- Split View
-
Views
-
Cite
Cite
Kelsey Vaughan, Ngozi Akwataghibe, Babatunde Fakunle, Liezel Wolmarans, Who benefits from the Obio Community Health Insurance Scheme in Rivers State, Nigeria? A benefit incidence analysis, International Health, Volume 8, Issue 6, 1 November 2016, Pages 405–412, https://doi.org/10.1093/inthealth/ihw040
- Share Icon Share
A key aspect of monitoring and evaluating health programs is ensuring that benefits are reaching their target population. We conducted a benefit incidence analysis (BIA) of a Shell-sponsored community health insurance scheme in Nigeria to determine the extent to which the target group (the poor) was benefitting.
We examined a sample of 616 patients’ hospital attendance, financial and administrative records from 2012–2013. We estimated annual utilization rates and average unit costs for inpatient and outpatient services. We multiplied the two to produce a total cost per patient, then deducted annual out-of-pocket expenditures to estimate the total community-based health insurance scheme benefit per person. Benefits were multiplied by the total number of persons in each socioeconomic group to aggregate benefits. We used concentration curves and dominance tests to determine statistical significance at 5% and 10% levels of significance.
Collectively, the poorest 20% of the population received 12% of benefits while the richest quintile received the largest share (23%). Inpatient and outpatient benefits are weakly regressive (pro-rich), statistically significant at a 10% level of significance.
Although the poor were found to benefit, this BIA revealed a tendency towards pro-rich distributions. Removing co-payments for the poorest, reducing long wait and visit times and using community volunteers to help increase access to health services may improve benefits for the poor.
Introduction
People in Nigeria, Africa's most densely populated country with a 2014 population of nearly 178 million, face cultural, resource and geographical or physical barriers to accessing healthcare.1 With regards to resource barriers, over 63% of the population pays out-of-pocket (OOP) to access healthcare, and lack of financial means has been reported as a key barrier to accessing care.2–4 One strategy for addressing this barrier that has recently been gaining importance in low income settings is community-based health insurance schemes (CHIS). Membership is usually voluntary and scheme members regularly pay small premiums into a collective fund which is then used to pay for health services that members require.5,6 CHIS are generally designed for people working in the informal sector or in rural areas with low or irregular income and who do not have access to public, private, or employer-sponsored health insurance.5,7 CHIS can enhance financial access and security to the poor, protecting them from indebtedness and insolvency resulting from medical expenditures.5
In February 2010, Shell Petroleum Development Company in partnership with the Shell Industrial Area Cluster Communities Development Foundation, the Obio-Akpor local government area and the Rivers State Government formed the Obio CHIS. The insurance scheme is the first in the Niger Delta and aims to reduce and possibly eliminate OOP healthcare expenditures for 40 000 persons living in four communities in Obio-Akpor local government area (Rumuobiokani, Rumuezeolu, Oginigba and Rumuomasi). The scheme particularly targets the poor, native (indigene) dwellers in the area (approximately 25% of the target population) but non-indigenes are eligible as well. At the time of the study Shell subsidized 50% of the monthly premium for indigenes under the assumption that the remaining 50% co-payment would be affordable to them. Enrolees are entitled to a range of primary and secondary care services (including consultations, medicines, laboratory and diagnostic procedures, childhood immunization, annual physical examinations, hospitalization up to 15 days per year, maternity care, optical services, minor surgeries and more), some with minimal co-payment at point of service. Insurance enrolment rates for the poor and indigenes are unknown.
The primary health care provider for the scheme is Obio Cottage Hospital (formerly known as Obio Primary Health Centre prior to a recent upgrading), a government-owned health facility that serves nine communities in Obio-Akpor local government area of Rivers State. The hospital offers a wide range of primary care services and a limited range of secondary care services as well as strong referral networks for patients requiring other specialized care.
In 2012 Shell requested a benefit incidence analysis (BIA) study to ascertain what proportion of the target populations of the Obio CHIS were actually benefiting from the scheme. A BIA describes the distribution of public spending or health sector subsidies across a population ranked by their socioeconomic status.8,9 In other words, it assesses the efficiency of health spending and compares health spending with the needs of the most vulnerable.8,9 Equity is considered to be attained when social/economic welfare benefits are distributed appropriately.10,11 BIAs have been used in Nigeria and elsewhere in Africa with regards to both insurance and government spending on health. A 2013 study from Enugu, Southeast Nigeria of federal employees enrolled in the country's National Health Insurance Scheme found the poorest received the highest net annual benefit for outpatient and inpatient care.12 For both outpatient and inpatient care, the richest quartile accounted for a smaller than proportional (16.1 and 7.1%, respectively) share of outpatient and inpatient visits, but a disproportionately high (37.5%) share of deliveries. A study from Tanzania by Mtei et al. looked at benefits derived from various forms of health financing, including National Health Insurance Fund and Community Health Fund contributions.13 That study found National Health Insurance Fund contributions (and income taxes) to be the most progressive while Community Health Fund contributions (and OOP payments) were found most regressive. Public primary care facility use was found to be pro-poor while higher levels of care were generally pro-rich. Finally, Castro-Leal et al. estimated the benefit incidence of government spending in seven countries – Ghana, Guinea, Ivory Coast, Kenya, Madagascar, South Africa and Tanzania – and found that curative health care is overall not well targeted to the poorest: the poorest 20% received less than 20% of the benefits and in all countries except South Africa, the richest 20% receive far more than their proportional share of benefits.14
Given the limited applications of BIAs to evaluate health spending on CHIS in Africa, as well as the paucity of evidence on target efficiency of public or any other subsidies in health in Nigeria, this study is a valuable addition to the meagre evidence base on targeting of CHIS in Nigeria and around the world.15,16
Methods
Data sources and collection
This BIA includes data from March 2012 to February 2013 and was collected in April 2013 by three researchers. Data sources included hospital patients’ outpatient and inpatient, financial and monitoring and evaluation records.
The sampling frame consisted of 1195 scheme members from lists provided by the CHIS Health Maintenance Organizations. Ethical approval was obtained to extract hospital records for these members for the indicated time periods. Only the study team had access to data collected and all identifying links to the individuals were removed. Only 838 folders could be retrieved from the filing system. Upon review of these folders, 216 members were excluded because they did not meet the criterion of the time range of the research. Of these, 23 had no record to show that they had ever accessed health care in the facility. Data extraction was carried out on the remaining 622 folders. We developed a questionnaire to extract data from the sampled hospital outpatient or inpatient records about outpatient health service utilization in the preceding 5 months and inpatient visits in the preceding 12 months. Questions on age, gender, indigenous status and occupation (as a proxy measure for socioeconomic status) were also included.
Hospital financial records provided information on hospital revenue (OOP payments, capitation for scheme members as well as payments made by the insurance company for secondary care) and recurrent expenditure and together were used to estimate the cost of hospital services. Hospital monitoring and evaluation records provided detailed data on utilization of services by scheme and non-scheme clients.
Data were analysed using SPSS version 20 (IBM, Armonk, NY, USA), STATA version 12 (StataCorp LP, College Station, TX, USA) and Microsoft Excel 2010 (v14.0) (Microsoft Corp., Redmond, WA, USA). For categorical variables, χ2 tests were used to test for significance. For continuous variables, analysis of variance (ANOVA) was used to compare groups. Significance tests were kept at a 5% level of significance. However, we commented at a 10% level of significance in the analysis involving concentration curves. Nigeria Naira were converted to US dollars at a rate of N150=US$1.
Socioeconomic status
As mentioned before, occupational data was extracted from hospital records, and coded using 1220.0 ANZSCO for sample patients and their spouse, if applicable.17 Other data sources to measure socioeconomic status were not available in the hospital folders and records. According to Burgard et al., ‘occupational status is one component of socioeconomic status (SES), summarizing the power, income and educational requirements associated with various positions in the occupational structure. Occupational status has several advantages over the other major indicators of SES, which are most commonly educational attainment and personal or family income’.18 We recognize the limitation of applying a developed country's classification to Nigeria. However, assessors of the ANZSCO classification reported that the prestige of occupations tends to be quite stable over time and across countries, and, on a pragmatic level, since the scale is continuous, it is easy to identify the lowest socioeconomic group.18 A score ranging from 0 to 100 and based on an international classification for occupations was then allocated for the patient and the patient's spouse, where 0 indicates low socioeconomic status and 100 indicates high.19 An age correction was added to take into account historical increases in average levels of educational attainment across cohorts and a life course effect whereby older persons tend to have higher earnings. Patients or spouses who indicated that they were unemployed without a working spouse were allocated the lowest possible score; otherwise, the spouse's score was used. The majority of students were married (46 out of 59, 78%) and could therefore be classified according to their spouses’ occupation. Unmarried students (n=13), due to the allocation of one score, ended in the lowest quintile. Minors (n=40/616, 6.5%) were excluded from the socioeconomic status analysis as no data on occupation of a parent or caretaker was available. In 26 cases the socioeconomic status could not be determined as the occupation of the patient or for married patients, the occupation of both the client and the spouse were unknown, or could not be classified due to a too vague occupational description. The total number of patients for whom a proxy measure using occupation could be determined was 547 (88.8% of the sample). The sum of the scores for the patient and spouse was used to generate a combined socioeconomic status score, which was then divided into quintiles. The quintiles were labelled from quintile 1, the poorest, to quintile 5, the richest.
Benefits of utilization
From hospital folders of the sampled patients we extracted the number of inpatient and outpatient visits by CHIS members as well as non-members. Due to the large number of outpatient visits and time and resource constraints, the number of outpatient visits during the period October 2012 to February 2013 was extracted and converted to an annual number by multiplying by 2.4 (12/5). An attempt to adjust for seasonal variations made very little difference to the total number of yearly visits, as has been found in other studies, so the simpler method was adopted.9 The number of inpatient visits for the one year period of March 2012 to February 2013 was used without adjustment.
The average unit cost of each type of health service (inpatient and outpatient) was calculated by dividing hospital revenue and expenditure data by total hospital utilization. Ideally unit costs should be calculated on the basis of recurrent expenditure data9; however, because of how financial records were kept at the hospital, the total hospital recurrent expenditure could not be accurately allocated to inpatient and outpatient departments. We applied the proportional distribution of hospital revenue from OOP payments which was disaggregated by health service (inpatient and outpatient) (0.68) for the period March 2012 to February 2013 to total recurrent expenditure from the same time period to allocate between inpatient and outpatient expenditures.
Inpatient expenditure was then divided by the total number of inpatient visits for the same time period, obtained from the hospital monitoring and evaluation data, to determine the cost per inpatient visit. The same was done for outpatients. Using this method, the cost of an outpatient visit was found to be slightly more than one-fourth the cost of an inpatient day (i.e., ratio of 0.27), within the range (0.25–0.33) that has been documented in literature.9,20,21
Yearly utilization rates were then multiplied by the average unit costs for each type of health service (inpatient and outpatient). Annual OOP expenditures were deducted to produce the total CHIS monetary benefit per person (in Naira); for outpatient OOP expenditure, we converted expenditure for the period October 2012 to February 2013 to an annual expenditure by multiplying by 2.4 (12/5). Finally, monetary benefits were multiplied by the total number of persons in each socioeconomic group to aggregate benefits by group, separately for outpatient and inpatient visits and then as a whole. Proportions of total benefit were then calculated using these figures and compared against each group's share of the population size. To assess whether distribution of benefits is appropriate, we compared whether each socioeconomic quintile's percentage of share of benefits corresponded to their share of the population.
Quintile shares of utilization and benefits
We look at utilization in terms of inpatient (hospital admissions) and outpatient visits. For each, the percentage and cumulative percentage of total utilization per quintile of socioeconomic status were calculated.
The standard error for the concentration index was derived from the raw/micro data in the case of health care utilization using a nonlinear function of totals for the concentration index and applying the delta method to obtain the standard error.23 The nonlinear regression was carried out in STATA through the nlcom command. A t-test, based on the concentration index and its SE, was used to test for significance from zero of the concentration index per category of health care utilization (and benefits) at a 5% level of significance. Concentration curves per category of health care utilization (and benefits) were graphed in Microsoft Excel.
The standard error for the concentration index in the case of share of benefits was inferred from grouped data as benefit calculations were not performed at an individual level.23
Results
Utilization of health services
There were a total of 38 281 outpatient visits by both CHIS members and non-CHIS clients in the period March 2012 to February 2013. Overall 69.4% (n=26 571) of outpatient visits were made by CHIS members; non-CHIS clients accounted for the rest. There were a total of 6322 inpatient visits with an average stay of 2.4 days each.
CHIS members made an estimated average of 6.5 outpatient visits during this 12 month period, ranging from a low of 5.5 visits by the lowest socioeconomic group to a high of 7.0 visits by the highest socioeconomic group. Scheme members made an average of 0.24 inpatient visits during the same period, ranging from 0.17 visits in the lowest socioeconomic group to a high of 0.28 visits by the 2nd and 5th quintiles. Estimated average number of yearly outpatient visits showed statistically significant differences across socioeconomic groups (at a 10% level of significance, F=2.336, ANOVA, p=0.054), whilst estimated average number of impatient visits showed no differences (F=1.047, ANOVA, p=0.382). Table 1 shows details of utilization by socioeconomic status.
. | Socioeconomic status group (quintile) . | |||||
---|---|---|---|---|---|---|
Q1 (poorest) . | Q2 . | Q3 . | Q4 . | Q5 (richest) . | Average . | |
Estimated average number of yearly outpatient visits | 5.50 | 6.41 | 6.89 | 6.54 | 7.04 | 6.47 |
Average number of yearly inpatient visits | 0.17 | 0.28 | 0.23 | 0.27 | 0.28 | 0.24 |
. | Socioeconomic status group (quintile) . | |||||
---|---|---|---|---|---|---|
Q1 (poorest) . | Q2 . | Q3 . | Q4 . | Q5 (richest) . | Average . | |
Estimated average number of yearly outpatient visits | 5.50 | 6.41 | 6.89 | 6.54 | 7.04 | 6.47 |
Average number of yearly inpatient visits | 0.17 | 0.28 | 0.23 | 0.27 | 0.28 | 0.24 |
. | Socioeconomic status group (quintile) . | |||||
---|---|---|---|---|---|---|
Q1 (poorest) . | Q2 . | Q3 . | Q4 . | Q5 (richest) . | Average . | |
Estimated average number of yearly outpatient visits | 5.50 | 6.41 | 6.89 | 6.54 | 7.04 | 6.47 |
Average number of yearly inpatient visits | 0.17 | 0.28 | 0.23 | 0.27 | 0.28 | 0.24 |
. | Socioeconomic status group (quintile) . | |||||
---|---|---|---|---|---|---|
Q1 (poorest) . | Q2 . | Q3 . | Q4 . | Q5 (richest) . | Average . | |
Estimated average number of yearly outpatient visits | 5.50 | 6.41 | 6.89 | 6.54 | 7.04 | 6.47 |
Average number of yearly inpatient visits | 0.17 | 0.28 | 0.23 | 0.27 | 0.28 | 0.24 |
Table 2 shows the percentages of total utilization per quintile of socioeconomic status for each category of care. Poorer groups utilized less than their population shares of health services, i.e., a quintile representing 20% of the population accounted for less than 20% of the use of health services.
SES group . | Total visitsa . | Total outpatient visitsb . | Total inpatient daysc . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | Sum . | % . | Cum % . | Sum . | % . | Cum % . | Sum . | % . | Cum % . |
Q1 (poorest) | 641 | 17.5 | 17.5 | 622 | 17.6 | 17.6 | 19 | 14.2 | 14.2 |
Q2 | 709 | 19.3 | 36.8 | 679 | 19.2 | 36.8 | 30 | 22.4 | 36.6 |
Q3 | 783 | 21.4 | 58.1 | 758 | 21.5 | 58.2 | 25 | 18.7 | 55.2 |
Q4 | 762 | 20.8 | 78.9 | 732 | 20.7 | 79.0 | 30 | 22.4 | 77.6 |
Q5 (richest) | 774 | 21.1 | 100.0 | 744 | 21.1 | 100.0 | 30 | 22.4 | 100.0 |
Total | 3669 | 100.0 | 100.0 | 3535 | 100.0 | 100.0 | 134 | 100.0 | 100.0 |
SES group . | Total visitsa . | Total outpatient visitsb . | Total inpatient daysc . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | Sum . | % . | Cum % . | Sum . | % . | Cum % . | Sum . | % . | Cum % . |
Q1 (poorest) | 641 | 17.5 | 17.5 | 622 | 17.6 | 17.6 | 19 | 14.2 | 14.2 |
Q2 | 709 | 19.3 | 36.8 | 679 | 19.2 | 36.8 | 30 | 22.4 | 36.6 |
Q3 | 783 | 21.4 | 58.1 | 758 | 21.5 | 58.2 | 25 | 18.7 | 55.2 |
Q4 | 762 | 20.8 | 78.9 | 732 | 20.7 | 79.0 | 30 | 22.4 | 77.6 |
Q5 (richest) | 774 | 21.1 | 100.0 | 744 | 21.1 | 100.0 | 30 | 22.4 | 100.0 |
Total | 3669 | 100.0 | 100.0 | 3535 | 100.0 | 100.0 | 134 | 100.0 | 100.0 |
Cum: cumulative; Q: quintile; SES: socioeconomic status.
a Concentration index 0.0348; SE 0.0152; t-test 2.2986; p-value>0.05.
b Concentration index 0.0337; SE 0.014; t-test 2.4057; p-value>0.05.
c Concentration index 0.0657; SE 0.0707; t-test 0.9283; p-value>0.05.
SES group . | Total visitsa . | Total outpatient visitsb . | Total inpatient daysc . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | Sum . | % . | Cum % . | Sum . | % . | Cum % . | Sum . | % . | Cum % . |
Q1 (poorest) | 641 | 17.5 | 17.5 | 622 | 17.6 | 17.6 | 19 | 14.2 | 14.2 |
Q2 | 709 | 19.3 | 36.8 | 679 | 19.2 | 36.8 | 30 | 22.4 | 36.6 |
Q3 | 783 | 21.4 | 58.1 | 758 | 21.5 | 58.2 | 25 | 18.7 | 55.2 |
Q4 | 762 | 20.8 | 78.9 | 732 | 20.7 | 79.0 | 30 | 22.4 | 77.6 |
Q5 (richest) | 774 | 21.1 | 100.0 | 744 | 21.1 | 100.0 | 30 | 22.4 | 100.0 |
Total | 3669 | 100.0 | 100.0 | 3535 | 100.0 | 100.0 | 134 | 100.0 | 100.0 |
SES group . | Total visitsa . | Total outpatient visitsb . | Total inpatient daysc . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | Sum . | % . | Cum % . | Sum . | % . | Cum % . | Sum . | % . | Cum % . |
Q1 (poorest) | 641 | 17.5 | 17.5 | 622 | 17.6 | 17.6 | 19 | 14.2 | 14.2 |
Q2 | 709 | 19.3 | 36.8 | 679 | 19.2 | 36.8 | 30 | 22.4 | 36.6 |
Q3 | 783 | 21.4 | 58.1 | 758 | 21.5 | 58.2 | 25 | 18.7 | 55.2 |
Q4 | 762 | 20.8 | 78.9 | 732 | 20.7 | 79.0 | 30 | 22.4 | 77.6 |
Q5 (richest) | 774 | 21.1 | 100.0 | 744 | 21.1 | 100.0 | 30 | 22.4 | 100.0 |
Total | 3669 | 100.0 | 100.0 | 3535 | 100.0 | 100.0 | 134 | 100.0 | 100.0 |
Cum: cumulative; Q: quintile; SES: socioeconomic status.
a Concentration index 0.0348; SE 0.0152; t-test 2.2986; p-value>0.05.
b Concentration index 0.0337; SE 0.014; t-test 2.4057; p-value>0.05.
c Concentration index 0.0657; SE 0.0707; t-test 0.9283; p-value>0.05.
Aggregation of the benefits of utilization
Unit cost of services
The average unit cost of an outpatient visit, based on an estimated outpatient department expenditure of N50 750 878 (US$338 339) and 38 281 outpatient visits was found to be N1326 (US$8.84). The average unit cost of an inpatient visit, assuming estimated inpatient expenditure of N74 521 564 (US$496 810) and 6322 visits, was found to be N11 788 (US$78.59). Assuming an average inpatient stay of 2.4 days, the cost per inpatient bed per day is N4906 (US$32.71).
OOP payments
The large majority of outpatients incurred OOP expenditures (427/547, 78.1%), while only 19 patients from our sample were found to have incurred OOP expenditures associated with an inpatient visit. The average annual OOP expenditure for outpatient visits was found to be N3680 (US$24.53). There was considerable variation between socioeconomic groups; notably, the lowest socioeconomic group had the highest average annual OOP expenditure at N4848 (US$32.32), with the middle group incurring the lowest of N3081 (US$20.54). The analysis of inpatient OOP expenditure by socioeconomic group showed that the majority of patients incurring OOP expenditures were located in the fourth quintile.
Distribution of benefits
The poorest quintile (Q1) had the least benefit (both for outpatient and inpatient visits) (12%), while the richest quintile (Q5) received the greatest share of benefits (23%) when compared to its population share (Table 3). For the poorest quintile, particularly notable is their outpatient benefit, which is approximately half of what it should be if proportional to their population share. The second and fifth quintiles consumed more benefits for both outpatient and inpatient visits compared with their share of the sample population, while the third and fourth quintiles consumed proportionally more outpatient benefits but less inpatient benefits than their share of the sample population.
SES group . | Total benefitsa . | Outpatient benefitsb . | Inpatient benefitsc . | |||
---|---|---|---|---|---|---|
. | % . | Cum % . | % . | Cum % . | % . | Cum % . |
Q1 (poorest) | 12 | 12 | 10 | 10 | 15 | 15 |
Q2 | 21 | 33 | 20 | 30 | 24 | 39 |
Q3 | 23 | 56 | 25 | 55 | 19 | 58 |
Q4 | 21 | 77 | 22 | 77 | 18 | 76 |
Q5 (richest) | 23 | 100 | 23 | 100 | 24 | 100 |
Total | 100 | 100 | 100 | 100 | 100 | 100 |
SES group . | Total benefitsa . | Outpatient benefitsb . | Inpatient benefitsc . | |||
---|---|---|---|---|---|---|
. | % . | Cum % . | % . | Cum % . | % . | Cum % . |
Q1 (poorest) | 12 | 12 | 10 | 10 | 15 | 15 |
Q2 | 21 | 33 | 20 | 30 | 24 | 39 |
Q3 | 23 | 56 | 25 | 55 | 19 | 58 |
Q4 | 21 | 77 | 22 | 77 | 18 | 76 |
Q5 (richest) | 23 | 100 | 23 | 100 | 24 | 100 |
Total | 100 | 100 | 100 | 100 | 100 | 100 |
Cum: cumulative; Q: quintile; SES: socioeconomic status.
a Concentration index 0.088; SE 0.0364; p-value>0.05 and <0.10*;
b Concentration index 0.112; SE 0.047; p-value>0.05 and <0.10*;
c Concentration index 0.048; SE 0.0196; p-value>0.05 and <0.10*;
* Statistically significant at 10%.
SES group . | Total benefitsa . | Outpatient benefitsb . | Inpatient benefitsc . | |||
---|---|---|---|---|---|---|
. | % . | Cum % . | % . | Cum % . | % . | Cum % . |
Q1 (poorest) | 12 | 12 | 10 | 10 | 15 | 15 |
Q2 | 21 | 33 | 20 | 30 | 24 | 39 |
Q3 | 23 | 56 | 25 | 55 | 19 | 58 |
Q4 | 21 | 77 | 22 | 77 | 18 | 76 |
Q5 (richest) | 23 | 100 | 23 | 100 | 24 | 100 |
Total | 100 | 100 | 100 | 100 | 100 | 100 |
SES group . | Total benefitsa . | Outpatient benefitsb . | Inpatient benefitsc . | |||
---|---|---|---|---|---|---|
. | % . | Cum % . | % . | Cum % . | % . | Cum % . |
Q1 (poorest) | 12 | 12 | 10 | 10 | 15 | 15 |
Q2 | 21 | 33 | 20 | 30 | 24 | 39 |
Q3 | 23 | 56 | 25 | 55 | 19 | 58 |
Q4 | 21 | 77 | 22 | 77 | 18 | 76 |
Q5 (richest) | 23 | 100 | 23 | 100 | 24 | 100 |
Total | 100 | 100 | 100 | 100 | 100 | 100 |
Cum: cumulative; Q: quintile; SES: socioeconomic status.
a Concentration index 0.088; SE 0.0364; p-value>0.05 and <0.10*;
b Concentration index 0.112; SE 0.047; p-value>0.05 and <0.10*;
c Concentration index 0.048; SE 0.0196; p-value>0.05 and <0.10*;
* Statistically significant at 10%.
According to the concentration indices and their dominance tests, the three benefit categories are weakly regressive (pro-rich), but not significantly so at a 5% level of significance. However, at a 10% level of significance, the concentration indexes for three categories of benefits are statistically significant, indicative of a tendency therefore towards pro-rich distributions.
Discussion
Utilization of services by socioeconomic status showed that especially for outpatient services, the poorest quintile utilized health services the least and the richest quintile utilized services the most. This pattern is in line with the findings from many studies: the poorest population groups usually have the higher disease burden, so the expectation is that they should use health services more.11,14,24 However, the converse is usually true as wealthier groups tend to use health services more frequently.9,14,24,25 These differences in utilization may also reflect the health seeking behavior of the different groups.
The distribution of health benefits seen in this study highlights that the richer quintiles are accessing care at Obio Cottage Hospital more than the poorer ones. Evidence shows that a common failure of subsidized health care is the inequitable distribution of benefits, with the richest groups usually consuming a disproportionate share of benefits, especially in low- and middle-income countries such as Ghana, Kenya, Tanzania and others.9,10,13,14,21,23,25-31 Community health financing schemes in particular have been found to often fail to achieve equitable coverage.32,33 This study had similar findings, with the 2nd to 5th socioeconomic status quintiles enjoying more than their share of benefits; the poorest quintile incurred the least benefits (12% of total benefits) while the richest quintile also had the highest share of benefits (23%). Our findings differ from one study in Kenya which found larger pro-rich disparities for inpatient as compared to outpatient care.30
This could be linked to health seeking behaviour as previously mentioned; given the high OOP payments found in the poorest quintile both related to drugs for chronic diseases and scans or x-rays, it could also be an affordability issue. Other studies have recorded similar difficulties for the poor. For instance, Polonsky et al. noted relatively low participation in the Armenian scheme due to a lack of affordability and a package of primary care that did not include coverage of chronic diseases.15 Poletti and Balabanova found that the primary cause of the poor not enrolling in schemes was affordability but also found dissatisfaction with the health care package offered as an important reason.34 Additionally, Limwattananon et al., in their Thailand study, partially attributed a pro-poor outcome of public subsidies to health care to zero co-payment at the point of services.35
Another reason given in literature why the poor do not access health care as they should is the inclination of the poor to regard illness as a normal way of life and therefore not important enough to report.14
Other issues of importance with regards to utilization may include limited understanding of entitlements and transportation costs, as was observed in Ghana, Tanzania and South Africa.31 Further, the poor usually experience greater loss in terms of time spent away from economic activity in order to obtain health care compared to salaried workers, who usually receive their pay irrespective of time spent in the same pursuit.14 Gertler and van de Gaag found that lower income earners were more sensitive to the time required to obtain health care than higher income earners, which in effect creates a rationing effect, with the poor at the lower end of utilization.36 Many of those served by the Obio hospital live within the four Shell Industrial Area cluster communities; however, it is a recognized fact that physical proximity to a health service does not ensure access to it.37 Waiting times at the facilities can also contribute to significant time losses for patients.
There are a number of possible ways to improve the scheme's ability to target and benefit the poor. Since cost may be an issue, one solution may be to consider zero co-payment at the point of service for services with high co-payments, such as those for chronic diseases. To address issues related to opportunity costs of seeking healthcare, exploring further ways to reduce the time spent in the hospital by the clients during outpatient visits is critical. Several issues could be looked at, for instance, tracking visits of patients with a view to locating the points at which there might be bottlenecks and assessing patient to health worker ratio in order to determine if there is a shortage of health workers to cater to the number of clients. Qualitative research would be well-suited to further explore some of the issues identified in this study in order to understand why the poorest are utilizing services less and how the CHIS and Obio Cottage Hospital can address these issues to improve access and utilization in this group. Further, a community volunteer training program (such as those implemented by AMAR International Charitable Foundation: http://www.amarfoundation.org) to promote preventive education and patient education may help increase access to health care for the poor.
Study limitations include challenges with assigning occupational status, as previously discussed, and the need to make some assumptions due to limited data availability. Notably we had to make assumptions about the distribution of visits between scheme and non-scheme clients, the use of hospital services over a one year period and in allocating expenditures to different health services. Additionally, during sampling a number of hospital folders could not be retrieved, possibly due to filing errors or different name spellings. This means we cannot know if those clients excluded from the sample for this reason are different in terms of characteristics and utilization of services.
Conclusions
Although reaching and benefitting the target group, the poor, this study found the scheme to have a tendency towards pro-rich distributions. Removing co-payments for the poorest, reducing long wait and visit times and using community volunteers to help increase access to health services may improve benefits for the poor.
Authors' contributions: KV, NA and LW all contributed substantially to the study design, data analysis, interpretation of findings, drafting of the manuscript and critical revision of drafts. NA was the principal investigator for the study. BF initiated the study and contributed to drafting of earlier versions of the manuscript and critical revision of drafts. All authors read and approved the final version of the paper. NA is the guarantor of the paper.
Acknowledgements: The team acknowledges Dr Edet Edet for his support throughout the study implementation period. The help of Dr Akinwunmi Fajola in providing the team with required databases for review is also appreciated. The team thanks the entire staff of Obio Cottage Hospital for their cooperation and support during data collection—in particular, the help of Mr Babalola Idowu, Dr Ugo Ottih, Mrs Ngozi Adibe, Dr Folake Alamina and Dr Tondor Uzosike throughout the data collection period is acknowledged. Finally, the in-country data collectors Mr Mike Odu and Mr Jideofor Agbo are especially appreciated.
Funding: This work was supported by Shell Petroleum Development Company (SPDC) Nigeria.
Competing interests: None declared.
Ethical approval: Ethical approval for the research was obtained from the University of Port Harcourt Ethical Review Committee. Only the study team had access to data collected.
References
Author notes
Present address: Royal Tropical Institute (KIT), Amsterdam, The Netherlands 1090HA and ENAULD Health Research and Services, Leidschendam, The Netherlands 2264BC;
Present address: Centre for Sustainable Access to Health in Africa (C-HASA), Ontario, Canada LE8 6B7;
Comments