Two methods for assessing the risk-factor composition of the HIV-1 epidemic in heterosexual women: southeast England, 1988-1991

An agenda for future research on HIV and sexual behaviour among African migrant communities in the UK

The epidemiology of the recent rise in HIV cases in Britain highlights the need for more research among the heterosexual African migrant population. New research should not, however, only extend the limited number of studies that describe observable patterns in sexual health but should also seek to determine their underlying social causation. To achieve this, both methodological and ontological shifts are necessary in the existing research paradigm; we advocate that a broad range of qualitative techniques be deployed both to uncover the empirical details of specifically African sexual behaviours and to highlight and explore the 'relational' nature of sexual decision-making. Rather than fixing on individuals, analysis must situate them within the broader discursive and material frames that structure the boundaries of decision-making. In addition, researchers need to utilise the parallel literature on the social embeddedness of HIV in Africa to inform analysis of the British context. It would then be possible to address the crucial question of whether the social conditions known to cause high-risk behaviours and facilitate transmission in Africa persist, or are transformed, after migration to the UK. A key, and neglected, dimension of this is the role of spatial context in relational sexual decision-making and the constitution of social relationships in particular arenas. This needs further thought, particularly in relation to domestic space and gender identities. We believe that the research agenda proposed herein has much to contribute to interventions and service provision. Nevertheless, we are mindful of the need for self-reflexivity about our role in the production of powerful knowledges about sex. Our final proposal is that researchers seek ways to work with, not on, African communities in order to facilitate their own informed management of sexual health.

Factors associated with HIV infection are not the same for all women

STUDY OBJECTIVES

To determine if factors associated with HIV infection vary between subpopulations of women resident in Great Britain.

DESIGN

Case-control analyses on already existing datasets.

SETTING

Great Britain.

PARTICIPANTS

317 cases selected from a MRC Collaborative Study of HIV Infection in Women and 3635 controls selected from a National Survey of Sexual Attitudes and Lifestyles.

MAIN RESULTS

Factors associated with HIV infection varied among subpopulations of women. Among women heterosexually infected, factors associated with HIV infection were residence in London (odds ratio (OR) = 8.3; 95% confidence intervals (CI) 4.6 to 14.9), widowhood (OR = 47.6; CI 20.0 to 113.1), being black (OR = 25.2; CI 15.6 to 40.8) and particularly among white women, having 10 or more sexual partners (OR = 14.5; CI 5.1 to 41.3). Young age was important for black women heterosexually infected. Among women who shared needles, residence in London (OR = 19.0; CI 5.8 to 62.6) or Scotland (OR = 26.9; CI 8.0 to 90.4) and large numbers of sexual partners was important (OR = 19.6; CI 6.4 to 60.0); termination of pregnancy history was also important for those with fewer than 10 sexual partners (OR = 6.7; CI 3.4 to 13.1); and low social class was important for those with 10 or more sexual partners (OR = 4.1; CI 1.7 to 9.6).

CONCLUSIONS

Factors vary in importance and significance of association with HIV infection in different subpopulations of women resident in Great Britain. This diversity is also likely to occur in other populations. It is important to identify these differences between subpopulations. Prevention and control policies and activities for HIV/AIDS cannot treat all women as if they were the same.

Prevalence of HIV among pregnant women in Dundee 1988-1997: evidence to gauge the effectiveness of HIV prevention measures

OBJECTIVES

During November 1988-July 1990, an HIV prevalence survey of pregnant women in Dundee, which used a combined voluntary diagnostic testing and voluntary unlinked anonymous approach, revealed a rate of 0.3%, the highest recorded prevalence among such a population in the U.K. at the time. To determine if, and why, any changes in HIV prevalence had occurred during the early to mid 1990s, further studies were conducted.

METHODS

During January 1993-December 1997, antenatal patients of, and women undergoing therapeutic termination of pregnancy at, Ninewells Hospital, Dundee, were offered a diagnostic HIV antibody test. For those declining, residual sera from rubella specimens were tested for HIV antibodies using an unlinked anonymous approach which did not necessitate the securement of informed consent. Information about injecting drug use was obtained from all women and linked to their HIV test results.

RESULTS

For all pregnant women, a significant decline in HIV prevalence (P<0.05) from 0.3% (19/6228) during 1988-1990 to 0.12% (22/17899) during 1993-1997 was observed. For those who injected drugs, prevalence decreased significantly (P<0.05) from 27.5% (11/40) to 7% (6/85), while among women who had never injected drugs prevalence decreased slightly, but not significantly (P>0.05), from 0.13% (8/6188) to 0.09% (16/17814).Prevalence in the non-injectors who reported no sexual intercourse with an injector was low at 0.04% (8/17682) during 1993-1997. No significant trends in HIV prevalence were seen over the 5 years up to 1997.

CONCLUSION

The decline in HIV prevalence among pregnant women in Dundee during the early to mid-1990s can be explained predominantly by the control of HIV transmission among the city's injectors, and from them to the wider heterosexual population. It is likely that interventions designed to reduce needle/syringe sharing among injectors have been successful. It is imperative that the preventive effort is not allowed to be weakened.

Geographic variation of HIV infection in childbearing women with syphilis in the United States

OBJECTIVES

Substantial biologic and epidemiologic data indicate the importance of syphilis as a potential cofactor for sexual transmission of HIV infection, but few detailed data exist on the geographic covariation of these two important sexually transmitted infections.

DESIGN

HIV prevalence in childbearing women and primary and secondary (P&S) syphilis data from 29 states were examined to explore the importance of the epidemiology of syphilis as a factor in facilitating HIV transmission.

METHOD

The spatial relationship between P&S syphilis and HIV infection in the health districts of 29 states was analyzed and adjusted for demographic and socioeconomic factors such as racial composition, income, housing, education levels, and access to medical services using the 1990 US census, and geographic location.

RESULTS

In 29 states and the District of Colombia, 448 health districts, representing more than 75% of the US population, reported HIV prevalence rates for mothers' district of residence. The HIV seroprevalence ranged from 0 to 1258/10 000 in these health districts. The incidence of P&S syphilis from 1984-1994 in these districts ranged from 0 to 87/100 000. The P&S syphilis incidence was positively associated with the prevalence of HIV infection among childbearing women (P < 0.0001).

CONCLUSIONS

Syphilis that persists in communities in the United States appears to represent a 'sentinel public health event' reflecting risk for sexual HIV transmission. These findings, along with other biologic and epidemiologic information, reinforce the importance of syphilis as an indicator for targeting HIV prevention efforts generally, as well as syphilis control as a specific HIV-prevention strategy.

Effect of the worldwide epidemic on HIV prevalence in the United Kingdom: record linkage in anonymous neonatal seroprevalence surveys

OBJECTIVE

To assess the impact of the worldwide HIV/AIDS epidemic on the prevalence of HIV in women in the United Kingdom (UK), particularly in the large immigrant and ethnic minority communities.

METHOD

Unlinked anonymous neonatal seroprevalence survey with electronic record linkage of data from child health computers (maternal age and ethnic status) and birth registration (parent's country of birth).

RESULTS

Of a total 137456 samples collected in 1997-1998, 188 (0.14%) were anti-HIV-1 seropositive. Seroprevalence was highest in women born in East Africa (2.3%) and Central Africa (1.9%). 76.4% of seropositive newborns were delivered to mothers born in sub-Saharan Africa; a further 6.0% had fathers from sub-Saharan Africa. However, there was little evidence of HIV in women born in Southern Asia [prevalence 0.0081%; 95% confidence interval (CI) 0-0.04], and none within UK-born Asian communities. Prevalence among the UK-born Black African community was low (0.14%; 95% CI 0-0.6). Among infants with both parents known to be born in the UK, seroprevalence was 0.023% (95% CI 0.01-0.04) in London, and zero (95% CI 0-0.007) in non-Metropolitan areas. Irrespective of mother's region of birth, seroprevalence was 4.2 times higher (95% CI 3.0-5.8) in newborns whose father's details were not recorded at birth registration, a marker for single unsupported mothers.

CONCLUSION

The risk of HIV among pregnant women from sub-Saharan Africa has been recognized. However, in southern England, HIV is very rare in women from Southern Asia and in UK-born women in ethnic minority communities, in spite of cultural and travel ties to high-prevalence countries. Data linkage in anonymous surveys assists in monitoring the impact of the worldwide epidemic on prevalence and incidence locally.

Selective versus universal antenatal HIV testing: epidemiological and implementational factors in policy choice

OBJECTIVE

To develop an epidemiological basis for economic analyses of selective and universal antenatal screening strategies, and to apply it to the UK.

METHODS

The prevalence of higher-risk women and the prevalence of undiagnosed infection within groups of high-risk and low-risk women was estimated from surveillance and survey data. The numbers of women tested and the numbers of infected women who would be identified by universal and selective strategies were then calculated under a range of assumptions about the identification of higher-risk women and acceptance of testing.

RESULTS

In higher-risk women estimated prevalence of undiagnosed infection was between 0.06% and 2.8%, comparing well with independent estimates. In low-risk women, estimates ranged from 0.014% in London to 0.002% in the rest of the UK. If uptake among the high-risk women was the same in selective and universal strategies, universal testing would entail testing between 7100 (London) and 50000 (rest of England) additional women to detect an additional case. However, if selective screening identified only 60% of those at high risk and achieved only 60% acceptance compared with a universal programme, then universal screening would require only 1150 additional women to identify one additional case in London, compared to 6470 in Scotland and 13140 in the rest of the UK.

CONCLUSIONS

Overall prevalence does not form an adequate basis for determining screening strategy. Instead, universal screening can be justified either because the prevalence of HIV in the low-risk group is sufficiently high, or because it achieves sufficiently higher uptake relative to selective screening among those at higher risk.

Surveillance for the impact in the UK of HIV epidemics in South Asia

OBJECTIVES

To determine whether, because of the extensive recent spread of HIV infection in South Asia, South Asians (those people who classify themselves as Indian, Pakistani, Bangladeshi or Sri Lankan in origin) resident in the UK were at increased risk of HIV infection and to review current surveillance systems for detecting any such increase.

DESIGN

Analysis of: ethnic grouping and probable country of infection recorded on voluntary confidential reports of AIDS cases and newly diagnosed HIV infections; blood donation testing data; reports of imported gonorrhoea infections; country of birth data from the unlinked anonymous (UA) survey of Sexually Transmitted Disease (STD) clinic attenders; district of residence data from the UA survey of pregnant women; ethnic grouping of prevalent diagnosed HIV infections.

RESULTS

Few reported AIDS cases or HIV infections were found in people of South Asian ethnic origin and few reported HIV or gonorrhoea infections were associated with exposure in South Asia. Data derived from the UA programme suggested as yet no increase in HIV prevalence in either STD clinic attenders born in South Asia or in pregnant women resident in districts containing substantial numbers of ethnic South Asians.

CONCLUSIONS

There was no evidence that South Asians resident in the UK are currently at greater risk of HIV infection than people of white ethnicity or, therefore, that south Asian heterosexuals are a group deserving priority in HIV prevention. However, as rapid spread of HIV infection is being recorded in the Indian subcontinent, continuous monitoring is necessary. This will be facilitated by improved collection of ethnic group information in all surveillance activities.

HIV prevalence in pregnant women in Europe: differences in assessment methods and prevalence levels across countries

OBJECTIVE

To describe methods used to assess HIV prevalence and to assess prevalence levels and time trends among pregnant women in various European countries.

METHODS

We used the European HIV Prevalence Database to examine annual HIV prevalence data in pregnant women for the years 1990 to 1996 (20 countries).

RESULTS

In Western Europe, prevalences were generally obtained through unlinked anonymous surveys, whereas in most Central, and Eastern European countries, they were based on testing programs (voluntary or mandatory). Prevalences (per 10,000) were highest (i.e., 10-30/10,000) in large western urban areas including Amsterdam, Barcelona, London, Milan, Paris, and Rome; between 1 and 2 in Scandinavian countries; and down to 0.5/10,000 in Central and Eastern European countries (except Ukraine, 1996: 5/10,000). Prevalences decreased in Rome, whereas they increased in London, the Czech Republic, and since 1995 in Russia and Ukraine; elsewhere, no time trends were detected.

CONCLUSIONS

Methodologic differences and potential biases should be considered when comparing these data. HIV prevalence in pregnant women is useful for monitoring the AIDS epidemic and for assessing and improving prevention. Efforts should be made to offer voluntary counseling and testing to women at risk for HIV and provide treatment to those who are infected.

The incidence and prevalence of HIV infection among childbearing women living in Edinburgh city, 1982-1995

OBJECTIVE

To track the complete course of the HIV epidemic among women from the city of Edinburgh who delivered babies during 1982-1995.

METHODS

The performance of the modified Serodia HIV test on dried blood spots from archived neonatal metabolic screening cards stored for up to 11 years was evaluated by testing 221 cards from neonates whose mothers' HIV infection status was already known (100 HIV-positive, 121 HIV-negative). Unlinked anonymous HIV testing of cards from neonates born during 1982-1989 was then performed and the resulting prevalence data were combined with existing data from 1990-1995. Maximum and minimum limits of HIV incidence among women during the 36-month period prior to delivery were calculated using data held on a clinical database of HIV-infected pregnant women that had been generated under strict conditions of confidentiality; these data included the date of the woman's first HIV-positive and, if available, last HIV-negative specimen.

RESULTS

The evaluation revealed a sensitivity of 91%, not clearly related to storage time, and a specificity of 100%. HIV infection first entered Edinburgh's childbearing population during the early 1980s with prevalence peaking at 0.4% in 1986 and then decreasing to 0.1% in 1995; a similar incidence profile was seen during this period. Since 1986, the first full year that HIV testing was available, 78% of all infections were known during the pregnancy, 13% were identified retrospectively, and only 10% (10 cases) remain unaccounted for. For infected cases during 1984-1987, 78% were injecting drug users (IDU) and only 22% acquired their infection sexually; this distribution had reversed by 1992-1995.

CONCLUSION

HIV testing of neonatal metabolic screening cards stored for up to 11 years can yield results of sufficient accuracy for epidemiological purposes. There has been a substantial decline in the prevalence and incidence of HIV since the mid-1980s. Although new infections are still occurring, the numbers are small. The decline may largely be explained by the impact of preventive measures on the spread of HIV amongst IDU, and thus from IDU to their sexual partners.

Interpreting HIV seroprevalence data from pregnant women

HIV prevalence estimates for pregnant women from unlinked anonymous surveys are becoming increasingly available and can be used to determine the prevalence of HIV in women in the same population. The ratio of prevalence in pregnant women to that in all women is influenced by HIV-related risk behaviours that are different for pregnant and nonpregnant women and also by differences in fertility level among infected and uninfected women. This ratio is affected by biases that are likely to be culturally and socially specific. A model is proposed for the qualification and quantification of these biases and hence the estimation of general female population prevalence from serosurveillance data on pregnant women.

Voluntary, named testing for HIV in a community based antenatal clinic: a pilot study

Despite the increasing advantages of identifying HIV infection in pregnant women, only some 12% of HIV positive women attending antenatal clinics in London have been identified by named testing. As virtually all antenatal care will be community based within the next two to three years, we assessed the problems of introducing named HIV testing during pregnancy into the primary care setting. Planning the service took a considerable time and required the production of educational material for both staff and pregnant women and some reorganisation of procedures. Over a one year period an uptake of 44% was noted. Several problems were encountered including an average of 21 minutes needed to give information on AIDS and HIV, an adverse effect on the midwife-mother relationship, and anxiety (affecting both women and midwives). Possible solutions to this difficult problem are discussed.

The epidemiology and transmission of AIDS: a hypothesis linking behavioural and biological determinants to time, person and place

Epidemiologically, the Acquired Immune Deficiency Syndrome, AIDS, is transmitted and distributed in the USA and Europe almost entirely in well-defined subsets of populations engaging in, or subjected to, the effects of behaviours which carry high risks of genital and systemic infections. The persons predominantly affected are those engaging in promiscuous homosexual and bisexual activity, regular use of addictive drugs, and their sexual and recreational partners. In such persons and in subsets of populations with corresponding life-styles, the risk of AIDS increases by orders of magnitude. Because of continuity of risk behaviour and of associated indicator infections, the incidence of AIDS over 3-5 year periods is predictable to within 10% of actual totals of registered cases in the USA and UK. Secondary transmission of AIDS beyond these groups is minimal or, in many locations, absent. There is no indication of appreciable spread by heterosexual transmission to the general population. The Human Immunodeficiency Virus, HIV, is transmissible to some extent in general populations, and more so among promiscuous persons. It may cause viraemia, lymphadenopathy and latent infection (HIV disease) in anyone. In persons engaging in risk behaviours which themselves alter or suppress immune responses, it can interact with MHC, antibodies to other organisms and to semen, and other allogenic antigens to initiate a programmed death of CD4 lymphocytes and other defensive cells, as in graft-host rejections. This occurs also in haemophiliacs receiving transfusions of blood products, and is more pronounced in persons with reactive HLA haplotypes. The susceptibility of particular subsets of populations to AIDS is thereby largely explained. But these changes occur in the absence of HIV, and so do Kaposi's sarcoma, lymphadenopathies and opportunistic infections which are regarded as main indicators of AIDS. The hypothesis that HIV-1 can do all this by itself and thereby cause AIDS is falsifiable on biological as well as epidemiological grounds. An alternative hypothesis is proposed, linking the incidence of AIDS to the evolution of contemporary risk behaviour in particular communities and locations in the USA, UK and probably in most of Europe. It does not pretend to explain the reported incidence of AIDS in Africa and other developing regions where data are insufficient to provide validation of the pattern of disease and contributory variables. The immediate, practical implication of this alternative hypothesis is that existing programmes for the control of AIDS are wrongly orientated, extremely wasteful of effort and expenditure, and in some respects harmful.

How much paediatric HIV infection could be prevented by antenatal HIV testing?

OBJECTIVE

To estimate the reduction in the number of children infected with HIV that might be achieved by extending the provision of voluntary antenatal HIV testing. This effect would be mediated by increased numbers of women infected with HIV who receive an intervention to reduce the risk of vertical transmission (for example, zidovudine or caesarean section delivery), who use an alternative to breast feeding, or whose pregnancy is terminated.

SETTING

London, United Kingdom.

METHODS

Relevant data were derived from neonatal seroprevalence studies, obstetric and paediatric reporting schemes, and review of external information. Sensitivity analyses were performed for certain parameters.

RESULTS

Of 106,000 births annually in London, an estimated 169 are to women infected with HIV whose infection is not currently recognised before pregnancy. An estimated 28-33 children born to these women will be infected. Precise prediction of the number of paediatric HIV infections that could be prevented is difficult because of uncertainty in certain factors, particularly the uptake of antenatal testing and the efficacy and acceptability of interventions to reduce prenatal or perinatal transmission. If a testing programme detected 70% of infected women, none of whom opted for a termination but all of whom exclusively bottle fed and received an intervention which halved the risk of transmission, about 12-16 (42-46%) paediatric HIV infections would be prevented annually.

CONCLUSIONS

The estimated cost of preventing each paediatric infection is high, but this should be seen in the context of the lifetime health and social care costs for a child infected with HIV. The feasibility of selective testing should be considered when formulating policies on antenatal HIV testing. Programmes that are introduced should be audited to obtain better estimates of costs and benefits.

Use of anonymous newborn serosurveys to evaluate antenatal HIV screening programmes

OBJECTIVE

To evaluate the extent to which antenatal HIV screening programmes identify HIV infected women who go to term.

DESIGN

Comparison of results of two surveillance systems. An anonymous neonatal HIV serosurvey was used to estimate the numbers of HIV infected women giving birth; reporting by obstetricians was used to assess the proportion who had been identified.

SETTING

Three Thames regions.

RESULTS

729,105 neonatal blood samples were tested, of which 484 were HIV seropositive. Newborn HIV seroprevalence is increasing, at different rates, in inner London, suburban London, and in non-metropolitan districts. During the past four years the proportion of infected women who have been identified before delivery is 16.9%, but less than half of these were diagnosed during pregnancy. In 1993 only five of the 128 (4%) previously undiagnosed infected women delivering babies were identified by antenatal screening.

CONCLUSION

Despite increased emphasis on antenatal testing for HIV in areas of higher prevalence the number of undiagnosed women delivering babies continues to increase. Consideration should be given to alternative strategies for offering antenatal HIV testing. Antenatal screening programmes should be monitored continuously by comparing anonymous neonatal seroprevalence with clinical reports from obstetricians.

Vertically transmitted HIV infection in the British Isles

OBJECTIVE

To describe the epidemiology of vertically acquired HIV infection in the British Isles, the level of underreporting, the vertical transmission rate, and clinical spectrum of paediatric AIDS.

DESIGN

Confidential, linked registers based on reporting from obstetricians and paediatricians; anonymous unlinked neonatal HIV serosurveys.

SETTING

British Isles.

SUBJECTS

Children born to mothers with HIV infection.

MAIN OUTCOME MEASURES

Trends in HIV infection and vertical transmission rate.

RESULTS

In Scotland and the Irish Republic, where most maternal HIV infection is related to drug misuse, the annual number of reports of children born to infected mothers has fallen since 1989. In England and Wales nearly half of maternal infections have been acquired overseas, and the number of children born to these women, and to women who became infected in Britain, is increasing. In south east England the proportion of live births to women whose infection was identified before delivery was only 17% (50/287), compared with 68% (26/38) in Scotland. The vertical transmission rate was 13.7% (23/168), and 23% of infected children developed AIDS in the first year of life. 41% (38/92) of children born to infected mothers who were ascertained after delivery were breast fed, compared with 5% (12/236) of those ascertained before delivery.

CONCLUSIONS

The incidence of vertically transmitted HIV infection is increasing in England and Wales. More extensive antenatal testing would enable infected women to be counselled against breast feeding, which could prevent a substantial proportion of vertical transmission in some areas, and would increase opportunities for early diagnosis and treatment of infected children.

Screening for HIV infection in pregnancy

Antenatal HIV screening policies throughout the British Isles were surveyed and results linked to data on HIV-infected pregnant women notified through the Royal College of Obstetricians and Gynaecologists. Units offering HIV testing to all pregnant women were compared with those offering it only to women considered to be at risk, and in both situations fewer than 50% of infected women were identified as HIV infected for the first time in antenatal clinics. Based on laboratory reports of HIV infection in women of childbearing age, paediatric reports of children born to HIV positive women and unlinked anonymous neonatal screening programmes, there was evidence of under-recognition of HIV infection in pregnancy, particularly in England and Wales.