utilization is synonymous with expenditures. In developing countries, the relationship between utilization and expenditures is not as straightforward. For example, people who use a public facility may or may not have expenditures at that facility. The question is: what group of people who are ill is the relevant comparison to the respondents to the contingent valuation questions? First consider who the respondents to the contingent valuation questions in the national sample are. Ninety-eight percent of the respondents in the national sample said that they would use modern care if the quality improvements were instituted. So, the 20-36% of people who refused to pay the cost of the seven quality improvements include the 2% of people who would not use modern care, some people who would use modern care only if it was free, and people who would pay for modern care but less than the cost of the quality improvements. The percentage of people who would not use modern care seems negligible. We do not know the breakdown of those who would pay zero and those who would pay more than zero but less than the cost. The issue then is whether the comparison group should be between all people who were ill and used modern care or only those who had expenditures. Given that the contingent valuation respondents included all people who would use modern care, the comparison presented here is of all people who used modern care including those who had zero expenditures. This comparison assumes that the majority of people who refused to pay the cost of the quality improvements would use modern care only if it was free. In fact, 20% of the people who had modern expenditures had zero expenditures. This is comparable to the 20-36% who refused to pay the cost of the quality improvements if the absence of expenditures does not reflect differences between the severity of the actual illness and the illnesses that would be treated by the quality improvements (or other factors).
31. 31. Two percent of the people who were ill visited a private facility and purchased drugs from a pharmacy.
32. 32. A private communication from an anonymous reviewer states that the World Bank reports that 38% of health care expenditure in the Central African Republic are for private providers of care. If the World Bank statistic is for expenditures at facilities, our estimate of 48% is somewhat higher. If the World Bank statistic is for total expenditures for modern care and the definition of providers excludes pharmacies, our estimate of 14% is much lower; if it includes pharmacies, our estimate of 85% is much higher. The World Bank report has not been released to the public, so it was not possible to reconcile differences in definitions and statistics.
33. 33. As with health care expenditures, all of the analyses were conducted with the median; the distribution of consumption is skewed by a few households with unusually high consumption and the median is a more representative statistic under these circumstances. 34. World Bank. World Development Report, p. 238. Oxford University Press, New York, 1993. There are two explanations for the difference: (1) World Bank estimates are less comparable and reliable in countries with high levels of subsistence farming (p. 308), and (2) the consumption data were collected right before the planting season when the market prices may be higher than average.
34. 35. The natural log of consumption is the preferred variable in analysis of variance because the natural log transformation produces a more normal distribution when the distribution of consumption is skewed by a few households with unusually high consumption.
35. 36. The confidence interval for the estimate depends on whether the estimate was an extreme value (e.g. 20 or 80%) or a central value (e.g. 50%). For example, a sample size of 250 for each version of the questionnaire results in a confidence interval of ±5% for an estimate of 20 or 80%, and a confidence interval of less ± 10% for an estimate of 50%.