Gonorrhoea in inner London: results of a cross sectional studyBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7096.1719 (Published 14 June 1997) Cite this as: BMJ 1997;314:1719
- Nicola Low, health services research training fellowa,
- Gavin Daker-White, researcherb,
- David Barlow, consultant physicianb,
- Anton L Pozniak, senior lecturer in genitourinary medicinea
- a Academic Unit, Department of Genitourinary Medicine, King's College School of Medicine and Dentistry, The Caldecot Centre, London SE5 9RS
- b Department of Genitourinary Medicine, Guy's and St Thomas's NHS Trust, London SE1 7EH
- Correspondence to: Dr Low
- Accepted 19 May 1997
Objectives: To estimate population based incidence rates of gonorrhoea in an inner London area and examine relations with age, ethnic group, and socioeconomic deprivation.
Design: Cross sectional study.
Setting: 11 departments of genitourinary medicine in south and central London.
Subjects: 1978 first episodes of gonorrhoea diagnosed in 1994 and 1995 in residents of 73 electoral wards in the boroughs of Lambeth, Southwark, and Lewisham who attended any of the departments of genitourinary medicine.
Main outcome measures: Yearly age, sex, and ethnic group specific rates of gonorrhoea per 100 000 population aged 15-59 years; rate ratios for the effects of age and ethnic group on gonorrhoea rates in women and men before and after adjustment for confounding factors.
Results: Overall incidence rates of gonorrhoea in residents of Lambeth, Southwark, and Lewisham were 138.3 cases yearly per 100 000 women and 291.9 cases yearly per 100 000 men aged 15-59 years. At all ages gonorrhoea rates were higher in non-white minority ethnic groups. Rate ratios for the effect of age adjusted for ethnic group and underprivilege were 15.2 (95% confidence interval 11.6 to 19.7) for women and 2.0 (1.7 to 2.5) for men aged 15-19 years compared with those over 30. After deprivation score and age were taken into account, women from black minority groups were 10.5 (8.6 to 12.8) times as likely and men 11.0 (9.7 to 12.6) times as likely as white people to experience gonorrhoea.
Conclusions: Gonorrhoea rates in Lambeth, Southwark, and Lewisham in 1994-5 were six to seven times higher than for England and Wales one year earlier. The presentation of national trends thus hides the disproportionate contribution of ongoing endemic transmission in the study area. Teenage women and young adult men, particularly those from black minority ethnic groups, are the most heavily affected, even when socioeconomic underprivilege is taken into account. There is urgent need for resources for culturally appropriate research and effective intervention to prevent gonococcal infections and their long term sequelae in this population.
The sexual health of the nation is a key topic prioritised for improvement by the British government.1 The incidence of gonorrhoea was chosen as a proxy indicator for trends in condom use and new cases of HIV infection.2 Gonorrhoea merits prevention efforts in its own right because its sequelae include pelvic inflammatory disease resulting in tubal infertility and ectopic pregnancy, fetal prematurity, and ophthalmia neonatorum.3 The target reduction of 20% in the national rate of gonorrhoea from the 1990 level was reached by 1992.4 This has been welcomed as an achievement of the NHS,5 6 though recent statistics show a 5% increase in the total number of episodes of gonococcal infection treated at departments of genitourinary medicine in England between 1994 and 1995.7
The focus on trends in national rates of infection for the whole adult population masks known variations in the incidence of gonorrhoea by geography, age, and sex,4 and it has been suggested that local targets defined by health authorities would be more useful.8 Departments of genitourinary medicine, however, are open access clinics treating patients regardless of where they live. Hence the incidence of sexually transmitted infections in the population of a health authority cannot be calculated directly from the number of episodes recorded by local clinics. This protects individual patient confidentiality but limits the availability of basic epidemiological data about the scale of the problem, those who are at most risk, and where interventions are required.
In the United States, where some sexually transmitted infections are statutorily notifiable9 and data are reported by age, sex, and ethnic group, there is at least a 10-fold excess of cases of gonorrhoea in African Americans compared with white, Hispanic, and all other ethnic groups. In urban areas the disparity is even greater.10 11 Though socioeconomic deprivation is known to favour the spread of sexually transmitted infections12 and disproportionate numbers of African Americans live in poverty,13 the inequality in gonorrhoea rates persists after adjustment for socioeconomic confounding.10
In the inner London boroughs of Lambeth, Southwark, and Lewisham high numbers of gonococcal infections have been reported in heterosexual men of black African and black Caribbean ethnic groups.14 15 We studied the epidemiology of gonorrhoea in Lambeth, Southwark, and Lewisham by using population based rates in order to aid the setting of appropriate local targets for interventions aimed at improving sexual health.
Ascertainment of cases
Sixteen departments of genitourinary medicine in south and central London were contacted (fig 1). A deadline for data collection was set three months after the end of the study period. By 31 March 1996 permission had been obtained and data collected from eight out of 10 clinics in south London and three out of six in central London.
Episodes of uncomplicated and complicated gonococcal infection diagnosed from 1 January 1994 to 31 December 1995 were identified from computerised or manual records by the diagnostic coding system used to report cases of sexually transmitted infections to the Department of Health (form KC60). KC60 codes B1, B1.1-3, B1.4a-c, and B5 were included, taking into account a change in the diagnostic categories introduced on 1 January 1995. Repeat episodes were defined as a new diagnosis made four weeks or more after a previous diagnosis in records with the same sex, date of birth, and postcode. When repeat episodes were detected only the first diagnosis in each 12 month calendar period was included in the analysis.
First episodes in adults aged 15-59 years were included in the final dataset only if the postcode (checked against a printed list provided by Lambeth, Southwark, and Lewisham Health Authority) was within the administrative boundaries of the boroughs. Records with no address were excluded. Incomplete postcodes were completed when possible from available information by using the Post Office directory.16 Postcode, date of episode, sex, date of birth, ethnic group, country of birth or nationality, and whether episodes in men were acquired homosexually were entered as individual records in spreadsheet files protected by a password. To anonymise the records and ensure that deductive disclosure of individual identity was not possible date of birth and postcode were deleted from each record immediately after the exclusion of duplicate and repeat episodes and computerised allocation to an electoral ward.
Categorisation of cases by ethnic group into 1991 census categories17 was attempted, but differences between clinics in classification systems recording race, ethnic group, nationality, or country of birth meant that only three broad groupings could be used. “White” included people recorded as white or Caucasian. Fourteen people with European, North American, or Australian country of birth were assumed to be white; “black” comprised those recorded as black, black African, black Caribbean, or black “other” and included three people with African or Caribbean country of birth or nationality; “other” comprised people from all other ethnic groups and nationalities. “UK” nationality or country of birth and those with no information were coded as unknown ethnic group.
Local base statistics from the 1991 census were used (table L06 (Crown copyright)) to estimate the population at risk in the 73 wards of Lambeth, Southwark, and Lewisham. The ward population was stratified by sex, age (15-19, 20-24, 25-29, 30-59 years), and ethnic group (white, black (black African, black Caribbean, black “other”), other (all other ethnic groups)). Undercoverage of the population by the census in inner London was corrected by using published adjustment factors specific for age and sex (appendix).18
Yearly incidence rates of gonorrhoea specific for age, sex, and ethnic group in residents of Lambeth, Southwark, and Lewisham per 100 000 population were calculated with 95% confidence intervals. Poisson regression models19 were used to examine the relation between gonorrhoea, ethnic group, age, and sex before and after adjustment for confounding by socioeconomic deprivation. Individual measures of socioeconomic status were not available, so a ward level measure of underprivilege used by Lambeth, Southwark, and Lewisham Health Authority for the allocation of deprivation payments to general practitioners was applied to each case resident in the ward. The Jarman score is a weighted average of eight census derived variables which correlates with self reported morbidity20 and all cause mortality.21 A three level category corresponding to the standard cut off points for deprivation payments was used as a proxy for socioeconomic underprivilege–namely, 0-29.99 (no deprivation), 30-39.99 (low deprivation), and 40 or more (medium or high deprivation).
Evidence for effect modification by sex was examined by using a likelihood ratio χ2 test. Rate ratios with 95% confidence intervals for the effect of age and ethnic group on gonorrhoea rates in men and women were calculated before and after adjustment for each other and for socioeconomic deprivation.
All statistical analyses were conducted with stata (version 4.0, Austin, Texas).
From 1 January 1994 to 31 December 1995, 2256 episodes of gonorrhoea were enumerated from 11 departments of genitourinary medicine. Data from 76 episodes with south London postcodes that could not be completed were excluded, leaving 2180 episodes in 1932 adults aged 15-59 years who were resident in Lambeth, Southwark, and Lewisham at the time of diagnosis. Of the 248 repeat episodes, two thirds occurred in men and three quarters in people from a black ethnic group. Forty six occurred in people who also had a recorded episode in the previous calendar year. This analysis therefore included 1978 first episodes of gonorrhoea in two calendar years. Information about heterosexual or homosexual acquisition of infection was absent from 46% of male records, so all cases in men were considered together. Table 1) shows the distribution of cases in women and men.
Tables 2 and 3 give the age, sex, and ethnic specific rates of gonorrhoea in Lambeth, Southwark, and Lewisham. Overall rates were 138.3 and 291.9 cases yearly per 100 000 population aged 15-59 years for women and men respectively. For women in white and black ethnic groups the highest rates of infection were seen in teenagers (table 2). The highest age specific rates in men were for black men aged 20-24 years and white men aged 25-29 years (table 3). Rates for men and women from other ethnic groups were based on small numbers and not considered in detail. Rates were higher in men than in women at all ages except 15-19 years. In all age groups gonococcal infection was less common in white men and women than in men and women from black minority ethnic groups.
Age, ethnic grouping, and deprivation
As suggested by the stratified gonorrhoea rates there was strong statistical evidence for an interaction between sex and ethnic group (P=0.0006) and between sex and age (P<0.0001) (table 4). Regression analyses are therefore presented separately for women and men.
Women–Rates of gonorrhoea were strongly associated with age and ethnic group. Young age was the strongest risk factor, teenagers being 15 times more likely than women over 30 to experience gonorrhoea after adjustment for confounders. Adjusting for age and deprivation score had only a small effect on the effect of ethnic group. Women from black minority ethnic groups had around 10 times the rate of gonococcal infection seen in white women.
Men–As in women, men over the age of 30 had the lowest rate of gonorrhoea but the adjusted rate ratios for younger men did not show a strong gradient, ranging from 2.0 to 2.6. The effect of ethnic group in men was similar to that for women. After adjustment for age and Jarman score men from black minority ethnic groups were 11 times more likely than white men to experience gonorrhoea.
For both men and women the effect of living in a ward with any deprivation compared with wards with no deprivation (defined by Jarman score) was modest and unaffected by further adjustment for age and ethnic group.
These analyses underestimate the frequency of gonorrhoea in the study population. Firstly, patients with no address and those attending non-participating clinics could not be included. Secondly, only genitourinary clinics complete form KC60. Microbiology records for 1994-5 from a local hospital with no genitourinary department identified only 45 episodes of gonorrhoea from all other clinical settings (G Rao, personal communication). Thirdly, repeat episodes were excluded from multivariate analyses because these observations are not independent of one another. For consistency they were also excluded from descriptive analyses, and crude rates represent the number of people affected per 100 000 yearly. Repeat episodes occurred more often in people from black ethnic groups,14 15 so exclusion may reduce the magnitude of effect of ethnic group but does not alter the conclusions.
The Jarman score was a convenient proxy for underprivilege at ward level. The proportion of households headed by a person born in the new Commonwealth or Pakistan is one component of the score, even if with the lowest weighting.21 Thus the ecological association between gonorrhoea and the Jarman score could simply reflect the ethnic composition of ward populations. Analyses of the relation between gonorrhoea rates and single indicators of socioeconomic status and other validated deprivation scores21 are now planned.
The hidden epidemic
Geography–In 1993 national reported rates of 22 and 43 gonorrhoea cases per 100 000 women and men aged 15-59 years respectively were below the Health of the Nation target.4 Among residents of Lambeth, Southwark, and Lewisham in 1994-5 the corresponding rates were 138 and 292 cases per 100 000 yearly. Roughly 11% of all episodes of gonorrhoea reported from departments of genitourinary medicine in England in 1994-57 occurred in residents of Lambeth, Southwark, and Lewisham, who accounted for 1.4% of the population aged 15-59 years (1991 census data). High rates have also been recorded in the West Midlands22 and Leeds (C J N Lacey, D Bensley, D Merrick, I Fairley, paper presented at the 11th meeting of the International Society for Sexually Transmitted Diseases Research, New Orleans, August 1995). Presentation of national rates of gonorrhoea thus hides the disproportionate contributions of a few geographic foci. These findings should not be extrapolated to other, particularly viral sexually transmitted infections, which are more uniformly distributed.23
Sex–Twice as many cases of gonorrhoea in men as in women were identified. Including infections acquired through sex between men does not account for this disparity. About a quarter of episodes for which information was available were acquired homosexually. When this proportion of all male cases is subtracted the male to female ratio is still 1.5:1, indicating that women are remaining untreated in the community. The highest rates of diagnosed gonorrhoea–that is, in teenage women from black ethnic groups–would be two to three times higher if the denominator was restricted to women who were sexually experienced.24 These young women are also at risk of pelvic inflammatory disease, tubal infertility, and ectopic pregnancy.
Ethnic grouping–Information about ethnic grouping was too inconsistent to allow distinctions to be made within broad groups. The results should not therefore be applied to individual minority populations but they highlight inequalities in sexual health which require investigation. Increased monitoring of ethnicity in health service settings using census categories17 should improve the quality of these data. Knowledge of cultural, social, and economic influences on sexual health is scarce in the United Kingdom. Pejorative studies of gonorrhoea in immigrants from the Commonwealth in the 1950s and 1960s25 26 27 have probably inhibited useful research in more recent years.
In the United States, Healthy People 2000–the equivalent of Health of the Nation–has set ethnic specific targets9 and directed resources towards reducing the disparity in rates of sexually transmitted infections. As a start in the United Kingdom accurate information should be disseminated to the community, health service providers, purchasers, and policy makers. The detection of inequalities in rates of gonorrhoea experienced by members of minority ethnic groups should be used positively to argue for resources to investigate the reasons for these disparities and to intervene appropriately and effectively.
An earlier analysis of these data was presented at the spring meeting of the Medical Society for the Study of Venereal Diseases, Edinburgh, 9-11 May 1996. We thank Dr Jonathan Sterne for statistical advice, Mr Dal Sandhu for help with preparation of the dataset, and Dr Kevin de Cock for comments on previous versions of this paper. Mr B Cappi, Mr R Halai, Mr M Moreno, Ms C Shergold, and Ms A Smith helped in the retrieval of data. The following consultants allowed us to use information from their clinics: Dr S Barton, Dr F Davidson, Dr B A Evans, Dr P French, Dr P E Hay, Dr R Lau, Dr A Lawrence, Dr P Lister, Dr D Mercey, Dr S Mitchell, Dr J Russell.
Funding: NL was funded by a South Thames NHS Executive health services research training fellowship; GDW was part funded by a grant from Lambeth, Southwark, and Lewisham Health Authority.
Conflict of interest: None.