Slowing the spread of human immunodeficiency virus in developing countries

608

REVIEW ARTICLE

Slowing the spread of human immunodeficiency virus in developing countries

immunodeficiency viruses type 1 and type 2 (HIV-1, HIV-2), the aetiological agents of AIDS, continue to spread unabated in most countries. The extent to which these viruses will spread further in specific countries will depend on the rapidity with which measures are introduced Human

decrease transmission and on the effectiveness of those In some areas of sub-Saharan Africa, between one tenth and one third of all sexually active adults harbour HIV-1 (fig 1). In certain urban centres in Africa, AIDS is already the leading cause of adult mortality and a major cause of infant deaths.1 Levels of infection have increased rapidly over the past 5 years in many other regions of the developing world--eg, among intravenous drug users attending Thanyorak Hospital in Bangkok, levels of infection rose from a few per cent to over 30% between August, 1987, and August, 1988. A 1990 serological survey of HIV-1in 26 000 young men (age 20-22 years) registering for military service in Thailand showed that 18% were positive (5-6% in the northern provinces and 11-5% in the provincial capital of Chang Mai) .3 Moreover, the long incubation period between HIV infection and the development of AIDS must mean that this global pandemic is still in its early stages. Cheap and effective therapies or vaccines are unlikely to slow the inexorable progress of this disease in the next decade, so it is all the more important to change human behaviour and to introduce other preventive measures to slow HIV transmission. The rapidity of spread of HIV has outstripped the response of some institutions. Thus pledges to the World Health Organisation Global Programme on AIDS for 1991 are 30% below the planned budget of$100 million.4 We now discuss several public health policies based on the natural history of the infection, field experience of AIDS prevention programmes, and insights from mathematical modelling. We have not considered interventions to control HIV among intravenous drug users or transmission through blood transfusions-the former can be important in the early spread of HIV but only in some countries, while the latter, by itself, is unlikely to give rise to a self-sustaining to

measures.

epidemic.

Background epidemiological observations tell us? Estimates of the doubling time of the epidemic in Africa range from 1 to 3 years in sexually active adults in the worst afflicted regions to 5 years or more in areas where the prevalence is currently low. One of the most important routes of spread into the general population in both urban

and rural areas-eg, in Africa, India, and Thailand-is sexual contact by males with female prostitutes who, as a result of their occupation, often acquire the disease early in the history of the epidemic. Horizontal studies reveal a steady rise in the prevalence of HIV-1as teenagers join the sexually active age classes. That maximum levels are in women aged 20-25years and men aged 25-35 years reflects sexual contact between older men and younger women. The rate of vertical transmission from mothers to babies is about 30-40% in developing countriesalthough much lower figures have been reported in European studies (13%).66 Estimates of transmission probabilities (per partnership) are about 20% from male to female and 11 % from female to male. Observed sex ratios of HIV-1 infection in Africa-1 male to 1 ’4 females- accord with a two-fold difference in the magnitude of transmission probabilities between the sexes. Condom use greatly slows HIV transmission between partners.7 The higher prevalence of HIV infection in some developing countries than in developed countries may be a consequence of several factors: (a) sexually transmitted diseases (STDs, often untreated in developing countries) as co-factors to promote transmission; (b) differences in sexual behaviour (eg, number of sexual partners, higher use of condoms, and more oral sex among prostitutes in the west); and (c) networks of sexual contact (eg, between low-risk and high-risk groups). Social and economic processes are important determinants of mixing within or between different strata of the population (eg, age classes, sexual activity classes, urban and rural location).

possibly acting

What do mathematical models tell us? Mathematical models can improve on judgments based experience and intuition, help identify data needs, assist in interpretation of observed patterns, and generate projections under defined assumptions.8,9 Models can be especially helpful in guiding policy formulation about the likely impact of different interventions given various assumptions about their effectiveness and timing of introduction during the course of the epidemic. Models suggest that much is to be gained by targeting intervention strategies at high-risk segments of the on

What do

ADDRESSES: International Family Health, London (M. Potts, MB, PhD), and Parasite Epidemiology Research Group, Imperial College, London University, London, UK (Prof R. Anderson, FRS, M-C Boily, BSc). Correspondence to Dr M. Potts, International Family Health, First Floor, Margaret Pyke Centre, 15 Bateman’s Buildings, London W1V 5TW, UK.

609

determining rate of spread of HIV by sexual contact Rate of acquisition of new sexual partners. - A- Type and frequency of sexual contact within a partnership. * Pattern of mixing between groups with various degrees of sexual activity or other risk behaviours (eg, intravenous drug use). * Probability of transmission from an infected man to susceptible women, and vice versa. * Timing of sexual contact during the long and variable incubation period of AIDS. Factors

widely disseminated epidemic, developing over many decades, can arise even when rates of sexual partner change within the general population are insufficient to maintain the transmission of HIV. The basic reproductive rate of infection, Ro, within low-risk groups in many areas is probably less than unity in value, where Ro defines the average number of secondary cases generated by one primary case, with Ro > 1 needed for an epidemic to occur. However, the overall reproductive rate for low-risk and high-risk (female prostitutes and their male patners) groups combined will be sufficient to generate a widely disseminated epidemic provided low-risk groups have a small degree of contact with high-risk groups. 12 Thus altering contact patterns between high-risk and low-risk groups (wives and girlfriends of men who have regular contact with prostitutes) is clearly important. Small changes in the pattern of mixing between low and high sexual activity classes can significantly slow, or even prevent, a much more widely disseminated epidemic. Effective treatment of STDs is likely to have a beneficial effect. Fig 1-HIV-1 seroprevalence in low-risk (above) (below) urban populations in Africa.

and

high-risk

Estimates of recent HIV-1 seroprevalence in Africa based on data recorded in reference 3. Low-risk samples drawn from populations of pregnant women and blood donors; high-risk samples drawn from prostitutes, bar girls, clients of prostitutes, and patients attending STD clinics.

population, especially in the early stage of HIV spread in the general population (eg, Nigeria at present). Models also suggest that in the worst afflicted regions AIDS can significantly slow or even reverse the sign of population growth rates from positive to negative, if there is no change in existing patterns of behaviour. Models that mirror different mixing or contact patterns between sexual activity classes (high or low activity, based on rates of sexual partner change) suggest very non-linear patterns of epidemic growth which may, under certain circumstances, show two peaks in seroprevalence (fig 2). The first is an early but small peak in the high-activity class (prostitutes and their clients), with a very high level of seroprevalence in this small proportion of the total sexually active population; the second is a much larger peak in the low-activity classes who constitute the majority of the population. In between these peaks, which may be separated by several decades, seroprevalence may show a small decline or remain stable.1O The early part of this process has been observed in some urban centres in Africa—eg, Nairobi." Models highlight the importance of contact patterns between sexual activity classes as the major determinant of the overall pattern of the epidemic.1,9 They also show that a

What additional information is needed? We need to know more about the effectiveness of different intervention strategies. Monitoring STDs in target populations may be a rapid and simple way to track the combined effects of partner change, STD treatment, and condom use in large populations. 13 Study of sexual behaviour and evaluation of interventions will have to be organised locally, but the use of common protocols should be encouraged. Generic research on the scientific assessment of various interventions in widely different

Interventions

Implement education programmes about HIV and AIDS and encourage fewer sexual partners and more safe-sex practices. * Use multiple reinforcing communication channels (eg, radio, community organisations, peer group, counselling) and include target audiences in design of educational materials. Promote and distribute condoms. Improve diagnosis and treatment services (and subsequent condom use) of individuals with STDs. * Improve drug abuse treatment programmes and access to clean needles. * Establish HIV screening for blood transfusion services and promote use of clean needles in immunisation programmes and medical treatment.

*

610

communities is both possible and necessary. It is unwise to rely on changes in the rate of acquisition of HIV infection (assessed by longitudinal cohort studies of seroprevalence) in view of the highly non-linear character of epidemic growth (fig 3). Such information must be complemented by reported changes in sexual activity, condom use, and mixing patterns, ideally assessed by repeated interviews of targeted individuals. With respect to prevention, there is an urgent need for clinical information about barrier methods or virucidal agents that a woman might use if she believes she is at risk of HIV but cannot compel her partner to use a condom. More epidemiological data are also needed to decide if the offer of circumcision is a plausible additional preventive intervention.14 Genetic change should be monitored world wide, as is done for influenza viruses.

Interventions The following generalisations are based on a review of Family Health International AIDSTECH (North Carolina) work, the social marketing programmes of Population Services International (Washington, DC), and

Impact of interventions * Timing-early interventions have a disproportionately greater effect than similar changes introduced later in the course of the epidemic. of the non-linear character of the epidemic a 50% reduction in the rate of transmission of HIV (induced by behaviour change and/or condom use) has more than twice the effect of a 25% reduction. * Targeting-in areas with low prevalence of HIV in the general population ( < 2-5%), behaviour treatment of and condom use in STDs, changes, and their female prostitutes male partners can delay, or even prevent, widespread dissemination of the infection into the general population. * Scale--in areas with medium to high prevalence of H IV in the general population (5-30%) interventions must be targeted at people with high-risk behaviours and the general sexually active population. * Effectiveness-because

selected other programmes.

Timing Money spent on interventions at an early stage in the growth of the epidemic has a much greater impact than similar expenditure at a later stage in terms of the number of cases of infection prevented. Models that chart the slow but continuous development of the epidemic over many decades (fig 3) illustrate the wisdom of designating considerable resources to control the spread of HIV in order to prevent a widely disseminated lethal epidemic over the coming decades. Unfortunately, the most important interventions are sometimes the last to be implemented—eg, programmes to prevent transmission through contaminated blood are commonly designed and funded before plans to promote condoms to groups with high-risk behaviour. Decision makers

reluctant to shift resources for health care or medicine among such marginalised groups,

seem

preventive

U

GV

1.

uu

OV

I uu

TIME IN YEARS FROM START OF EPIDEMIC

Fig 2-Projections, based on a mathematical model, of spread of HIV-1 in a heterosexual population according to sexual activity class.

High-activity men (20% of male population) assumed to have frequent contact (assortative mixing) with high-activity women (prostitutes, who form 1 % of the female population) and limited contact with low-activity women (99% of the population). Trajectories record changes in the proportion mfected over time in the four classes of the sexually active population (from reference 10, where details of model are documented) Total populatIOn = temporal changes in overall proportion infected in sexually active population sexual

whose members can be distrustful of outside help. It can be difficult to convince governments to sanction heavy expenditures on HIV prevention when seroprevalence is low and when there are few or no reported AIDS cases. Many countries have gone through a cycle of denial, followed by what may be called "soft options", and finally by realistic policy setting. For example, one south Asian country, where 20% or more of the prostitutes in some big cities are HIV positive, is not promoting educational or service programmes to this obvious group. Designing and implementing interventions for groups with high-risk behaviour almost always takes less time than securing bureaucratic approval, commonly by a factor of 10 to 1. Among six projects providing education and condoms for prostitutes in Africa and Mexico, the shortest time between developing a concept and getting the programme under way was 3 months, and the longest was about 2 years. Developing a budget and work plan was usually straightforward. The time taken to identify a group who could implement the project and find staff varied from 1 week to 1 year. Some projects were delayed by lack of condoms or difficulty in obtaining a vehicle, and interventions were also hindered by unreliable telephone or facsimile links and by poorly organised local banking services. However, the greatest portion of time was always spent in securing funds, and in obtaining approval from the ministry of health, especially from the national AIDS committee. Thus, an intervention in East Africa, targeted at a group of men known to visit prostitutes frequently, took more than 18 months to gain approval of the committee. Social marketing projects designed to promote and supply condoms to part of or all of a country have proved easier to establish, although even here design is usually more rapid than approval; in Africa it has taken 6-10 months between first discussing a social marketing project to the sale of the first condom.

Effectiveness

Self-reports of condom use by prostitutes for several developing countries (K. Puthikanon, personal communication, and refs 15, 16) show that use increases after education and supply of condoms, even in societies

611

Expansion of interventions will be largely limited by cost by the availability of both governmental and nongovernmental professionals. Prostitute education and condom supply cost$2 50-$9 70 per woman per month. Although costs can be expected to fall as the volume of work rises and relatively less is spent on evaluation, many prostitute populations are highly mobile, new recruits join all the time, and early projects did not include the cost of STD treatment. Some degree of cost-recovery for condoms is likely but will not be substantial (the full cost of a condom can be as high as 20% of the customer’s fee) and the annual cost of providing all African prostitutes with condoms free of charge would be$10-20 million per annum (A. Bhat, B. Muskovitz, personal communication). Experience in family planning suggests that individuals will spend up to 2-3 days’ annual disposable income on condoms for contraception, but condoms cost 2 3-1-5 US cents each and a year’s supply (including distribution and promotion costs) can equal 6-10 days, disposable income. The Bangladesh contraceptive social marketing programme requires a subsidy of$625 per couple year of protection against pregnancy, and this is well below average for family planning programmes. If half the men in Nigeria were to be supplied by a social marketing programme of equal cost-effectiveness, it would cost$18 million a year. US Agency for International Development (which provides 80% of all donated condoms used in developing countries) spends$35 million annually on purchasing condoms. Nationwide programmes of education, condom access, and STD treatment, rather than projects focused solely on groups with high-risk behaviours, present even more formidable problems of support. and

Fig 3-Projections of temporal changes in incidence of AIDS in sexually active population. Projections based on model referred to in fig

2.

The first mmor peak in the incidence of AIDS is generated by very rapid spread m a small proportion of high-risk females (prostitutes) and males (clients of female prostitutes), subsequently, the virus spreads more widely in the population as a result of limited sexual contact between low-risk men and women with high-risk partners of the opposite sex.

with little or no previous experience of this method. Nevertheless, rates of use are invariably insufficient to make a major epidemiological impact. The effectiveness of family planning social marketing projects may be applicable to HIV prevention. The Bangladesh programme, begun in 1975, now has over 100 000 retail outlets and sells over 90 million condoms annually, reaching about 8% of all married couples of reproductive age. In Zaire, a social marketing programme sold just under 1 million condoms in 1988, 4 million in 1989, and 8 million in 1990 and has a target of 18 million by 1992. The programme focuses on AIDS prevention and is accompanied by a big mass-media campaign. Based on the assumption of 100 condoms per man

per year, the programme protects about 2-25% of the active male population. Social marketing

sexually

programmes can target populations or geographical areas known to have high rates of HIV transmission. For example, 1-2 million condoms have been sold in two industrialised

of Zaire where factory workers higher than usual risk of infection. areas

are

thought to have a

Scale The scale of most existing programmes is exceedingly limited. AIDSTECH has implemented projects for groups with high-risk behaviours in nine cities in Africa, and about five other projects are underway supported by other institutions.

Policies To slow the global spread of HIV several policy issues will have to be addressed.

Need to focus resources Where the prevalence of HIV remains low (fig 1), scarce should be focused on groups with high-risk behaviours. These groups (predominantly prostitutes and their clients) warrant preventive interventions in their own right and they represent the most cost-effective application of resources early in the epidemic. Focused interventions do not need much coordination with other aspects of health care, and if such coordination causes delay then it is harmful. Vertical programmes can always be integrated into other services at some future date. However, once the prevalence rises above a few per cent in the general population (1 -1-5 % resources

in

fig 1), nationwide programmes are essential and integration within the medical infrastructure becomes more important. Need for multiple interventions

Scale of interventions

Existing programmes reach about 24 000 prostitutes in seven African countries. * 103 cities with population > 100 000 in subSaharan Africa alone, each needing a project directed at high-risk groups. * Outside Africa only about thirty prostitute intervention programmes, all small scale. * In the Dominican Republic target for intervention is 3000 prostitutes-about a tenth of the female population with high-risk behaviour.

None of the major interventions—education, access to condoms, and STD control-has the power by itself to slow HIV transmission substantially. All three must be implemented as much as resources allow. Modelling suggests that different interventions will act synergistically.

Need for speed

Programmes must be designed and managed so that they can be put in place as rapidly as possible and organisers must have freedom

to

choose the

most

Non-governmental organisations rapidly in areas of perceived

cost-effective solution. often move more controversy than can

can

612

governments. International consideration is being given to a new agency that might channel resources to national

whether they will recommend condoms in addition to pills and other methods, or simply instead of other methods.

non-governmental organisations. In some developing countries the family planning infrastructure is well established and can be a potential partner in HIV control. 17

Prevention

reach some of those on the threshold of sexual activity. National AIDS committees now exist in most countries and can have a useful role in coordination, but they must not be allowed to become a barrier to rapid action. Some committees are too big and include such a wide range of political and religious views that they find it difficult to sanction focused projects-eg, for prostitutes.

Schools

can

Increased funding Resources needed to create a measurable impact on the spread of HIV will have to be an order of magnitude greater than those currently designated for AID S prevention. Some of the money will come from those directly affected: there be cost recovery on condoms and antibiotics for STD to amount to more than 10 or 20% of the total costs in some badly afflicted countries. The shortfall will have to be met by governments’ own tax bases and from the international donor community. A rising proportion of future budgets will have to go towards condom purchase and promotion and STD treatment. Condoms cannot be divided into those used for family planning and those used for AIDS prevention. A competitive global industry can meet the rising demand for high-quality inexpensive condoms; the difficulty will be to find the cash needed to subsidise their availability. Discussions are taking place under the leadership of the United Nations Population Fund to set up a global system, funded by industrialised governments, to purchase contraceptives in the world market.

must

control, but this is unlikely

Population growth and HIV In the next 10-20 years HIV will have little impact on population growth, partly because so many healthy young people are waiting to cross the threshold of post-pubescent sexual activity. Immediate efforts to improve access to family planning will strengthen the ability of most developing nations to fight AIDS. Slower population growth increases the rate of accumulation of family and national wealth. The gross national product (GNP) of several countries in sub-Saharan Africa is falling. If the per caput income continues to fall, as is likely if rapid population growth is not curtailed, by the year 2040 (in the absence of any successful interventions), AIDS treatment could be swallowing up to 4% of the total GNP. However, if the GNP per caput increases, as is likely if voluntary family planning is made available, the same number of AIDS cases in 50 years’ time would consume only about 1-5% of the GNP.8 If effective prevention measures are not implemented and a vaccine is not developed, in 50-100 years the current AIDS epidemic could virtually halve the population of sub-Saharan Africa. An additional factor is the impact of enhanced STD control on net fertility. Effective control of gonorrhoea in sub-Saharan Africa is likely to induce a measurable rise in overall fertility; effective STD control should therefore be linked with family planning initiatives (Brunham RC, et al unpublished). HIV will also have a major impact on family planning policies in high-prevalence areas. Policy makers will have to decide (a) at what level of HIV prevalence they will explicitly promote condoms to all sexually active individuals and not merely to high-risk groups and (b)

vs care

Only about 10% of those infected with HIV have symptoms of AIDS or have died. Even so, the burden of care is already overwhelming some health systems. The total per caput health budget of Zimbabwe is$30 a year, and in some countries government health expenditure is as low as $1 a year. The harsh reality is that zidovudine or expensive antibiotic therapies for intercurrent infections are not going to be readily available in developing countries. Nevertheless, there will be a need for simple therapies and the association between high prevalences of HIV and serious increases in tuberculosis and probably also malaria cannot be ignored.18 Humane and dignified care of AIDS patients is expensive, and there are growing numbers of orphans. How are immediate human needs to be balanced against the fact that short-changing prevention programmes will make things even worse in the future? We believe that donors should earmark money for each purpose separately, rather than set recipients the difficult task of dividing resources between prevention and care.

Conclusions In the next few years, it is unlikely that any new knowledge will emerge to change our understanding of how best to control HIV. Moreover, knowledge that has been available for 5-6 years is still not being fully applied to policy setting and programme design; opportunities to slow the spread of HIV, which can never return, have been missed.

Focused and nationwide interventions will have to be greatly expanded before they can have an epidemiological impact. Large increases in budgets, both from the tax base of the countries concerned and from the donor community, are imperative. Some costs can be recovered from those willing to protect themselves against the risk of HIV (eg, through the social marketing of condoms). Where necessary, development budgets will need to be reallocated. Since AIDS has a major impact on adults during their most economically productive years, the potential scale of the epidemic is such that HIV prevention expenditures may need to be categorised as part of the development budget rather than as merely another health item. For example some African governments spend less than 1% of their GNP on health. If a disease has the potential to kill a fifth of adults it might be reasonable to assign a twentieth or even a tenth of development budgets to its prevention. If interventions are to have any chance of making a measurable impact at a global level, existing projects must be replicated with great speed. If, as analysis suggests, one rate-limiting factor in initiating new programmes is bureaucratic approval, new administrative procedures must be adopted nationally and internationally. Action taken this year will be far cheaper than waiting until next. We thank Ms Susan Crane, Dr Peter Lamptey, Prof Robert May, Prof Peter Piot, Mr Jason Smith, and Dr David Sokal for their cooperation.

REFERENCES 1. Anderson RM, May RM, Boily MC, Garnett GP, Rowley JT. The spread of HIV-1 in Africa: sexual contact patterns and the predicted demographic impact of AIDS. Nature 1991; 352: 581-89. 2. Thangcharven, P, Wasi C, Louisorolchonakul S, et al. HIV infection in Thailand. Bangkok: Mahidol University Publications, 1989. 3. US Bureau of the Census. Recent HIV seroprevalence levels by country. Washington, DC, February, 1991. 4. Ewing T, AIDS programme faces donor fatigue. Nature 1990; 346: 595.

613

5. Prazuck T, Malkin JE, Lechuga P, et al. Mother-to-child transmission of HIV viruses: a cohort study in west Africa, Burkina Faso. VIth International Conference on AIDS. San Francisco, 1990: Th C 610. 6. European Collaborative Study. Children born to women with HIV-1 infection: natural history and risk of transmission. Lancet 1991; 337:

253-60. 7. Potts DM, Short RV. Condoms for the prevention of HIV transmission: cultural dimensions. AIDS 1989; (suppl 1): S259-63. 8. Rowley JT, Anderson RM, Ng TW. Reducing the spread of HIV infection in sub-Saharan Africa: some demographic and economic

implications. AIDS 1990; 4: 47-56. 9. Auvert B, Moore M, Bertrand WE, et al. Dynamics of HIV infection and AIDS in central African cities. Int J Epidemiol 1990; 19: 417-28. 10. Boily M-C, Anderson RM. Sexual contact patterns between men and women and the spread of HIV-1 in urban centres in Africa. IMA J Math Med Biol (in press). 11. Nagel Kerke NJ, Plummer FA, Holton D, et al. Transition dynamics of HIV disease in a cohort of African prostitutes: a Markov model

12.

approach. AIDS 1990; 4: 743-47. May RM, Anderson RM. Transmission dynamics of HIV infection.

Nature 1981; 326: 137-42. 13. Sokal D, Lankoade S, Mugrditchian D, et al. The use of male STD history as an indicator for impact assessment. International Conference on Assessing AIDS Prevention. Montreux, October, 1990. 14. Ronald A, Ndinya-Achola JO, Ngugi EN, Moses S, Brunham R, Plummer FA. Slowing the heterosexual transmission of AIDS. AIDS Soc 1991; Jan/Feb: 1-8. 15. Mhalu F, Hirji K, Ijumba P, et al. A cross-sectional study of a programme for HIV infection control among public house workers. J AIDS 1991; 4: 290-96. 16. Padian NS. Prostitute women and AIDS: epidemiology. AIDS 1988; 2: 413-19. 17. Williamson N, Potts DM. Family planning workers and AIDS prevention. AIDSED Newsletter 1991; 2: 28-32. 18. Goodgame RW. AIDS in Uganda—clinical and social features. N Engl J Med 1990; 323: 383-89.

SCREENING

Pilot study of screening for prostate cancer in

general practice

The

screening programme for cancer depends sensitivity of the tests used and on the of the target population that comes proportion forward for screening. To assess the value of digital rectal screening and prostate-specific antigen (PSA) measurement as screening measures, the 814 men in a city general practice aged between 55 and 70 were success

of

a

on the

recruited in

one

of five different ways. Men with

a

palpably suspicious prostate or a serum PSA greater ng/ml were referred for transrectal ultrasonography and, if indicated, biopsy. 472 men (58%)

than 4 were

screened; of these 68 underwent transrectal

ultrasonography and 29 biopsy. In 7 the biopsy specimen showed carcinoma. Serum PSA was better than digital examination as a screening test—all men with prostate cancer had raised concentrations of serum PSA, whereas only 1 had a palpably abnormal prostate. All 7 had localised disease, and 5 underwent radical prostatectomy. The best methods of patient recruitment were to send an appointment for screening and to "tag" the patient’s notes.

antigen, a more sensitive serum marker than acid phosphatase for prostate cancer, now has an established role in the monitoring of patients known to have prostate cancer.9,10 Its value in screening has only recently received attention-measurement of serum prostate specific antigen was a useful adjunct to rectal examination and transrectal ultrasonography in the detection of prostate cancer, and of the three tests measurement of prostate specific antigen had the greatest predictive value." Unpublished data from our department also suggest that the serum concentration of prostate specific antigen can be used to differentiate prostate cancer from benign disease with acceptable sensitivity and specificity. Transrectal ultrasonography is more sensitive than digital examination for the diagnosis of prostate cancer .12,13 Despite enabling immediate ultrasound-guided biopsy, its use as a primary screening test for prostate cancer has been limited by cost and availability. We have conducted a prostate cancer screening programme based at a large city general practice. The aims of this pilot study14,15 are to assess different methods of patient recruitment, the relative value of the screening tests used in the detection of early prostate cancer, and the costs of the programme.

Introduction Prostate cancer is the commonest cancer in men and its incidence is increasing. Since cure of organ-confmed disease is possible with radical prostatectomy,2 it is worthwhile screening for patients whose disease is at a curable stage. The simplest, and a cost-effective, method of screening for early-stage prostate cancer is by digital rectal examination.3-6 The lack of sensitivity of acid phosphatase concentrations means that measuring this enzyme adds little to screening by digital examination alone.7.8 Prostate specific

Methods At the Horfield Health Centre, which is situated in a residential of North Bristol, seven partners serve a population of 13 200. All 863 men who were aged between 55 and 69, inclusive, on Aug 1, area

ADDRESSES: Horfield Health Centre, Bristol (T. Kemple, MRCGP); and Departments of Urology (D. J. Chadwick, FRCS, D. A. Gillatt, FRCS, P. Abrams, FRCS, J. C. Gingell, FRCS), Clinical Chemistry (J. P. Astley, FRCPath), and Histopathology, Southmead Hospital, Bristol, UK (A. G. Maclver, FRCPath). Correspondence to Mr D. J. Chadwick, Department of Urology, Norfolk and Norwich Hospital, Norwich NR1 3SR.