The UK HIV epidemic predominantly affects two distinct populations; men who have sex with men, the majority of whom are born in the United Kingdom, and heterosexual men and women largely born in countries within sub-Saharan Africa where the prevalence of HIV and tuberculosis is high [1,2]. As a consequence, a substantial number of heterosexual men and women living with HIV in the United Kingdom are likely to have been exposed to tuberculosis.
Despite recommendations for HIV patients in the United Kingdom to be screened for tuberculosis [3,4], the incidence of active tuberculosis among this group is not known. A cohort study of HIV clinic attendees in large urban areas of the United Kingdom (London, Brighton, Bristol and Edinburgh) reported an incidence rate of tuberculosis between 1996 and 2006 of 3.3 per 1000 person-years [95% confidence interval (CI) 2.94, 3.61] . Only new episodes of tuberculosis occurring 3 or more months after first attendance at the HIV clinic were considered in the analyses. The true incidence of tuberculosis among persons living with diagnosed HIV would be expected to be higher as it is likely that a large number of tuberculosis cases are diagnosed at the same time as, or within 3 months of HIV diagnosis.
In this article, we estimate annual tuberculosis incidence rates between 2002 and 2010 among heterosexual adults living with diagnosed HIV in England and Wales (E&W). We also examine the relationship between time of HIV and tuberculosis diagnosis and investigate risk factors for HIV and tuberculosis co-infection.
The national cross-sectional Survey of Prevalent HIV Infections Diagnosed (SOPHID) collects information on all adults (≥15 years of age) seen for HIV care in England, Wales and Northern Ireland in a calendar year . The survey includes adults previously seen for care as well as those newly diagnosed with HIV. The national HIV and AIDS New Diagnoses database, established in 1982, receives voluntary and confidential reports of new HIV and AIDS diagnoses in England, Wales and Northern Ireland . The national tuberculosis register collects information on all persons diagnosed with active tuberculosis in E&W .
Records of heterosexual adults seen for HIV care in E&W from 2002 to 2010 (referred to here as adults living with diagnosed HIV) were linked to records of adults diagnosed with active tuberculosis in E&W between 1999 and 2010 (referred to here as ‘adults diagnosed with tuberculosis’). Through data linkage between SOPHID and the national HIV and AIDS New Diagnoses database, diagnoses of tuberculosis were supplemented by reports of tuberculosis as an AIDS indicator disease.
All three national databases contain minimum patient identifiers, are under strict data protection, and are held at the Health Protection Agency (HPA) . Data were as reported to end June 2011.
Probabilistic matching, developed using Transact-SQL, was used for data linkage. Based on the agreement, partial agreement, or disagreement between a soundex code (a four character coding of the surname) , sex, country of birth, residential postcode, and date of birth, pairs of records between the national HIV and tuberculosis systems were determined as matches, nonmatches, or possible matches. Possible matches were subsequently reviewed by a member of the research team and assigned as matches or nonmatches.
A first diagnosis of active tuberculosis at the time of an HIV diagnosis (‘simultaneous diagnosis’) was defined as having a tuberculosis notification date up to 3 months before or after the HIV diagnosis date. The analyses were conducted in four stages.
In the first stage, descriptive analysis of all heterosexual adults living with diagnosed HIV with a first episode of tuberculosis prior to, at the time of, or subsequent to, their HIV diagnosis was conducted. In the second stage, adults with a first episode of active tuberculosis at the time of, or subsequent to, their HIV diagnosis were included in univariate and multivariate analysis investigating risk factors for co-infection. In the third stage, annual incidence rates of active tuberculosis were calculated based on the number of adults living with diagnosed HIV in a given year (denominator) and the number, in that same year, with a first episode of active tuberculosis at the time of, or subsequent to, their HIV diagnosis (numerator). In the fourth and final stage, annual incidence rates of active tuberculosis were calculated among adults newly diagnosed with HIV each year, using the number diagnosed with HIV in a given year (denominator) and the number of these same adults who had a first episode of active tuberculosis up to 1 year after their HIV diagnosis (numerator).
Of the 45 322 heterosexual adults living with diagnosed HIV in E&W between 2002 and 2010, sex and age were available for all, ethnicity for 99% (45 032), likely place of HIV acquisition for 94% (42 670), place of birth for 79% (35 685), and blood CD4 cell count at HIV diagnosis for 78% (35 248). A CD4 cell count at HIV diagnosis was defined as a count taken within 3 months of diagnosis. A late HIV diagnosis was defined by a CD4 cell count less than 350 cells/μl within 3 months of diagnosis. A person's HIV antiretroviral therapy (ART) status and CD4 cell count at the time of tuberculosis diagnosis were as reported up to 3 months prior to their diagnosis of tuberculosis.
Continuous variables were compared using Wilcoxon–Mann–Whitney test. Rates were compared using Pearson's chi-square test. Cochran-Armitage chi-square trend test was used for time-trend analyses. Proportions were calculated on the basis of available data. Variables found to be significantly associated with HIV tuberculosis co-infection in univariate analysis were included in a multivariate logistic regression model. Testing values and all confidence intervals (CIs) are at the 99% significant level. STATA 12.0 (Stata Corp., College Station, Texas, USA) was used for analyses.
Of the 45 322 heterosexual adults living with diagnosed HIV in E&W between 2002 and 2010, 4266 (9.4%) had a diagnosis of active tuberculosis reported [3780 (89%) cases were reported to the national tuberculosis register and an additional 486 (11%) were reported to the national HIV and AIDS New Diagnoses database as an AIDS indicator disease]. Among these adults, 54% (2308/4266) had a simultaneous diagnosis of HIV and tuberculosis, 38% (1623) were diagnosed with tuberculosis more than 3 months after their HIV diagnosis, and 8% (335) were diagnosed with tuberculosis more than 3 months before their HIV diagnosis (Fig. 1). The percentage of heterosexual adults with a diagnosis of HIV and tuberculosis who were simultaneously diagnosed with both infections was significantly higher in 2010 than in 2002 (76%; 116/153 versus 52%; 264/509; P < 0.01).
Median CD4 cell count at HIV diagnosis was significantly lower among adults simultaneously diagnosed with both infections (90 cells/μl) compared with those diagnosed with tuberculosis prior to (162 cells/μl) or after their HIV diagnosis (220 cells/μl) (both P < 0.01). Of adults diagnosed simultaneously with both infections, 92% (1882/2042) were diagnosed late with HIV (CD4 cell count < 350 cells/μl), and 53% (1087) had a blood CD4 cell count less than 100 cells/μl.
Of the 3931 heterosexual adults living with diagnosed HIV co-infected with tuberculosis (2308 with simultaneous diagnoses and 1623 diagnosed with tuberculosis more than 3 months after their HIV diagnosis), 84% (3307/3918) were black African, 96% (3163/3310) were born abroad, and 95% (3604/3785) probably acquired HIV infection prior to arrival in the United Kingdom. The median CD4 cell count at the time of tuberculosis diagnosis (available for 3016 co-infected adults) was 137 cells/μl for all years, increasing from 130 cells/μl in 2002 to 162 cells/μl in 2010 (P < 0.01). Among adults diagnosed simultaneously with HIV and tuberculosis, 0.8% (19/2257) were known to be prescribed ART at the time of their tuberculosis diagnosis. This rose to 42% (556/1344) among adults diagnosed with tuberculosis more than 3 months after their HIV diagnosis. In a multivariate analysis, compared with adults without tuberculosis, co-infected adults were more likely to be male, black African or Indian, Pakistani and Bangladeshi ethnicity, diagnosed late with HIV, or to have acquired HIV abroad (Table 1).
Annual tuberculosis incidence rates among heterosexual adults living with diagnosed HIV
Between 2002 and 2010, the average annual incidence rate of tuberculosis among heterosexual adults living with diagnosed HIV was 19 per 1000, falling from 30.2 per 1000 in 2002 to 8.8 per 1000 in 2010 (P < 0.01) (Table 2). The tuberculosis incidence rate among heterosexual adults who were newly diagnosed with HIV within a given calendar year was elevated, but fell from 61.5 per 1000 in 2002 to 40.4 per 1000 in 2010 (P < 0.01). The overall decline in annual tuberculosis incidence rates over the study period was observed across most subgroups, with the largest decline seen among adults with a CD4 cell count less than 100 cells/μl at HIV diagnosis (Table 2).
The highest tuberculosis incidence rates in both 2002 and 2010 were among adults not on ART, those with a CD4 cell count less than 200 cells/μl at the time of HIV diagnosis, and those newly diagnosed with HIV (Table 2). Over the study period, the overall tuberculosis incidence rate among adults not on ART (56 per 1000) was over fourteen times higher than among those in receipt of ART (3.9 per 1000) (P < 0.01). Considering only diagnoses of tuberculosis that occurred 6 or more months after an HIV diagnosis (thus allowing time for ART to be prescribed as indicated), the overall tuberculosis incidence rate remained almost four times higher among those not on ART than among those in receipt of ART (14 versus 3.6 per 1000) (P < 0.01).
Annual tuberculosis incidence rates among all heterosexual adults not on ART differed significantly by CD4 cell count strata at HIV diagnosis (all P < 0.01) (Fig. 2a). In 2010, the tuberculosis incidence rate among adults not on ART and diagnosed with a CD4 cell count less than 100 cells/μl was 309 per 1000 adults, compared with 10.1 per 1000 among those with a count at least 350 cells/μl (P < 0.01). Between 2002 and 2010, there was a marked decline in the tuberculosis incidence rate among adults not on ART with a CD4 cell count less than 100 cells/μl (from 541 in 2002 to 309 in 2010; P < 0.01). In all years, tuberculosis incidence rates in heterosexual adults in receipt of ART did not differ significantly by CD4 cell count at HIV diagnosis (2010: P = 0.23) (Fig. 2b). The tuberculosis incidence rate among adults not on ART was higher than among those in receipt of ART for all CD4 categories in all years (all P < 0.01) (Fig. 2a and b).
Almost one in 10 heterosexual adults living with diagnosed HIV in England and Wales between 2002 and 2010 were also diagnosed with active tuberculosis, half of whom were diagnosed at the same time as, or close to their HIV diagnosis. The majority of co-infected heterosexual adults were of black African ethnicity, born abroad and had advanced HIV infection at the time of their HIV and/or tuberculosis diagnosis. The annual tuberculosis incidence rate among our study population fell from 30 per 1000 in 2002 to 8.8 per 1000 in 2010. This decline is due to a modest fall in the absolute number of incident tuberculosis diagnoses (from 384 in 2002 to 288 in 2010) and a large increase in the total number of heterosexual adults living with diagnosed HIV (from 12 720 in 2002 to 32 621 in 2010).
There are two likely explanations why susceptibility to tuberculosis infection among our study population has declined. Firstly, the fall in tuberculosis diagnoses has coincided with a decline in the overall number of heterosexuals in the United Kingdom acquiring their HIV infection in sub-Saharan Africa, where there is a high prevalence of tuberculosis. Secondly, in recent years, the uptake of ART among persons with a CD4 cell count less than 350 cells/μl has increased in line with national HIV treatment guidelines .
The tuberculosis incidence rate of 8.8 per 1000 heterosexual adults living with diagnosed HIV in 2010 remains substantially higher than the rate of 0.14 per 1000 in the general UK population . Elevated incidence rates of tuberculosis among heterosexual adults living with diagnosed HIV in 2010 were observed among the following groups: aged 35 year and above at HIV diagnosis; male; born abroad; infected abroad; not on ART (including those diagnosed with tuberculosis 6 or more months after HIV); with a CD4 cell count <200 cells/μl at HIV diagnosis. The tuberculosis incidence rate reported among black Africans living with diagnosed HIV in 2010 (10.9 per 1000) remains four times higher than the UK rate among all black Africans born abroad (2.7 per 1000), and 25 times that among black Africans born in the United Kingdom (0.43 per 1000) .
The average annual tuberculosis incidence rate in our study population (19 per 1000) was also considerably higher than the rate reported in a multicentre study of persons accessing HIV care in four large UK cities between 1996 and 2006 (3 per 1000) . The authors of this study acknowledge that their rate was likely to be a minimum estimate as tuberculosis episodes within 3 months of first HIV clinic attendance were excluded from the analyses .
Just over half of the study population who were diagnosed with tuberculosis were diagnosed with HIV and tuberculosis simultaneously, and the overwhelming majority of tuberculosis diagnoses in 2010 were among adults with evidence of severe immunosuppression at HIV diagnosis. It is likely that symptoms of tuberculosis were the trigger for both the tuberculosis and HIV diagnoses although it is not possible to confirm this in our study due to imprecise HIV diagnosis and tuberculosis notification dates.
We report a substantial number of heterosexual adults who were diagnosed with tuberculosis more than 6 months after their HIV diagnosis. Active tuberculosis may have been prevented in some cases had testing and treatment for latent tuberculosis been a routine part of initial assessment at the time of HIV diagnosis. Although current guidelines recommend screening HIV patients for latent tuberculosis following a risk assessment , there is little evidence that this is routine among persons born in high endemic countries.
We also report a small number of people who were diagnosed with tuberculosis 6 or more months before being diagnosed with HIV. Since HIV often remains undiagnosed for several years , it is likely that some of these adults were living with HIV when diagnosed with tuberculosis. Missed opportunities for earlier HIV testing and more timely initiation of treatment are of obvious concern given that the majority of heterosexual adults are diagnosed with HIV at a point beyond which treatment should be initiated .
In the United Kingdom, it is recommended that all tuberculosis patients be offered an HIV test , and that people identified for a BCG vaccination considered to be at increased risk of HIV are offered an HIV test before vaccination . A study of tuberculosis patients in Greater London in 2003 and 2004 highlighted missed opportunities for diagnosing HIV earlier , and a subsequent audit of UK HIV and tuberculosis service providers found many services were still not routinely testing tuberculosis patients for HIV . Across Europe, the level of routine testing of tuberculosis patients for HIV remains low .
Our findings reinforce the need for stricter implementation of existing testing guidelines in the HIV and tuberculosis clinic setting [3,4], and support the recommendation for primary care practitioners to consider the risk of both diseases in migrant patients from high prevalence countries . Our findings also highlight the benefits of ART in reducing rates of tuberculosis among heterosexual adults living with diagnosed HIV.
A possible limitation to our analysis is that annual tuberculosis incidence rates have been underestimated due to problems with data linkage between systems. To increase the likelihood of linking records relating to the same adult we applied probabilistic matching and all possible matches were subsequently reviewed by a member of the research team. It is also possible that some adults were wrongly assigned with respect to ART use due to either only recently having started treatment or previously having received treatment. It is unlikely that this significantly biased our results.
In conclusion, we have reported a significant decline between 2002 and 2010 in the annual tuberculosis incidence rate among heterosexual adults living with diagnosed HIV in E&W. However, the tuberculosis incidence rate in 2010 remains higher than previously reported among HIV clinic attendees, and greatly exceeds that in the general population.
To promote optimal care for patients with tuberculosis, and to ensure HIV ART is initiated appropriately, it is essential that the universal offer of an HIV test to all persons diagnosed with tuberculosis continues to be extended in all relevant healthcare settings. To reduce the incidence of active tuberculosis among people diagnosed with HIV, ART must be initiated on time and the recommendation for HIV-infected patients to be screened and possibly treated for latent tuberculosis infection implemented. Integrated tuberculosis and HIV services are key to ensuring that this is a routine and successful part of UK healthcare.
We would like to thank the SOPHID advisory group and all the SOPHID facilitators across England, Wales and Northern Ireland, for their continuing help and support with the survey. National HIV and tuberculosis surveillance are funded by the Department of Health. The views expressed in the publication are those of the authors and not necessarily those of the Department of Health.
The HIV and AIDS Reporting System and Enhanced Tuberculosis Surveillance are funded by the Department of Health.
No article resembling the enclosed article has been or will be published elsewhere.
Conflicts of interest
The authors have no conflicts of interest to declare.
2. Health Protection Agency. HIV in the United Kingdom: 2011 report.
London: Health Protection Services, Colindale. November 2011.
3. Pozniak AL, Coyne KM, Miller RF, Lipman MCI, Freedman AR, Ormerod LP, et al. British HIV Association guidelines for the treatment of TB/HIV coinfection 2011
. HIV Med
5. Grant AD, Bansi L, Ainsworth J, Anderson J, Delpech V, Easterbrook P, et al. Tuberculosis among people with HIV infection in the United Kingdom: opportunities for prevention?
9. Mortimer JY, Salathiel J. ‘Soundex’ codes of surnames provide confidentiality and accuracy in a national HIV database
. Commun Dis Rep CDR Rev
10. Brown AE, Kall MM, Smith RD, Yin Z, Hunter A, Delpech VC. Auditing, National HIV Guidelines and Policies: the United Kingdom CD4 surveillance scheme
. Open AIDS J
2012; 6 (Suppl 1 M12):149–155.
11. Health Protection Agency. Tuberculosis in the UK: 2012 report.
London: Health Protection Agency, July 2012.
13. Rodger AJ, Story A, Fox Z, Hayward A. HIV prevalence and testing practices among tuberculosis cases in London: a missed opportunity for HIV diagnosis?Thorax
14. Backx M, Curtis H, Freedman A, Johnson M. British HIV Association; BHIVA Clinical Audit Sub-Committee. British HIV Association national audit on the management of patients co-infected with tuberculosis and HIV
. Clin Med
15. Kruijshaar ME, Pimpin L, Abubakar I, Rice B, Delpech V, Drumright LN, et al. The burden of TB-HIV in the EU: how much do we know? A survey of surveillance practices and results
. Eur Respir J