Botswana’s new training programme for the revised syndromic management protocols was associated with significant improvements in four outcomes: (1) routine HIV testing; (2) physical examination; (3) risk-reduction counselling and (4) patient education about HIV risk. The providers were significantly more likely to offer an HIV test and overall patients were significantly more likely to have an HIV test at training than at comparison clinics. The likelihood of accepting an HIV test among patients who were offered one was the same for both samples, so having an HIV test depended on whether or not a provider offered it.
This is among the first reports of routine HIV testing for STI patients in a resource-limited setting since June 2004, when the World Health Organisation and the Joint United Nations Programme on HIV/AIDS recommended routine HIV testing in STI clinics or other clinics that provide STI care.22
In Botswana, the difference in the percentage of patients who had an HIV test between training (33%) and comparison clinics (14%) was within the range reported in England by Day et al
Providers who attended the new STI training programme were significantly more likely to conduct a physical examination, help patients to make a plan to avoid acquiring STI in the future and discuss HIV risks with patients. Our findings add substantially to the relatively limited body of research demonstrating an effect of STI training in resource-limited settings on conducting a physical examination1 7
and counselling STI patients on their HIV risk.5 7 24
This study is among the first to use patient reports on whether or not tasks were performed by providers. Although patient exit interviews are a well-known method for evaluating the quality of STI care,25
previous researchers have only used them to collect information on patient education,7 26 27
opinions on waiting time and satisfaction with care,27
and contact slips.26
Some researchers consider unannounced (or blinded) standardised patient encounters to be the “gold standard” for measuring the quality of clinical practice.28
Standardised patients can not be used, however, to evaluate some aspects of quality, such as physical examination and the diagnosis of an STI based on clinical symptoms.
The study had several limitations. First, the design was limited to a single time period after training providers at the training sites, so differences between districts could potentially have been confounded with the effects of the training programme. Health districts were, however, selected to minimise differences among clinics and patients and multivariate analyses adjusted for differences between the samples. Second, it is possible that patient visits at training clinics that were not observed were influenced by the recent presence of a clinical specialist. The clinical specialist, however, was not at training clinics on the days when observations were not performed. Third, the analysis may not have fully adjusted for unobserved differences among providers and clinics; random effects analysis of variance regressions would be necessary to adjust fully for these differences. Given the magnitude of the differences between patient reports at training and comparison clinics, however, it is unlikely that the additional analysis would alter the conclusions for the four main outcomes. Fourth, HIV testing results do not include whether or not patients learned their HIV test results. Fifth, the data were collected within two months of training and can not show whether or not the effects of training persisted over time. Future activities will include mentoring and supervision visits for trainees to reinforce the training programme and learn whether or not the results persisted. Finally, the percentage of patients who strongly agreed or agreed with statements about quality of care was high, which could accurately reflect patient experience or could reflect acquiescent response bias.29 30
To the extent that bias existed, it would not be more likely to occur in training than comparison clinics.
Future studies on the outcomes of training programmes should have a more rigorous quasi-experimental design, with baseline and post-training data. In these studies, the provider or clinic should be the unit of analysis, with a sample of a relatively large number of providers or clinics and a small number of patients per provider or clinic. The analysis should account for clustering at the provider and clinic level.
In conclusion, a multifaceted training programme was associated with higher rates of HIV testing, physical examination, risk-reduction counselling and better HIV risk education.