463
EPIDEMIOLOGY Impact of user fees on attendance at a referral centre for
sexually transmitted diseases in Kenya
We investigated the impact of a short-lived policy of charging fees to patients attending public-sector outpatient health facilities in Kenya by collecting data on attendance at Nairobi’s Special Treatment Clinic for sexually transmitted diseases (STDs) before (23 months), during (9 months), and after (15 months) the user-charge period.
During the user-charge period, the seasonally adjusted total mean monthly attendance of men decreased significantly to 40% (95% Cl 36-45) of that before fees
levied. Attendance rose in the post-user-charge period, but reached only 64% (59-68) of the pre-user-charge level. For women, the adjusted total mean monthly attendance during the user-charge period was reduced significantly to 65% (55-77) of the pre-user-charge level. Mean monthly attendance by women rose in the postuser-charge period to 22% (9-37) above the preuser-charge level. There was no evidence of an increase in attendance over the course of the user-charge period among either men or women. The introduction of user fees probably increased the number of untreated STDs in the population, with potentially serious long-term health implications. The user-fee experience in Kenya should be carefully evaluated before similar measures are introduced elsewhere. were
Lancet 1992; 340: 463-66.
Introduction In Kenya, sexually transmitted diseases (STDs) are among the most common presenting complaints of adults at outpatient health facilities, representing 5-10% of the case-load at many clinics.1 Gonorrhoea, syphilis, chancroid, chlamydia, genital herpes, and human immunodeficiency virus type 1 (HIV-1) infections are the most important STDs in Kenya. The prevalence of these infections in the general population is unknown, but in an unselected antenatal population in Nairobi the prevalence of STDs was over 15%, with rates for gonorrhoea, syphilis, and HIV-1 of 5%, 5%, and 10%, respectively.2 The main public-sector health-service provider in Nairobi is the Nairobi City Commission (NCC), which operates 54 health facilities. Patients presenting at NCC primary-level health centres with genital-tract complaints ’with the usual exception of women with pelvic inflammatory disease) are routinely referred to the NCC’s Special Treatment Clinic (STC). This referral pattern has
accounted for almost all patients attending the STC. A programme to decentralise STD treatment to the primary level started at the end of 1991 in three NCC health centres, and it is planned to further decentralise STD services. The Government of Kenya began charging fees for patients attending public-sector outpatient facilities in December, 1989. Although it was intended that STD patients would be exempt from such charges at primarylevel facilities, patients generally paid before diagnosis. In practice, therefore, patients with STDs were only exempt on follow-up visits. Furthermore, it was decided to levy an initial visit fee at the STC. The outpatient charge at government health centres was 10 Kenya shillings (Ksh), about US$0-35, and the fee for an initial visit to the STC was 50 Ksh. Thus, a patient who had been diagnosed and prescribed treatment would pay 60 Ksh, leaving aside the costs incurred in seeking care (eg, loss of income, transport). These are sizeable sums for many people. In Nairobi, more than 70% of households (3-6 people on average) have a monthly income less than 3000 Ksh (US$107-15), and more than 20% have a monthly income less than 1000 Ksh.3 It is these poorer people who are served by the STC, so the introduction of user fees might be expected to discourage some people from seeking care. In other words, some "demand elasticity" might be expected, especially for conditions such as STDs that are not immediately
incapacitating. On September 1, 1990, after 9 months, charging of outpatient at government health facilities was stopped. The reasons given were public dissatisfaction and administrative difficulties. The introduction and subsequent withdrawal of fees has provided an opportunity to examine the impact of user charges on attendance at Nairobi’s principal treatment and referral
centre
for STDs.
Methods Information on attendance at the STC was collected from routine clinic records for: January, 1988, to November, 1989 (period one, ADDRESSES. Departments of Community Health and Medical Microbiology, University of Nairobi, Nairobi, Kenya (S Moses, MD, F. A Plummer, MD); Departments of Community Health Sciences, Medical Microbiology, and Medicine, University of Manitoba, Winnipeg, Canada (S. Moses, F. A Plummer); Health Sciences Division, Regional Office for Eastern and Southern Africa, International Development Research Centre, Nairobi (F. Manji, PhD); International Projects Assistance Services, Nairobi (J. E. Bradley, MA); International Statistical Institute, The Hague, Netherlands (N J. D. Nagelkerke, PhD); and Special Treatment Clinic, Nairobi City Commission, Nairobi (M. A. Malisa, MB). Correspondence to Dr Stephen Moses, Department of Community Health, University of Nairobi, PO Box 19676, Nairobi, Kenya.
464
Fig 1-New
attendances
by
men
and
women at
the STC
by
month.
Fig 2-New attendances by
men
at the STC
by diagnosis by
month. .
=
men, .
=
women, .
=
men
and
women.
0 = gonorrhoea, . = non-STD, . = chancroid.
the 23 months immediately before the introduction of user fees); December, 1989, to August, 1990 (period two, the 9-month user-fee period); and September, 1990, to November, 1991 (period three, the 15 months after the withdrawal of user fees). Between January, 1988, and November, 1991, the pattern of routine referral of STD patients from primary level to the STC was in force. At the STC, diagnoses were made clinically and were not routinely confirmed by laboratory investigations. Data were collected on patient sex, date of presentation, and the diagnosis made. The common diagnoses in men were gonorrhoea, syphilis, non-gonococcal urethritis, and chancroid, and in women gonorrhoea, syphilis, trichomoniasis, candidiasis, and chancroid. Genital herpes was uncommon, so we have excluded this condition from analysis. Because suspected STD cases were routinely referred to the STC from primary facilities, some patients were diagnosed as having a condition other than an STD, and these non-STD attendances were also recorded. Statistical analysis was based on the assumption that total monthly attendances are generated as a random Poisson process, with a mean value that varies systematically with month of year (to take into account possible seasonal patterns) and with time of attendance (periods one, two, and three). For the mean attendance in calendar monthi and period j, the following model was postulated: log (Ilij) a + month (1) + period 0), where a is the natural logarithm (e) of the mean attendance in month 1 during period one (ie, January, 1988). This then became the reference value for estimating changes in attendance. Variations between the same month in different years within the same period were thus assumed to be random. The GLIM program (Royal Statistical Society, London, UK) was used to estimate a andi for the remaining 11 months (February to December), and j for the periods during and =
SEASONALLY ADJUSTED MEAN
(95% CI) MONTHLY
’Slgnlflcantly different (p = 0 05) from 100%
after user charges. Since the model is additive on a logarithmic scale, the effect of the parameters is multiplicative. For example, ifi for March (month 3) was 1, then clinic attendance for that month would be 2-72 (el) times as high as during the reference month (January). Because the data represent a time series there may be some extra-Poisson variation caused by, for example, specific disease outbreaks lasting for a finite period. Since such occurrences could partly invalidate the standard error estimates, we adjusted by automatic scaling of the variances.
Results Actual numbers of new attendances by month for men and women before, during, and after the user-charge period are shown in fig 1, and the table gives the seasonally adjusted mean monthly attendance by sex and diagnosis for the three periods. The table also shows the mean monthly attendance in periods two and three as a percentage of the mean monthly attendance in period one. The adjusted mean monthly attendance for men in period two was 40% of that in period one, and although the adjusted monthly attendance rose in period three, it was still only 64% of that in period one (table). The adjusted mean monthly attendance of women also fell significantly during the user-charge period, but increased to 122 % of the pre-charge attendance during the post-charge period. Men showed a more striking drop in attendance than women during the user-charge period. New attendances by men were about double those of women during period one,
PATIENT ATTENDANCE AT THE STC BEFORE, DURING, AND AFTER USER CHARGES
465
Discussion
Monm
Fig 3-New attendances by by month. I:8J = trichomoniasis,. . = chancroid.
women
at the
STC by diagnosis
= non-STD,. = candidiasis, 0 = gonorrhoea,
dropped to roughly the same level as women during period two, and have since remained at about the same level as women (fig 1). For both women and men there was no evidence of any trend towards an increase in attendance over the course of the user-charge period. Monthly actual new attendances by men for gonorrhoea, chancroid, and non-STDs are shown in fig 2. The most striking fall in adjusted mean monthly attendance during period two, to 5 % of the level in period one, was among men diagnosed as not having an STD. Statistically significant decreases also occurred among men with diagnoses of chancroid, gonorrhoea, non-gonococcal urethritis, and syphilis (table). Attendance by men with non-STD diagnoses remained low during period three, whereas attendance for the other diagnoses rose, although not to the levels observed in period one (fig 2, table). Monthly actual new attendances for the most common diagnoses in women are shown in fig 3. As with men, there was a striking drop in non-STD diagnoses during period two and this fall was sustained during period three. Mean monthly trichomoniasis diagnoses during period two fell to 69% of the period-one level, and then rose to nearly the period-one level during period three. Mean monthly candidiasis diagnoses remained constant during periods one and two, but rose strikingly during period three to 353 % of the period-one level. In period two, mean monthly chancroid diagnoses in women dropped to 72% of the pre-charge level then rose to significantly above the period-one level in period three. Mean monthly gonorrhoea diagnoses increased during period two to 132% of the period-one level and continued to rise during period three to over four times the level in period one. There was also a significant fall in the adjusted mean monthly attendance during period two for women diagnosed as having syphilis (to 55% of the period-one level), although attendance rose in period three to 76% of the period one-level (table).
User charges for health services in Kenya were introduced on the assumption that the demand for essential health services is inelastic.l°4 It was believed that charging would lead to very little decrease in the use of health services and that the policy would generate income for the health sector. Furthermore, it was hoped that demand for health services for trivial complaints would decrease, thereby freeing scarce resources for more serious problems. The notion of "need" as the criterion for allocation of services was largely replaced by the concept of "demand". 5-7 If we assume that non-STD diagnoses represent unimportant problems, then our data suggest that the demand for inessential health services is indeed elastic, because there was a striking reduction in attendance for non-STD diagnoses during the user-charge period. However, the introduction of user fees also resulted in reductions in attendance for almost all STDs. Furthermore, there was no evidence of recovery in attendances until after user charges were discontinued. This finding calls into question the assumption that the demand for essential health services is largely inelastic. Our data show only the impact of user charges on the effective demand for STC services; we have no clear knowledge of the many factors that influence decisions about seeking care for STDs, although this issue has been looked at in more general terms in Kenyan populations.8 Some insight into the process involved is essential for adequate planning of health services. For example, it is possible that part of the drop in attendance at the STC during the user-charge period is attributable to changes in clinical practice at primary-health centres. Staff at these centres may have referred fewer patients to the STC because of the high fees levied there, which would account for part of the decrease observed in non-STD as well as STD diagnoses. It is unlikely, however, that staff at primaryhealth centres would have provided much effective treatment, because up to and including the period under study these staff had little training in STD management, and appropriate drugs for STD treatment were generally not available. The reasons for sex differences in health-service use over the three study periods are unclear. It is possible that the initially smaller numbers of women than men attending the STC reflects a larger "opportunity cost" for women (ie, it may be more costly socially and economically for women to attend a health facility or to seek care). This may also explain why the fall in attendance after the introduction of user fees was less striking for women, since the relative increase in total costs (opportunity costs plus user charges) was probably smaller for women. After user fees were withdrawn, attendance by men remained substantially lower than in the pre-charge period, indicating a sustained change in health-seeking behaviour among male STD
patients. Under conditions where STDs are endemic, a decrease in the cure rate for a given STD could bring about a substantial increase in disease incidence and prevalence. This might explain why the number of women diagnosed with gonorrhoea increased over time, particularly after the withdrawal of user charges. It is unclear, however, why candidiasis diagnoses also increased, and we do not know whether this is a coincidental finding or has some other explanation, such as women with an infection that may not have been very severe taking advantage of the reintroduction
466
of free treatment. It is also possible that patterns of diagnosis and disease classification have changed over the study period, although there is no evidence for this. Were patients no longer seen at the STC provided with appropriate care in the private sector? Since private-sector fees are generally higher than those charged at the STC, it seems unlikely that the balance between public and private sectors would have changed to the extent required to account for the drop in attendances we observed. It is possible that traditional healers, drug peddlers, and other informal operators may have absorbed some of the fall-out from the STC. However, as far as can be ascertained, these members of the private sector cannot treat STDs
adequately. It has been argued that about half a day’s pay is an affordable fee for a medical consultation in developing countries and that two consultations per year should be an affordable average,9,10 implying that such fees should not affect the demand for health services, at least not for essential services. The fees charged at the Nairobi STC were of this order of magnitude, but they were clearly sufficient to deter attendance. The advocates of user charges usually stipulate that there should be exemptions for the poor. But deciding who is eligible foi -cli relief causes difficulties even in industrialised countries, and introducing a formal system for identifying the poor would be costly and impractical in countries sucn as Kenya.l1.1Z Most of those attending the STC are poor, and they are therefore among the most vulnerable to ill-health. Studies in the developing countries Bangladesh,13 Zaire,l4 Ghana,15.16 Swaziland,l’ and Lesotho18 have shown that, as might be expected, use of health services by the poor is more sensitive to and more affected by price increases than use by the rich. Similar conclusions have also been drawn from studies in industrialised countries.6.19 The impact on attendance at the STC of user charges was clearly undesirable. Unless patients received adequate care elsewhere, it is probable that untreated STDs in the population, and consequently STD transmission, increased as a result of the user-fees policy. Of particular concern is the decline in attendance by men and women with chancroid. If, as is believed, chancroid facilitates HIV transmission,10,21 then an increase in the numbers of cases of untreated or inadequately treated chancroid will pose a major obstacle to HIV control. User charges for health services are being increasingly adopted by developing countries, often at the behest of the or World Bank International Monetary Fund. Unfortunately, the arguments in favour of user fees are not supported by empirical data. As Griffin22 concluded in a review of the user-charge issue, none of the demand studies refer to "actual field experiments in which user fees were introduced and their effects studied: all are computer exercises". Such a field "experiment" was undertaken in Kenya during the 9 months between December, 1989, and August, 1990, when user fees were levied for outpatient services. We have looked at only one aspect of a system that was implemented on a national scale. National data should prove an invaluable source of information on the response of a health system to changes in its costing procedures. We thank the Public Health Department, for their assistance and cooperation.
NCC, and the staff of the STC
REFERENCES 1. Resources for Child Health Project (REACH). Nairobi area study. Nairobi: United States Agency for International Development, 1988.
2. Temmerman M, Maitha G, Ndinya-Achola JO, et al. HIV-1and syphilis infection in women in Nairobi, Kenya (abstr WA146). 6th International Conference on AIDS in Africa; 1991 Dec 16-19; Dakar,
Senegal. 3. Odada JE, Otieno JO, eds. Socio-economic profiles. Nairobi: Government of Kenya, 1990. 4. World Bank. Financing health services in developing countries. Washington: World Bank, 1987. 5. Kanji N, Kan)i N, Manji F. From development to sustained crisis: structural adjustment, equity and health. Soc Sci Med 1991; 33: 985-93. 6. Creese AL. User charges for health care: a review of recent experience. Current Concerns no 1. Geneva: World Health Organisation WHO/
SHS/CC/90.1, 1990. 7.
Kanji N. Charging for drugs in Africa: UNICEF’s "Bamako Initiative". Health Policy Planning 1989; 4: 110-20.
8. Mwabu G, Mwangi WM. Health care financing in Kenya: a simulation of the welfare effects of user fees. Soc Sci Med 1986; 22: 763-67. 9. de Ferranti D. Paying for health services in developing countries: an overview. World Bank Staff Working Papers, No 721. Washington: World Bank, 1985. 10. de Ferranti D. Strategies for paying for health services. World Health Stat Q 1984; 37: 428-50. 11. Abel-Smith B. Financing health for all. World Health Forum 1991; 12: 191-200. 12. Abel-Smith B. Discussion paper on issues and options in health financing (in Kenya). Consultant report to World Bank/SIDA. Nairobi: World
Bank, 1989. 13. Stanton B, Clemens J. User fees for health-care in developing countries a case study of Bangladesh. Soc Sci Med 1989; 29: 1119-205. 14. de Bethune X. The influence of an abrupt price increase on health service utilization: evidence from Zaire. Health Policy Planning 1989; 4: 76-81 15. Waddington C, Enyinmayew K. A price to pay: the impact of user charges in Ashanti-Akim District, Ghana. Int J Health Planning Management 1989; 4: 17-47. 16. Waddington C, Enyinmayew K. A price to pay: the impact of user charges in Ashanti-Akim District, Ghana; part 2: the impact of charges in the Volta region of Ghana. Int J Health Planning Management 1990; 5: 287-312. 17. Yoder RA. Are people willing and able to pay for health services? Soc Sci Med 1989; 29: 35-42. 18. Bennet S. The impact of the increase in user fees: a preliminary investigation. Lesotho Epidemiol Bull 1989; 4. 19. Beck RG, Horne JM. Utilization of publicly insured health services in Saskatchewan before, during and after copayment. Med Care 1980; 18: 787-805. 20. Cameron DW, Simonsen JN, D’Costa LJ, et al. Female to male transmission of human immunodeficiency virus type 1: risk factors for seroconversion in men. Lancet 1989; ii: 403-07. 21. Pépin J, Plummer FA, Brunham RC, et al. Editonal review: the interaction of HIV infection and other sexually transmitted diseases: an opportunity for intervention. AIDS 1989; 3: 3-6. 22. Griffin C. User charges for health care in principle and in practice. Economic Development Institute Seminar Paper No 37. Washington: World Bank, 1988.
From The Lancet "Heads" must roll
Systems of government, control, and mangement, have too often their foundations in ignorance and corruption, and, being generated and fostered in such loathsome soils, they rapidly and continuously produce fruit of a truly appropriate description. If the medical institutions of this country were founded upon a rational basis, if the principles which regulate their government were sound, we should only have to give activity to those principles by an appropriate arrangement of details; but now all is defective; there are no just principles of government in operation.... But do the "heads" of the profession, as they are stupidly denominated, enter into strong and powerful combinations for the purpose of establishing the operation of new principles of election and management in our medical colleges and hospitals? Nothing of the kind is to be heard or seen. Slothfully and negligently do they witness the foul workings of the rotten system of which they form a part; and indifferently do they gaze on it, while their professional brethren and the community alike suffer from its pernicious effects. Such men deserve no forbearance. (Oct 22, 1842)
EPIDEMIOLOGY Impact of user fees on attendance at a referral centre for
sexually transmitted diseases in Kenya
We investigated the impact of a short-lived policy of charging fees to patients attending public-sector outpatient health facilities in Kenya by collecting data on attendance at Nairobi’s Special Treatment Clinic for sexually transmitted diseases (STDs) before (23 months), during (9 months), and after (15 months) the user-charge period.
During the user-charge period, the seasonally adjusted total mean monthly attendance of men decreased significantly to 40% (95% Cl 36-45) of that before fees
levied. Attendance rose in the post-user-charge period, but reached only 64% (59-68) of the pre-user-charge level. For women, the adjusted total mean monthly attendance during the user-charge period was reduced significantly to 65% (55-77) of the pre-user-charge level. Mean monthly attendance by women rose in the postuser-charge period to 22% (9-37) above the preuser-charge level. There was no evidence of an increase in attendance over the course of the user-charge period among either men or women. The introduction of user fees probably increased the number of untreated STDs in the population, with potentially serious long-term health implications. The user-fee experience in Kenya should be carefully evaluated before similar measures are introduced elsewhere. were
Lancet 1992; 340: 463-66.
Introduction In Kenya, sexually transmitted diseases (STDs) are among the most common presenting complaints of adults at outpatient health facilities, representing 5-10% of the case-load at many clinics.1 Gonorrhoea, syphilis, chancroid, chlamydia, genital herpes, and human immunodeficiency virus type 1 (HIV-1) infections are the most important STDs in Kenya. The prevalence of these infections in the general population is unknown, but in an unselected antenatal population in Nairobi the prevalence of STDs was over 15%, with rates for gonorrhoea, syphilis, and HIV-1 of 5%, 5%, and 10%, respectively.2 The main public-sector health-service provider in Nairobi is the Nairobi City Commission (NCC), which operates 54 health facilities. Patients presenting at NCC primary-level health centres with genital-tract complaints ’with the usual exception of women with pelvic inflammatory disease) are routinely referred to the NCC’s Special Treatment Clinic (STC). This referral pattern has
accounted for almost all patients attending the STC. A programme to decentralise STD treatment to the primary level started at the end of 1991 in three NCC health centres, and it is planned to further decentralise STD services. The Government of Kenya began charging fees for patients attending public-sector outpatient facilities in December, 1989. Although it was intended that STD patients would be exempt from such charges at primarylevel facilities, patients generally paid before diagnosis. In practice, therefore, patients with STDs were only exempt on follow-up visits. Furthermore, it was decided to levy an initial visit fee at the STC. The outpatient charge at government health centres was 10 Kenya shillings (Ksh), about US$0-35, and the fee for an initial visit to the STC was 50 Ksh. Thus, a patient who had been diagnosed and prescribed treatment would pay 60 Ksh, leaving aside the costs incurred in seeking care (eg, loss of income, transport). These are sizeable sums for many people. In Nairobi, more than 70% of households (3-6 people on average) have a monthly income less than 3000 Ksh (US$107-15), and more than 20% have a monthly income less than 1000 Ksh.3 It is these poorer people who are served by the STC, so the introduction of user fees might be expected to discourage some people from seeking care. In other words, some "demand elasticity" might be expected, especially for conditions such as STDs that are not immediately
incapacitating. On September 1, 1990, after 9 months, charging of outpatient at government health facilities was stopped. The reasons given were public dissatisfaction and administrative difficulties. The introduction and subsequent withdrawal of fees has provided an opportunity to examine the impact of user charges on attendance at Nairobi’s principal treatment and referral
centre
for STDs.
Methods Information on attendance at the STC was collected from routine clinic records for: January, 1988, to November, 1989 (period one, ADDRESSES. Departments of Community Health and Medical Microbiology, University of Nairobi, Nairobi, Kenya (S Moses, MD, F. A Plummer, MD); Departments of Community Health Sciences, Medical Microbiology, and Medicine, University of Manitoba, Winnipeg, Canada (S. Moses, F. A Plummer); Health Sciences Division, Regional Office for Eastern and Southern Africa, International Development Research Centre, Nairobi (F. Manji, PhD); International Projects Assistance Services, Nairobi (J. E. Bradley, MA); International Statistical Institute, The Hague, Netherlands (N J. D. Nagelkerke, PhD); and Special Treatment Clinic, Nairobi City Commission, Nairobi (M. A. Malisa, MB). Correspondence to Dr Stephen Moses, Department of Community Health, University of Nairobi, PO Box 19676, Nairobi, Kenya.
464
Fig 1-New
attendances
by
men
and
women at
the STC
by
month.
Fig 2-New attendances by
men
at the STC
by diagnosis by
month. .
=
men, .
=
women, .
=
men
and
women.
0 = gonorrhoea, . = non-STD, . = chancroid.
the 23 months immediately before the introduction of user fees); December, 1989, to August, 1990 (period two, the 9-month user-fee period); and September, 1990, to November, 1991 (period three, the 15 months after the withdrawal of user fees). Between January, 1988, and November, 1991, the pattern of routine referral of STD patients from primary level to the STC was in force. At the STC, diagnoses were made clinically and were not routinely confirmed by laboratory investigations. Data were collected on patient sex, date of presentation, and the diagnosis made. The common diagnoses in men were gonorrhoea, syphilis, non-gonococcal urethritis, and chancroid, and in women gonorrhoea, syphilis, trichomoniasis, candidiasis, and chancroid. Genital herpes was uncommon, so we have excluded this condition from analysis. Because suspected STD cases were routinely referred to the STC from primary facilities, some patients were diagnosed as having a condition other than an STD, and these non-STD attendances were also recorded. Statistical analysis was based on the assumption that total monthly attendances are generated as a random Poisson process, with a mean value that varies systematically with month of year (to take into account possible seasonal patterns) and with time of attendance (periods one, two, and three). For the mean attendance in calendar monthi and period j, the following model was postulated: log (Ilij) a + month (1) + period 0), where a is the natural logarithm (e) of the mean attendance in month 1 during period one (ie, January, 1988). This then became the reference value for estimating changes in attendance. Variations between the same month in different years within the same period were thus assumed to be random. The GLIM program (Royal Statistical Society, London, UK) was used to estimate a andi for the remaining 11 months (February to December), and j for the periods during and =
SEASONALLY ADJUSTED MEAN
(95% CI) MONTHLY
’Slgnlflcantly different (p = 0 05) from 100%
after user charges. Since the model is additive on a logarithmic scale, the effect of the parameters is multiplicative. For example, ifi for March (month 3) was 1, then clinic attendance for that month would be 2-72 (el) times as high as during the reference month (January). Because the data represent a time series there may be some extra-Poisson variation caused by, for example, specific disease outbreaks lasting for a finite period. Since such occurrences could partly invalidate the standard error estimates, we adjusted by automatic scaling of the variances.
Results Actual numbers of new attendances by month for men and women before, during, and after the user-charge period are shown in fig 1, and the table gives the seasonally adjusted mean monthly attendance by sex and diagnosis for the three periods. The table also shows the mean monthly attendance in periods two and three as a percentage of the mean monthly attendance in period one. The adjusted mean monthly attendance for men in period two was 40% of that in period one, and although the adjusted monthly attendance rose in period three, it was still only 64% of that in period one (table). The adjusted mean monthly attendance of women also fell significantly during the user-charge period, but increased to 122 % of the pre-charge attendance during the post-charge period. Men showed a more striking drop in attendance than women during the user-charge period. New attendances by men were about double those of women during period one,
PATIENT ATTENDANCE AT THE STC BEFORE, DURING, AND AFTER USER CHARGES
465
Discussion
Monm
Fig 3-New attendances by by month. I:8J = trichomoniasis,. . = chancroid.
women
at the
STC by diagnosis
= non-STD,. = candidiasis, 0 = gonorrhoea,
dropped to roughly the same level as women during period two, and have since remained at about the same level as women (fig 1). For both women and men there was no evidence of any trend towards an increase in attendance over the course of the user-charge period. Monthly actual new attendances by men for gonorrhoea, chancroid, and non-STDs are shown in fig 2. The most striking fall in adjusted mean monthly attendance during period two, to 5 % of the level in period one, was among men diagnosed as not having an STD. Statistically significant decreases also occurred among men with diagnoses of chancroid, gonorrhoea, non-gonococcal urethritis, and syphilis (table). Attendance by men with non-STD diagnoses remained low during period three, whereas attendance for the other diagnoses rose, although not to the levels observed in period one (fig 2, table). Monthly actual new attendances for the most common diagnoses in women are shown in fig 3. As with men, there was a striking drop in non-STD diagnoses during period two and this fall was sustained during period three. Mean monthly trichomoniasis diagnoses during period two fell to 69% of the period-one level, and then rose to nearly the period-one level during period three. Mean monthly candidiasis diagnoses remained constant during periods one and two, but rose strikingly during period three to 353 % of the period-one level. In period two, mean monthly chancroid diagnoses in women dropped to 72% of the pre-charge level then rose to significantly above the period-one level in period three. Mean monthly gonorrhoea diagnoses increased during period two to 132% of the period-one level and continued to rise during period three to over four times the level in period one. There was also a significant fall in the adjusted mean monthly attendance during period two for women diagnosed as having syphilis (to 55% of the period-one level), although attendance rose in period three to 76% of the period one-level (table).
User charges for health services in Kenya were introduced on the assumption that the demand for essential health services is inelastic.l°4 It was believed that charging would lead to very little decrease in the use of health services and that the policy would generate income for the health sector. Furthermore, it was hoped that demand for health services for trivial complaints would decrease, thereby freeing scarce resources for more serious problems. The notion of "need" as the criterion for allocation of services was largely replaced by the concept of "demand". 5-7 If we assume that non-STD diagnoses represent unimportant problems, then our data suggest that the demand for inessential health services is indeed elastic, because there was a striking reduction in attendance for non-STD diagnoses during the user-charge period. However, the introduction of user fees also resulted in reductions in attendance for almost all STDs. Furthermore, there was no evidence of recovery in attendances until after user charges were discontinued. This finding calls into question the assumption that the demand for essential health services is largely inelastic. Our data show only the impact of user charges on the effective demand for STC services; we have no clear knowledge of the many factors that influence decisions about seeking care for STDs, although this issue has been looked at in more general terms in Kenyan populations.8 Some insight into the process involved is essential for adequate planning of health services. For example, it is possible that part of the drop in attendance at the STC during the user-charge period is attributable to changes in clinical practice at primary-health centres. Staff at these centres may have referred fewer patients to the STC because of the high fees levied there, which would account for part of the decrease observed in non-STD as well as STD diagnoses. It is unlikely, however, that staff at primaryhealth centres would have provided much effective treatment, because up to and including the period under study these staff had little training in STD management, and appropriate drugs for STD treatment were generally not available. The reasons for sex differences in health-service use over the three study periods are unclear. It is possible that the initially smaller numbers of women than men attending the STC reflects a larger "opportunity cost" for women (ie, it may be more costly socially and economically for women to attend a health facility or to seek care). This may also explain why the fall in attendance after the introduction of user fees was less striking for women, since the relative increase in total costs (opportunity costs plus user charges) was probably smaller for women. After user fees were withdrawn, attendance by men remained substantially lower than in the pre-charge period, indicating a sustained change in health-seeking behaviour among male STD
patients. Under conditions where STDs are endemic, a decrease in the cure rate for a given STD could bring about a substantial increase in disease incidence and prevalence. This might explain why the number of women diagnosed with gonorrhoea increased over time, particularly after the withdrawal of user charges. It is unclear, however, why candidiasis diagnoses also increased, and we do not know whether this is a coincidental finding or has some other explanation, such as women with an infection that may not have been very severe taking advantage of the reintroduction
466
of free treatment. It is also possible that patterns of diagnosis and disease classification have changed over the study period, although there is no evidence for this. Were patients no longer seen at the STC provided with appropriate care in the private sector? Since private-sector fees are generally higher than those charged at the STC, it seems unlikely that the balance between public and private sectors would have changed to the extent required to account for the drop in attendances we observed. It is possible that traditional healers, drug peddlers, and other informal operators may have absorbed some of the fall-out from the STC. However, as far as can be ascertained, these members of the private sector cannot treat STDs
adequately. It has been argued that about half a day’s pay is an affordable fee for a medical consultation in developing countries and that two consultations per year should be an affordable average,9,10 implying that such fees should not affect the demand for health services, at least not for essential services. The fees charged at the Nairobi STC were of this order of magnitude, but they were clearly sufficient to deter attendance. The advocates of user charges usually stipulate that there should be exemptions for the poor. But deciding who is eligible foi -cli relief causes difficulties even in industrialised countries, and introducing a formal system for identifying the poor would be costly and impractical in countries sucn as Kenya.l1.1Z Most of those attending the STC are poor, and they are therefore among the most vulnerable to ill-health. Studies in the developing countries Bangladesh,13 Zaire,l4 Ghana,15.16 Swaziland,l’ and Lesotho18 have shown that, as might be expected, use of health services by the poor is more sensitive to and more affected by price increases than use by the rich. Similar conclusions have also been drawn from studies in industrialised countries.6.19 The impact on attendance at the STC of user charges was clearly undesirable. Unless patients received adequate care elsewhere, it is probable that untreated STDs in the population, and consequently STD transmission, increased as a result of the user-fees policy. Of particular concern is the decline in attendance by men and women with chancroid. If, as is believed, chancroid facilitates HIV transmission,10,21 then an increase in the numbers of cases of untreated or inadequately treated chancroid will pose a major obstacle to HIV control. User charges for health services are being increasingly adopted by developing countries, often at the behest of the or World Bank International Monetary Fund. Unfortunately, the arguments in favour of user fees are not supported by empirical data. As Griffin22 concluded in a review of the user-charge issue, none of the demand studies refer to "actual field experiments in which user fees were introduced and their effects studied: all are computer exercises". Such a field "experiment" was undertaken in Kenya during the 9 months between December, 1989, and August, 1990, when user fees were levied for outpatient services. We have looked at only one aspect of a system that was implemented on a national scale. National data should prove an invaluable source of information on the response of a health system to changes in its costing procedures. We thank the Public Health Department, for their assistance and cooperation.
NCC, and the staff of the STC
REFERENCES 1. Resources for Child Health Project (REACH). Nairobi area study. Nairobi: United States Agency for International Development, 1988.
2. Temmerman M, Maitha G, Ndinya-Achola JO, et al. HIV-1and syphilis infection in women in Nairobi, Kenya (abstr WA146). 6th International Conference on AIDS in Africa; 1991 Dec 16-19; Dakar,
Senegal. 3. Odada JE, Otieno JO, eds. Socio-economic profiles. Nairobi: Government of Kenya, 1990. 4. World Bank. Financing health services in developing countries. Washington: World Bank, 1987. 5. Kanji N, Kan)i N, Manji F. From development to sustained crisis: structural adjustment, equity and health. Soc Sci Med 1991; 33: 985-93. 6. Creese AL. User charges for health care: a review of recent experience. Current Concerns no 1. Geneva: World Health Organisation WHO/
SHS/CC/90.1, 1990. 7.
Kanji N. Charging for drugs in Africa: UNICEF’s "Bamako Initiative". Health Policy Planning 1989; 4: 110-20.
8. Mwabu G, Mwangi WM. Health care financing in Kenya: a simulation of the welfare effects of user fees. Soc Sci Med 1986; 22: 763-67. 9. de Ferranti D. Paying for health services in developing countries: an overview. World Bank Staff Working Papers, No 721. Washington: World Bank, 1985. 10. de Ferranti D. Strategies for paying for health services. World Health Stat Q 1984; 37: 428-50. 11. Abel-Smith B. Financing health for all. World Health Forum 1991; 12: 191-200. 12. Abel-Smith B. Discussion paper on issues and options in health financing (in Kenya). Consultant report to World Bank/SIDA. Nairobi: World
Bank, 1989. 13. Stanton B, Clemens J. User fees for health-care in developing countries a case study of Bangladesh. Soc Sci Med 1989; 29: 1119-205. 14. de Bethune X. The influence of an abrupt price increase on health service utilization: evidence from Zaire. Health Policy Planning 1989; 4: 76-81 15. Waddington C, Enyinmayew K. A price to pay: the impact of user charges in Ashanti-Akim District, Ghana. Int J Health Planning Management 1989; 4: 17-47. 16. Waddington C, Enyinmayew K. A price to pay: the impact of user charges in Ashanti-Akim District, Ghana; part 2: the impact of charges in the Volta region of Ghana. Int J Health Planning Management 1990; 5: 287-312. 17. Yoder RA. Are people willing and able to pay for health services? Soc Sci Med 1989; 29: 35-42. 18. Bennet S. The impact of the increase in user fees: a preliminary investigation. Lesotho Epidemiol Bull 1989; 4. 19. Beck RG, Horne JM. Utilization of publicly insured health services in Saskatchewan before, during and after copayment. Med Care 1980; 18: 787-805. 20. Cameron DW, Simonsen JN, D’Costa LJ, et al. Female to male transmission of human immunodeficiency virus type 1: risk factors for seroconversion in men. Lancet 1989; ii: 403-07. 21. Pépin J, Plummer FA, Brunham RC, et al. Editonal review: the interaction of HIV infection and other sexually transmitted diseases: an opportunity for intervention. AIDS 1989; 3: 3-6. 22. Griffin C. User charges for health care in principle and in practice. Economic Development Institute Seminar Paper No 37. Washington: World Bank, 1988.
From The Lancet "Heads" must roll
Systems of government, control, and mangement, have too often their foundations in ignorance and corruption, and, being generated and fostered in such loathsome soils, they rapidly and continuously produce fruit of a truly appropriate description. If the medical institutions of this country were founded upon a rational basis, if the principles which regulate their government were sound, we should only have to give activity to those principles by an appropriate arrangement of details; but now all is defective; there are no just principles of government in operation.... But do the "heads" of the profession, as they are stupidly denominated, enter into strong and powerful combinations for the purpose of establishing the operation of new principles of election and management in our medical colleges and hospitals? Nothing of the kind is to be heard or seen. Slothfully and negligently do they witness the foul workings of the rotten system of which they form a part; and indifferently do they gaze on it, while their professional brethren and the community alike suffer from its pernicious effects. Such men deserve no forbearance. (Oct 22, 1842)
