AVAILABLE EVIDENCE AND REASONS FOR OTHER STUDIES

Population based studies provide the best method of assessing the epidemiology of STIs that are not generally confined to core groups, but such studies are rare. In the United Kingdom, the only investigation of this type was the second National Survey of Sexual Attitudes and Lifestyles (NATSAL 2000), which explored the reported population experience of STI infection, and the prevalence of C trachomatis.⁷ This showed that 69% of women diagnosed with C trachomatis had attended GUM services, whereas only 30% diagnosed with PID had attended GUM services. This emphasises the role of services other than GUM in the management of STIs, and highlights the need to widen the scope of STI surveillance. However, the complex, expensive probability sampling technique used in NATSAL is not feasible for routine data collection. While not as accurate as population studies, primary care surveillance could provide estimates of the burden of morbidity within a group that closely corresponds to the general population. Almost 80% of the UK population consult a general practitioner at least once a year, and other primary care settings, such as family planning clinics are regularly attended by young people.⁸

A number of studies have provided information on the burden of STIs in primary care.^9–11 One recent study, the Chlamydia Screening Studies (ClaSS) project, explored C trachomatis prevalence in a general practice registered population aged 16–39 between 2001 and 2002, and included the collection of a limited range of variables for each patient.¹² However, since these studies are unique they cannot be used for surveillance.

Infrequent panel surveys are a source of surveillance data. The Morbidity Statistics from General Practice (MSGP) was undertaken every 10 years and the last study, MSGP4, was undertaken in 1992.⁸ The dataset was derived from attendances over a 1 year period at 60 general practices in England and Wales, and represented a 1% population sample. Although not a random sample, the dataset is broadly representative of the general population in terms of age, sex, marital status, socioeconomic status, smoking behaviour, and burden of disease.⁸ However, although this is a potentially good source of data, it only gives an insight into the prevalence and determinants of incidence within the population at the time of the study. Nationally representative, prospective data are required to produce the timely insights into STI epidemiology that are needed to guide health policy.

Large prospective primary care surveillance datasets have become available over the past decade. The General Practice Research Database (GPRD) and MediPlus UK Primary Care Database are based on general practice attendances. The GPRD covers 6.4% of the UK population, is broadly similar to the Office for National Statistics (ONS) census in terms of age and sex, and includes data on diagnosis and treatment. MediPlus contains clinical and prescribing data from approximately 1.7 million patients who have attended 140 UK general practices since 1991 and is representative of the general population in terms of age, sex, and regional distribution. In addition, the Royal College of General Practitioners’ General Practice Research Unit undertakes continuous sentinel surveillance in 72 practices covering 600 000 people in England and Wales and collects demographic and diagnostic data. However, these datasets cannot be easily adapted to the specific requirements of STI surveillance because they do not capture data on reproductive history, contraceptive history, or sexual behaviour. Clinical samples are not taken, and the methodologies cannot be modified. In addition, the cost of using these datasets can be high. Information from family planning clinics is only consistently available from NHS family planning clinics and Brook Advisory Centres. The data, which are not validated and confined to contraceptive method, age, and sex of each patient at the first contact in the financial year, are reported to the Department of Health on the KT31 return.

The Public Health Laboratory Service (PHLS) laboratory reporting surveillance system managed by the Communicable Disease Surveillance Centre would be an alternative method of obtaining information on STIs diagnosed in primary care. The recent Department of Health (England) publication “Getting ahead of the curve, a strategy for combatting infectious diseases” recommended the mandatory collection of detailed data with all laboratory diagnoses and the PHLS system has been redeveloped to include more information on each report, such as ethnicity and source of report.¹³ This will potentially provide universal laboratory surveillance data for C trachomatis, Neisseria gonorrhoeae, HSV, and syphilis. Soon all positive and negative samples of C trachomatis will be reported, which will allow the diagnostic rate to be estimated in a variety of clinical settings including primary care. Data from the roll out of the English chlamydia screening programme will also provide measures of screening uptake, coverage, and acceptability. However, both datasets are likely to be limited because of the restricted ability to collect relevant demographic and behavioural data. Ideally, surveillance should collect a range of information on factors that influence STI incidence (box 2), so that the diagnoses can be place within the context of the patient’s reproductive and contraceptive history, sexual behaviour, demographic characteristics, and health seeking behaviour. Points of reference should also be included to allow comparisons with related studies.

Overall available primary care surveillance data are limited because many are either unique studies or collected once every 10 years; prospective studies are designed for general disease surveillance; little or no behavioural, demographic, and reproductive health data are collected; and many studies do not use standard diagnostic criteria. A new primary care STI surveillance system would help resolve these problems and create a resource to inform sexual health planning. Here we explore the attributes of such a system.

REPEATED PANEL SURVEYS OF PRIMARY CARE ATTENDERS: A WAY FORWARD?

The aims of the survey would be to estimate prevalence, and change in prevalence over time. The need for the timely collection of detailed data, together with a clinical sample suggests that a methodology based on a substantial population sample is inappropriate. A point prevalence sampling technique would increase efficiency and would be surprisingly small: a random sample of about 4000 for each sex would give a representative view of disease prevalence over the range 2% to 20% (table 1). The methodology and sample size calculations would have to be evaluated in a pilot study and the sampling strategy tailored to surveillance priorities, such as young people or geographic variation in prevalence. Patients could be invited to participate at clinic visits or by post. Measuring change in prevalence is an important method of evaluating intervention, but changes are difficult to predict and may be small.¹⁴ Again, a sample of 4000 would allow year on year trends to be detected accurately.

A self administered patient questionnaire that can be scanned directly into a computer database would be used to reduce data entry errors and allow a database to be constructed and “cleaned” within a few days. Information could thus be collected, interpreted, and disseminated quickly, and would allow surveillance to move towards the proactive stance required for health service planning.

Diagnosis would be the main outcome measure and tests and specimen types would need to be selected carefully. Molecular techniques, such as multiplex or microarray assays, could be used to diagnose several organisms, such as C trachomatis, N gonorrhoeae (including antibiotic resistance) using self obtained vaginal and vulval swabs or non-invasive specimens, such as urine.¹⁵Mycoplasma genitalium, genital HSV infection, and human papillomavirus could also be included in the testing strategy but this may raise ethical concerns as screening for these infections is not generally recommended.

Potentially there are problems associated with a panel survey, the main concern being the comparability between the general practice age/sex register and the general population. Young men are less likely to be registered with a general practitioner than young women. For patients between the ages of 15 and 44 attending general practice the median contact days remain relatively constant over time at 6.4 per 100 people per week and 9.8/100 people per week, for males and females respectively.¹⁶ This panel survey technique is an idealised model and would have to be adapted to specific situations as it may not be feasible or appropriate to collect detailed data in some situations. In addition, the focus on general practitioners may make such a study difficult to implement at a time of increasing demands, although primary care research networks could be used as the framework for such an initiative. This system could also be used as the basis of a harmonised surveillance system that would allow a rapid evaluation of STI epidemiology and sexual health within and between countries.

CONCLUSIONS

The close relation between surveillance intelligence, disease prevention, and public health policy was highlighted by the Department of Health’s “Getting ahead of the curve.” At present, STI surveillance is limited by the relative lack of information from primary care: initiatives capable of identifying and evaluating emerging health care need to be developed. The method discussed here could allow an efficient assessment of STI prevalence, sexual health, and health service access behaviour. Such developments will be invaluable to future primary care STI surveillance, but no single method can effectively undertake surveillance. Other opportunities are evolving such as NHS electronic health records, a longitudinal record of patient health and health care, developments that have to be evaluated and assimilated into STI surveillance to complement existing knowledge. Creativity, innovation, and vision will be needed but it is only by doing this that we will be able to get ahead and stay ahead of the curve.