- Email: [email protected]
Drop-out analysis of community-based health insurance membership at Nouna, Burkina Faso
Health Policy 92 (2009) 174–179
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Drop-out analysis of community-based health insurance membership at Nouna, Burkina Faso Hengjin Dong ∗ , Manuela De Allegri, Devendra Gnawali, Aurélia Souares, Rainer Sauerborn Department of Tropical Hygiene and Public Health, University of Heidelberg, INF 324, D-69120 Heidelberg, Germany
a r t i c l e
i n f o
Keywords: Community-based health insurance Drop-out Burkina Faso Enrolment
a b s t r a c t Objectives: This study aims to identify the reasons why enrolled people decide not to renew their membership in following years. Methods: Household survey is used to collect information on the factors influencing dropping out from community-based health insurance (CBI). Information from CBI agency databank is used to describe the general situation of enrolment and drop-out. Results: Since the launch of CBI the enrolment rate has been low ranging from 5.2% to 6.3%. The drop-out rate, however, has been high ranging from 30.9% to 45.7%. It is found, by the multivariate analysis, that female household head, higher age or lower education of a household head, lower number of illness episodes in the past three months, fewer children or elderly in a household, poor perceived health care quality, less seeking care in the past month positively effected on drop-out, increasing the rate. However, the household six-month expenditure and the distance to the contracted health facility did not have the hypothesised sign. In contrast, a higher household expenditure and a shorter distance to the contracted health facility increased the drop-out. Conclusions: High drop-out rates endanger the sustainability of CBI not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. The drop-out rate in the scheme of the Nouna Health District, Burkina Faso, is very high. The reasons for drop-out may be related to affordability, healthneeds and health demand, quality of care, household head and household characteristics. This study represents a valuable attempt towards further increasing the sustainability of CBI schemes, by understanding not what motivates people to first enrol in CBI, but what motivates them to renew membership year after year. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In the past decades, community-based health insurance (CBI) has been seen as a promising new tool of health system improvement for rural populations in low-income countries, particularly in sub-Saharan Africa (SSA) [1,2]. In CBI, community members pool their resources to share the financial risks of health care. They own the scheme
∗ Corresponding author. Tel.: +49 6221 566823; fax: +49 6221 564886. E-mail address: [email protected] (H. Dong). 0168-8510/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.healthpol.2009.03.013
and control its management, including the collection of premiums, the payment of health care providers, and the negotiation of the benefits package. Unlike private insurance, premiums are community rated and not based on individual risk assessments. CBI has the advantage of dissociating the time of payment from the time of service use. This system is clearly better adapted than user fees to the seasonal fluctuations in revenue and expenditure flows of rural households [3]. There is strong evidence that CBI provides financial protection by reducing out-of-pocket health care spending and moderate evidence that it improves costrecovery [4]. There is no evidence that CBI improves the quality of care [4].
H. Dong et al. / Health Policy 92 (2009) 174–179
Particularly in SSA, however, the successful development of CBI schemes is severely hampered by low enrolment rates [5–10]. Low enrolment rates have been found to be related to affordability of premiums, adverse selection, non-cooperative attitudes of health providers and poor quality of care [8,9,11–14], lack of understanding of the risk pooling concept, [11,15,16], poor marketing of schemes [7,8], and lack of basic information on the design and operations of schemes as well as lack of trust in their management [8–10]. In addition, even schemes with relatively high enrolment rates often suffer from substantial fluctuations in membership [7,17]. While authors have focus in understanding what motivates people to first enrol in CBI, the problem of membership retention has been repeatedly reported, but not thoroughly analysed in the literature. Criel and Waelkens provide a qualitative analysis of declining subscription rates in the Maliando scheme in Guinea Conakry, but complementary quantitative analyses of such patterns are absent from the literature [7]. Since 2004, a CBI scheme has been operative in the catchment area of the demographic surveillance system (DSS) of the Nouna Health District, rural Burkina Faso. The scheme has been designed according to the scientific evidence emerging from a number of studies which preceded its implementation. These included a study on community preferences for a benefit package, a study on willingnessto-pay for the selected package, a study on health service cost and premium estimation [18–20]. In order to scientifically assess the sustainability of CBI and its effects on health service utilization, the area where the scheme is operative has been divided into 33 clusters and CBI has progressively been offered to 11 additional clusters every year. Since 2006, the residents of all 33 clusters have enjoyed the opportunity to enrol in CBI. Field experience, however, shows that the enrolment rate is low. In addition, drop-out rates, meaning people who discontinue membership after enrolling one year, are very high. The study presented in this paper aims to identify the reasons why previously enrolled people decide not to renew their membership in following years. This analysis is justified by a need to understand the specific reasons motivating people to drop-out of schemes. The aim is to empower decision makers with the information necessary to design measures that can enhance retention in CBI, thus increasing the sustainability of schemes in Burkina Faso as elsewhere in SSA. Given that similar analyses of dropout behaviour are absent from the literature, this article contributes to the wealth of knowledge on CBI. 2. Method The study site is the Nouna Health District which has roughly 230,000 inhabitants. The district is located in the Northwest of Burkina Faso, about 300 Km from the capital Ouagadougou. The area is a dry orchard Savannah, populated almost exclusively by subsistence farmers of different ethnic groups [18,19]. A household survey was used to collect information on the factors influencing the decision to drop-out of CBI. Specifically, we looked at which socio-demographic,
175
economic, and health care related factors influenced households, who had enrolled in CBI in 2005 or before, not to renew their membership in 2006. In addition, household heads were asked to motivate his decision not to renew membership. The survey was carried out in May 2006. Seven hundred fifty six households from the rural area and 553 households from the town of Nouna were randomly selected by using a two-stage cluster sampling procedure. Since enrolment was lower than what expected at the time when the original sample was drawn, re-adjustment occurred to include all households who were or had ever been insured in the CBI scheme. The household was defined as the basic socioeconomic unit to which the various members are related. In general, household members live together in houses or compounds for at least six months, share their resources and jointly satisfy their needs, under the authority of a household head. The management unit of the CBI scheme has built a databank to keep all enrolment and drop-out information year by year. In the present study, these data were used to describe the general situation of enrolment and drop-out. These data nicely complemented survey data as they were comprehensive and reflected the overall situation in terms of enrolment and drop-out in the scheme. The analysis first explored differences in general characteristics between drop-out and non-drop-out households. Chi-square tests were used to assess differences in proportions and analysis of variance was used to assess differences in means. Then, logistic regression was used to study the influence of individual and household factors on drop-out. The presentation of results and the discussion focused on the analysis of the trends and patterns without restricting the focus only to statistically significant results. In the logistic regression, the dependent variable was drop-out (yes = 1 and no = 0). The independent variables were selected based on the study hypotheses and on relevant knowledge of the field. In Burkina Faso, the household head is the one to decide to join or drop-out of CBI [13,21]. Therefore, the analysis included: household head age, sex, and education. In addition, household level characteristics were included, such as household six-month expenditure (as a proxy of household socio-economic status), household size and structure as well as previous health care seeking behaviour. It was assumed that both household head and household characteristics would influence the decision to renew or not to renew membership. The hypothesis was that the following factors would increase the probability to drop-out of CBI: female household head, poorly educated household head, older household head, poor perception of the quality of care by the household head, lower socioeconomic status, larger household, living in rural area, having fewer children and elderly people, having had fewer health needs in the 3 months prior to the survey and not having needed health care services in the month prior to the survey, and living further away from the contracted health facility. This study intended to study drop-out conditional to first even being enrolled in CBI. Compared to the full sample, selection bias may have occurred because the group
176
H. Dong et al. / Health Policy 92 (2009) 174–179 Table 1 General information of the CBI enrolled households. Characteristics
Drop-out Yes (n = 77)
HH heads’ age Male HH heads’ HH head’s education
49.48 67
47.55 109
74.03 10.39 15.58
45.08 21.31 33.61
11.69 15.58 72.73
7.38 18.03 74.59
0.19 9.36 533055 640882
0.18 7.23 351927 661892
66.23 33.77
65.57 34.43
49.35 44.16 6.49
64.75 34.43 0.82
Never schooling (%) Primary school (%) Middle school or above (%) Care quality judged by HH heads Poor ≤2 (%) Average 3 (%) Good ≥4 (%) HH disease episodes of past 3 months HH size HH 6-month expenditurea HH 6-month incomea Location of residency Urban (%) Rural (%) Religion Muslim (%) Catholicity (%) Others (%) a
No (n = 122)
P 0.2736 0.6530 0.0000
0.5730
0.8299 0.0015 0.2814 0.9000 0.0000
0.0160
FCFA (D 1 = 655 FCFA), HH means household (Data source: household survey 2006).
might have been a restricted, non-random sample. To investigate whether conditional estimates suffered from selection bias, a probit model with sample selection was applied [22]. Since the correlation between error terms was found to be insignificant, the possibility of a sample selection bias was ruled out.
Table 2 Enrolment and drop-out. Year
No. eligible households
No. enrolled households (%)
Drop-out (%)
2006 2005 2004
7122 4936 3125
370 (5.2) 311 (6.3) 162 (5.2)
142 (45.7) 50 (30.9)
3. Results
Data source: CBI agency 2006.
3.1. General information on drop-out and non-drop-out households
‘could no longer afford’ and ‘I did not like the behaviour of the medical staff’ (Table 3). In the multivariate analysis, almost all explainable variables had the hypothesised sign. The following factors all had a positive effect on drop-out, meaning that they increased the probability that a household did not renew its membership in CBI: female household head, higher household head’s age, lower household head’s education, larger household size, living in rural area, lower number of illness episodes in the past three months, fewer children or elderly in a household, poor perceived quality of care, less health care seeking in the month prior to the survey. The household six-month expenditure and the distance to the contracted health facility, however, did not have the hypothesised sign. In contrast, a higher household expenditure and a shorter distance to the contracted health facility
The household heads in the drop-out group had a significantly lower education than in the non-drop-out group (Table 1). The households in the drop-out group also had a significantly higher household size, were more likely to live in the town of Nouna, and more likely not to be Muslim (Table 1). No significant difference between the two groups was detected in terms of the age and sex of the household head, the judgement on the quality of care, the household illness reporting pattern, the household six-month expenditure, and the household six-month income (Table 1). 3.2. Enrolment and drop-out Since the launch of CBI in Nouna in 2004 the enrolment rate has been low ranging from 5.2% to 6.3% during the period from 2004 to 2006 (Table 2). The drop-out rate, however, has been high ranging from 30.9% in 2005 to 45.7% in 2006 (Table 2). 3.3. Factors influencing drop-out The main reasons provided by respondents to motivate their decision to discontinue membership in CBI were
Table 3 Reasons for drop-out. Reasons
%
Could afford no longer I didn’t like medical staff behavior Not satisfied with services received I was not given good drugs Nobody got sick last year
28.4 19.0 7.4 6.3 3.2
Data source: household survey 2006.
H. Dong et al. / Health Policy 92 (2009) 174–179
177
Table 4 Factors influencing drop-out (logistic regression analysis). Variables
Reference category
Sex of household head Age of household head
Female
−0.4423 0.0108
0.5843 0.0182
0.6426 1.0109
0.449 0.553
Level of education (household head): primary school Middle school or above
No schooling No schooling
−1.1971 −0.5603
0.5398 0.5672
0.3021 0.5710
0.027 0.323
Religion: Catholicity Muslim
Others Others
−2.0495 −2.6854
1.2438 1.2369
0.1288 0.0682
0.099 0.030
Living in Nouna town Household size
Rural area
−2.2231 0.0809
0.6037 0.0415
0.1083 1.0843
0.000 0.051
Household 6-month expenditure: quarter 2 Quarter 3 Quarter 4
Quarter 1 Quarter 1 Quarter 1
−0.2974 0.9261 0.4458
0.7373 0.6856 0.7069
0.7427 2.5246 1.5617
0.687 0.177 0.528
−0.1296 −3.2131 −1.2242
0.4342 1.6346 1.7978
0.8784 0.0402 0.2940
0.765 0.049 0.496
−1.0102 −1.2250
0.7106 0.6341
0.3641 0.2938
0.155 0.053
−0.5686 −1.0030 4.2152 199 0.2272
0.4745 0.5196 1.8115
0.5663 0.3668 67.7077
0.231 0.054 0.020
Number of episodes in past 3 months % of children (<5) in household % of elders (>65) in household CSPS quality judged by household head: average Good
Poor Poor
Seeking care in the past month Distance to the CBI contracted health facility (Constant) n R2
Not seeking care
B
S.E.
Odds ratio
P
Model test: LR 2 (18) = 60.35, p < 0.001 (Data source: household survey 2006).
increased the drop-out (Table 4). Moreover, only education, religion, place of residence, and the percentage of children in a household were found to be statistically significant. 4. Discussion To our knowledge, this is one of very few studies looking at the reasons motivating drop-out of CBI [7]. High drop-out rates endanger the sustainability of CBI schemes not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. Proof of this statement is the everyday experience of schemes which cease to exist a few years after their inception exactly because they do not even manage to retain the members which they had originally acquired [7,17]. A good documented example in the literature comes from the Maliando scheme in Guinea Conkakry, a scheme which, in spite of all the scientific efforts at the core of its design, was discontinued only two years after its foundation [7]. As the experience of the Maliando scheme demonstrates, high drop-out rates endanger CBI sustainability because they enhance the already existing problem of low enrolment. Having too few people in a scheme, either because people do not enrol at all or because they do not renew their membership year after year, inevitably translates into limited resource mobilization. In turn, limited resource mobilisation represents a menace for the long term viability of schemes and the stabilisation of the financial resources made available to providers, forcing schemes which cannot cover basic costs to suspend their operations. Unrevealing the reasons behind high drop-out is therefore an essential factor to sustain the development of CBI
in SSA. Most of the existing research has focused on understanding why people choose to enrol or not to enrol in CBI [7–11,13], but rarely the reasons why people choose to renew or not to renew their membership year after year [7]. Therefore, our study represents a valuable attempt to complement the existing evidence, offering an important contribution to the wealth of knowledge on CBI. Our series of studies in Burkina Faso show that factors resulting in high drop-out rates are similar to those resulting in low enrolment. Previous research had already identified affordability and concerns with the quality of health services, including the attitude of health providers, as important factors influencing enrolment in CBI [13,23]. The present study confirms that concerns with the quality of care are also an important factor influencing the decision to drop-out of CBI. Multivariate analysis further supports this point: the better perception a household head holds of the quality of the care available, the less likely he/she will choose to discontinue the household membership in CBI. Likewise, previous research on enrolment had already identified a surprisingly positive association between enrolment and distance to the health facility [23]. A qualitative study conducted at the time had helped to explain what might have appeared as a contradictory finding, by highlighting the fact that distant communities had come to appreciate CBI more than communities living near the facility. The qualitative study revealed that those living far from the facility felt that they faced higher non medical costs when seeking care (due both to the cost of transport and to the opportunity cost of seeking care) and had therefore come to value CBI as a tool to relieve them at least of part of the financial burden they faced when sick [23]. Since our study identified that people living closer to the facility were also more likely to drop out, it is very likely that the same
178
H. Dong et al. / Health Policy 92 (2009) 174–179
pattern holds true when choosing whether to renew one’s membership or not. Since the effect cannot be attributed to the perceived quality of care, as this is controlled for in the multivariate analysis, we are left to suppose that when sick, people living close to a facility face primarily only medical costs and may therefore be more willing and able to cope with them directly. People living close to a facility may see fewer advantages in CBI membership than what they had originally expected. In line with our hypothesis, is the fact that household size was positively correlated with dropping out of CBI: the larger the household, the higher the probability of discontinuing membership in the scheme. Although the previous quantitative analysis of enrolment had not investigated the relationship between household size and enrolment, the results of the qualitative study mentioned above had already suggested that given current payment modalities, requiring the premium to be paid in one single instalment, large households, although in principle often wealthier, may find it difficult to mobilise large amounts of cash all at once. This difficulty may help to explain why large households more easily move out of CBI after having originally enrolled. In addition, the education level of a household head is found to be another important factor influencing drop-out. Compared to ‘no schooling’, ‘primary school’ and ‘middle school or above’ can decrease the drop-out. People may have higher expectation on health status when they have a better education. The indicators related to health-needs and health demands all had the hypothesised signs. Households with more episodes of illness in the past three months and with a higher utilisation of health care in the past month had a lower probability of dropping out of CBI. From one side, this situation has positive effects on protecting unhealthy people. From another side, however, this situation may tell us that adverse selection was taking place, meaning that only people with poor health were likely to renew their membership. These findings are in line with evidence emerging from the evaluation of the Nkoranza scheme in Ghana [24]. Although the Nkoranza evaluation did not entail a similar use of multivariate analysis, it still suggested that fluctuations in membership were closely linked to adverse selection, meaning that the healthiest individuals or households tended to move out of the scheme, after having originally enrolled. In contradiction both with our hypothesis and with previous research on enrolment, we found that the higher household economic status (proxied by a higher household expenditure) was positively correlated with dropping out of CBI. Although not significant in the multivariate analysis, this finding may indicate that richer households may be able to afford and demand a better quality of health care by paying for services out-of-pocket. Further research may be needed here. This study implies that low enrolment and high dropout of membership may endanger the sustainability of CBI. Measures to improve affordability and the quality of health services, including the attitude of health providers, should be taken in Burkina Faso in order to increase enrolment and reduce drop-out. Bridging community-based health insurance and social protection for health care may be one way
to keep the sustainability and to make a step towards universal coverage [25]. More detailed drop-out analyses, using a mixture of quantitative and qualitative methods, are needed to complement the results of this study. This study could not develop into a more detailed analysis in the reasons for drop-out by household characteristics because due to the low enrolment rate, the sample size was simply not big enough for further classification. 5. Conclusion High drop-out rates endanger the sustainability of CBI not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. The drop-out rate in the scheme of the Nouna Health District, Burkina Faso, is very high. The reasons for drop-out may be related to affordability, healthneeds and health demand, quality of care, household head and household characteristics. This study represents a valuable attempt towards further increasing the sustainability of CBI schemes, by understanding not what motivates people to first enrol in CBI, but what motivates them to renew membership year after year. Acknowledgements This work was supported by the collaborative research grant ‘SFB 544’ of the German Research Society (DFG). The views expressed in this paper are those of the authors. The authors would like to thank all staff from the Nouna Health Research Centre for their valuable help during the data collection process. References [1] Asenso-Okyere WK, Osei-Akoto I, Anum A, Appiah EN. Willingness to pay for health insurance in a developing economy. A pilot study of the informal sector of Ghana using contingent valuation. Health Policy 1997;42:223–37. [2] Creese A, Bennett S. Rural risk-sharing strategies. In: George J. Schieber, ed. Innovations in health care financing. World Bank Discussion paper no. 365. Washington, DC: The World Bank; 1997. pp. 163–182. [3] Sauerborn R, Nougtara A, Hien M, Diesfeld HJ. Seasonal variations of household costs of illness in Burkina Faso. Social Science & Medicine 1996;43:281–90. [4] Ekman B. Community-based health insurance in low-income countries: a systematic review of the evidence. Health Policy and Planning 2004;19:249–70. [5] Arhin DC. Rural health insurance: a viable alternative to user fees? A review and evidence from three countries. Public Health and Policy Publications, No. 19. UK: London School of Hygiene and Tropical Medicine; 1995. [6] Jütting J. Health insurance for the poor? Determinants of participation in community-based health insurance schemes in rural Senegal. Paris: OECD Development Centre; 2003. [7] Waelkens MP, Criel B. Les mutuelles de santé en Afrique subSaharienne—Etat des Lieux et Reflexions sur un Agenda de Recherche. Health, Nutrition and Population Discussion Paper. Washington: The World Bank; 2004. [8] Basaza R, Criel B, Van der Stuyft P. Community health insurance in Uganda: why does enrolment remain low? A view from beneath. Health Policy 2008;87(2):172–84. [9] Basaza R, Criel B, Van der Stuyft P. Low enrolment in Ugandan community health insurance schemes: underlying causes and policy implications. BMC Health Services Research 2007;7:105.
H. Dong et al. / Health Policy 92 (2009) 174–179 [10] De Allegri M, Sauerborn R, Kouyaté B, Flessa S. Community health insurance in sub-Saharan Africa: what operational difficulties hamper its successful development? Tropical Medicine and International Health 2009;15(5), in press. [11] Arhinful DK. The solidarity of self-interest: Social and cultural feasibility of rural health insurance in Ghana. African Studies centre, ASC Research Report No. 71. The Netherlands: University of Amsterdam; 2003. [12] Baltussen RM, Ye Y, Haddad S, Sauerborn R. Perceived quality of care of primary health care services in Burkina Faso. Health Policy and Planning 2002;17:42–8. [13] De Allegri M, Kouyaté B, Becher H, Gbangou A, Pokhrel S, Sanon M, et al. Understanding enrolment in community health insurance in subSaharan Africa: population-based case-control study in rural Burkina Faso. Bulletin of the World Health Organisation 2006;84:852–8. [14] Kouyaté B, Sanon M, Mugisha F. Community preference for benefit package under community-based insurance. Burkina Faso: CRSN; 2001. [15] Jütting J. Do community based health insurance schemes improve poor people’s access to health care: Evidence from rural Senegal. World Development 2004;32:273–88. [16] Sommerfeld J, Sanon M, Kouyate BA, Sauerborn R. Informal risksharing arrangements (IRSAs) in rural Burkina Faso: lessons for the development of community-based insurance (CBI). International Journal of Health Planning and Management 2002;17:147–63. [17] La Concertation. Inventaire des mutuelles de santé en Afrique— synthèse des travaux de recherche dans le 11 Pays. Dakar, Senegal: La Concertation; 2004.
179
[18] Dong H, Kouyate B, Cairns J, Mugisha F, Sauerborn R. Willingnessto-pay for community-based insurance in Burkina Faso. Health Economics 2003;12:849–62. [19] Dong H, Mugisha F, Gbangou A, Kouyate B, Sauerborn R. The feasibility of community-based health insurance in Burkina Faso. Health Policy 2004;69:45–53. [20] Mugisha F, Kouyate B, Dong H, Sauerborn R. Costing health care interventions at primary health facilities in Nouna, Burkina Faso. African Journal of Health Sciences 2002;9:69–79. [21] Dong H, Gbangou A, De Allegri M, Pokhrel S, Sauerborn R. The differences in characteristics between health-care users and non-users: implication for introducing community-based health insurance in Burkina Faso. European Journal of Health Economics 2008;9: 41–50. [22] Schwartz JB, Akin JS, Popkin BM. Price and income elasticities of demand for modern health care: the case of infant delivery in the Philippines. World Bank Economic Review 1988;2:49–76. [23] De Allegri M, Sanon M, Sauerborn R. To enrol or not to enrol? A qualitative investigation of demand for health insurance in rural West Africa. Social Science & Medicine 2006;62:1520–7. [24] Atim C, Sock M. An external evaluation of the nkoranza community financing health insurance scheme, Ghana. Bethesda, MD: Partnerships for Health Reform Project, Abt Associates, Inc.; 2000. [25] Hidayat B, Thabrany H, Dong H, Sauerborn R. The effects of mandatory health insurance on equity in access to outpatient care in Indonesia. Health Policy and Planning 2004;19:322–35.
Contents lists available at ScienceDirect
Health Policy journal homepage: www.elsevier.com/locate/healthpol
Drop-out analysis of community-based health insurance membership at Nouna, Burkina Faso Hengjin Dong ∗ , Manuela De Allegri, Devendra Gnawali, Aurélia Souares, Rainer Sauerborn Department of Tropical Hygiene and Public Health, University of Heidelberg, INF 324, D-69120 Heidelberg, Germany
a r t i c l e
i n f o
Keywords: Community-based health insurance Drop-out Burkina Faso Enrolment
a b s t r a c t Objectives: This study aims to identify the reasons why enrolled people decide not to renew their membership in following years. Methods: Household survey is used to collect information on the factors influencing dropping out from community-based health insurance (CBI). Information from CBI agency databank is used to describe the general situation of enrolment and drop-out. Results: Since the launch of CBI the enrolment rate has been low ranging from 5.2% to 6.3%. The drop-out rate, however, has been high ranging from 30.9% to 45.7%. It is found, by the multivariate analysis, that female household head, higher age or lower education of a household head, lower number of illness episodes in the past three months, fewer children or elderly in a household, poor perceived health care quality, less seeking care in the past month positively effected on drop-out, increasing the rate. However, the household six-month expenditure and the distance to the contracted health facility did not have the hypothesised sign. In contrast, a higher household expenditure and a shorter distance to the contracted health facility increased the drop-out. Conclusions: High drop-out rates endanger the sustainability of CBI not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. The drop-out rate in the scheme of the Nouna Health District, Burkina Faso, is very high. The reasons for drop-out may be related to affordability, healthneeds and health demand, quality of care, household head and household characteristics. This study represents a valuable attempt towards further increasing the sustainability of CBI schemes, by understanding not what motivates people to first enrol in CBI, but what motivates them to renew membership year after year. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction In the past decades, community-based health insurance (CBI) has been seen as a promising new tool of health system improvement for rural populations in low-income countries, particularly in sub-Saharan Africa (SSA) [1,2]. In CBI, community members pool their resources to share the financial risks of health care. They own the scheme
∗ Corresponding author. Tel.: +49 6221 566823; fax: +49 6221 564886. E-mail address: [email protected] (H. Dong). 0168-8510/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.healthpol.2009.03.013
and control its management, including the collection of premiums, the payment of health care providers, and the negotiation of the benefits package. Unlike private insurance, premiums are community rated and not based on individual risk assessments. CBI has the advantage of dissociating the time of payment from the time of service use. This system is clearly better adapted than user fees to the seasonal fluctuations in revenue and expenditure flows of rural households [3]. There is strong evidence that CBI provides financial protection by reducing out-of-pocket health care spending and moderate evidence that it improves costrecovery [4]. There is no evidence that CBI improves the quality of care [4].
H. Dong et al. / Health Policy 92 (2009) 174–179
Particularly in SSA, however, the successful development of CBI schemes is severely hampered by low enrolment rates [5–10]. Low enrolment rates have been found to be related to affordability of premiums, adverse selection, non-cooperative attitudes of health providers and poor quality of care [8,9,11–14], lack of understanding of the risk pooling concept, [11,15,16], poor marketing of schemes [7,8], and lack of basic information on the design and operations of schemes as well as lack of trust in their management [8–10]. In addition, even schemes with relatively high enrolment rates often suffer from substantial fluctuations in membership [7,17]. While authors have focus in understanding what motivates people to first enrol in CBI, the problem of membership retention has been repeatedly reported, but not thoroughly analysed in the literature. Criel and Waelkens provide a qualitative analysis of declining subscription rates in the Maliando scheme in Guinea Conakry, but complementary quantitative analyses of such patterns are absent from the literature [7]. Since 2004, a CBI scheme has been operative in the catchment area of the demographic surveillance system (DSS) of the Nouna Health District, rural Burkina Faso. The scheme has been designed according to the scientific evidence emerging from a number of studies which preceded its implementation. These included a study on community preferences for a benefit package, a study on willingnessto-pay for the selected package, a study on health service cost and premium estimation [18–20]. In order to scientifically assess the sustainability of CBI and its effects on health service utilization, the area where the scheme is operative has been divided into 33 clusters and CBI has progressively been offered to 11 additional clusters every year. Since 2006, the residents of all 33 clusters have enjoyed the opportunity to enrol in CBI. Field experience, however, shows that the enrolment rate is low. In addition, drop-out rates, meaning people who discontinue membership after enrolling one year, are very high. The study presented in this paper aims to identify the reasons why previously enrolled people decide not to renew their membership in following years. This analysis is justified by a need to understand the specific reasons motivating people to drop-out of schemes. The aim is to empower decision makers with the information necessary to design measures that can enhance retention in CBI, thus increasing the sustainability of schemes in Burkina Faso as elsewhere in SSA. Given that similar analyses of dropout behaviour are absent from the literature, this article contributes to the wealth of knowledge on CBI. 2. Method The study site is the Nouna Health District which has roughly 230,000 inhabitants. The district is located in the Northwest of Burkina Faso, about 300 Km from the capital Ouagadougou. The area is a dry orchard Savannah, populated almost exclusively by subsistence farmers of different ethnic groups [18,19]. A household survey was used to collect information on the factors influencing the decision to drop-out of CBI. Specifically, we looked at which socio-demographic,
175
economic, and health care related factors influenced households, who had enrolled in CBI in 2005 or before, not to renew their membership in 2006. In addition, household heads were asked to motivate his decision not to renew membership. The survey was carried out in May 2006. Seven hundred fifty six households from the rural area and 553 households from the town of Nouna were randomly selected by using a two-stage cluster sampling procedure. Since enrolment was lower than what expected at the time when the original sample was drawn, re-adjustment occurred to include all households who were or had ever been insured in the CBI scheme. The household was defined as the basic socioeconomic unit to which the various members are related. In general, household members live together in houses or compounds for at least six months, share their resources and jointly satisfy their needs, under the authority of a household head. The management unit of the CBI scheme has built a databank to keep all enrolment and drop-out information year by year. In the present study, these data were used to describe the general situation of enrolment and drop-out. These data nicely complemented survey data as they were comprehensive and reflected the overall situation in terms of enrolment and drop-out in the scheme. The analysis first explored differences in general characteristics between drop-out and non-drop-out households. Chi-square tests were used to assess differences in proportions and analysis of variance was used to assess differences in means. Then, logistic regression was used to study the influence of individual and household factors on drop-out. The presentation of results and the discussion focused on the analysis of the trends and patterns without restricting the focus only to statistically significant results. In the logistic regression, the dependent variable was drop-out (yes = 1 and no = 0). The independent variables were selected based on the study hypotheses and on relevant knowledge of the field. In Burkina Faso, the household head is the one to decide to join or drop-out of CBI [13,21]. Therefore, the analysis included: household head age, sex, and education. In addition, household level characteristics were included, such as household six-month expenditure (as a proxy of household socio-economic status), household size and structure as well as previous health care seeking behaviour. It was assumed that both household head and household characteristics would influence the decision to renew or not to renew membership. The hypothesis was that the following factors would increase the probability to drop-out of CBI: female household head, poorly educated household head, older household head, poor perception of the quality of care by the household head, lower socioeconomic status, larger household, living in rural area, having fewer children and elderly people, having had fewer health needs in the 3 months prior to the survey and not having needed health care services in the month prior to the survey, and living further away from the contracted health facility. This study intended to study drop-out conditional to first even being enrolled in CBI. Compared to the full sample, selection bias may have occurred because the group
176
H. Dong et al. / Health Policy 92 (2009) 174–179 Table 1 General information of the CBI enrolled households. Characteristics
Drop-out Yes (n = 77)
HH heads’ age Male HH heads’ HH head’s education
49.48 67
47.55 109
74.03 10.39 15.58
45.08 21.31 33.61
11.69 15.58 72.73
7.38 18.03 74.59
0.19 9.36 533055 640882
0.18 7.23 351927 661892
66.23 33.77
65.57 34.43
49.35 44.16 6.49
64.75 34.43 0.82
Never schooling (%) Primary school (%) Middle school or above (%) Care quality judged by HH heads Poor ≤2 (%) Average 3 (%) Good ≥4 (%) HH disease episodes of past 3 months HH size HH 6-month expenditurea HH 6-month incomea Location of residency Urban (%) Rural (%) Religion Muslim (%) Catholicity (%) Others (%) a
No (n = 122)
P 0.2736 0.6530 0.0000
0.5730
0.8299 0.0015 0.2814 0.9000 0.0000
0.0160
FCFA (D 1 = 655 FCFA), HH means household (Data source: household survey 2006).
might have been a restricted, non-random sample. To investigate whether conditional estimates suffered from selection bias, a probit model with sample selection was applied [22]. Since the correlation between error terms was found to be insignificant, the possibility of a sample selection bias was ruled out.
Table 2 Enrolment and drop-out. Year
No. eligible households
No. enrolled households (%)
Drop-out (%)
2006 2005 2004
7122 4936 3125
370 (5.2) 311 (6.3) 162 (5.2)
142 (45.7) 50 (30.9)
3. Results
Data source: CBI agency 2006.
3.1. General information on drop-out and non-drop-out households
‘could no longer afford’ and ‘I did not like the behaviour of the medical staff’ (Table 3). In the multivariate analysis, almost all explainable variables had the hypothesised sign. The following factors all had a positive effect on drop-out, meaning that they increased the probability that a household did not renew its membership in CBI: female household head, higher household head’s age, lower household head’s education, larger household size, living in rural area, lower number of illness episodes in the past three months, fewer children or elderly in a household, poor perceived quality of care, less health care seeking in the month prior to the survey. The household six-month expenditure and the distance to the contracted health facility, however, did not have the hypothesised sign. In contrast, a higher household expenditure and a shorter distance to the contracted health facility
The household heads in the drop-out group had a significantly lower education than in the non-drop-out group (Table 1). The households in the drop-out group also had a significantly higher household size, were more likely to live in the town of Nouna, and more likely not to be Muslim (Table 1). No significant difference between the two groups was detected in terms of the age and sex of the household head, the judgement on the quality of care, the household illness reporting pattern, the household six-month expenditure, and the household six-month income (Table 1). 3.2. Enrolment and drop-out Since the launch of CBI in Nouna in 2004 the enrolment rate has been low ranging from 5.2% to 6.3% during the period from 2004 to 2006 (Table 2). The drop-out rate, however, has been high ranging from 30.9% in 2005 to 45.7% in 2006 (Table 2). 3.3. Factors influencing drop-out The main reasons provided by respondents to motivate their decision to discontinue membership in CBI were
Table 3 Reasons for drop-out. Reasons
%
Could afford no longer I didn’t like medical staff behavior Not satisfied with services received I was not given good drugs Nobody got sick last year
28.4 19.0 7.4 6.3 3.2
Data source: household survey 2006.
H. Dong et al. / Health Policy 92 (2009) 174–179
177
Table 4 Factors influencing drop-out (logistic regression analysis). Variables
Reference category
Sex of household head Age of household head
Female
−0.4423 0.0108
0.5843 0.0182
0.6426 1.0109
0.449 0.553
Level of education (household head): primary school Middle school or above
No schooling No schooling
−1.1971 −0.5603
0.5398 0.5672
0.3021 0.5710
0.027 0.323
Religion: Catholicity Muslim
Others Others
−2.0495 −2.6854
1.2438 1.2369
0.1288 0.0682
0.099 0.030
Living in Nouna town Household size
Rural area
−2.2231 0.0809
0.6037 0.0415
0.1083 1.0843
0.000 0.051
Household 6-month expenditure: quarter 2 Quarter 3 Quarter 4
Quarter 1 Quarter 1 Quarter 1
−0.2974 0.9261 0.4458
0.7373 0.6856 0.7069
0.7427 2.5246 1.5617
0.687 0.177 0.528
−0.1296 −3.2131 −1.2242
0.4342 1.6346 1.7978
0.8784 0.0402 0.2940
0.765 0.049 0.496
−1.0102 −1.2250
0.7106 0.6341
0.3641 0.2938
0.155 0.053
−0.5686 −1.0030 4.2152 199 0.2272
0.4745 0.5196 1.8115
0.5663 0.3668 67.7077
0.231 0.054 0.020
Number of episodes in past 3 months % of children (<5) in household % of elders (>65) in household CSPS quality judged by household head: average Good
Poor Poor
Seeking care in the past month Distance to the CBI contracted health facility (Constant) n R2
Not seeking care
B
S.E.
Odds ratio
P
Model test: LR 2 (18) = 60.35, p < 0.001 (Data source: household survey 2006).
increased the drop-out (Table 4). Moreover, only education, religion, place of residence, and the percentage of children in a household were found to be statistically significant. 4. Discussion To our knowledge, this is one of very few studies looking at the reasons motivating drop-out of CBI [7]. High drop-out rates endanger the sustainability of CBI schemes not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. Proof of this statement is the everyday experience of schemes which cease to exist a few years after their inception exactly because they do not even manage to retain the members which they had originally acquired [7,17]. A good documented example in the literature comes from the Maliando scheme in Guinea Conkakry, a scheme which, in spite of all the scientific efforts at the core of its design, was discontinued only two years after its foundation [7]. As the experience of the Maliando scheme demonstrates, high drop-out rates endanger CBI sustainability because they enhance the already existing problem of low enrolment. Having too few people in a scheme, either because people do not enrol at all or because they do not renew their membership year after year, inevitably translates into limited resource mobilization. In turn, limited resource mobilisation represents a menace for the long term viability of schemes and the stabilisation of the financial resources made available to providers, forcing schemes which cannot cover basic costs to suspend their operations. Unrevealing the reasons behind high drop-out is therefore an essential factor to sustain the development of CBI
in SSA. Most of the existing research has focused on understanding why people choose to enrol or not to enrol in CBI [7–11,13], but rarely the reasons why people choose to renew or not to renew their membership year after year [7]. Therefore, our study represents a valuable attempt to complement the existing evidence, offering an important contribution to the wealth of knowledge on CBI. Our series of studies in Burkina Faso show that factors resulting in high drop-out rates are similar to those resulting in low enrolment. Previous research had already identified affordability and concerns with the quality of health services, including the attitude of health providers, as important factors influencing enrolment in CBI [13,23]. The present study confirms that concerns with the quality of care are also an important factor influencing the decision to drop-out of CBI. Multivariate analysis further supports this point: the better perception a household head holds of the quality of the care available, the less likely he/she will choose to discontinue the household membership in CBI. Likewise, previous research on enrolment had already identified a surprisingly positive association between enrolment and distance to the health facility [23]. A qualitative study conducted at the time had helped to explain what might have appeared as a contradictory finding, by highlighting the fact that distant communities had come to appreciate CBI more than communities living near the facility. The qualitative study revealed that those living far from the facility felt that they faced higher non medical costs when seeking care (due both to the cost of transport and to the opportunity cost of seeking care) and had therefore come to value CBI as a tool to relieve them at least of part of the financial burden they faced when sick [23]. Since our study identified that people living closer to the facility were also more likely to drop out, it is very likely that the same
178
H. Dong et al. / Health Policy 92 (2009) 174–179
pattern holds true when choosing whether to renew one’s membership or not. Since the effect cannot be attributed to the perceived quality of care, as this is controlled for in the multivariate analysis, we are left to suppose that when sick, people living close to a facility face primarily only medical costs and may therefore be more willing and able to cope with them directly. People living close to a facility may see fewer advantages in CBI membership than what they had originally expected. In line with our hypothesis, is the fact that household size was positively correlated with dropping out of CBI: the larger the household, the higher the probability of discontinuing membership in the scheme. Although the previous quantitative analysis of enrolment had not investigated the relationship between household size and enrolment, the results of the qualitative study mentioned above had already suggested that given current payment modalities, requiring the premium to be paid in one single instalment, large households, although in principle often wealthier, may find it difficult to mobilise large amounts of cash all at once. This difficulty may help to explain why large households more easily move out of CBI after having originally enrolled. In addition, the education level of a household head is found to be another important factor influencing drop-out. Compared to ‘no schooling’, ‘primary school’ and ‘middle school or above’ can decrease the drop-out. People may have higher expectation on health status when they have a better education. The indicators related to health-needs and health demands all had the hypothesised signs. Households with more episodes of illness in the past three months and with a higher utilisation of health care in the past month had a lower probability of dropping out of CBI. From one side, this situation has positive effects on protecting unhealthy people. From another side, however, this situation may tell us that adverse selection was taking place, meaning that only people with poor health were likely to renew their membership. These findings are in line with evidence emerging from the evaluation of the Nkoranza scheme in Ghana [24]. Although the Nkoranza evaluation did not entail a similar use of multivariate analysis, it still suggested that fluctuations in membership were closely linked to adverse selection, meaning that the healthiest individuals or households tended to move out of the scheme, after having originally enrolled. In contradiction both with our hypothesis and with previous research on enrolment, we found that the higher household economic status (proxied by a higher household expenditure) was positively correlated with dropping out of CBI. Although not significant in the multivariate analysis, this finding may indicate that richer households may be able to afford and demand a better quality of health care by paying for services out-of-pocket. Further research may be needed here. This study implies that low enrolment and high dropout of membership may endanger the sustainability of CBI. Measures to improve affordability and the quality of health services, including the attitude of health providers, should be taken in Burkina Faso in order to increase enrolment and reduce drop-out. Bridging community-based health insurance and social protection for health care may be one way
to keep the sustainability and to make a step towards universal coverage [25]. More detailed drop-out analyses, using a mixture of quantitative and qualitative methods, are needed to complement the results of this study. This study could not develop into a more detailed analysis in the reasons for drop-out by household characteristics because due to the low enrolment rate, the sample size was simply not big enough for further classification. 5. Conclusion High drop-out rates endanger the sustainability of CBI not only because they reduce the size of the insurance pool, but also because they bear a negative impact on further enrolment and drop-out. The drop-out rate in the scheme of the Nouna Health District, Burkina Faso, is very high. The reasons for drop-out may be related to affordability, healthneeds and health demand, quality of care, household head and household characteristics. This study represents a valuable attempt towards further increasing the sustainability of CBI schemes, by understanding not what motivates people to first enrol in CBI, but what motivates them to renew membership year after year. Acknowledgements This work was supported by the collaborative research grant ‘SFB 544’ of the German Research Society (DFG). The views expressed in this paper are those of the authors. The authors would like to thank all staff from the Nouna Health Research Centre for their valuable help during the data collection process. References [1] Asenso-Okyere WK, Osei-Akoto I, Anum A, Appiah EN. Willingness to pay for health insurance in a developing economy. A pilot study of the informal sector of Ghana using contingent valuation. Health Policy 1997;42:223–37. [2] Creese A, Bennett S. Rural risk-sharing strategies. In: George J. Schieber, ed. Innovations in health care financing. World Bank Discussion paper no. 365. Washington, DC: The World Bank; 1997. pp. 163–182. [3] Sauerborn R, Nougtara A, Hien M, Diesfeld HJ. Seasonal variations of household costs of illness in Burkina Faso. Social Science & Medicine 1996;43:281–90. [4] Ekman B. Community-based health insurance in low-income countries: a systematic review of the evidence. Health Policy and Planning 2004;19:249–70. [5] Arhin DC. Rural health insurance: a viable alternative to user fees? A review and evidence from three countries. Public Health and Policy Publications, No. 19. UK: London School of Hygiene and Tropical Medicine; 1995. [6] Jütting J. Health insurance for the poor? Determinants of participation in community-based health insurance schemes in rural Senegal. Paris: OECD Development Centre; 2003. [7] Waelkens MP, Criel B. Les mutuelles de santé en Afrique subSaharienne—Etat des Lieux et Reflexions sur un Agenda de Recherche. Health, Nutrition and Population Discussion Paper. Washington: The World Bank; 2004. [8] Basaza R, Criel B, Van der Stuyft P. Community health insurance in Uganda: why does enrolment remain low? A view from beneath. Health Policy 2008;87(2):172–84. [9] Basaza R, Criel B, Van der Stuyft P. Low enrolment in Ugandan community health insurance schemes: underlying causes and policy implications. BMC Health Services Research 2007;7:105.
H. Dong et al. / Health Policy 92 (2009) 174–179 [10] De Allegri M, Sauerborn R, Kouyaté B, Flessa S. Community health insurance in sub-Saharan Africa: what operational difficulties hamper its successful development? Tropical Medicine and International Health 2009;15(5), in press. [11] Arhinful DK. The solidarity of self-interest: Social and cultural feasibility of rural health insurance in Ghana. African Studies centre, ASC Research Report No. 71. The Netherlands: University of Amsterdam; 2003. [12] Baltussen RM, Ye Y, Haddad S, Sauerborn R. Perceived quality of care of primary health care services in Burkina Faso. Health Policy and Planning 2002;17:42–8. [13] De Allegri M, Kouyaté B, Becher H, Gbangou A, Pokhrel S, Sanon M, et al. Understanding enrolment in community health insurance in subSaharan Africa: population-based case-control study in rural Burkina Faso. Bulletin of the World Health Organisation 2006;84:852–8. [14] Kouyaté B, Sanon M, Mugisha F. Community preference for benefit package under community-based insurance. Burkina Faso: CRSN; 2001. [15] Jütting J. Do community based health insurance schemes improve poor people’s access to health care: Evidence from rural Senegal. World Development 2004;32:273–88. [16] Sommerfeld J, Sanon M, Kouyate BA, Sauerborn R. Informal risksharing arrangements (IRSAs) in rural Burkina Faso: lessons for the development of community-based insurance (CBI). International Journal of Health Planning and Management 2002;17:147–63. [17] La Concertation. Inventaire des mutuelles de santé en Afrique— synthèse des travaux de recherche dans le 11 Pays. Dakar, Senegal: La Concertation; 2004.
179
[18] Dong H, Kouyate B, Cairns J, Mugisha F, Sauerborn R. Willingnessto-pay for community-based insurance in Burkina Faso. Health Economics 2003;12:849–62. [19] Dong H, Mugisha F, Gbangou A, Kouyate B, Sauerborn R. The feasibility of community-based health insurance in Burkina Faso. Health Policy 2004;69:45–53. [20] Mugisha F, Kouyate B, Dong H, Sauerborn R. Costing health care interventions at primary health facilities in Nouna, Burkina Faso. African Journal of Health Sciences 2002;9:69–79. [21] Dong H, Gbangou A, De Allegri M, Pokhrel S, Sauerborn R. The differences in characteristics between health-care users and non-users: implication for introducing community-based health insurance in Burkina Faso. European Journal of Health Economics 2008;9: 41–50. [22] Schwartz JB, Akin JS, Popkin BM. Price and income elasticities of demand for modern health care: the case of infant delivery in the Philippines. World Bank Economic Review 1988;2:49–76. [23] De Allegri M, Sanon M, Sauerborn R. To enrol or not to enrol? A qualitative investigation of demand for health insurance in rural West Africa. Social Science & Medicine 2006;62:1520–7. [24] Atim C, Sock M. An external evaluation of the nkoranza community financing health insurance scheme, Ghana. Bethesda, MD: Partnerships for Health Reform Project, Abt Associates, Inc.; 2000. [25] Hidayat B, Thabrany H, Dong H, Sauerborn R. The effects of mandatory health insurance on equity in access to outpatient care in Indonesia. Health Policy and Planning 2004;19:322–35.