The changing demographics of women living with HIV/AIDS in from to The changing demographics of women living with HIV/AIDS in southern Alberta from 1982 to 2006

obJeCTIVe: To describe the female population presenting for HIV care in southern Alberta and to examine the impact of opt-out pregnancy screening. MeTHoDS: A retrospective review of demographic and clinical characteristics of all patients presenting to the Southern Alberta HIV Clinic (SAC) care program from 1982 to 2006, was performed. ReSuLTS: The proportion of newly diagnosed patients who were female increased from 7.5% before 1998 to 21.5% after 1998. Women were more likely to be from vulnerable populations, such as intravenous drug users (31.3% versus 13.7%, P<0.001), aboriginals/Métis (21.5% versus 8.7%, P<0.001), blacks (28.9% versus 4.9%, P<0.001) and immigrants (36.6% versus 14.7%, P<0.001). Heterosexual intercourse was the main risk factor for HIV acquisition (43.7%). Women were less likely than men to have requested HIV testing (20.9% versus 37.8%, P<0.001). Opt-out pregnancy screening accounted for 12.7% of HIV-positive tests in women, following its introduction in 1998. Of the women diagnosed by pregnancy screening, 62.1% were from HIV-endemic countries. There was an association between reason for testing and CD4 count at presentation; women who requested their HIV test had higher median CD4 counts than those diagnosed because of illness (478 cells/mL, interquartile range [IQR]=370 cells/mL versus 174 cells/mL, IQR=328 cells/mL, P<0.001) or pregnancy screening (478 cells/mL, IQR=370 cells/mL versus 271 cells/mL, IQR=256 cells/mL, P=0.001). ConCLuSIonS: Women were less likely than men to have requested HIV testing and were more likely to be diagnosed by population-based screening methods. Women, especially vulnerable groups, account for a growing number and proportion of newly diagnosed HIV infections in Alberta. The implications of expanded screening in this population merit further consideration. higher median CD4 counts than women diagnosed as a result of pregnancy screening (478 cells/mL, interquartile range [IQR]=370 cells/mL versus 271 cells/mL, IQR=256 cells/mL, P=0.001) or illness (478 cells/mL, IQR=370 cells/mL versus 174 cells/mL, IQR=328 cells/mL, P<0.001).

Similar to the rest of the world (5), Canada is experiencing a feminization of the HIV epidemic, with women accounting for 28% of new positive HIV test reports in 2006 (1) compared with 12% in 1985 (6). The same trend is observed in Alberta, which has the fourth highest number of HIV cases among the provinces and territories. Since 1985, 21% of Alberta's 5237 positive HIV test reports have been reported among women (1). More recently, that proportion has increased, with women accounting for 35% of new positive HIV test reports in 2008 (7).
Only 30% of Canadian women have ever been tested for HIV, and 54% of women are diagnosed with HIV less than 12 months before AIDS diagnosis, with those born outside Canada, Aboriginal, black reVIew and Asian women more likely to test late (8). Poverty, marginalization, social isolation and sexual violence increase women's vulnerability to HIV infection (5,8) and likely also influence testing behaviours.
The introduction of opt-out pregnancy screening in Alberta in 1998 led to a dramatic increase in HIV testing among women. More than 95% of women accessing prenatal care in Alberta are screened for HIV, leading to a positive test rate of 3.3 in 10,000 in 2000 (9). The sociodemographic and clinical characteristics of women diagnosed as a result of pregnancy screening in northern Alberta have been described recently (10).
Although few Canadian studies have focused on women, factors associated with HIV testing have been described previously, with only 35% of Canadians having ever been tested for HIV and less than 17% reporting voluntary testing. Among those who had been tested, the highest rates were seen among those reporting at least one risk factor for HIV. Even so, 40% of injection drug users (IDUs), and 70% of men and 62% of women reporting high-risk sex partners had never been tested (11). Similarly, laboratory surveillance data in Alberta suggests that HIV testing is accessed infrequently, with 70% of newly diagnosed individuals discovering their status at their first HIV test. Even repeat testers, many of whom had risk factors for HIV, had a median interval of almost two years between their last negative and first positive test (12).
A longitudinal database has been maintained on all patients receiving HIV care in southern Alberta since 1982, presenting a unique opportunity to describe epidemiological trends, reasons for testing and the impact of pregnancy screening on the female population seeking HIV/AIDS treatment over the entire history of HIV within a complete geographical region. We wished to define the changing clinical and sociodemographic characteristics of the female population presenting for HIV care in southern Alberta over time with particular focus on reasons for HIV testing and the impact of opt-out pregnancy screening. We also examined the clinical and sociodemographic differences between women and men with HIV in southern Alberta.

Study participants and design
All HIV-positive individuals referred for HIV assessment and care are automatically enrolled in the Southern Alberta Cohort. The cohort includes all patients in southern Alberta who receive or have received HIV care through the regional HIV care program in Calgary. Access to provincially funded HIV care, including HIV specific laboratory tests and free antiretroviral therapy, is provided exclusively through this program. Routine administrative, demographic, clinical, inpatient and outpatient data are collected at all visits and are maintained within the Southern Alberta Cohort database. The present study included all HIV-infected patients initiating care within the regional centralized HIV care program from January 1982 to December 2006. Ethics approval was obtained for analysis of non-nominal data.
A retrospective review of demographic and clinical characteristics, as a series of secondary analyses in this cohort over the study period, was performed.

outcomes of interest
The objective of the present retrospective review was to describe the female population presenting for HIV care in southern Alberta from 1982 to 2006 and to examine demographic changes over time. The characteristics at first presentation with HIV that were focused on were: age, self-reported ethnicity, marital status, highest achieved education level, reason for HIV testing, hepatitis C virus coinfection, country of birth and initial CD4 cell count at presentation to the Southern Alberta HIV Clinic (SAC). Likely risk activity for HIV acquisition was assigned by a clinician according to a standard hierarchy. Data were stratified according to date of presentation (Pre-HAART [defined   Additionally, female and male populations were compared using the same parameters. Testing behaviours among women with different risk factors for HIV, including the impact of opt-out pregnancy screening, were reviewed. The median CD4 counts at presentation of women with different reasons for HIV testing were also examined.

Statistical analyses
Tests for association were performed using a χ 2 test, or a Fisher's exact test as appropriate. Differences between sexes in continuous variables, such as baseline median CD4 counts and age at presentation, were evaluated using a Mann-Whitney and two-sample t test, respectively. In all analyses, patients with missing data were excluded from that particular analysis but were included for all others where data were present. All analyses were two-tailed, with statistical significance evaluated at the 0.05 level. The demographic characteristics of women presenting to SAC from 1982 to 2006 are described in Table 1. Over time, the female population presenting for HIV care in southern Alberta has become more diverse. Before 1997, most women diagnosed with HIV were Caucasian, and the main risk factor for HIV was heterosexual activity. From 1997 onward, women presenting for HIV care were significantly more likely to be black (35.2% versus 12.6%, P<0.001) or Aboriginal/Métis (24.5% versus 13.7%, P=0.03). Heterosexual intercourse (excluding women from endemic areas) remained the most common risk factor for women but accounted for a significantly smaller proportion of women presenting post-HAART than previously (37.7% versus 58.9%, P<0.001).

Demographic characteristics of women presenting to SAC
Compared with men, women tended to present at a younger age (32.9 versus 35.5 years of age, P<0.001) (

Reasons for HIV testing
Women in this population had a wide variety of reasons for HIV testing (Table 1). Reasons for testing changed as opt-out pregnancy screening was introduced in 1998 and mandatory HIV screening, as part of the immigration process, was implemented in 2002 (1,13). Among women with different risk factors, there was considerable heterogeneity of reasons for undergoing testing (Table 3). Female IDUs were more likely to discover their HIV-positive status due to patient requested testing (38.7% versus 12.9%, P<0.001) or physician requested testing (25.5% versus 16.3%, P=0.049). In contrast, heterosexual women from endemic areas were less likely than other women to discover their status as a result of patient-requested testing (3.1% versus 25.4%, P=0.004). Compared with men, women were significantly less likely to be diagnosed as a result of patient-initiated testing (20.9% versus 37.8%, P<0.001) or illness (18.0% versus 26.2%, P<0.001) ( Table 2). Even when pregnancy screening was excluded, women were more likely to be diagnosed through population-based screening (18.1% versus 7.7%, P<0.001).
CD4 counts were generally associated with the reason for HIV testing ( Figure 2). Women diagnosed via patient request had significantly

Impact of opt-out pregnancy screening
Opt-out pregnancy screening accounted for 12.7% of HIV-positive tests in women after its introduction in 1998. Women diagnosed as a result of pregnancy screening tended to be younger than women diagnosed for other reasons (26.6 versus 32.9 years of age, P<0.001) and were more likely to be black (62.1% versus 25.8%, P<0.001) ( Table 4).

DISCuSSIon
In the population reviewed, the absolute number and proportion of women being diagnosed with HIV have increased substantially over the past decade. Before 1997, the majority of HIV-positive women were heterosexual Caucasians. Since then, our female HIV-positive population has become more diverse, with newly diagnosed women more likely to be from vulnerable groups. In keeping with previous studies (8), black and Aboriginal women are overrepresented. Additionally, more than one-third of our female patients were born outside of Canada, with more than one-half of these women immigrating from HIV-endemic countries. Women diagnosed with HIV after requesting testing had a median CD4 count of 478 cells/mL, which was higher than most women diagnosed by other means. However, only one in five women in our population discovered their HIV-positive status as a result of a patientinitiated test, a proportion which did not change over the study period. Previous studies suggest that self-perception of HIV risk is one determinant of HIV testing (11,14,15) and in keeping with this, female IDUs were more likely than other women to learn their HIV status as a result of patient-initiated testing. Since HIV infection in our population was most commonly diagnosed in women with no risk factors for HIV, aside from heterosexual exposure, the low rate of patient-initiated testing may suggest that many women may be in denial or may perceive themselves erroneously to be low-risk for HIV. Other studies have suggested that even individuals who are aware of their risk factors for HIV often delay testing (12). Social stigmatization, language barriers, financial limitations and cultural beliefs may also contribute to low rates of patient-initiated testing. The factors influencing HIV testing among women from HIV-endemic countries merit further consideration, given that <4% of these women initiated HIV testing.
In the United States, the Centers for Disease Control and Prevention recommends routine, voluntary HIV screening for patients in all health care settings between 13 and 64 years of age and repeat screening at least annually for those at high risk (16). Previous guidelines had focused on testing high-risk individuals and patients in areas where the prevalence of HIV was >1%. The new recommendations were intended to increase HIV screening, diagnose HIV earlier, identify patients who are unaware of their HIVpositive status so that they can receive treatment and reduce HIV transmission. By contrast, current Canadian guidelines recommend voluntary counselling and risk-based testing for HIV (17), but this may lead to delayed diagnosis for many women in our population. Although women diagnosed via physician-initiated testing tended to have CD4 counts that were comparable with women who initiated HIV testing, only one in five women were diagnosed as a result of a physician-requested test. Nearly the same number of women were diagnosed by healthy population-based screening, most notably optout pregnancy screening.
Data from pregnancy screening are often used for sentinel serosurveillance (8,18) but likely underestimate the prevalence of HIV infection in the heterosexual population. Although >95% of pregnant women in Alberta are screened, women who opt out are actually three times more likely to be HIV positive (19) and often do so based on an erroneous assessment of their risk for HIV (20). Obviously, women who never become pregnant are also missed by this screening method. Pregnancy screening in our population mainly diagnosed a specific subset of young, black women from HIV-endemic countries. We were not able to determine whether these patients came to Canada before the implementation of mandatory HIV testing during the immigration process in January 2002 (1,13), were diagnosed initially at immigration and lost to follow-up, or were infected after arrival. The low median CD4 count of women in our population diagnosed via pregnancy screening is concerning, because it may suggest a delay in diagnosis. A recent study suggests that initiating HAART when CD4 counts are between 351 cells/mL to 500 cells/mL improves survival rates by 70% (3); the median CD4 count in our female cohort after the introduction of HAART was only 347 cells/mL, which indicates that one-half of our women were unable to benefit from early treatment. Aside from the health implications, previous studies in our population have shown that the direct costs associated with HIV infection are inversely related to CD4 count at presentation (4). Data from the United States suggests that expanded screening is cost-effective even when the prevalence of undiagnosed HIV infection is as low as 0.20% (21). Given these factors, there may be a rationale for expanded HIV screening in our population.
There are some limitations that should be considered in generalizing these findings to the rest of Canada. Our results are those of a  Figure 2) Median CD4 counts at presentation among HIV-positive women stratified according to reason for testing retrospective, observational study. There was a small number of women compared with men in our population and we may not have had sufficient power to detect small differences. Additionally, our population may be unique in terms of the high proportion of immigrants from subsaharan Africa; even within Alberta, there are considerable demographic differences between northern and southern Alberta (10,12,22). The length of our study period and the comprehensive nature of a geographically defined population may mitigate some of these limitations.
In southern Alberta, women from vulnerable groups appear to be at disproportionate risk for HIV infection. Women are less likely than men to be diagnosed due to patient request, and this may be related to lower CD4 counts at presentation. Further research is needed to determine strategies for prevention and diagnosis in women, especially women from HIV-endemic countries. In particular, the impact of expanded HIV screening merits further consideration.