Results of the insurance scheme

The interest shown by the Bwamanda community in this voluntary insurance scheme for hospital care was overwhelming and beyond most expectations. In 1986, 32,600 people - i.e. 28% of the total district population - joined the scheme within four weeks. The financial balance after the first year of operation was positive, with a small surplus of approximately U.S.$1300. Each year, the subscription charge was adjusted in line with inflation, but it remained approximately equivalent to the purchasing price of 2 kg of soybeans- approximately one-third of a U.S.$ - though with small variations over the years.

In the following years, the membership rates steadily increased, indicating a high degree of social acceptation. In 1987, 60,000 people joined the scheme, and, in 1988, 80,000. The membership rate stabilised around approximately 60-65%. However, the membership dropped significantly from 66% in 1991 to about 40% in 1992 and from 66% in 1993 to 41% in 1994 (Figure 3). In 1992, severe ethnic tensions in the Bwamanda area, and a climate of social unrest were probably responsible for this fall in subscriptions. In 1994, the enrolment period was preceded by the nation-wide change in currency from anciens to nouveaux Zaire Consequently, cash availability was a problem for many people.

But overall interest remained even in the period of dramatic social and political turmoil, which Zaire experienced from the beginning of the nineties onwards. This is a priori somewhat surprising, since one would expect expenditure for a hospital insurance scheme to drop on people's priority list in an environment where the daily search for food became a major challenge for very many people

Figure 3 (Subscription rates (%) to the Bwamanda hospital insurance scheme (period 1986-97)

The general performance of the health centre, as well as the quality of its interaction with the community in its area, were major factors for obtaining these relatively high subscription rates. A survey carried out in 1988 among 518 households living in the district shows that socio-economic characteristics of insured and non-insured households were very similar (Moens 1990). There was no difference in terms of housing, education level, family size and religion. The same survey, however, indicated a difference in terms of monetary income The very low income group (<U.S.$20/month) and the high income group (>U.S.$200/month) were more represented in the non- member population (14.9% very low income households among members versus 18.7% among non-members, 5.9% high income households among members versus 10.5% among non-members).

The size of the population joining the scheme made genuine risk-sharing arrangements possible. The high membership rates, together with the option to have the household as subscription unit, greatly reduced the risk of a preferential selection of high-risk cases (adverse selection)⁵.

5 These subscription rates are in fact a slight underestimate of the real subscription rates, since a sub-population of a few thousand people in the Bwamanda health district most of them employees of the different CDI project services are covered by mandatory employer-organised health insurance schemes These provide them and their families with free health care, so they did not have an immediate incentive to join the scheme If some of them did pay the insurance premium out of their own pocket, it was with the objective of being insured if they ever lost their job and thus the benefit of free care.

HOSPITAL REVENUE

The evolution and sources of the Bwamanda hospital revenue for the period 1985-1989 are presented in Table 3. The revenue raised from payments for hospital care (internal or locally generated revenue) doubled from U.S.$21, 180 in 1985, the year before the start of the insurance plan, to U.S.$44,475 in 1989 (Figure 4). The internal revenue is made up of direct payments by non-insured patients, prepayment of employer-organised health care schemes (covering a few thousand of people), reimbursements to the hospital by the insurance fund, and co-payments by insured patients them- selves. Between 1986 and 1989, revenue from the insurance scheme (reimbursements and co-payments) increased. The insurance ensures the hospital a stable source of income with less non-paying patients. Direct payments by non-insured persons halved from U.S.$11,655 in 1986 (when 72% of the district population was not insured) to U.S $6,135 in 1989 (when only 39% of the population was not insured).

Table 3 Hospital revenue from 1986 to 1989, in US$

Source of hospital revenue	1985	1986	1987	1988	1989
Internal revenue
Refunding by insurance fund for insured: i.e. 80% of regular hospital fees	0	10,670	8,620	14,700	19,630
Co-payment by insured: i.e. 20% of regular hospital fees	0	2,670	2,155	3,675	4,900
Prepayment by employers for employees and their families	0	6,465	10,990	9,635	13,810
Direct revenue from patients a	21,180	11,655	10,870	7,010	6,135
Total internal revenue (% of total hospital revenue)	21,180 (41%)	31,460 (61%)	32,635 (82%)	35,020 (75%)	44,475 (79%)
Subsidiesb and gifts (% of total hospital revenue)	30,635 (59%)	20,040 (39%)	7,200 (18%)	11,515 (25%)	11,910 (21%)
Total hospital revenue	51,815	51,500	39,835	46,535	56,385

NOTES: Source of data (Tshinko 1992) and annual reports of the Bwamanda health district. Average annual exchange rates US$1 for 50 Zaire in 1985, 61 Zaire in 1986, 128 Zaire in 1987, 187 Zaire in 1988, 400 Zaire in 1989. ^aNon-insured self-employed patients. ^bThe last government subsidies for the Bwamanda hospital were in 1984 From then on all external hospital funding came through the CDI project.

Figure 4: Internal and external revenue in Bwamanda hospital

An a posteriori analysis of the evolution of hospital fees shows that the fees for non-insured persons- and the 20% co-payments for the insured- have dramatically increased over the same period. A Caesarean section, for in- stance, was charged at approximately U.S.$5 in 1985, $15 in 1986, $14 in 1987, $19 in 1988 and $28 in 1989 (Figure 5).

Figure 5 (Hospital fees 1985-89)

On the other hand, hospital subsidies (external revenue) decreased in 1989 to about one-third of the 1985 level (from U.S.$30, 635 to $11,910), whereas total hospital revenue increased from $51,815 in 1985 to $56,385 in 1989. Table 3 clearly shows that the relative proportion of internal revenue in total hospital income increased dramatically from 41% in 1985 to 79% in 1989.

HOSPITAL UTILISATION DATA

In 1986, hospital admission rates for the insured and non-insured population were 36.2 and 24.8‰ respectively. In 1988, these rates were 35.6 and 24.6‰ respectively (Table 4). These differences are statistically highly significant. In 1989 hospital admission rates⁶ were between 1.9 and 6.7 times higher for insured than for non-insured patients (other than the ones covered by employer-organised schemes) (Shepard et al. 1990).

6 Based on a one in ten sample from the hospital register.

Table 4: Hospital admission rates of insured and non-insured populations

	Admissions/1000 insured populationa	Admissions/1000 non-insuredb populationa
1986	1,181/32,614 (36.2‰)	2,133/85,998 (24.8‰)	x2=113; P<0.001
1988	2,863/80,495 (35.6‰)	1200/48,749 (24.6‰)	x2=119; P<0.001

a These ratios are considered true proportions although the numerator may contain several admissions for one individual.

^b Admissions for non-insured population includes admissions of patients covered by employer- organised schemes.

More recent data for the 12-month period April 1993 -March 1994 give admission rates of 49‰ for the insured and 24.9‰ for the non-insured. The latter figure can be split further into 17‰ for non-insured self-employed persons and an estimated 184‰ for people covered by an employer-organised scheme (Table 5).

Table 5: Hospital admissions 1/4/93 - 31/3/94

	Insured pop=101,352	Non-insured pop=50,131	Employer-organised schemes pop=2,500a	Admissions from outside district	Total
Paediatrics	1,267	168	221	132	1,788
Gynaecology	278	39	21	68	406
Internal medicine (M+F)	547	201	42	356	1,146
Surgery men	452	20	17	78	567
Surgery women	370	15	32	87	504
Maternity	1,119	82	29	35	1,265
Intensive care	939	326	99	322	1,686
Total admissions	4,972	851	461	1,078	7,362
Admission rate	49‰	17‰	184‰

NOTES: Patients of the trypanosomiasis ward are not included; most of the patients in intensive care are transferred to other wards after a few days: these admissions counted twice and the real number of admissions is therefore lower. ^a estimate admitted are thus

During the latest three or four years of operation of the insurance plan, people from outside the district frequently claimed to live within the district boundaries so as to be eligible for subscription to the insurance plan during the enrolment period. They had their names added on the family file of a 'host' family (which was often composed of relatives). This happened mainly in the areas of the two Bwamanda town health centres, as well as in the areas of two health centres situated at the edges of the district. Hence the admission rate for insured persons from the district of 49‰ is probably a (slight) overestimate of the real figure. Table 5 also shows that in the period 1993-1994 about 15% of all admissions (1,078 out of 7,362) were patients living in neighbouring districts This is not a new finding Bwamanda hospital has always been a facility with a substantial proportion of users from other districts The annual hospital reports indicate that in 1987 17% of admissions (691 out of 4090) were patients from outside the district, in 1995 this figure increased to 20 4% (1599 out of 7843) (Ministry of Health 1987, Ministry of Health 1995)

This pattern of higher hospital admission rates for the insured population is probably due to a combination of moral hazard and better access for those who need it Within the limits of these data, it is difficult to assess the relative importance of each single possible cause The fact that insured patients can benefit only from the insurance scheme when referred by a health centre, and the system of co-payment at hospital level are factors which a priori tend to counteract any substantial degree of inappropriate hospital utilisation It is important to acknowledge the fact that the increment of hospital utilisation by the insured population seems to be a highly variable phenomenon The data in Table 6 indicate that excess use is particularly high for surgical services, both female and male, but that it is hardly apparent for internal medicine services The very high admission rates for the (small) population covered by employer-organised pre-paid health care schemes are not surprising, for these patients the majority of whom live in and around Bwamanda township have no financial cost to bear in case of hospital admission.

Table 6: Hospital admissions 1/4/93-31/3/94

	Insured population	Non insured population	Population covered by employer's prepaid schemes	Ratio admission rate insured/ admission rate non insured
Paediatrics	125‰	3.3‰	88.4‰	3.8
Gynaecology	2.7‰	0.8‰	8.4‰	3.4
Internal medicine	5.4‰	4.0‰	16.8‰	1.35
Surgery men	4.4‰	0.4‰	6.8‰	11.0
Surgery female	3.6‰	0.3‰	12.8‰	12.0
Maternity	11.0‰	1.6‰	11.6‰	6.9
Maternity, per 100 expected deliveries	27.6%	4.1%	29%	6.7