AIDS and Behavior

, Volume 17, Issue 2, pp 471–478

Incidence of Pregnancy Among Women Accessing Antiretroviral Therapy in Urban Malawi: A Retrospective Cohort Study

Authors

    • The International Union Against Tuberculosis and Lung Disease
    • The Lighthouse Trust
  • Caryl Feldacker
    • The Lighthouse Trust
    • International Training and Education Center for Health (I-TECH)
    • University of Washington
  • Elizabeth Breeze
    • London School of Hygiene and Tropical Medicine
  • Andreas Jahn
    • International Training and Education Center for Health (I-TECH)
    • University of Washington
    • CMED/Department for HIV and AIDSMinistry of Health
  • Lisa Blake Haddad
    • Department of Obstetrics and GynecologyEmory University
  • Anne Ben-Smith
    • Maame Akua
  • Thom Chaweza
    • The Lighthouse Trust
  • Sam Phiri
    • The Lighthouse Trust
Original Paper

DOI: 10.1007/s10461-012-0150-0

Cite this article as:
Tweya, H., Feldacker, C., Breeze, E. et al. AIDS Behav (2013) 17: 471. doi:10.1007/s10461-012-0150-0

Abstract

Although previous studies investigated pregnancy rates among women on antiretroviral therapy (ART), incidence of, and factors associated with pregnancy among these women remain poorly understood. We, therefore, conducted a retrospective cohort study at a large public HIV clinic in Lilongwe, Malawi, between July 2007 and December 2010. At each clinic visit, pregnancy status was assessed. Time to event analysis was conducted using Poisson regression. Among 4,738 women, 589 pregnancies were observed. Pregnancy incidence was 9.3/100 person-years. After 6 months on ART, women on ART had similar total fertility rates to women in the urban population. In multivariable analysis, increasing age and advanced WHO clinical stage were associated with decreased probability of becoming pregnant while higher body mass index and longer time on ART were associated with increased probability of becoming pregnant. We recommend that ART clinics integrate comprehensive family planning services to address reproductive health needs among women on ART.

Keywords

Antiretroviral therapyPregnancy rateFamily planningMalawi

Introduction

Women in sub-Saharan Africa are disproportionately affected by HIV [1]. Almost 60% of the region’s HIV infections are among women aged 15–49 years [2]. Substantial progress is underway in sub–Saharan Africa to increase access to antiretroviral therapy (ART) and to prevent mother to child transmission (PMTCT) of HIV. In 2009, about 40% of women in need of ART accessed treatment in sub-Saharan Africa and approximately 54% of HIV-infected pregnant women received PMTCT services with country-level access ranging from 40 to 84% [3]. Although Malawi’s PMTCT program falls behind those of some of countries in the region, its rapid ART scale-up is positive. Approximately 60% of registered HIV-infected people on ART in Malawi, are women and PMTCT programs reached 41% of eligible women in 2010 [4]. Furthermore, the number of patients alive and on treatment reached 250,987 at the end of December 2010 compared to only 10,761 in 2004 [5]. With the Malawian government’s commitment to the national response to HIV and AIDS, the HIV prevalence rate in Malawi stabilized at 12%.

As access to ART increases throughout the region, there is need for increased understanding of issues that affect HIV-infected women in their reproductive years. ART use is associated with increased pregnancy rates [6, 7], and there is growing interest in exploring the effects of ART among pregnant women. Previous studies in sub-Saharan Africa reported conflicting findings on the associations between ART use during pregnancy and adverse pregnancy outcomes. Some studies reported increased risk of adverse pregnancy outcomes among women on ART such as low birth weight [8, 9], while others found reduced [10] or similar risk of adverse pregnancy outcomes among women on ART {Ekouevi, 2011 #276}. With a high fertility rate in Malawi, estimated at 5.5 births per woman, and low modern contraceptive usage, estimated at 41% [11], the relationship between ART and pregnancy merits further attention.

To date, there are few studies that explore pregnancy incidence or examine factors associated with pregnancy among women on ART in sub-Saharan Africa. Much of the previous literature describes fertility intentions among those who are HIV-infected and provides insight into the drivers of pregnancy [1214]. Other previous research on pregnancy rates and determinants in the region come from longitudinal studies conducted in controlled research settings. Although these studies contribute towards our understanding of these issues, they do not necessarily represent more typical public health settings as the studies were conducted in clinical research settings [15], and/or in clinics with existing comprehensive reproductive health services [6]. Additional studies were conducted before ART was readily available in the region, limiting the time women in the studies were on ART hence the ability to look at how ART affects pregnancy rates over time [16, 17]. Lastly, other previous research compared pregnancy rates of HIV-infected women to rates in the general population, but the HIV-positive women included in the study were not yet on ART [1820].

To the authors’ knowledge, no study examined pregnancy rates among those on ART in large public ART clinics nor compared these more representative rates to the pregnancy rates of the urban population in general. In response, we conducted a study to explore the incidence and potential predictors of pregnancy among women accessing ART at a large, public, urban health clinic in Lilongwe, Malawi. Furthermore, we also compare the pregnancy rates among these women on ART to the pregnancy rates of urban women in the country as a whole. As ART access increases and healthy HIV-infected women return to childbearing, recognizing predictors of pregnancy can help inform future programmes and policies to meet the specific needs of this population.

Methods

Setting

The study was conducted at the Martin Preuss Centre (MPC) ART clinic in Lilongwe, Malawi. MPC uses a real-time, touch screen-based, electronic data system (EDS) for patient management [21]. All diagnosed HIV-positive women who report to the clinic are registered in the EDS. During clinic registration, patient demographic details are entered, and patients are initiated on ART based on WHO clinical staging (stages 3–4) or CD4 cell count <250 cells/mm3. Patients are seen 2 weeks after ART initiation and then monthly for clinical assessment and drug dispensing. At each clinic visit, nurses review all patients and refer those with clinical problems to a clinician. Patient’s height and weight are recorded to determine body mass index (BMI), categorized either as normal (18.5 kg/m2 and above) or malnourished (below 18.5 kg/m2). Visit intervals are extended to 2 months if patients complete 6 months of ART without complications. CD4 cell counts are assessed every 6 months; test results are entered by trained data clerks or electronically uploaded. Outcomes for the national ART program are recorded at every visit and classified as “alive on ART”, “died”, “lost to follow up (no visit in 2 months)”, “transferred out”, and “stopped ART”. Program outcomes are ascertained through active follow-up.

Self-reported pregnancy status is ascertained at initial staging and at each follow-up visit. Women typically report pregnancy at 4 or 5 months of gestation period (unpublished data). Currently, there is no standardised counseling procedure on pregnancy and use of contraception and risks associated with pregnancy while on ART. However, nurses and clinicians provide general advice on pregnancy prevention and contraception in addition to distributing male and female condoms at every visit.

Study Design and Participants

This was a retrospective cohort study using routine programmatic data. All HIV-infected women aged 15–49 years who registered for ART at the clinic and had at least one ARV drug dispensing visit between July 2007 and December 2010 were eligible for inclusion in the study.

Statistical Analysis

Pregnancy rates were calculated based on conversion of current pregnancy status from ‘non-pregnant’ to ‘pregnant’ at subsequent ART visits. Due to the way information is captured in the EDS, we could not estimate gestation age. Incident pregnancy was defined as either an episode of pregnancy reported during ART treatment for women who were not pregnant at ART initiation, or a subsequent episode of pregnancy for women who were pregnant at ART initiation. For calculation of the person-years denominator, women were entered in the analysis at the time of first ART clinic registration (if not pregnant) or 3 months after delivery (if registered while pregnant). Women who transferred in on ART from another facility were considered to come under observation when first registered at the study clinic, but their previous time on ART was accounted for when stratifying time on ART. Observation of ‘time at risk’ of getting pregnant on ART ended either at the time of stopping ART follow-up at the study clinic (death, stopping ART, lost to follow-up, transfer to other ART facility), when reported pregnant, or at the censoring date (31st December 2010). Gaps in observation time were accounted for in women who returned after transfer to another ART facility, or who re-started ART after interrupting.

Interval pregnancy rates were calculated for duration of 0–6, 7–12, 13–24, 25–36 and ≥36 months on ART. To compare pregnancy rates in the ART cohort to pregnancy rates of urban women in the general population, we calculated the total fertility rate (TFR) of the study population and compared it to the TFR in the general population as estimated by the 2004 Malawi Demographic Health Survey (MDHS) [22], the most recent published MDHS data. We use TFR rather than standardized fertility rate because it is the most widely used measure of fertility and unaffected by age composition. TFR was defined, according to MDHS, as the number of children that would be born per woman if she were to pass through the childbearing years according to the current age-specific fertility rates (ASFR). Since ASFR were calculated per year in 5 years age categories, the TFR was calculated as: TFR = 5Σ (ASFR). The TFR of the general urban female population includes both HIV-positive and HIV-negative women.

Poisson regression models were used to investigate the primary outcome, “incidence of pregnancy”. Poisson was appropriate as it is applied to time to event data where the rates are not expected to vary rapidly as was the case in this cohort. Multivariable Poisson regression was used to model the association between pregnancy and covariates of interest. Covariates included age, WHO clinical staging, body mass index (BMI), occupation, change in CD4, illness during follow-up and time on ART. We treated age, BMI, CD4 count and illness as time-varying variables. Change in CD4 count was defined as a 300/mm3 increase between any two consecutive CD4 testing visits among the study women. Illness during follow-up was defined as any episode of the following: diarrhea, peripheral neuropathy, jaundice or skin rash. Assuming that each respective measurement applied for the period up to the next visit, the value of the time-varying variable was determined by the most recent prior measurement for BMI, age and illness. The final multivariable model was determined by including explanatory variables that were known risk factors of pregnancy or confounders of associations between other explanatory variables and pregnancy according to the literature. Unadjusted rate ratios (RR) and adjusted RR were reported, including the 95% confidence intervals (CI).

Ethics Considerations

This study was approved by the Malawi National Health Science Research committee, the Ethics committee of the London School of Hygiene & Tropical Medicine, UK and the Ethics Advisory Group of The International Union Against Tuberculosis and Lung Disease, Paris, France. As the study used routine programme data, all ethics boards provided a waiver of individual informed consent.

Results

Baseline Characteristics and Incidence of Pregnancy

Of 4,968 eligible women, 4,738 (95%) were included in the study sample. A total of 230 women were excluded from the study either because their ART follow-up time was less than 1 month or they were pregnant at ART registration but lost to follow-up before the end of 3 months after delivery. Among 4,738 women included in the study, the median age at ART initiation was 31 years [inter-quartile range (IQR 27–37)]. At ART initiation; 2,126 (45%) were in WHO clinical stage 1 or 2 with CD4 count less than 250 cells/mm3; 2,159 (45%) were in stage 3; and 453 (10%) were in stage 4. Of the 3,313 (70%) with baseline CD4 cell count data, the median CD4 count at ART initiation was 165 cells/mm3 (IQR 1–1977). Median BMI at ART initiation was 22 kg/m2 (IQR 19.8–24.4). Fifty-four percent were unemployed.

Among 4,738 women, 589 pregnancies occurred during 6,343 person-years of follow-up. Overall incidence of pregnancy was 9.3 per 100 person-years of observation time with median follow-up of 14 months (IQR 7–24) in the study (Table 1). Within the first 6 months after initiation of ART pregnancy rate was 4 (95% CI 3.40–5.28). This increased to 11 (95% CI 9.49–13.0) for months 7 through 12. Pregnancy rates remained stable at 11–12 for years 1–3 years and for those on ART beyond 3 years (12 (95% CI 10.27–13.45) 11 (95% CI 9.20–13.7) and 11 (95% CI 8.3–14.6) per 100 person-years during intervals of 13–24, 25–36 and ≥37 months after ART initiation, respectively). At 6, 12, 24 and 36 months after ART initiation, cumulative pregnancy rates were 2% (95% CI 1.31–2.19), 7% (95% CI 5.89–7.66), 17% (95% 15.64–18.67) and 25% (95% CI 23.69–27.97), respectively.
Table 1

Baseline characteristics of women who accessed ART at Martin Preuss Centre between July 2007 and December 2010 (n = 4,760)

Characteristic

Women who became pregnant

Women who did not become pregnant

P value

n

%

n

%

All women

589

12

4,149

88

 

Age at ART initiation

 15–19

17

3

99

2

<0.001

 20–24

98

17

487

12

 25–29

223

38

1,089

26

 30–34

153

26

1,094

26

 35–39

64

11

724

17

 40–44

30

5

417

10

 45–49

4

<1

239

6

WHO clinical stage

 1 or 2

322

55

1,804

43

<0.001

 3

252

43

1,907

46

 4

15

2

438

11

Body mass index (kg/m2)a

 <18.5

43

7

570

14

<0.001

 ≥18.5

542

93

3,544

86

CD4 count (cells/mm3) at ART initiationb

 <250

82

18

396

14

0.038

 ≥250

385

82

2,450

86

Occupation

 Unemployed

366

62

2,170

52

<0.001

 Business/wages

93

16

634

15

 Professional

18

3

171

4

 Other employment

112

19

1,174

28

aBody mass index = (weight/(height/100)2)and 39 had missing BMI

b1,426 had missing baseline CD4

Comparison of Pregnancy Rates between Women Accessing ART and Urban Women in the General Population

After 6 months on ART, women on ART had similar age-specific pregnancy rates to those observed among urban women in the general population (Fig. 1). Overall, the TFR for women on ART (TFR 3.1) was lower than for women in the general urban population (TFR 4.2). However, women who were on ART for at least 6 months had a TFR of 3.9, similar to that of the general urban population. Figure 2 shows changes in pregnancy rates by age category after ART initiation.
https://static-content.springer.com/image/art%3A10.1007%2Fs10461-012-0150-0/MediaObjects/10461_2012_150_Fig1_HTML.gif
Fig. 1

Comparison of age-specific fertility rates between women on ART and women in the general urban population aged 15–49 years

https://static-content.springer.com/image/art%3A10.1007%2Fs10461-012-0150-0/MediaObjects/10461_2012_150_Fig2_HTML.gif
Fig. 2

Comparison of age-specific fertility rates over time among women aged 15–49 years

Factors Associated with Incidence of Pregnancy

In univariable analysis, current age, occupation at ART initiation, WHO stage at ART initiation, illness, and time on ART were associated with incidence of pregnancy (Table 2). Apart from in women aged 15–19 years, there was a decreasing likelihood of pregnancy with increasing age. The probability of becoming pregnant was also higher in women who were in WHO clinical stage 1 or 2 compared to those who were in WHO clinical stage 3 (RR = 1.37; 95% CI 1.16–1.61). Women in WHO clinical stage 4 had lower probability of becoming pregnant than women in stage 3 (RR = 0.25; 95% CI 0.15–0.43). Patients with BMI below 18.5 kg/m2 were 61% less likely to become pregnant than women with a BMI of 18.5 kg/m2 and above. Women who reported illness during follow-up time were 22% less likely to become pregnant than women who did not report any illness. The probability of becoming pregnant was also higher among women on ART for at least 6 months compared to women on ART for less than 6 months (RR = 3.25; 95% CI 2.47–4.28). A 300 cell count increase in CD4 cell count was not associated with rate of pregnancy (RR = 1.29; 95% CI 0.94–1.77).
Table 2

Factors associated with incidence of pregnancy among women accessing ART at Martin Preuss centre

Characteristics

Pregnancies/person-years

Unadjusted rate ratio (95% CI)

P valuea

Adjustedb rate ratio (95% CI)

P valuea

Current age category

  

<0.001

 

<0.001

 15–19

8/854

0.93 (0.46–1.89)

 

1.22 (0.59–2.48)

 

 20–24

71/473

1.50 (1.14–0.97)

 

1.61 (1.22–2.12)

 

 25–29

190/1,430

1.32 (1.08–1.62)

 

1.31 (1.07–1.62)

 

 30–34

181/1,803

1.00 (Reference)

 

1.00 (Reference)

 

 35–39

95/1,314

0.72 (0.56–0.92)

 

0.72 (0.56–0.92)

 

 40–44

32/783

0.41 (0.28–0.59)

 

0.41 (0.28–0.60)

 

 45–49

9/453

0.20 (0.10–0.39)

 

0.21 (0.10–0.40)

 

WHO clinical stage at ART initiation

  

<0.001

 

<0.001

 1 or 2

322/2,759

1.37 (1.16–1.61)

 

1.22 (1.03–1.44)

 

 3

252/2,950

1.00 (Reference)

 

1.00 (Reference)

 

 4

15/693

0.25 (0.15–0.43)

 

0.23 (0.14–0.39)

 

Body mass index during follow-up (kg/m2)c

 <18.5

15/398

0.39 (0.24–0.66)

<0.001

0.46 (0.27–0.77)

<0.001

 18.5+

574/6,000

1.00 (Reference)

 

1.00 (Reference)

 

Occupation at ART initiation

  

<0.001

 

0.130

 Unemployed

366/3,467

1.00 (Reference)

 

1.00 (Reference)

 

 Business/wages

93/1,160

0.76 (0.60–0.95)

 

0.93 (0.73–1.17)

 

 Professional

18/348

0.49 (0.31–0.79)

 

0.62 (0.38–1.00)

 

 Other

112/1,426

0.74 (0.60–0.92)

 

0.84 (0.67–1.04)

 

Illness during follow-up

 Yes

132/1,745

0.78 (0.64–0.94)

0.008

0.84 (0.68–1.04)

0.376

 No

454/4,656

1.00 (Reference)

 

1.00 (Reference)

 

Time on ART

 6 months

57/1,655

3.25 (2.47–4.28)

<0.001

3.62 (2.75–4.77)

<0.001

 <6 months

532/4,747

1.00 (Reference)

 

1.00 (Reference)

 

aP value for likelihood ratio test

bAdjusted for current age, WHO clinical stage, body mass index, occupation, illness during follow-up and time on ART

cBody mass index = (weight/(height/100)2)

In multivariable analysis, incidence of pregnancy was significantly associated with current age, WHO clinical stage at ART initiation, BMI during follow-up, and time on ART (Table 2). No significant association was found between incidence of pregnancy and either occupation or illness at follow-up.

Discussion

This is one of the first studies providing empirical data on fertility for women on ART in sub-Saharan Africa, demonstrating a high overall pregnancy rate of 9.3 per 100 person-years among women on ART. After 6 months on ART, fertility in our study cohort reverted to that of the average urban population in Malawi: the TFR in the ART cohort was 3.9 as compared to 4.2 among urban women population. Incidence of pregnancy was negatively associated with both increasing age and WHO clinical staging at ART initiation.

Our study confirmed previous findings that improved health status among women on ART is associated with increased fertility. We found lower pregnancy rates among women with advanced WHO clinical stage and among malnourished women. Complementing this, we found that time on ART was a strong predictor of becoming pregnant: women on ART for more than 6 months were three times more likely to become pregnant than their peers on ART for shorter time periods. These results are consistent with previous studies that showed lower pregnancy rates among women with advanced AIDS-related symptoms [23, 24] and increased fertility among HIV-infected women whose health returned while on ART [25, 26]. Similar to other studies [27, 28], we found that young women were at high risk of becoming pregnant.

Although the high rates of pregnancy among both women on ART and those in the general population are of interest, the complications and consequences of pregnancy to women on ART require special attention. Comprehensive, patient-focused sexual and reproductive health services are needed for women on ART as they return to fertility levels of their non HIV-infected peers. It is imperative that we address both the need for contraceptives for women who desire to prevent pregnancy while also helping reduce the risks associated with pregnancy for women on ART who wish to become pregnant. For all women on ART, counseling about pregnancy, family planning, PMTCT, and risk reduction to avoid transmitting sexually transmitted infections (STIs), including HIV, in combination with condom promotion is critical to protect their health throughout their reproductive years.

Although we highlight an increase in fertility on ART, we do not know if these were intended or unintended pregnancies. Chronic illness can alter reproductive physiology leading to reduced fertility [23]. This process can be reversed with ART and improved health status, returning fertility to those were not at previous risk of pregnancy [7]. For those women on ART who wish to delay or prevent pregnancy, expanded family planning services, including an appropriate method mix and comprehensive counseling, should be made available. Although the proportion of unwanted pregnancy among ART women in our study is unknown, according to the Malawi Demographic Health Survey (MDHS) about 20% of pregnancies in Malawi are unwanted [22]. MPC currently provides family planning counseling; however, provision of contraceptives is currently limited to male and female condoms which are sometimes not used at all. Despite availability and education, condom use remains limited; even among discordant couples seeking family planning, condoms are not consistently used [29]. Efforts to expand access to reproductive health services are currently underway with a specific focus on long term reversible contraceptive methods. The new Malawi National ART Guidelines launched in July 2011, include promotion of provider-initiated family planning services focused on depo-provera, improving access to contraceptives for women on ART across the country [30].

Some studies in sub-Saharan Africa note that a substantial proportion of HIV-infected women desire to have more children [17, 31, 32], partly reflecting the cultural significance of motherhood and family life [33]. In such contexts, provision and promotion of family planning services to avoid or delay pregnancy may be insufficient to address the reproductive health needs of women on ART. Although critical for women who want to get pregnant, several steps may help reduce transmission risk to infants and partners among all women on ART. First, in low resource setting such as Malawi, counseling couples about timing unprotected intercourse to coincide with the most fertile periods, and using family planning methods outside these times, may decrease the number of unprotected sex acts [34]. As unprotected sex increases risk of HIV transmission, especially among discordant couples with low CD4 count [35], reducing the frequency of unprotected sex would reduce HIV transmission risk among discordant couples [34]. Second, unprotected sex can lead to transmission of other STIs that can worsen HIV infection [36, 37]; therefore, inclusion of screening and pre-treatment for STIs could prevent other STIs and protect the health and wellness of women on ART. Third, discordant couples must be counseled with appropriate information and strategies including early ART initiation to reduce HIV transmission risk [38].

Our results should be viewed in light of the following limitations. First, since pregnancies were self-reported, an underestimate of true pregnancies or delay in reporting might have occurred. Delays in reporting might lead to misclassification of period within which the pregnancy occurred. Second, some pregnancies in the ART cohort may have ended in miscarriage or spontaneous abortion; pregnancy rate estimates in the MDHS were based on births. Third, the ART cohort has few patients below the age of 20 years as this is lower than the average age of HIV infection in Malawi. This might have decreased the power to detect meaningful differences among younger age groups. Lastly, several variables to measure established factors associated with childbearing were not available for analysis, including marital status, parity, gravity, death of a child, women’s education level, cultural and religious beliefs, sexual behavior, fertility intention and contraceptive practices. Despite these limitations, our findings likely better reflect the reality of urban women on ART in sub-Saharan Africa than previous studies as we use patient-level, longitudinal data from a large sample of women attending a public ART clinic. We believe these strengths outweigh the limitations.

Conclusion

Although improved access to ART is a success story in the region, it is time to move the conversation forward to a more holistic approach to the health and reproductive needs of women on ART. Our findings suggest the need to expand from efforts concentrated on improving access to ART to raising awareness of the growing need for comprehensive, reproductive health services for women already on ART. A multifaceted approach to increase access to, and use of, contraceptives among women on ART in Malawi is warranted. Improving access to both ART and integrated family planning services could decrease both unwanted pregnancies and transmission of other STIs, while reducing risks for women who wish to become pregnant, leading to reduced morbidity and mortality of women on ART.

Acknowledgments

We are grateful to all the clinic staff at Martin Preuss Centre who collected data. We thank numerous donors supporting the Lighthouse and the Malawi national ART program. Hannock Tweya is supported as an Operational Research Fellow by the International Union Against Tuberculosis and Lung Disease, Paris, France.

Copyright information

© Springer Science+Business Media, LLC 2012