Article Text

Original article
Trends in the incidence of HIV in Scotland, 1988–2009
  1. Scott A McDonald1,2,
  2. Sharon J Hutchinson1,2,
  3. Lesley A Wallace1,
  4. Sheila O Cameron3,
  5. Kate Templeton4,
  6. Paul McIntyre5,
  7. Pamela Molyneaux6,
  8. Amanda Weir1,
  9. Glenn Codere1,
  10. David J Goldberg1
  1. 1Health Protection Scotland, Glasgow, UK
  2. 2Department of Mathematics and Statistics, University of Strathclyde, Glasgow, UK
  3. 3West of Scotland Specialist Virology Centre, Gartnavel General Hospital, Glasgow, UK
  4. 4East of Scotland Specialist Virology Centre, Royal Infirmary of Edinburgh, Edinburgh, UK
  5. 5Department of Medical Microbiology, Ninewells Hospital and Medical School, Dundee, UK
  6. 6Department of Medical Microbiology, University Medical School, Aberdeen, UK
  1. Correspondence to Dr Scott A McDonald, Health Protection Scotland, Meridian Court, 5 Cadogan Street, Glasgow G2 6QE, UK; smcdonald4{at}nhs.net

Abstract

Objectives To estimate temporal trends in HIV incidence and prevalence in Scotland, according to three main risk groups for infection: men who have sex with men (MSM), injecting drug users (IDUs) and heterosexuals.

Methods The authors extracted data for all single- and multiple-tested individuals from the national HIV test database covering the period 1980–2009 and calculated the incidence of HIV infection in each risk group and estimated RRs by fitting Poisson regression models.

Results 620 of 59 807 individuals tested positive following an initial negative HIV test, generating an overall incidence rate of 3.7/1000 person-years (95% CI 3.4 to 4.0); 60%, 20% and 37% of the 620 were associated with the risk behaviour categories MSM, IDU and heterosexual, respectively. The incidence rate among MSM in Scotland remained relatively stable between the periods <1995 and 2005–2009 (overall: 15.3/1000 person-years, 95% CI 13.8 to 17.0), whereas the incidence among IDUs decreased between the periods <1995 and 2005–2009, from 5.1/1000 to 1.7/1000 person-years, and also decreased among heterosexuals, from 2.9/1000 to 1.4/1000 person-years.

Conclusions The reduction in the incidence rate among IDUs suggests that harm reduction measures initiated from the late 1980s were effective in reducing HIV transmission in this risk group; however, the absence of a reduction in HIV incidence rates among MSM is disappointing and highlights the need for renewed efforts in the prevention of HIV in this major risk group.

  • HIV
  • men who have sex with men
  • heterosexuals
  • injecting drug users
  • incidence
  • epidemiology (general)
  • hepatitis C
  • liver
  • epidemiology
  • hepatitis
  • virology clinical

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Infection with HIV is a significant public health problem in Scotland; as of 31 December 2009, 6247 HIV-infected individuals had been diagnosed.1 Onward transmission among men who have sex with men (MSM), who represent 37% of the total diagnosed to date, is currently of concern and so constitutes the focus of this paper, although transmission rates were historically high among injecting drug user (IDU) populations in Scotland, with prevalence as high as 51% among Edinburgh's IDUs.2 An increasing trend in the annual number of MSM diagnosed as having HIV was observed until 2007, peaking at 209 new diagnoses,1 which could reflect a number of factors including increases in HIV testing in genitourinary medicine (GUM) clinics or more worryingly, a rise in the incidence of new infection in this population.

Harm reduction interventions aimed at reducing the spread of HIV and other bloodborne viruses among IDUs—chiefly needle/syringe exchanges and methadone maintenance treatment—were introduced in Glasgow and other Scottish urban areas from 1987. Harm reduction services in Glasgow and elsewhere in Scotland were expanded during the 1990s, and a decrease in HIV prevalence among Glasgow's injectors from 4.8% in 1987 to 1.0% in 1992 has been reported,3 suggesting that these measures helped to reduce the spread of HIV in this population.4

Since the mid-1980s, education campaigns on safer sexual practices have been initiated within the MSM and heterosexual communities,5 and attempts to improve case identification via routine, opt-out HIV testing in GUM clinics, have been introduced in the UK as part of the strategy for reducing transmission among susceptible populations. Laboratory methods (serological testing algorithms such as STARHS6) for estimating HIV incidence have recently been implemented in several countries, including England and Wales7 and the USA.8 In Scotland, the uptake of voluntary confidential HIV testing among MSM increased from 47% to 59% between the periods 1997 and 2002.9 Determining the effectiveness of these initiatives to prevent the spread of HIV is of public health interest.

Using Scotland's national HIV test database, our goal was to estimate the temporal trends in HIV incidence within the three key risk groups: MSM, heterosexual and IDU. Such information is essential for evaluating the effectiveness of public health interventions and is valuable for guiding future policy regarding education, awareness and prevention initiatives.

Methods

Record linkage within a national HIV test database covering the period 1980–2009 was used to estimate trends in HIV incidence according to reported risk activities.

Data source and study population

Health Protection Scotland maintains a database of all HIV tests conducted within Scotland,4 excluding routine screening (eg, antenatal, renal, travel/insurance; the database was developed to assess prevalence among those individuals actively seeking or who had been offered, a diagnostic HIV test) and testing of previously known HIV positives and persons under 15 years of age. Data are provided by all National Health Service laboratories in Scotland that perform HIV testing. Individual records on this database contain the following non-named information: sex, date of birth, surname and forename initials, as well as data concerning risk activities, health board of residence (HB; National Health Service Scotland administrative area). Risk activity(s) was assessed at time of consultation for HIV testing by the healthcare professional with the patient; this was based on the major risk factor for possible exposure to the virus at that visit. We mapped test records to distinct individuals using a deterministic approach (ie, the procedure required a complete match on either the set of identifiers sex, date of birth and initials or the set of sex, date of birth and a hospital/clinic number, generated from GUM clinic/hospital referrals only). The HIV test database contained records for 523 251 HIV tests conducted between 1 January 1988 and 31 December 2009 (including the testing of named stored sera back to 1980, for which consent was given), of which 3.7% lacked sufficient identifiers for record-linkage. Sex, date of birth, hospital/clinic number and initials were missing in 1.1%, 0.8%, 12.8% and 61.4% of records, respectively. HB was updated at subsequent HIV tests if the individual was inferred to have moved.

The study population consisted of those individuals identified via linkage between records on the HIV test database with non-missing date of birth and with test dates within the period 1980–2009; N=412 994 unique individuals (table 1). An individual was classified into four possible risk categories (MSM, heterosexual, IDU and blood product/accidental exposure) if the risk was mentioned as a risk activity in either their first test record or in any of their subsequent test records; thus, an individual could belong to more than one risk group. Only MSM and IDU could be coded in a single test record; however, this rarely occurred and so the vast majority of individuals with multiple risk activities were defined as such across multiple records. Incidence rates were estimated from the subpopulation, who initially tested HIV negative and had at least one subsequent test within the same period (N=59 807, after additionally excluding 41 persons with missing sex information).

Table 1

Characteristics of the individuals on the HIV test database (initially tested in the period 1 January 1988 to 31 December 2009, including testing of stored sera back to 1980), according to initial test result and multiple-tested status

Data analysis

We computed incidence rates using the person-years method, by defining follow-up for each individual as the time from initial (HIV negative) test to either (1) the midpoint of the interval between the date of the last HIV-negative test and the date of the first HIV-positive test or (2) the date of the last HIV-negative test (for those without a subsequent HIV-positive test). CIs for rates were computed using the exact Poisson method. Incidence was calculated separately for subcohorts for whom the set of linked test records reported one of the three major risks (heterosexual, MSM or IDU).

We fitted Poisson regression models to midpoint-imputed follow-up times to estimate the RR of HIV infection in individuals who initially tested HIV negative, adjusting for sex, age group at initial test (<25, 25–34, 35–44, 45+ years), health board (Greater Glasgow, Lothian, Tayside, Grampian, all other HBs/HB not known), other major reported risk (ie, heterosexual, MSM, IDU, accidental exposure/blood product) and a time-dependent covariate for calendar period of test (<1995, 1995–1999, 2000–2004, 2005–2009) (ie, both person-years at risk and outcome of the second test were assigned to the period in which they fell). All statistical analyses were conducted using R V.2.7.2.

Annual positivity rates were derived using the number of individuals first tested in a given year as the denominator and the number of individuals first tested in that year with a positive first test result as the numerator.

Results

Characteristics of HIV-tested individuals

HIV testing in Scotland increased substantially over the last two decades, from 8938 in 1990 to 55 500 in 2009. The corresponding per capita testing rate increased from 1.7/1000 population in 1990 to 10.7/1000 in 2009 (for ages 25–44 years only: 4747 to 26 500; 3.2/1000 to 19.1/1000). The largest increase in the uptake of HIV testing since 1990 was observed for the heterosexual risk category (from 4759 in 1990 to 41 758 in 2009), followed by MSM risk (from 704 in 1990 to 4782 in 2009) (online supplement figure S1).

Of the 412 994 individuals identified through record linkage as having ever been tested for HIV in Scotland within the period 1980–2009, 1.2% (4943 of 412 994) tested positive at first test (table 1). The distribution of reported risk activities among all those tested indicated heterosexual as the most common risk (69.2% of individuals), followed by IDU (7.2%) and MSM (6.7%). Among those who reported risk as MSM, 4.6% (1272 of 27 705) tested positive at initial test compared with 1.9% (584 of 29 973) among those who reported IDU and 0.6% (1780 of 283 918) among those who reported heterosexual risk.

The positivity rate at initial test in persons with IDU risk declined over the study period (from 1.9% in 1990 to 0.3%–1.6% in 1998–2009) (figure 1). The positivity rate in MSM declined from 5.3% (yearly range: 4.2%–7.0%) in the 1990s to 4.2% (yearly range: 3.2%–5.0%) in 2000–2009, and among heterosexuals, the positivity rate was low and stable throughout the study period (1990–2009: 0.4%–1.0%). Persons who initially tested HIV positive were older on average (34.0 years, SD=11.4) than those initially testing negative (30.7 years, SD=12.0).

Figure 1

Number of HIV tests conducted by calendar year (1990–2009), shown separately for the three main reported risk groups.

The proportions of all testees who had a single HIV test record only (n=353 814) according to calendar period of test were 1.3% (<1990), 9.4% (1990–1994), 13.4% (1995–1999), 21.9%, (2000–2004) and 39.7% (2005–2009). The proportions of all once-tested individuals who had MSM, heterosexual, IDU and accidental exposure/blood product risk were 5.3%, 69.2%, 6% and 3.5%, respectively, and the median time between test date and 31 December 2009 was 5.5 years (IQR: 2.1–10.9 years).

Overall HIV incidence rate

Of the study population who initially tested negative and had at least one follow-up HIV test, 620 of 59 807 (1.04%) subsequently tested positive. The estimated HIV incidence rate was 3.7/1000 person-years (95% CI 3.4 to 4.0), based on a total of 168 746 person-years (mean of 2.8 years per individual) of follow-up time.

HIV incidence rate among MSM

The highest incidence rate of new HIV-positive cases was seen in MSM, with 15.3/1000 person-years (95% CI 13.8 to 17.0). Incidence was relatively stable between the periods <1995 and 2005–2009 (table 2). Statistically significant decreased RRs were associated with older age at first test (RR for 45+ years of 0.5, 95% CI 0.3 to 0.8, compared with <25 years). HIV incidence in MSM aged 35+ years slightly increased between the periods <1995 and 2005–2009, from 6.0/1000 to 15.2/1000 person-years, but for the younger age groups, it remained relatively stable (16.8 to 14.4/1000 person-years). However, including a term in the Poisson regression model for an age group (dichotomised as <35 and 35+ years) by calendar year period interaction did not yield a statistically significant improvement in model fit (p=0.33). There was also a reduced RR of testing HIV positive if either co-risk heterosexual or IDU was also present (RR=0.7, 95% CI 0.5 to 0.9 and RR=0.5, 95% CI 0.3 to 0.9, respectively).

Table 2

Incidence of HIV infection in individuals who initially tested HIV negative during the period 1988–2009 (including testing of stored sera back to 1980) and had at least one subsequent test in the same period (n=59 807)

HIV incidence rate among heterosexuals

Although of the three major risk groups investigated, the largest proportion of individuals reported heterosexual risk (66.3%, 39 671/59 807); the incidence rate of new positive cases in this group was the lowest, at 2.1/1000 person-years (95% CI 1.8 to 2.4). Incidence declined slightly between the periods <1995 (2.9/1000 person-years) and 2005–2009 (1.4/1000 person-years) (table 2). There were statistically significant increased RRs associated with older age at first test (25–34 years: RR=1.6, 95% CI 1.2 to 2.1; 35–44 years: RR=2.6, 95% CI 1.8 to 3.7; and 45+ years: RR=1.9, 95% CI 1.1 to 3.3) compared with <25 years. Residing in Tayside HB was associated with a 1.6 times greater risk (95% CI 1.1 to 2.4) compared with the other HB/NK category. The risk of testing HIV positive was 6.1 (95% CI 4.3 to 8.5) and 1.5 (95% CI 1.1 to 2.0) times higher if MSM and IDU co-risk were reported, respectively.

HIV incidence rate among IDUs

The incidence rate of new HIV-positive cases among those reporting IDU was 2.7/1000 person-years (95% CI 2.2 to 3.2). Incidence decreased between the periods <1995 and 2005–2009 (table 2C), from 5.1/1000 (95% CI 3.7 to 6.9) to 1.7/1000 (95% CI 0.9 to 2.7) person-years. There was a statistically significant increased RR associated with having MSM as co-risk (RR=3.7, 95% CI 2.0 to 6.8) and a decreased risk for individuals residing in Grampian HB (RR=0.4, 95% CI 0.2 to 0.9) compared with other HB/not known.

Discussion

Our principal finding is that the incidence of HIV in MSM has remained relatively constant between the periods <1995 and 2005–2009, despite education and prevention initiatives. Our discussion primarily focuses on the MSM risk group, but we note that incidence rates for individuals who reported heterosexual risk (2.1/1000 overall) are the lowest of the three main risk groups and have slightly decreased since <1995. For those reporting IDU, incidence rates have decreased even more so, from 5.1 to 1.7/1000 person-years in 2005–2009, which is consistent with the development and expansion of harm reduction measures initiated since the earlier period. In the most recent quinquennium of our study period (2005–2009), we observed a 10-fold difference in HIV incidence between MSM (15/1000 person-years) and either IDUs (1.7/1000 person-years) or heterosexuals (1.4/1000 person-years); this relatively high stable incidence in MSM suggests that renewed policy efforts for preventing HIV transmission within MSM are needed.

The current study represents the first investigation into long-term trends regarding HIV incidence rates among MSM, heterosexual and IDU risk groups in Scotland. Our retrospective cohort design had considerable power (assuming the data linkage achieved high precision), as we were able to follow-up a large sample (n=8667) over a span of up to 21 years, which equated to 24 455 person-years of observation. Our approach is original in its use of a extensive national HIV test database and can be seen as complementary to other recently developed approaches for inferring trends in the incidence of HIV, such as STARHS (seriological testing algorithm for recent HIV seroconversion),6 with advantages of low cost and the ability to generate historical estimates (ie, back to the early 1990s) without reliance on the availability of stored samples. A continuing high incidence of HIV in MSM relative to other risk groups suggests ongoing risky sexual practices in this subpopulation, which is broadly consistent with reports from other countries of continuing or even increasing high-risk behaviour over comparable time periods.10 11 Both stable and increasing trends12 13 in HIV incidence among MSM have been reported over a comparable calendar year period in the USA and Canada, respectively. The period of analysis ended at 2000 in the Canadian study, however; therefore, it is not known whether the increasing trend was maintained over the next decade.

Due to its retrospective design, our database study has several limitations. Because the definition of our study population required at least one test subsequent to initial HIV test, estimated incidence rates may not be representative of testees who have undergone a single HIV test only if referral for this subsequent test was prompted by awareness or report of recent risk-taking behaviour. Observed incidence rates therefore may be higher than for a once-tested population, and the incidence of HIV among the repeat-tested population may differ from the never-tested population. If the incidence rate in the underlying MSM population had actually increased over time but the increase in HIV testing had resulted in more low-risk individuals contributing to the HIV test database over the same period, then these data could potentially generate a stable incidence trend. A stable trend would also be obtained if the incidence rate had actually decreased over time but the increase in HIV testing had resulted in more high-risk individuals contributing to the HIV test database. Consistent with the latter point, the proportion of HIV tests referred from GUM clinics increased over time, from 2.5% in 1990–1994 to 8.6% in 2005–2009. Unmeasured changes in demographics (eg, due to immigration/emigration) or risk behaviour of the at-risk population over time cannot be ruled out as a (partial) explanation of the stable incidence.

Our database linkage indicated testing as a whole to have increased over the study period (online supplement figure S1), which is corroborated by research showing increased HIV testing in MSM GUM clinic attendees in Scotland, with a 60% increase in the number of people tested (451–1157) observed between the periods 1997 and 2004.14 However, the likelihood of subsequent testing among MSM increased only moderately across time, from 19.1% in 1990–1994 to 24.1% in 2000–2004 (online supplement figure S2), which suggests that the stable incidence in MSM is not due to a rise in test uptake.

Incidence rates in the most recent period (ie, 2008–2009) may be artificially low due to data being available until 2009 only—persons initially tested in this period may not yet have had the opportunity for a second test. Given that the median inter-test interval was 1.7 years, it is plausible that roughly half of those initially tested in 2008 will not have been subsequently tested by the end of 2009.

We estimated time of seroconversion as the midpoint between the date of the last negative test and first positive test. The most conservative approach would have been to assume this occurred immediately following the date of last negative test. Results of sensitivity analyses (not reported) implementing this alternative were very similar to those presented.

Finally, record linkage using limited identifying information, and with a database constructed over a long timespan (recording identifying information that can change over time, such as names), means that the correspondence between HIV test records and individuals will be subject to error; if multiple test records for a particular testee failed to be identified as corresponding to the same person, then the existence of a subsequent test may not be recognised. Due to the strict deterministic linkage criteria, we are confident that high specificity was achieved; however, sensitivity could not be usefully estimated.

Our finding of relatively stable HIV incidence among MSM in Scotland, despite prevention-orientated sexual health interventions, is consistent with data from other sources. High-risk sexual behaviour among MSM in Scotland as measured by cross-sectional survey (eg, the Gay Men's Task Force, a peer-led sexual health prevention intervention) increased between the periods 1999 and 200215 and no difference was observed between the periods 2002 and 2008.16 In addition, the positivity rate among MSM in our study (4.6% among those tested during 1988–2009) was very similar to prevalence reported in a community sample of 1350 gay men tested in Edinburgh and Glasgow during 2005, for whom HIV prevalence was 4.4% (95% CI 3.5% to 5.7%).17

Our results are consistent with continuing high-risk sexual behaviour among MSM and with findings of stable recent trends in HIV incidence in the UK. In England, a decrease observed in the modelled annual numbers of new HIV infections and diagnoses of other STIs from the time (1983–1984) that HIV/AIDS awareness campaigns were initiated suggests a beneficial impact of these initiatives.18 However, this study showed that the annual numbers of HIV diagnoses among MSM did not decrease further between the periods 1987 and 1999 (the study endpoint), consistent with Scottish data.19

In conclusion, our study—achieving considerable statistical power through the use of linkage among records in a large national HIV test database—has highlighted the absence of a reduction in HIV incidence among MSM in Scotland over time, despite public health interventions and the availability of antiretroviral therapy in this period. Accordingly, other approaches need to be developed; the release of the Scottish government's Action Plan on HIV in 2009 signalled a renewed focus on HIV in Scotland, with plans to reduce transmission through identification and implementation of effective interventions for MSM.20 The marked drop in HIV incidence among IDUs from approximately 1995 is consistent with the establishment of effective harm reduction initiatives before that time. Diagnosis of HIV at an early stage provides the opportunity for commencement on antiretroviral therapy and thus suppression of viral load.

Key messages

  • The incidence of HIV in men who have sex with men in Scotland has remained stable over the past 15 years, despite awareness/education initiatives.

  • The decreasing incidence in the same period among injecting drug users is consistent with harm reduction measures reducing transmission in this group.

  • Results highlight the need for renewed efforts in the prevention of HIV in men who have sex with men.

Acknowledgments

We thank Louise Primrose-Shaw, Danny Nicholson and Stewart Robison, of the BBV/STI team at Health Protection Scotland, and Hazel Paterson, Joan Findlay and Pamela Munro for their support with the HIV test database. We acknowledge the Scottish government for supporting information-generating initiatives on HIV.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Competing interests None.

  • Ethics approval It is a database study, involving anonymised records.

  • Provenance and peer review Not commissioned; externally peer reviewed.