High prevalence of unintended pregnancies in HIV-positive women of reproductive age in Ontario, Canada: a retrospective study
*See Appendix S1.
Abstract
Objectives
There is speculation, but there are few data, on the high rates of unintended pregnancies in HIV-positive women. We investigated rates and correlates of unintended pregnancies among HIV-positive women of reproductive age.
Methods
A cross-sectional study was conducted with recruitment stratified to match the geographical distribution of HIV-positive women of reproductive age (18–52 years) living in Ontario, Canada. Women, recruited from 38 sites between October 2007 and April 2009, were invited to complete a 189-item self-administered survey. This analysis focused on questions relating to pregnancy and whether the last pregnancy was intended. Logistic regression models were fitted to calculate unadjusted and adjusted odds ratios of correlates of unintended pregnancies occurring after HIV diagnosis. Happiness with unintended pregnancies was also assessed.
Results
The median age at the time of the survey of the 416 participating HIV-positive women who were previously pregnant (53% before and 47% after HIV diagnosis) was 38 years [interquartile range (IQR) 33–44 years] and their last pregnancy was a median of 8 years (IQR 3–14 years) prior to the survey (n=283). Fifty-nine per cent were born outside Canada and 47% were of African ethnicity. Of the 416, 56% [95% confidence interval (CI) 51–61%] identified that their last pregnancy was unintended (57% before and 54% after HIV diagnosis). In the multivariable model, significant correlates of unintended pregnancy after HIV diagnosis were: marital status (P=0.01) and never having given birth (P=0.01). Women were less happy if their pregnancy was unintended (P<0.01).
Conclusions
The prevalence of unintended pregnancy was high in this cohort. Pregnancy planning programmes are needed for this population to decrease fetal and maternal complications and reduce vertical and horizontal transmission.
Introduction
Over the past two decades, significant breakthroughs have occurred in the area of HIV and pregnancy, largely centred on the prevention of vertical transmission [1,2]. However, there are other important factors to consider for an HIV-positive woman wanting to become pregnant, including the prevention of horizontal transmission between partners, the optimization of antiretroviral therapy (ART), including the discontinuation of potentially teratogenic drugs, and the promotion of a healthy pre-conception lifestyle to reduce maternal and fetal complications [2,3]. While promotion of a healthy pre-conception lifestyle is applicable to all pregnancies, the additional considerations in the context of HIV infection make planning pregnancies of vital importance. This has been demonstrated by the release and updating of guidelines on the management of HIV infection and pregnancy by many countries and, most recently, by the World Health Organization (WHO) [2–6].
Despite the importance of planning pregnancies in the context of HIV infection, many remain unplanned [7,8]. A study conducted in Argentina reported that 55% of HIV-positive women had at least one pregnancy after their HIV diagnosis and that more than half of those pregnancies were accidental [7]. In a survey of 42 sub-Saharan African countries, where the prevalence of HIV infection is high, 10–65% of women responded that their last pregnancy had been unintended [9]. In the United States of America (USA), Koenig and colleagues found that, of 1183 births to 1090 adolescent HIV-positive girls, only 50% knew their HIV status prior to the pregnancy, 67% had been previously pregnant and 83.3% of the pregnancies were unplanned [8].
Unintended pregnancies are similarly common in the general population [10–13]. The 2002 National Survey of Family Growth showed that 49% of pregnancies to women aged 18–44 years old in 2001 in the USA were unintended [10]. The U.S. Behavioral Risk Factor Surveillance System survey data showed that 29% of 18- to 44-year-old fertile women were at high risk for unintended pregnancy, based on the report of failure to use any form of contraception [11]. A 19% pregnancy rate was observed among a cohort of women seen in a sexually transmitted disease clinic in the USA, all of whom reported ‘no intention of becoming pregnant’ at their previous visit [12]. The 2008 Preconception Health Survey of 200 pregnant women and 151 women with a child under the age of 7 years living in Ontario, Canada, revealed that 30% of pregnancies were unplanned and 67% of women were happy with their last pregnancy [13].
To explore rates and correlates of unintended pregnancies among adult HIV-positive women in Canada, we conducted a secondary analysis of a cross-sectional study of HIV-positive women of reproductive age living in Ontario, which collected information about the primary outcome of fertility intentions along with pregnancy history data and whether pregnancies were intended [14]. This analysis aimed to determine the prevalence of unintended pregnancies in an HIV-positive female population before and after their HIV diagnosis and to identify potential correlated sociodemographic and clinical variables for those unintended pregnancies after HIV diagnosis. By highlighting these results, our aim is to make recommendations that will positively impact the behaviour of HIV-positive women and their healthcare providers, by ensuring that the discussion of pregnancy planning is a part of routine HIV care, thereby increasing the likelihood of more planned pregnancies and providing an opportunity for optimal management.
Methods
Study design and population
This was a secondary analysis of a larger study, the details of which are reported elsewhere [14]. The main data set was from a cross-sectional study using a survey instrument which was conducted with participants who met the following inclusion criteria: (1) HIV-positive, (2) biologically female, (3) of reproductive age (between the ages of 18 and 52 years), (4) living in Ontario, Canada, and (5) able to read English or French. The upper age limit was chosen to reflect the cut-off for fertility clinic consultation in Canada. As the primary outcome of interest for this analysis was unintended pregnancies, two supplementary inclusion criteria were added: (1) women had to have been previously pregnant and (2) women had to have answered the question on whether their last pregnancy was intended or not.
Recruitment and ethics
Recruitment was conducted from 5 October 2007 to 31 March 2009 through 38 sites across the province of Ontario and is reviewed in detail elsewhere [14]. An attempt was made to stratify recruitment by provincial regions described by the provincial Public Health Departments such that the study sample would be proportional to the geographical distribution of the HIV-positive female population in Ontario [14,15]. Each research site received ethics approval from their local institutional research ethics board. Written informed consent was obtained from every participant.
Survey instrument and validation
A 189-item survey instrument, The HIV Pregnancy Planning Questionnaire, was created using the methods of Fowler for instrument development and has been previously described in detail elsewhere (full survey instrument available upon request) [14,16]. The survey was first developed in English and translated into French using the back translation method. Content and face validity were achieved as previously described [14].
Statistical analysis
Baseline characteristics of the study population were summarized using medians and interquartile ranges (IQRs) for continuous variables and frequencies and proportions for categorical variables.
The primary outcome of interest for this analysis was unintended pregnancies. The question in the survey used to represent unintended pregnancy was ‘Was your last pregnancy planned?’ The variable was dichotomized into ‘unintended pregnancy’ if answered ‘No’ and ‘intended pregnancy’ if answered ‘Yes’. Women who had never been pregnant were excluded from the analysis. Women who had been pregnant but did not answer this question or answered ‘I don't know’ were also excluded from the analysis. Additional analyses were carried out limiting the sample to those with pregnancies before and after HIV diagnosis. Other outcomes of interest included the total number of births, the proportion of women who gave birth before and after their HIV diagnosis and the timing of births.
Univariate logistic regression models were fitted to determine the unadjusted odds ratios with 95% confidence intervals (CIs) for correlates of unintended pregnancy after HIV diagnosis. Current CD4 cell count, viral load, employment status, household income, sexual relations and contraceptive use were not considered in the regression models as they corresponded to the time of administration of the survey and not the time of the last unintended pregnancy. Other variables, namely, age, ethnicity, number of years since immigration for those born outside Canada, religion, sexual orientation, marital status, duration of HIV diagnosis, HIV risk factor, hepatitis B virus (HBV) and hepatitis C virus (HCV) coinfection and fertility history, were included in the modelling as they were considered to be correlated to those at the time of the pregnancy. Variables with P<0.20 in the univariate analyses were candidates for inclusion in final multivariable logistic regression models for unintended pregnancies. When multiple covariates measured similar phenomena, the variable representing each construct with the most statistical significance was chosen.
We carried out an additional analysis of interest to determine the level of happiness with the participants' last pregnancy analysed by whether the pregnancy was intended or unintended, including all pregnancies with an a priori hypothesis that HIV status at the time of the pregnancy and ethnicity may be predictors of happiness with an unintended pregnancy. The question used to represent the level of happiness with the participants' last pregnancy asked ‘How happy were you with being pregnant the LAST time you were pregnant?’ A five-point Likert scale was used for the answer from ‘not happy at all’ to ‘very happy’ and ‘neither happy nor unhappy’ in the middle. The Cochran–Armitage test for trend was used for the comparison of the degree of happiness with the participants’ last pregnancy based on whether it was intended or unintended. Levels of happiness according to whether or not the last pregnancy was unintended were compared among ethnic groups (African, Caribbean, European-British or French-Canadian, Aboriginal and Other). Also, univariate and multivariable logistic regression models were fitted to predict happiness with the last unintended pregnancy. Only women who indicated that their last pregnancy was unintended were included in this analysis. HIV diagnosis at the time of the pregnancy and ethnicity were included as covariates of interest to assess whether they influenced happiness with unintended pregnancy. Statistical analyses were performed using sas version 9.2 (SAS Institute, Cary, NC, USA).
Results
Participants and survey validation
A total of 504 HIV-positive women living in Ontario, Canada were recruited. Four participants did not meet the inclusion criteria (two were over the age of 52 years, and two were not living in Ontario). Fifty-nine women had never been pregnant, 13 did not answer and 12 answered ‘I don't know’ to the question used to represent unintended pregnancy. Therefore, 416 surveys were included in the final analysis. There was a small amount of missing data for a number of survey questions, resulting in different denominators for percentages and Ns used to calculate medians.
The final study sample had a median age of 38 years (IQR 33–44 years; range 18–52 years) at the time of the survey. The respondents' last pregnancy had been a median of 8 years (IQR 3–14 years) prior to the completion of the survey (n=283 for those with data). Of the 416 women included in the study, 60% (246/411) were born outside Canada, 51% (211/416) were living in Toronto, 47% (187/400) defined themselves as being of African ethnicity and 74% (303/408) were currently on ART. Twenty-five per cent (97/388) indicated that they had never been married and 51% (212/413) stated that they were in a monogamous relationship. Sixteen per cent (63/400) of our population were coinfected with HCV, consistent with Ontario statistics [15]. The results of a comparison of the geographical distribution of the study population to that of HIV-positive women living in Ontario were similar to those previously reported [14,15]. The demographic characteristics of the study population are presented in Table 1.
| Characteristics | N with valid reponses* | |
|---|---|---|
| Age (years): | 405 | 38 (33–44) |
| 18–25 | 18 (4%) | |
| 26–40 | 223 (55%) | |
| >40 | 164 (40%) | |
| Ethnic background: | 400 | |
| African | 187 (47%) | |
| Caribbean | 39 (10%) | |
| European-British/French-Canadian | 92 (23%) | |
| Aboriginal | 37 (9%) | |
| Other | 45 (11%) | |
| Born in Canada | 405 | 159 (39%) |
| Years in Canada (Not Born in Canada): | 243 | 5 (2–13) |
| Religion: | 407 | |
| Christian/Catholic/Protestant | 309 (76%) | |
| Other | 98 (24%) | |
| Sexual Orientation: | 393 | |
| Heterosexual | 348 (89%) | |
| Lesbian/Bisexual | 37 (9%) | |
| Other | 8 (2%) | |
| Work: | 405 | |
| Working | 153 (38%) | |
| On Government Assistance | 205 (51%) | |
| Marital Status: | 388 | |
| Never Married | 97 (25%) | |
| Married/Common-Law/Living with a Partner | 183 (47%) | |
| Divorced/Widowed | 108 (28%) | |
| Education: | 362 | |
| Less than High School | 116 (32%) | |
| High school or higher | 246 (68%) | |
| Annual Household Income: | 354 | |
| <20K | 165 (47%) | |
| 20–40K | 107 (30%) | |
| >40K | 82 (23%) | |
| Years since Diagnosis of HIV Positive: | 367 | 7 (4–12) |
| HIV Risk Factors: | 402 | |
| Sex with men | 275 (68%) | |
| IDU | 40 (10%) | |
| Blood Transfusion/Product | 33 (8%) | |
| Vertical Transmission | 1 (0.2%) | |
| Other | 41 (10%) | |
| Unknown | 53 (13%) | |
| Hepatitis B | 400 | 13 (3%) |
| Hepatitis C | 400 | 63 (16%) |
| Recent CD4 Count (cells/μL)**: | 298 | 481 (325–700) |
| Recent Viral Load (log10copies/mL)**: | 73 | 3.9 (2.9–4.5) |
| Ever on HIV Medication: | 411 | 356 (87%) |
| In Sexual Relations: | 412 | 224 (54%) |
| Current Contraceptive Use: | 409 | 144 (35%) |
| Lifetime Pregnancies: | 411 | |
| 1 | 81 (20%) | |
| 2 | 95 (23%) | |
| ≥3 | 235 (57%) | |
| Lifetime Births: | 411 | |
| 0 | 58 (14%) | |
| 1 | 125 (30%) | |
| 2 | 94 (23%) | |
| ≥3 | 134 (33%) | |
| Ever had miscarriage/stillbirth: | 407 | 135 (33%) |
| Ever had VPT: | 410 | 197 (48%) |
- Continuous variables are summarized with medians and interquartile range; categorical variables are summarized with N (%). Denominator varied slightly for each characteristic, missing data was generally <5%; lower response rates occured for education, income, years since HIV diagnosis, CD4 count and viral load.
- IDU, injection drug use; VL, viral load; VPT, voluntary pregnancy termination.
- * N with valid responses acting as denominator for % and N for calculating medians.
- ** Recent CD4 cell count and VL refers to self-reported values closest to the time of the survey.
Pregnancy and birthing history
For the 416 respondents, the median number of pregnancies was 3 (IQR=2–4). Eighty-three per cent of women (339/410) were pregnant before their HIV diagnosis, with a median number of 2 (IQR 2–4) pregnancies. Forty-seven per cent of women (195/411) had been pregnant after their HIV diagnosis, with a median number of 1 (IQR 1–2) pregnancy. More women were pregnant before their HIV diagnosis than after (P<0.0001). The pregnancy history of the sample is presented in more detail in Figure 1.
Number of times the participants had given birth ever in their lifetime and before and after HIV diagnosis. Participants were asked the question ‘How many times have you EVER given birth in your lifetime?’ and whether the birth was before or after their HIV diagnosis. Five women had not answered the number of times they had been pregnant and given birth; therefore the N used in the denominator was 411 for these analyses. For women who were pregnant before HIV diagnosis, the N used for the denominator was 410. Three hundred and fifty-three (86%) of the participants had given birth at some time in their life and the median number of children was 2 [interquartile range (IQR) 1–3]; 78% gave birth to at least one child before HIV diagnosis with a median of 2 children (IQR 1–3) and 42% gave birth to at least one child after HIV diagnosis with a median of 1 child (IQR 1–2). More women gave birth before their HIV diagnosis than after (P<0.0001).
Three hundred and fifty-three (86%) of 411 respondents had previously given birth. Of 410 respondents, 197 (48%) had a voluntary pregnancy termination (VPT) and 135 of 407 (33%) had a spontaneous abortion or stillbirth. For those women who had given birth, the median number of children was 2 (IQR 1–3); 78% (274/353) gave birth to at least one child before HIV diagnosis, with a median of 2 children (IQR 1–3), and 42% (149/353) gave birth to at least one child after HIV diagnosis, with a median of 1 child (IQR 1–2). More women gave birth before their HIV diagnosis than after (P<0.0001). Birthing histories are presented in Figure 1.
Unintended pregnancies
Of the 416 participants, 233 (56%; 95% CI 51–61%) indicated that their last pregnancy was unintended. Of the 195 participants who were pregnant after their HIV diagnosis, 106 (54%) of their last pregnancies were unintended. Of the 216 participants who were only pregnant before being diagnosed with HIV, 124 (57%) of their last pregnancies were unintended (Fig. 2). The proportions of unintended pregnancies before and after HIV diagnosis were similar (P=0.53). The overall proportion of unintended pregnancies was higher than the 30% reported in the general Ontario population in 2008 and the 49% reported in the general U.S. population in 2001 (P<0.0001 and <0.01 by binomial proportion test, respectively).
Proportion of participants whose last pregnancy was unintended ever in their lifetime and before and after HIV diagnosis. Participants were asked the question ‘Was your last pregnancy planned?’ The variable was dichotomized into ‘unintended pregnancy’ if the answer was ‘No’ and ‘intended pregnancy’ if the answer was ‘Yes’. They were asked if the pregnancy was before or after their HIV diagnosis. Of 416 participants, 233 [56%; 95% confidence interval (CI) 51–61%] indicated that their last pregnancy was unintended ever in their lifetime. Of the 195 participants who were pregnant after their HIV diagnosis, 106 (54%) of their last pregnancies were unintended. Of the 216 participants who were pregnant before being diagnosed with HIV, 124 (57%) of their last pregnancies were unintended.
The results from the univariate and multivariable logistic regression modelling revealed that significant correlates of unintended pregnancy after HIV diagnosis in our cohort of HIV-positive women were never having been married and never having given birth (Table 2). Covariates with unadjusted odds ratios <0.67 or >1.5 for unintended pregnancy lacking statistical significance included ethnic background, years in Canada, education level, HIV risk factor, HBV or HCV coinfection. Covariates with no significant impact on unintended pregnancies included age, religion, sexual orientation and duration of HIV diagnosis.
| Characteristic | Unintended pregnancies* | Unadjusted | Adjusted | |||
|---|---|---|---|---|---|---|
| Yes (n=106) | No (n=89) | Odds ratio for unintended pregnancy (95% CI) | P-value | Odds ratio for unintended pregnancy (95% CI) | P-value | |
| Age | ||||||
| 18–25 years | 9 (9%) | 6 (7%) | 1 | |||
| 26–40 years | 66 (63%) | 62 (71%) | 0.71 (0.24, 2.11) | 0.54 | ||
| >40 years | 29 (28%) | 19 (22%) | 1.02 (0.31, 3.32) | 0.98 | ||
| Ethnic background | ||||||
| African | 45 (44%) | 43 (49%) | 1 | |||
| Caribbean | 12 (12%) | 7 (8%) | 1.64 (0.59, 4.55) | 0.34 | ||
| European-British/French-Canadian | 25 (24%) | 16 (18%) | 1.49 (0.70, 3.17) | 0.30 | ||
| Aboriginal | 13 (13%) | 7 (8%) | 1.77 (0.65, 4.87) | 0.27 | ||
| Other | 8 (8%) | 14 (16%) | 0.55 (0.21, 1.43) | 0.22 | ||
| Not born in Canada | 60 (57%) | 55 (62%) | 0.81 (0.45, 1.43) | 0.46 | ||
| Years in Canada (for those not born in Canada) (odds ratio per additional year) | 8 (3–14) | 5 (2–10) | 1.05 (1.00, 1.10) | 0.07 | ||
| Religion | ||||||
| Christian, Catholic or Protestant | 75 (74%) | 70 (79%) | 0.75 (0.39, 1.48) | 0.41 | ||
| Other | 27 (26%) | 19 (21%) | 1 | |||
| Sexual orientation | ||||||
| Heterosexual | 91 (89%) | 76 (90%) | 1 | |||
| Lesbian, bisexual or other | 11 (11%) | 8 (10%) | 0.87 (0.33, 2.28) | 0.77 | ||
| Marital status | ||||||
| Never married | 20 (20%) | 19 (22%) | 1 | 1 | ||
| Married, common-law partner or living with a partner (neither married nor common-law) | 46 (47%) | 55 (65%) | 0.29 (0.13, 0.63) | <.01 | 0.35 (0.15, 0.78) | 0.01 |
| Divorced/widowed | 20 (20%) | 19 (22%) | 0.36 (0.14, 0.92) | 0.03 | 0.49 (0.19, 1.28) | 0.15 |
| Education: high school or higher | 59 (65%) | 64 (76%) | 0.58 (0.30, 1.12) | 0.10 | ||
| Years since HIV diagnosis | 9.0 (5.0–12.0) | 7.0 (5.0–12.0) | 1.00 (0.95, 1.06) | 0.87 | ||
| HIV risk factor | ||||||
| Blood transfusion/product | 8 (8%) | 5 (6%) | 0.50 (0.16, 1.60) | 0.25 | ||
| Hepatitis B virus coinfected | 5 (5%) | 1 (1%) | 4.48 (0.51, 39.09) | 0.17 | ||
| Hepatitis C virus coinfected | 15 (15%) | 8 (9%) | 1.72 (0.69, 4.28) | 0.24 | ||
| Fertility history | ||||||
| Lifetime pregnancies ≥2 | 86 (81%) | 78 (88%) | 0.61 (0.27, 1.35) | 0.22 | ||
| Lifetime births ≥1 | 50 (48%) | 58 (65%) | 0.19 (0.06, 0.57) | <.01 | 0.22 (0.07, 0.69) | 0.01 |
- CI, confidence interval.
- * Continuous variables are summarized as the median and interquartile range; categorical variables are summarized as n (%).
- † Marital status and at least one lifetime birth were the variables retained in the multivariate analysis.
Level of happiness with last pregnancy
When asked about their level of happiness with their last pregnancy, 92% (168/183) of those with intended pregnancies reported being happy or very happy compared with only 49% (113/230) of those whose last pregnancy was unintended (P<0.0001) (Table 3). The rates of happiness were similar between women who were HIV positive and HIV negative at the time of their last pregnancy, whether it was intended [93% (83/89) vs. 90% (83/92), p=0.46] or unintended [46% (48/125) vs. 51% (63/123), p=0.41]. When level of happiness and intention of last pregnancy were assessed in women of different ethnic backgrounds, only 43% (38/89) of African women were found to be happy or very happy with the last unintended pregnancy compared with 93% (88/95) who had an intended pregnancy (P<0.0001). Similar findings were noted with the other ethnic groups.
| Was your LAST pregnancy planned? | How happy were you with being pregnant the LAST time you were pregnant? | |||||
|---|---|---|---|---|---|---|
| Not happy at all | Not happy | Neither | Happy | Very happy | P-value* | |
| Overall cohort | ||||||
| Unintended pregnancy | 48 (21%) | 10 (4%) | 59 (26%) | 25 (11%) | 88 (38%) | <.0001 |
| Intended pregnancy | 2 (1%) | 1 (1%) | 12 (7%) | 13 (7%) | 155 (85%) | |
| Ethnic background | ||||||
| African women | ||||||
| Unintended pregnancy | 26 (29%) | 3 (3%) | 22 (25%) | 6 (7%) | 32 (36%) | <.0001 |
| Intended pregnancy | 1 (1%) | 0 | 6 (6%) | 6 (6%) | 82 (86%) | |
| Caribbean women | ||||||
| Unintended pregnancy | 3 (12%) | 1 (4%) | 6 (24%) | 3 (12%) | 12 (48%) | 0.54 |
| Intended pregnancy | 1 (7%) | 0 | 4 (29%) | 1 (7%) | 8 (57%) | |
| European-British/French-Canadian | ||||||
| Unintended pregnancy | 13 (23%) | 3 (5%) | 18 (32%) | 6 (11%) | 16 (29%) | <.0001 |
| Intended pregnancy | 0 | 1 (3%) | 2 (6%) | 2 (6%) | 31 (86%) | |
| Aboriginal women | ||||||
| Unintended pregnancy | 2 (8%) | 0 | 7 (28%) | 6 (24%) | 10 (40%) | 0.01 |
| Intended pregnancy | 0 | 0 | 0 | 2 (17%) | 10 (83%) | |
| Other women | ||||||
| Unintended pregnancy | 3 (13%) | 2 (8%) | 3 (13%) | 3 (13%) | 13 (54%) | 0.004 |
| Intended pregnancy | 0 | 0 | 0 | 2 (10%) | 19 (90%) | |
- Participants were asked about the level of happiness with their last pregnancy using a five-point Likert scale. We analysed this based on the last pregnancy being unintended or intended for the overall cohort and by different ethnic groups. Ethnicities in the ‘Other’ category (n=45) included Asian (13%), Hispanic (16%), Middle Eastern (4%) and other (67%).
- * Cochran–Armitage trend test.
The results from the multivariable analysis revealed that women who were happy with their last unintended pregnancy were more likely to be married or have a common-law partner and have given birth at least once (Table 4). HIV status at the time of pregnancy and ethnicity were not significant predictors of happiness with last unintended pregnancy.
| Covariates† | Unadjusted | Adjusted‡ | ||
|---|---|---|---|---|
| Odds ratio (95% CI) | P-value | Odds ratio (95% CI) | P-value | |
| Age | ||||
| 18–25 years | 1 | |||
| 26–40 years | 0.97 (0.30, 3.17) | 0.96 | ||
| >40 years | 0.89 (0.27, 2.98) | 0.85 | ||
| HIV positive | 0.80 (0.48, 1.35) | 0.41 | 0.74 (0.41–1.35) | 0.33 |
| Ethnic background | ||||
| African (n=92) | 1 | 1 | ||
| Caribbean (n=25) | 2.01 (0.82, 4.97) | 0.13 | 1.84 (0.67, 5.05) | 0.24 |
| European-British/French-Canadian (n=56) | 0.87 (0.44, 1.72) | 0.68 | 0.77 (0.37, 1.63) | 0.50 |
| Aboriginal (n=25) | 2.39 (0.95, 5.98) | 0.06 | 1.74 (0.65, 4.67) | 0.27 |
| Other§ (n=24) | 2.68 (1.04, 6.92) | 0.04 | 2.55 (0.87, 7.44) | 0.09 |
| Marital status | ||||
| Never married | 1 | 1 | ||
| Married/common-law | 2.25 (1.20, 4.24) | 0.01 | 2.01 (1.01, 4.00) | 0.05 |
| Divorced/widowed | 1.33 (0.63, 2.79) | 0.45 | 1.14 (0.50, 2.57) | 0.76 |
| Hepatitis C virus coinfected | 2.24 (1.10, 4.57) | 0.03 | ||
| Fertility history | ||||
| Lifetime pregnancies ≥2 | 2.23 (1.17, 4.24) | 0.01 | ||
| Lifetime births ≥1 | 4.43 (2.13, 9.22) | <.0001 | 3.79 (1.73, 8.31) | <.001 |
- CI, confidence interval.
- * Outcome was defined as ‘happy’ or ‘very happy’ with last pregnancy.
- † Covariates presented in this table include those deemed to be correlated at the times of the pregnancy and the survey.
- ‡ HIV status at the time of the last pregnancy, ethnic background, marital status and number of lifetime births were the variables retained in the multivariable model, as the two former covariates were the a priori covariates of interest and the latter two were statistically significant.
- § Ethnicities in the ‘Other’ category (n=24) included: Asian (4%), Hispanic (13%) and other (83%).
Discussion
In this study of 416 HIV-positive women of reproductive age living in Ontario, Canada, we documented an unintended pregnancy rate of 56% (95% CI 51–61%) for their most recent pregnancy; this proportion was similar before and after HIV diagnosis. This proportion is also similar to those presented in other international reports identifying unintended pregnancy rates in HIV-positive women [7,9]. Gogna et al. [7] found that 55% of women and 30% of men in their study had children after their HIV diagnosis and that half of those pregnancies had been unintended. Our study expands on these findings by exploring the correlates of unintended pregnancy in this population and by examining the degree of happiness with unintended pregnancies. Koenig and colleagues' finding that 83.3% of the pregnancies in HIV-positive adolescent girls were unplanned is of significant importance as the HIV epidemic increasingly affects younger individuals and women [8,17,18]. This is a group at significant risk of HIV infection and of unintended pregnancy, and these findings highlight the importance of public health programmes targeting these vulnerable adolescent girls [17,18].
We also concluded that the unintended pregnancy rate of 56% in our population was significantly higher than the rate in the U.S. and Ontario general populations (49 and 30%, respectively) [10,13]. Finer elegantly showed, in the 2002 National Survey of Family Growth, that unintended pregnancies resulted in higher rates of abortion (42%) but lower rates of fetal loss (14%) compared with those with intended pregnancies (0% abortion rate, 20% fetal loss) [10]. Finer also assessed correlates of unintended pregnancies and found that Black and Hispanic women had more unintended pregnancies than White women. Other significant correlates of unintended pregnancies were lower age, being unmarried, lower income and less education (with those having only a high school education being the group with the highest unintended pregnancy rate).
The significant correlates of unintended pregnancy after HIV diagnosis in our multivariable model were never being married and having given birth to no more than one child. No other studies that we identified assessed correlates of unintended pregnancies in HIV-positive women. Understanding the sociodemographic correlates of unintended pregnancies is clinically important, allowing clinicians to target HIV-positive women at higher risk of unintended pregnancies. There were additional clinically significant sociodemographic correlates of unintended pregnancies that Ontario clinicians may want to consider that lacked statistical significance because of a lack of power, including ethnic background, years in Canada, education level, HIV risk factor, and HBV or HCV coinfection. However, we assert that pregnancy planning, family planning and contraception discussion should be part of the standard discussion with all HIV-positive women and probably also men. For all women, HIV-infected or not, unintended pregnancies are associated with increased risks of poor maternal and fetal outcomes and this is reason enough to discuss family and pregnancy planning [19]. In the setting of HIV care, it is imperative that issues related to antiretroviral and other drugs that could be teratogenic and the risk of horizontal transmission to an uninfected sexual partner are discussed, considering the high rate of unintended pregnancies in this population [20]. Current therapeutic guidelines for first-line HIV treatment recommend use of tenofovir, emtricitabine and efavirenz, which are co-formulated in a single pill taken once daily (Atripla®; Gilead Sciences Inc, Foster City, CA, USA) [21]. Although other first-line HIV treatment options are available, Atripla® is a popular regimen because of its low pill burden. However, efavirenz is known to be teratogenic, emphasizing the need to discuss pregnancy intentions and contraception as well as alternative treatment options with HIV-positive women of reproductive age who are considering HIV treatment. Reducing the occurrence of unintended pregnancies among HIV-positive women may also reduce the occurrence of VPT. A recent Italian study compared 63 cases of VPT with 334 pregnancies not ending in a VPT among HIV-positive women. The authors found a significant correlation between unintended pregnancy and VPT (odds ratio 24.3; 95% CI 5.8–101.2), leading them to conclude that improved access to pregnancy planning in the context of HIV infection could reduce the occurrence of VPT. We also found a high rate of VPT in our cohort, with 47% reporting having had a VPT at some time in their life.
A landmark piece by Wilcher and Cates [23] about reproductive choices for women living with HIV was recently published in the WHO Bulletin. They stressed that access to reproductive health services for HIV-positive women is critical to ensuring that their reproductive needs are addressed and their rights protected. In addition, they stressed that preventing unintended pregnancies in women with HIV infection is an essential component of a comprehensive vertical transmission prevention programme. They called for stronger linkages between sexual and reproductive health and HIV policies, programmes and services. Although limited thus far, such linkages are starting to be developed in several international organizations and countries, including Canada [20–28].
Unintended pregnancies are not necessarily unwanted, but could vary with certain patient characteristics. To explore this idea, we asked about the women's level of happiness with their last pregnancy and observed that 92% of those with intended pregnancies reported being happy or very happy compared with only 49% of those whose last pregnancy was unintended. Despite our original hypothesis, ethnicity played a minimal role. Therefore, not only does planning pregnancies lead to better maternal and fetal outcomes, and better HIV care, but it may have the effect of promoting happier pregnancies.
Another noteworthy finding of our study is that, of women who had given birth, 78% of women gave birth to at least one child before HIV diagnosis while only 42% of women gave birth to at least one child after HIV diagnosis. It may be useful to explore this further to determine if the discrepancy between childbirths before and after HIV diagnosis is primarily explained by age and having reached one's parental goals, or whether living with HIV and its accompanying issues, such as stigmas [29], play a role in pregnancy decision making.
The present study has a number of limitations which include missing data, such as women who responded ‘I don't know’ to important questions. The missing data might potentially be explained by the high literacy level required for the survey and the fact that most of the women in Ontario living with HIV do not have English or French as their native language. Additional questions on pregnancies and birth were considered in the development phase of the survey but deleted because of the extensive survey length. The answers to these questions might have been informative in terms of the demographics and living situations of the women at the time of the pregnancies. The specific dates of the pregnancies were not available, but only the date of the last birth of a child who was cared for by the woman; this contributed to less information on the timing of pregnancies for women whose pregnancies ended in abortion or miscarriage or whose children did not live with them than for women whose last-born child lived with them. All questions on the number and details of, and happiness with, pregnancies are impacted by recall bias, as participants self-reported on their previous pregnancies from memory. Similarly, many clinical and demographic questions were reported at the time of the survey and not at the time of the last pregnancy, limiting the analysis of their impact. The question on happiness with the last pregnancy was rather simplistic and was not adapted from validated scales. Finally, the sample population was limited to adult women ≥18 years of age, which led to the exclusion of adolescent girls who are at particular high risk of unintended pregnancies [8].
Our findings have important implications for the healthcare management of HIV-positive women which providers and policy makers should consider. Healthcare providers ought to consider adding a discussion about pregnancy planning, healthy pre-conception lifestyle, and contraception into routine HIV care to support safer pregnancies, maximizing the health of the women and their partners and protecting future children by reducing vertical transmission. In Canada, we are in the process of developing national guidelines on pregnancy planning as well as provincial and national HIV Fertility Programs [20,30,31]. We hope that our research and ongoing projects will assist HIV-positive individuals, policy makers and healthcare providers globally to develop their programmes for safer, supportive pregnancy and family planning for HIV-positive individuals in their communities.
Acknowledgements
We are indebted to the frontline AIDS Service Organization staff and research co-ordinators for their dedication to this project; to the members of the Project Advisory Committee for their expertise; and to the participants whose involvement made this study possible.
