Late diagnosis of HIV infection in the era of highly active antiretroviral therapy

consequences for AIDS incidence

Abstract

Objective: 

To assess the repercussion of late diagnosis of HIV infection on AIDS incidence in the era of highly active antiretroviral therapy.

Design: 

Analysis of AIDS surveillance data.

Setting: 

Spain.

Patients: 

AIDS cases reported from 1994 though 2000.

Main outcome measures: 

Late testers were defined as persons who had a first positive HIV test in the month of or immediately preceding AIDS diagnosis. Their incidence trend was compared against that for the remaining cases, and the influence of demographic factors evaluated.

Results: 

Of 30 778 AIDS cases, 8499 (28%) were late testers, and of these, 1061 (12%) died within 3 months of diagnosis of HIV infection. From 1995 to 2000, AIDS diagnoses declined by 36% among late testers versus 67% for the remainder (P < 0.001). The percentage of late testers increased from 24% in 1994–1996 to 35% in 1998–2000 (P < 0.001). Among the 7825 AIDS cases diagnosed in 1998–2000, late testing was independently associated (P < 0.01) with male sex, age over 44 years, residence in provinces with a lower AIDS incidence, foreign origin, and no record of injecting drug use or prison stay.

Conclusions: 

A growing proportion of AIDS cases involves late diagnosis of HIV infection. Persons who are unaware of their HIV infection cannot benefit from antiretroviral therapy and, hence, early diagnosis would strengthen the impact of such therapy and so reduce AIDS incidence.

Introduction

Highly active antiretroviral therapy (HAART) has proved very effective in reducing the progression of HIV disease [¹]. The introduction of this treatment has led to substantial reductions in AIDS incidence and mortality [²]. The optimal time at which to initiate therapy remains controversial [^3,4], but in any event it should usually be before presentation of the first AIDS defining illness. Early diagnosis of HIV infection means that initiation of treatment can be timed to ensure maximum benefit and may, moreover, serve to bring about a reduction in HIV-transmission risk behaviours [⁵]. HIV infection usually precedes the first AIDS defining illness by a number of years, so that the appearance of any of these illnesses in persons unaware of having HIV infection must be considered a failure in early diagnosis.

Despite the widespread availability of antiretroviral treatment, Western European countries and the USA – using a different case definition – have reported, respectively, more than 9500 and more than 40 000 new AIDS cases in 2000, and their recent trends in AIDS incidence show that its decrease has slowed down significantly [^6,7]. Late diagnosis in the course of HIV infection accounts for some of these cases [^8,9], yet its true impact is unknown, as other circumstances, such as late entry into care [⁹], lack of compliance with therapy [¹⁰] and resistance to antiretroviral drugs [^11,12], also exert an influence.

AIDS incidence among persons who were previously unaware of their HIV infection is not directly affected by this therapy. This study compares the trend of late-testing versus remaining AIDS cases in Spain, to assess the repercussions of late diagnosis of HIV infection on AIDS incidence in the era of HAART. Secondly, it analyses the factors associated with late diagnosis of HIV infection among AIDS cases, before and after HAART became available.

Methods

Study population

In Spain (40 million inhabitants), free voluntary confidential HIV testing is available to the entire population at primary care centres, hospitals and genitourinary clinics. In addition, some cities provide sites offering anonymous HIV counselling and testing. Under current guidelines [^3,13,14], HAART has been available free of charge to all patients since 1997 and has led to > 60% reduction in AIDS incidence and mortality [⁶].

AIDS cases in Spain are reported to a national system. Since 1994 the 1993 revision of the European AIDS surveillance definition [¹⁵] has been used, according to which diagnosis of AIDS requires the presence of any one of the various AIDS defining conditions. Reporting is confidential, is undertaken by clinicians, and in some regions is completed by an active case search. A specific report form is used; among other variables, this includes the month and year of AIDS diagnosis and first positive HIV test result.

This study covered AIDS cases, aged over 14 years, reported in Spain until 30 June 2001, and with a first AIDS defining condition diagnosed in the period of January 1994 through December 2000.

Variables

AIDS cases were defined as ‘late testers’ if they reported the first HIV positive test result in the month in which AIDS was diagnosed or in the preceding month [^16,17]. We also used other variables recorded on the AIDS case report form, namely: sex, age, HIV transmission category (injecting drug users, homo/bisexual cases, heterosexual cases, and other or undetermined risk), prison record, province of residence and country of origin. With respect to prison record, a distinction was drawn between AIDS cases reported from prison, diagnosed outside prison but reporting previous stay in prison, and the remainder. Two categories were considered according to whether the province of residence had a cumulative rate of AIDS incidence (from the beginning of the epidemic) that was higher or lower than the national rate for Spain. Patients reported with a country of origin other than Spain were categorized as coming from industrialized countries (western Europe, Canada, USA, Australia, New Zealand and Japan) or developing countries.

Statistical analysis

Trends in the annual number of AIDS diagnoses, broken down into late testers and all remaining AIDS cases, were analysed for the 1994–2000 period with data adjusted for reporting delay [¹⁸].

The overall study period was divided into two according to year of AIDS diagnosis, namely, 1994–1996 and 1998–2000, respectively pre- and post-dating the widespread use of HAART; 1997 itself was not included because it was the year of transition during which these treatments became widely available. For each period, the proportion of late testers among AIDS cases was analysed by reference to the above variables. The χ² test was used to compare proportions and Wilcoxon's test to compare medians. The exploratory univariate analysis was followed by a multivariate logistic regression used to identify the isolated effect of each variable. The first-order interaction terms were tested and included into the models when P < 0.01. Statistical comparisons at P < 0.01 were considered significant. Association between variables was quantified by means of the odds ratio and its 99% confidence interval. Statistical analyses were performed in SAS (version 6.12; SAS Institute; Cary, North Carolina, USA).

Results

Study population

In the 1994–2000 period, a first AIDS defining condition was diagnosed in 33 923 persons aged > 14 years. A total of 3144 patients (9.3%) were excluded from the analysis for having the date of first diagnosis of HIV infection incomplete. There were small but statistically significant differences between analysed and excluded AIDS cases regarding sex, transmission category, prison record and country of origin, and more pronounced ones regarding province of residence. (Table 1). The median time between HIV and AIDS diagnoses was 37 months (range, 0–219 months), and 8499 cases (27.6%) reported a first positive HIV test result in the same or preceding month. Of these latter cases, 1061 (12.5%) were reported as having died within 3 months of diagnosis of HIV infection. Mortality reported in the month following AIDS diagnosis was slightly higher among those who had been late testers than among the other cases (10.2% versus 8.0%; P < 0.0001).

Table 1: Characteristics of the AIDS cases included in the analysis and of cases excluded because date of first HIV positive test result was incomplete. Spain, 1994–2000.

Median CD4 cell count at AIDS diagnosis was significantly lower among late testers than among all other cases (50 × 10⁶/l versus 81 × 10⁶/l; P < 0.0001). Pneumocystis carinii pneumonia and Kaposi's sarcoma were more frequent among late testers, whereas pulmonary tuberculosis and recurrent pneumonia were more frequent among the remainder. Nevertheless, the patterns of other AIDS defining diseases were similar within the two groups (Table 2).

Table 2: Clinical characteristics of AIDS cases diagnosed in the 1994–2000 period, by time since HIV diagnosis.

AIDS incidence trends among late testers and non-late testers

From 1995 to 2000, the number of AIDS diagnosis adjusted for reporting delay declined by 67% among persons previously aware of HIV infection versus 36% for late testers (P < 0.01) (Fig. 1). Among the former cases the greatest decline (52%) occurred from 1996 to 1998, coinciding with the spread of HAART, whereas the trend among late testers was not directly affected by the introduction of such therapies. For each transmission category, AIDS incidence adjusted for reporting delay showed a much smaller decline among late testers than among the remaining cases and, as a result, incidence of AIDS attributed to homo/bisexual contact or heterosexual contact among late testers now exceeds the incidence among non-late testers (Fig. 1).

Fig. 1.:  Annual incidence of AIDS by time between diagnosis of HIV infection and diagnosis of first AIDS defining condition. Results for all cases and main transmission categories.

Determinants of delayed diagnosis of HIV infection

Of the 18 755 AIDS cases diagnosed in the period 1994–1996, 24.1% were late testers, whereas among the 7825 diagnosed in the period 1998–2000, this percentage increased to 35.3% (P < 0.001). Of these late-testing AIDS cases, 12.7% and 11.3% (P = 0.079) were respectively reported as having died within 3 months of the date of first positive HIV test result.

The analysis of AIDS cases diagnosed in 1998–2000 showed that late testers were independently associated with male sex, age over 44 years, transmission categories other than injecting drug use, country of origin other than Spain and residence in provinces with low AIDS incidence. On the other hand, those either in prison or with a prison record had a far lower likelihood of being late testers than the remainder (Table 3). All of these variables, except for the province of residence, were associated with late-testing AIDS cases in the 1994–1996 period. As compared with the 1994–1996 period, in 1998–2000 the percentage of AIDS cases with late diagnosis of HIV infection increased significantly within all categories of variables analysed, except for AIDS cases diagnosed in prison (10.0–7.5%; P = 0.131).

Table 3: Proportion of AIDS cases unaware of their HIV infection until the month of or that preceding AIDS diagnosis, according to whether AIDS diagnosis was made before or after the introduction of highly active antiretroviral therapy (HAART).

We repeated the analyses, assuming all cases with an unknown date of HIV diagnosis to be non-late testers. However, while the percentage of late testers declined slightly (21.8% in 1994–1996 and 32.3% in 1998–2000), the main findings described above remained unaltered (data not shown).

Discussion

The results show that more than one-third of AIDS cases diagnosed in Spain between 1998 and 2000 had been unaware of their HIV infection until the month of diagnosis or the one immediately preceding it. Furthermore, > 10% of these cases died within 3 months of diagnosis of HIV infection. Many of these AIDS cases and deaths could have been prevented through early diagnosis of HIV infection and application of the appropriate antiretroviral treatment [^3,13,14]. Such persons had lived, probably for years, without realizing that they could transmit the HIV infection to others and without receiving counselling. Furthermore, indirect evidence suggests that the infectiousness of these persons could have been decreased by lowering their viral load [¹⁹].

The percentage of late testers that we find during the years preceding the introduction of HAART are in line with those described for other countries in population- [^16,20–22] and hospital-based studies [¹⁷], with small divergences that can be explained by differences in the definition of late testers. We detect an increase in the proportion of late testers among new AIDS diagnoses in Spain, coinciding with the introduction of HAART, and similar to that observed by Girardi et al. in an Italian hospital [⁸]. We therefore are able, by using population-based data, to say that the incidence of late-testing AIDS cases has shown a smaller decline in percentage than the incidence of non-late testers, who registered a clearly observable decrease coinciding with the spread HAART. The incidence of late-testing AIDS cases has also decreased by 36%, which may be explained by the remission in the epidemic and by the improvement of early diagnosis of HIV infection. Working on the assumption that earlier diagnosis of late-testing AIDS cases would have led to their plotting a trend similar to that of the remaining cases, total incidence of AIDS in 2000 would then have been 17% lower.

Despite its importance, undiagnosed HIV infection is a complex area of study. Our approach is based on the availability of HIV diagnosis dates in the AIDS case reports in Spain, which allows us to assess those persons who presented with any AIDS defining illness and were unaware of their HIV infection. The data source used was population based; its mode of operation has undergone no important changes since the AIDS case definition was expanded in 1994 [¹⁵]. In two Spanish provinces a record linkage of the AIDS and the HIV reporting systems was carried out, which enabled us to observe the high validity of the date of first positive HIV test result in the AIDS case reports. Any possible precision errors in the date of first positive HIV test result was minimized by categorizing it by reference to a single cut-off point in the 2 months preceding AIDS diagnosis. This date was unreported in 9.3% of cases and was not observed to have affected principal results. It is rare that AIDS diagnosis occurs during acute seroconversion; this would cause misclassification as a late tester. Among persons in whom HIV diagnosis coincides with the first AIDS defining condition, antiretroviral therapy can still bring about a better prognosis. Therefore, the consequence of late HIV diagnosis would probably be less appreciable in terms of mortality than it would in terms of AIDS incidence. Persons infected by HIV more recently have had less time to have a test done, although some may have developed AIDS (i.e. being a rapid progressor to AIDS). This is why the epidemiological characteristics of those who acquired the infection more recently may be over-represented in the group of late testers. The CD4 cell count, the pattern of AIDS defining diseases and the mortality in the month following AIDS diagnosis did not differ substantially between the late testers and the remaining AIDS cases, and the differences observed may well be justified by epidemiological and demographic characteristics of both groups.

The proportion of late testers was lower among women [^8,16,19,20], something that may be due to women's greater use of healthcare services, the standard practice of recommending HIV testing to all pregnant women [^23,24], and medical monitoring of women having sexual partners with known HIV infection. Persons with a record of injecting drug use registered a lower proportion of late HIV diagnoses [^8,16,19,20], possibly owing to their greater awareness of risk and the offer of testing included in addiction treatments and harm-reduction programs. On the contrary, among homo/bisexual, heterosexual and undetermined transmission categories late testers exceeded by 50% the number of AIDS cases in the era of HAART, probably reflecting their lower perception of risk.

Prison stay was the single factor most strongly associated with early diagnosis of HIV, even after adjustment for other variables. In Spanish prisons, voluntary HIV testing is offered to all persons on admission and annually thereafter to all inmates [²⁵], and antiretroviral treatment is administered to HIV-infected persons in accordance with current guidelines [^3,13]. The era of HAART has given a boost to these programs, which may explain why the proportion of late testers has not risen.

Late diagnosis was more frequent among AIDS patients who reported a country of origin other than Spain – whether developed or developing – a phenomenon that may be due to cultural barriers, alienation and language. It could also be the case that some of the AIDS diagnoses realized before arrival in Spain were not recorded because they had not been documented. On adjusting for other variables, late testing in the era of HAART was associated with residence in provinces with lower AIDS incidence. Probably, among the populations and clinicians of such provinces, suspicion of HIV infection is not as pronounced, and, since small towns or rural areas are usually involved, there are additional taboos of a cultural and social nature when it comes to requesting and prescribing the test.

In conclusion, the existence of a considerable proportion of undiagnosed HIV infections acts as a barrier, preventing HAART from having a greater impact. There are still marked shortcomings in early diagnosis of HIV infection, particularly with regard to men over the age 44 years, who have acquired the infection by sexual transmission and reside in areas where AIDS is not so prevalent, and to migrants. Across a broad spectrum of the population HIV testing still retains special connotations, which lead to reticence when it comes to requesting and, in the case of certain medical practitioners, also when it comes to recommending it [^26,27]. For any headway to be made in early diagnosis, the population must be informed as to its benefits, testing must be made generally available in pregnancy and other situations in which it is especially indicated, and the number and accessibility of HIV testing sites must be expanded and enhanced [²⁸]. Moreover, an attitude of normalcy to HIV testing must be fostered among the general population, and clinicians of all disciplines encouraged to incorporate it much more frequently – on a confidential basis and with patients’ informed consent [²⁷] – as part of their standard diagnostic procedures.

Acknowledgements

This study was made possible thanks to the collaboration of clinicians and public health professionals engaged in AIDS-case reporting in Spain.