Background
Postpartum depression is an important health problem for many new mothers [
1]-[
3], and has negative physical and psychological consequences for infants as well [
4]-[
9]. The prevalence of postpartum depression is estimated to be between 10%-15% in developed countries [
10], and similar or even higher rates have been reported for low- and middle-income countries, including countries in sub-Saharan Africa [
11]-[
18]. Many sub-Saharan countries also have high rates of HIV infection among pregnant and childbearing women [
19]. Although there have been efforts to improve care for women living with HIV in sub-Saharan Africa, postpartum depression among HIV-positive mothers remains understudied.
Pregnant women living with HIV may experience numerous additional stressors, including financial hardships [
20],[
21], reduced social support [
22], and concern for the physical wellbeing of their children [
23]-[
25]. These additional stressors may increase the risk for postpartum depression symptoms among HIV-positive women [
14],[
15],[
17],[
18]. Furthermore, women who discover their HIV-positive status during pregnancy may be more likely to develop depressive symptoms than those who were already aware of their HIV-positive status before pregnancy [
24].
HIV-related stigma may be another important factor contributing to depressive symptoms for pregnant women living with HIV. Stigma and discrimination around HIV are common in sub-Saharan Africa, and have been associated with adverse health outcomes. Specifically, among pregnant women, HIV-related stigma has repeatedly been shown to be a major barrier to HIV testing and counseling [
26], utilization of labor and delivery services [
27], and participation in prevention of mother-to-child transmission (PMTCT) programs [
26],[
28]-[
30]. HIV-related stigma has also been found to be associated with depression in the general population of persons living with HIV [
31]-[
41]. However, there is very little research examining stigma and depression simultaneously among pregnant or postpartum women in developing countries. In one cross-sectional study in South Africa, researchers found that experienced stigma (having experienced actual acts of discrimination) and internalized stigma (when persons living with HIV internalize negative attitudes perceived to be associated with HIV, resulting in feelings of low self-worth) were strong predictors of postpartum depression among HIV-positive women [
14]. Similarly in Thailand, researchers found that over 75% of HIV-positive postnatal women reported that they felt that HIV is a disease of which their family would be ashamed, and women who reported this familial shame were significantly more likely to report depressive symptoms and HIV-related worry during the postpartum period [
23].
While depression and stigma may impede linkage to HIV care, as well as adherence to antiretroviral therapy (ART) [
42], it is also possible that linkage to HIV care, including receipt of ART, impacts both HIV-related internalized stigma and depression. One potential mechanism for this effect may be that linkage to HIV care may change perceptions related to the implications of having HIV. It has been suggested that as perceived treatment efficacy increases for stigmatized illnesses, social stigma may become less intense [
43]. This would likely apply to persons living with HIV as well: Individuals linked to HIV care may have more positive attitudes regarding their HIV, and experience less internalized stigma, which may lead to higher self-esteem and feelings of being in control, which in turn may predict lower depression symptoms. It is also possible that linkage to care leads to lower stigma and depression by improving the physical well-being of people living with HIV [
44].
Research examining the relationship between linkage to care, stigma, and depression in the general population of persons living with HIV has mostly focused on one component of HIV care: access to ART. A 2009 systematic review of 21 publications (some using longitudinal designs around the time when ART was first made available) examining the impact of ART on various outcomes in developing countries revealed that individuals receiving ART experience less depressive symptoms than those not receiving ART, even after controlling for physical health improvements [
45]-[
49]. The association between ART and internalized stigma is less consistent. A few studies of stigma and ART in African countries have suggested that ART is associated with improved self-image and lower stigma among persons living with HIV [
44],[
50]-[
53]. The rollout of ART has had significant impact on health in Sub-Saharan Africa, and recent work has suggested that the ability to work and earn a living increases following ART initiation [
54],[
55], resulting in lower HIV-related stigma [
56]. However, other studies in sub-Saharan Africa have presented conflicting results about the extent to which ART predicts lower HIV-related stigma [
57]-[
60]. Furthermore, these relationships have not been examined in pregnant or postpartum women living with HIV.
It is important to note that linkage to comprehensive “HIV care and treatment” extends beyond access to ART medications, especially for pregnant women living with HIV. However, few studies have investigated linkage to HIV care itself in terms of impacts on mental health and internalized HIV stigma. Many HIV care facilities in resource-limited settings are designed to provide additional services such as prevention education, family planning counseling, infant feeding counseling, food by prescription, couples HIV counseling and testing, and the opportunity to participate in support groups. Understanding the unique roles of early linkage to care and ART initiation for pregnant women living with HIV has particular relevance as these countries move towards Option B+, a strategy which recommends early initiation of life-long ART for all pregnant and postpartum women testing HIV-positive [
61]. Despite the push for option B+, currently in many countries only those pregnant women meeting certain eligibility criteria are offered ART.
In this study, we investigated the associations among linkage to HIV care, depression, and internalized stigma in a population highly vulnerable to stigma and depression: newly diagnosed HIV-positive pregnant women in rural sub-Saharan Africa (women who tested HIV-positive for the first time at their first antenatal care visit). We hypothesized that linkage to HIV care after testing HIV-positive during pregnancy would be related to lower odds of having depression (and lower depressive symptoms) and lower internalized stigma at six weeks postpartum. Furthermore, we expected that the effects of linkage to HIV care on depression symptoms would be partially explained by internalized stigma. Finally, we examined effects on postpartum depression symptoms for women linked to Pre-ART (linked to HIV care but not on ART yet) versus ART care.
Results
Descriptive statistics for the main study variables are presented in Table
1. Of the 135 women included in the analyses, 34 (25.2%) were depressed according the EPDS criteria of a score of ≥ 13. Fifty-six participants had not linked to HIV care, whereas 79 participants had linked to HIV care (58.5%). As expected, depression scores were significantly correlated with internalized stigma scores (
r = .56,
p < .01).
Table 1
Characteristics of the sample (N = 135 HIV positive women)
Socio-demographic characteristics
| |
Mean age | 24.26 (4.87) |
Marital status (%) | |
Not currently married | 12.6 |
Currently married | 87.4 |
Educational level (%) | |
Primary or less | 88.1 |
More than primary | 11.9 |
Household goods (%) | |
Owns a mobile phone | 41.5 |
Does not own a mobile phone | 58.5 |
HIV-related characteristics
| |
Mean HASI-P Internalized HIV-related Stigma score | 0.64 (0.81) |
Linked to HIV care (%) | |
Yes | 58.5 |
No | 41.5 |
Family knows about HIV status (%) | |
No | 82.2 |
Yes | 17.8 |
Pregnancy-related characteristics
| |
Partner communication about birthplace (%) | |
No discussion of delivery with partner before the birth | 32.6 |
Discussion of delivery with partner | 67.4 |
Pre/postpartum intimate partner violence (%) | |
Yes | 28.1 |
No | 31.9 |
Unknown | 40.0 |
Mean number of births | 2.10 (1.61) |
Infant health status (%) | |
No serious health problems after birth | 80.7 |
Serious health problems after birth | 19.3 |
Mean postpartum depression score (EPDS) | 7.94 (5.59) |
EPDS category (%) | |
EPDS ≥13 possible depression | 25.2 |
EPDS <13 | 74.8 |
Linking to HIV care and postpartum depression
Non-linkage to HIV care was a significant predictor of postpartum depression symptoms: Women who had not linked to HIV care had higher levels of depression compared to women who had linked to HIV care (See Table
2 for statistical results). Other significant predictors of postpartum depressive symptoms included intimate partner violence (women experiencing partner violence and women who had missing data on violence had higher levels of depression), disclosure (women whose HIV status is known to someone in their family had higher levels of depression), number of births (women with fewer births had higher levels of depression), infant’s health problems (women whose babies had health problems had higher levels of depression), and partner communication about the birth place (women who did not discuss the location of the upcoming birth with their partner had higher levels of depression).
Table 2
Multivariable generalized estimating equation analysis predicting postpartum depression severity (N = 135)
Age | .08 (0.14) | .562 |
Marital status (ref: not currently married) | | |
Currently married | −0.01 (0.94) | .996 |
Educational level (ref: more than primary) | | |
Primary or less | −1.82 (1.76) | .301 |
Mobile phone ownership (ref: yes) | | |
No | -.59 (1.05) | .576 |
Pre/postpartum intimate partner violence (ref: none) | | |
Yes | 4.67 (0.84) | .000 |
Unknown | 2.26 (0.77) | .003 |
Infant health problems after birth (ref: no) | | |
Yes | 3.47 (0.72) | .000 |
Number of births | −0.66 (0.28) | .019 |
Family knows about HIV status (ref: no) | | |
Yes | 1.35 (0.66) | .040 |
Spousal communication about birthplace (ref: no) | | |
Yes | −2.31 (0.61) | .000 |
Linked to care (ref: no) | | |
Yes | −1.85 (0.91) | .043 |
Next, we repeated the GEE analyses using the dichotomized depression variable (EPDS ≥ 13, using binary logistic models). Linkage to care was significantly associated with lower odds of depression (odds ratio (OR) = 0.33, 95% CI = 0.16 - 0.69). Thus, this analysis also indicated that women who had not linked to HIV care were more likely to have symptoms consistent with a diagnosis of postpartum depression compared to women who had linked to HIV care.
Linkage to care and internalized stigma
Linkage to HIV care was also negatively associated with internalized stigma. Women who had not enrolled in care had higher levels of stigma at the time of the postpartum interview, compared to women who had enrolled, controlling for covariates (see Table
3 for statistical details). Other significant predictors of internalized stigma were intimate partner violence (women experiencing partner violence and women who had missing data on violence had higher levels of stigma), disclosure (women whose HIV status is known to someone in their family had higher levels of stigma), number of births (women with fewer births had higher levels of stigma), and partner communication about the birth place (women who did not discuss the location of the upcoming birth with their partner had higher levels of stigma).
Table 3
Multivariable generalized estimating equation analysis predicting internalized HIV-related stigma (N = 135)
Age | 0.02 (0.02) | .282 |
Marital status (ref: not currently married) | | |
Currently married | 0.05 (0.21) | .818 |
Educational level (ref: more than primary) | | |
Primary or less | −0.43 (0.33) | .188 |
Mobile phone ownership (ref: yes) | | |
No | −0.12 (0.07) | .086 |
Pre/postpartum intimate partner violence (ref: none) | | |
Yes | 0.35 (0.14) | .011 |
Unknown | 0.26 (0.16) | .022 |
Infant health problems after birth (ref: no) | | |
Yes | 0.15 (0.13) | .237 |
Number of births | −0.10 (0.05) | .037 |
Family knows about HIV status (ref: no) | | |
Yes | 0.37 (0.19) | .046 |
Spousal communication about birthplace (ref: no) | | |
Yes | −0.25 (0.12) | .039 |
Linked to care (ref: no) | | |
Yes | −0.34 (0.15) | .024 |
Next, we repeated the GEE analysis with symptoms of postpartum depression as the outcome and with internalized stigma added to the equation as a predictor (see Table
4 for statistical details). In this model, internalized stigma was a significant predictor of depression: Women with higher internalized stigma had higher levels of depression. With stigma in the model, linkage to HIV care was no longer a significant predictor of postpartum depression symptoms. Mediation analysis for clustered data revealed a significant indirect effect of linkage to HIV care on depression, mediated by stigma (coefficient = −.61,
p < .05, 95% bias-corrected confidence interval = −1.22 to -.11). Thus, this analysis provided evidence that stigma mediated the effect of linkage to HIV care on lower depression symptoms.
Table 4
Multivariable generalized estimating equation analysis predicting postpartum depression severity with internalized stigma added to the model (N = 135)
Age | 0.03 (0.10) | .764 |
Marital status (ref: not currently married) | | |
Currently married | −0.15 (0.48) | .765 |
Educational level (ref: more than primary) | | |
Primary or less | −0.57 (0.85) | .502 |
Mobile phone ownership (ref: yes) | | |
No | −0.24 (0.92) | .792 |
Pre/postpartum intimate partner violence (ref: none) | | |
Yes | 3.67 (0.87) | .000 |
Unknown | 1.50 (0.51) | .004 |
Infant health problems after birth (ref: no) | | |
Yes | 3.03 (0.83) | .000 |
Number of births | −0.38 (0.23) | .098 |
Family knows about HIV status (ref: no) | | |
Yes | 0.28 (0.88) | .746 |
Discussion of delivery with partner (ref: yes) | | |
No | 1.60 (0.43) | .000 |
Linked to care (ref: no) | | |
Yes | −0.86 (0.713) | .226 |
HASI-P internalized HIV-related stigma | 2.88 (.570) | .000 |
Next, we repeated the GEE analyses using the dichotomized depression variable (EPDS ≥ 13, using binary logistic models) with internalized stigma added as a predictor. In this model, stigma was a significant predictor of depression status (OR = 3.67, 95% CI = 1.87 - 7.22): Women with higher internalized stigma were more likely to be depressed. With stigma in the model, linkage to care was still a significant predictor of depression status; however, the effect of linkage was reduced (OR = 0.45, 95% CI = 0.22 - 0.90).
The role of ART
Of the 79 participants included in the GEE multivariate analyses who had linked to HIV care, 33 (42%) were on ART. We compared postpartum depression symptoms for the following three groups: a) not linked to HIV care, b) linked to HIV care but not on ART, c) linked to HIV care and on ART. GEE analyses with this three-category variable (and the control variables) suggested that both of the first two groups (not linked to HIV care and linked but not on ART) had significantly more postpartum depression symptoms than the third group (linked and on ART; B = 3.20 and B = 2.26, respectively, both p values <= 0.001). There were no significant differences between the first two groups (not linked to HIV care versus linked but not on ART; B = 0.94, p = 0.37). The estimated marginal means for postpartum depression symptoms for the three groups were: 11.95, 11.01, and 8.75, respectively for a) not linked to HIV care, b) linked but not on ART, and c) linked and on ART.
Binary logistic models using the dichotomized depression variable as the dependent variable (depressed defined as EPDS ≥ 13) led to similar results. Both of the first two groups (not linked to HIV care and linked but not on ART) were significantly more likely to be depressed than the third group (linked and on ART; OR = 7.91, 95% CI = 2.79 – 22.44 and OR = 4.33, 95% CI = 1.80 – 10.40, respectively, both p values <= 0.001). There were no significant differences between the first two groups (not linked to HIV care versus linked but not on ART; OR = .55, 95% CI = .25 – 1.18, p = 0.12).
Discussion
We examined postpartum depression symptoms and internalized stigma levels six weeks after birth for women who tested HIV-positive for the first time during their first ANC visit. We found that HIV-positive pregnant women who had linked with HIV care had significantly lower levels of depression symptoms (as well as lower postpartum depression rates using the recommended cutoff score for major depression) at around six weeks after the birth, as compared to HIV-positive women who did not link with HIV care. We also identified an important potential mediator for this effect: It appears that linkage to HIV care is associated with lower internalized HIV-related stigma, which in turn predicts lower postpartum depression symptom severity. ART seemed to be an important component of this effect on depression symptoms, with depression levels being lowest in women who were both linked to HIV care and on ART. This last finding (although based on a small sample size) lends further support to current efforts to roll out Option B+, the WHO-recommended strategy of initiating all HIV-positive pregnant women on lifelong ART, regardless of initial stage of HIV disease [
74].
The findings of the current study are particularly relevant as the global community struggles to achieve optimal maternal and child health outcomes in sub-Saharan Africa, including HIV-free infant survival. HIV-infected pregnant women in sub-Saharan Africa are a highly vulnerable group (especially those who become aware of their HIV infection during pregnancy), and it is very important to protect their mental as well as physical health [
75]. This study fills a gap in the current literature by examining the associations between linkage to HIV care and postpartum depression among pregnant women in a rural setting in sub-Saharan Africa.
This study also adds to the current body of literature by examining the effects of linkage to care itself (Pre-ART care, which includes a package of services that may benefit mental health) versus the effects of linkage to care and being on ART. There has been little previous research attempting to tease apart the separate effects of ART medication and other components of HIV care on these outcomes. Linkage to HIV care and the opportunity to take advantage of other supportive services may result in lower stigma and/or depression, especially for pregnant women, who are particularly vulnerable to depression [
14],[
22],[
23]. Although the current analyses are based on a relatively small sample size, the findings point to the possibility that ART is an important component of the effect of linkage to HIV care on postpartum depression.
Internalized HIV-related stigma and depression both include aspects such as self-blame and self-devaluation, and hence may be similar constructs. Although in this study we found these measures to be correlated, they appeared to be distinct aspects of the experiences of many newly diagnosed pregnant and postpartum women living with HIV in our sample. We found that internalized HIV-related stigma was a mediating factor in the association between linkage to HIV care and depression in this population.
The other variables that were found to be significantly related to postpartum depression in this setting are, for the most part, in agreement with existing literature. Intimate partner violence has consistently been shown to be related to adverse mental health outcomes in women [
76], and has been shown to be an important predictor of perinatal depression [
72]. Having an infant with health problems has also been found to be an important predictor in some settings [
25],[
77],[
78]. Number of births was found to be an important predictor of postpartum depression in HIV-infected women in Zimbabwe [
17], although in the opposite direction found in the current study. Our analyses also suggest important roles of family’s knowledge about the woman’s HIV-positive status and communication with the male partner in understanding postpartum depression for women living with HIV in this rural Kenyan setting. These same factors (with the exception of infant health status) also predicted HIV-related internalized stigma.
Our results should be evaluated in the context of the limitations of this study. Only 70% of the original sample could be located and participated in the follow-up assessment after the birth. However, as reported elsewhere, no important differences were detected between those women who could be located and interviewed postpartum and those lost-to-follow-up [
63]. Although we hypothesized that lower internalized HIV-related stigma leads to lower postpartum depression in our mediation analyses, these variables were measured at the same time point (i.e., during the postpartum interview). Because pregnant women were interviewed at baseline before their initial ANC visit and before they knew their HIV status, we were not able to collect baseline data on internalized HIV-related stigma. In addition, a large proportion of women in the sample did not link to HIV care, meaning that we were not able to obtain data on their stage of HIV disease, and we only had self-report data on women’s current ART use.
Our findings are correlational and cannot provide definitive evidence for a causal role of linkage to HIV care in affecting depression or stigma. However, we have adjusted our analyses for socio-demographics and several key predictors from the literature. Ancillary analyses on a limited subsample of women included in the current analyses who completed an additional questionnaire during late pregnancy (n = 67) suggested that depression symptom severity was not very stable from late pregnancy to six weeks after the birth (Pearson
r = 0.25). Thus, it is not likely that late pregnancy depression was a confounding variable in the association between linkage to care and postpartum depression. Repeating our main analysis on the limited subsample of women for whom we had late pregnancy data yielded a similar effect size for the association between linkage to care and postpartum depression controlling for depression at late pregnancy (
B = −1.51,
p = .07, n = 64 for the effect of linkage to care) compared to our main analyses where depression at late pregnancy was not controlled (
B = −1.85,
p = .04, N = 135; see Table
2).
Important strengths of the study include the prospective study design targeting a particularly vulnerable population: Participants were interviewed immediately before their first ANC visit and again around six weeks after the birth and only those participants who tested positive for HIV for the first time were included. Other strengths include the utilization of medical record data, rather than self-report, for our key predictor variable of linkage to care, and detailed analyses controlling for important variables that may be related to our outcome variables.
Acknowledgments
We thank the Kenyan women who participated in the study and shared their experiences with us. We acknowledge the important logistical support of the KEMRI-UCSF Collaborative Group and especially Family AIDS Care & Education Services (FACES). We thank the FACES CCHAs in Migori County for their diligent work in collecting the quantitative data. We gratefully acknowledge the Director of KEMRI, the Director of KEMRI’s Centre for Microbiology, and the Nyanza Provincial Ministries of Health for their support in conducting this research. We also thank Katie Schwartz, Nicole Schmidt, John Oguda, Merab Odero, Nancy Akinyi, Steven Ajuoga, Felix Owino, Rachel Steinfeld, and Zachary Kwena for their important contributions to this research. The project described was supported by Award Number K01MH081777 from the National Institute of Mental Health. Kristi Stringer’s contribution to this research was supported by grant 2 T32 HS013852 from the Agency for Healthcare Research and Quality, Rockville, MD, USA, and by the National Institute On Drug Abuse of the National Institutes of Health under Award Number F31DA037106. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
BT conducted statistical analyses and drafted the manuscript. KLS contributed to statistical analyses and drafting of the manuscript. MO participated in study design and coordination and helped to draft the manuscript. EAB participated in study design and coordination and helped to draft the manuscript. SDW helped to draft the manuscript. CRC participated in study design and coordination and helped to draft the manuscript. JMT conceived of the study, supervised the data collection, and helped to draft the manuscript. All authors read and approved the final manuscript.