Evaluating outreach clinics.

A notable feature of marginalised groups in society is their poor contact with healthcare systems. Disease control strategy among these groups requires a diVerent approach to measures developed for the general population. Sex workers and those falling within the broad definition of prostitution suggested by Day and Ward are one such group. The social prejudice they suVer causes many of them and their partners to avoid even the most appropriate and accessible specialist services. They are less likely to obtain preventative health care, and their marginalisation is an obstacle to screening. In an attempt to overcome this, and to provide occupational services that prostitutes want and use, the outreach clinic evolved. 7 These clinics are nevertheless diYcult to set up and their success within the target group diYcult to monitor, not least because demographics of the prostitute population, acquired indirectly, are likely to be inaccurate. Any system evaluating an outreach service must be linked to the epidemiological and social background. Horn et al search for a new method of evaluating the eVectiveness of their Edinburgh outreach clinic. To discover whether attendance was, firstly, “proactive, and health promoting” or, secondly, “reactive and symptom driven” they employed an hierarchical approach rather than a standard monitoring of trends in infection. Clients falling into the first category are given a positive score, those in the second, a negative. A weighting was added to represent the significance of diVerent issues: termination of pregnancy has a score of −10; hepatitis B vaccination, a score of +10. A strong positive score was given to fitting a contraceptive cap, as an additional barrier against infection—for example, pelvic inflammatory disease, which carries a score of −10. It is this arbitrary and subjective scoring that presents diYculties. Data measured by ordinal scales should be analysed by non-parametric methods. It is inappropriate to give the mean of the scores for the diVerent procedures before a ranking of procedures has been established. The authors’ approach is analogous to giving one stripe to a private and three to a sergeant. The private might equally well have four stripes, and the sergeant seven. Indeed the authors themselves admit that they repeated their calculations, varying the scores, and the overall results were unaVected. By listing a scoring system, as in their table 1, the authors lead the reader to expect that there will be an overall score for each year and that positive and negative scores will be combined to give an overall annual verdict. This does not seem to happen. In table 2, we see only a proactive score, no reactive score. It is also unclear whether the proactive score is a result of the positive score minus the negative score, or whether it is derived from the positive score alone. The authors have compared their scoring system with other systems such as the Glasgow coma scale and the APACHE score. Such comparisons are, however, invalid as the Glasgow coma scale and APACHE score relate to the clinical state of individual patients. The scoring system the authors propose is for evaluating the impact of outreach on health behaviour, quite diVerent from evaluating clinical outcome. If a scoring system is to work it must have real relevance to the eVectiveness of procedures in arriving at the authors’ goal: to contain and control sexually transmitted diseases (STDs) in Edinburgh’s prostitute population. Has contraceptive advice a higher, lower, or similar ranking in this context to hepatitis B vaccination, and why? These rankings are at the heart of identifying the performance of a clinic. Only with them will statistical analysis help in predicting an outreach clinic gold standard. To this end, epidemiological and demographic data, although diYcult to acquire, must be more thoroughly exploited. Clinical data refer only to those who attend the clinic. On its own, it tells little about the impact of the service on the sexual health and levels of STDs in the prostitute population or the population at large. Demographic data from clinic clients and the prostitute population are essential. The use of drugs and alcohol will aVect many aspects of sexual health including attendances, treatment adherence, and the use of condoms. Perhaps outreach workers and paid, trained, peer group worker prostitutes could be enlisted to help link the evaluating system described with epidemiological and social research. 11 The authors have suggested following a cohort of prostitutes to monitor the clinic’s eVectiveness. Cohort studies have great value but, in such a group, a tailing oV of attendance could be the normal pattern for STD clinic attendees in general. Clients are unlikely to return to the clinic on a regular basis unless they are specifically advised to do so. Cohort studies are problematic even with patients attending routine clinics, as their visits are transitory given the short term nature of their infections. Many STD clinics have increasing numbers of return patients but on an ad hoc basis. Such problems are likely to be compounded in a prostitute cohort as the occupation itself is transitory. There may be benefit in trying diVerent ways of evaluating STD services, but if this method is Sex Transm Inf 1999;75:198–199 198