XML Template (2015) [24.4.2015–5:41pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150065/APPFile/SG-STDJ150065.3d
(STD)
[1–10] [PREPRINTER stage]
Int J STD AIDS OnlineFirst, published on April 29, 2015 as doi:10.1177/0956462415584483
Original research article
Long-term postpartum adherence to antiretroviral drugs among women in Latin America
International Journal of STD & AIDS 0(0) 1–10 ! The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0956462415584483 std.sagepub.com
Regis Kreitchmann1, Debora Fernandes Coelho1, Fabiana Maria Kakehasi2, Cristina Barroso Hofer3, Jennifer S Read4, Marcelo Losso5, Jessica E Haberer6, George K Siberry4, D Robert Harris7 and Qilu Yu7
Summary Antiretroviral adherence in the postpartum period is crucial for maternal health and decreasing risks of mother-to-child HIV transmission and transmission to sexual partners. Self-reported antiretroviral adherence was examined between 6to 12-weeks and 30 months postpartum among 270 HIV-infected women enrolled in a prospective cohort study from 2008 to 2010 at multiple sites in Latin America. Adherence data were collected at each study visit to quantify the proportion of prescribed antiretrovirals taken during the previous three days, assess the timing of the last missed dose, and identify predictors of adherence. Mean adherence rates were 89.5% at 6–12 weeks and 92.4% at 30 months; the proportions with perfect adherence were 80.3% and 83.6%, respectively. The overall trend for perfect adherence was not significant (p ¼ 0.71). In adjusted regression modelling, younger age was associated with an increased probability of non-perfect adherence at 18 and 24 months postpartum. Other factors associated with increased probability of non-perfect adherence were higher parity, current use of alcohol and tobacco, and more advanced HIV disease. Women with perfect adherence had lower viral loads. Interventions for alcohol and tobacco use cessation, and support for young women and those with advanced HIV disease should be considered to improve postpartum adherence.
Keywords HIV, antiretroviral, adherence, postpartum Date received: 22 August 2014; accepted: 2 April 2015
Introduction
1
Globally, it is estimated that 35.3 million people are living with HIV, with 2.3 million new infections occurring each year.1 Among young individuals, half of these infections occur in women. As of December 2012, over 900,000 pregnant women living with HIV used antiretroviral (ARV) prophylaxis or treatment. Coverage of ARV programs for prevention of mother-to-child transmission increased from 57% (51–64%) in 2011 to 62% (57–70%) in 2012.1 For women who are eligible for treatment, highly active antiretroviral therapy (HAART) is continued during the postpartum period and is critical for preserving maternal health and increasing child survival. A recent multinational randomised trial, HPTN 052, found that initiation of ARV treatment of HIV-infected individuals with CD4 cell counts between 350 and 550
Irmandade da Santa Casa de Misericordia de Porto Alegre and Universidade Federal de Ciencias da Saude de Porto Alegre (UFCSPA), Porto Alegre, Brazil 2 Universidade Federal de Minas Gerais, Belo Horizonte, Brazil 3 Instituto de Puericultura e Pediatria Martaga˜o Gesteira and Department of Preventive Medicine Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil 4 Maternal and Pediatric Infectious Disease Branch; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Bethesda, MD, USA; Current affiliation for Dr. Read: National Institute of Allergy and Infectious Diseases; National Institutes of Health, Bethesda, MD, USA 5 Hospital General de Agudos Jose Maria Ramos Mejia, Buenos Aires, Argentina 6 Massachusetts General Hospital, Boston, MA, USA 7 Westat, Rockville, MD, USA Corresponding author: Regis Kreitchmann, Irmandade da Santa Casa de Misericordia de Porto Alegre, Prof. Annes Dias 285, 90020-090 Porto Alegre, Brazil. Email: [email protected]
Downloaded from std.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on July 12, 2015
XML Template (2015) [24.4.2015–5:41pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150065/APPFile/SG-STDJ150065.3d
(STD)
[1–10] [PREPRINTER stage]
2
International Journal of STD & AIDS 0(0)
cells/mm3 significantly reduced transmission to their uninfected sexual partners, compared to those in whom treatment was delayed.2 Given that rates of HIV serodiscordance reach 40 to 49% among couples affected by HIV in many settings, there is increasing impetus to continue ARVs (usually a HAART regimen) in HIV-infected women after delivery, even if they do not meet current indications for ARV treatment (ART).3,4 Indeed there are increased efforts to keep these women on HAART after delivery, with many countries having implemented this as policy (WHO Option Bþ).5 Continued adherence to HAART in the postpartum period is also critical for prevention of HIV transmission to sexual partners.6 The effectiveness of ART is a function of the extent and duration of viral suppression with high levels of adherence considered necessary to achieve optimal treatment outcomes.7 Factors associated with imperfect adherence to ARVs include patient-level factors (e.g. education about and commitment to ARV treatment, psychiatric illness, stigma, use of alcohol, or other drugs), treatment-level factors (e.g. adverse events, pill burden, and number of daily doses), and healthcare system-level factors (e.g. accessibility and confidentiality). Any intervention to improve adherence must consider the patient holistically, taking into account behavioral, cognitive, emotional, and social characteristics.8 However, relatively few studies of ARV adherence have included postpartum women,9–15 and fewer still have been conducted in Latin America, although an estimated 1.5 million people were living with HIV infection in Latin America at the end of 2012, and slightly more than half of them were women.1 In a prior analysis, we examined self-reported adherence to ARVs during pregnancy and the early postpartum period (6 months postpartum) in a population of HIV-infected women enrolled at multiple sites in Latin America in the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) International Site Development Initiative (NISDI) LILAC (Longitudinal Study in Latin American Countries) study.12 We found that selfreported adherence to ARVs was significantly higher during pregnancy compared to the early postpartum period (p < 0.01) but did not differ between the 6- to 12-week and 6-month postpartum visits (p > 0.4). Substance use (alcohol and tobacco) and older age were found to be important predictors of increased risk of ARV non-adherence in multivariable modelling at selected visits. The current analysis extends the former analysis to examine long-term adherence through the inclusion of time-points beyond the 6-month postpartum visit that were not previously available or had been completed by too few study participants to support formal analyses. This analysis is intended to provide long-term data on ARV adherence
during the postpartum period, which are currently very limited, and assist in identifying HIV-infected women at greatest risk for non-perfect adherence to ARVs to better inform the targeting of resources for adherence counselling and support.
Methods LILAC LILAC was a multi-center prospective cohort study of HIV-infected pregnant women and HIV-exposed, uninfected infants conducted at participating clinical sites in Latin America.16 In this study, women enrolled at 22 weeks of pregnancy and were followed through delivery and postpartum. Maternal study visits were conducted at hospital discharge after delivery, at 6–12 weeks and six months postpartum, and every six months thereafter, for up to five years after delivery. The medical history and laboratory samples were collected and a physical examination was performed at six-month intervals. ARVs were prescribed according to local guidelines and adherence data were collected at each study visit to quantify the proportion of prescribed ARVs actually taken during the previous three days, to assess the timing of the last missed dose, and to identify adherence difficulties experienced by study subjects. The protocol was approved by the ethical review board at each participating site, as well as the institutional review boards at the sponsoring institution (NICHD) and at the data management and statistical center (Westat). Prior to enrollment, all participants provided written informed consent for study participation.
Study population and definitions for this analysis The study population for this analysis comprised women enrolled in the LILAC protocol from June 2008 to June 2010 during their first on-study pregnancy, who were prescribed ARVs for at least one of the six targeted postpartum visits (6–12 weeks and 6, 12, 18, 24, and 30 months postpartum), and who completed the selfreport adherence form for the visit. Self-reported adherence to ARVs was derived based on all ARVs included in the subject’s regimen at a particular study visit and considered doses missed yesterday, two days ago and three days ago. For the three-day time period the total number of missed and expected doses were used to calculate percent adherence as follows: Percent adherence ¼ 100 1 P
P
number of doses missed expected number of doses prescribed
Downloaded from std.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on July 12, 2015
XML Template (2015) [24.4.2015–5:41pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150065/APPFile/SG-STDJ150065.3d
(STD)
[1–10] [PREPRINTER stage]
Kreitchmann et al.
3
For each of the six study visits, an indicator of nonperfect adherence was derived from percent adherence to distinguish those who were less than 100% adherent over the three-day period from those who were 100% adherent.
Statistical analysis Summary statistics, such as means, standard deviations, and medians were used to describe continuous-scaled characteristics of the study population and to describe percent adherence at each study visit; frequencies and proportions were used to describe categorical-scaled characteristics of the study population and perfect adherence at each study visit. The trend according to study visit in the proportion with non-perfect adherence was modelled using a generalised estimating equation (GEE) method to account for within-subject correlation and to provide population-averaged estimates of parameter effects. In order to allow for a possible non-linear trend in adherence according to study visit and allow for pair-wise comparisons between visits, a categorical time variable was used to model study visits. All subjects with at least one adherence measure were included in the analyses. Associations between subject characteristics and non-perfect adherence were examined in bivariate GEE models that examined the effect of one characteristic at a time across the six postpartum visits. Based on our prior analyses of these data,12 characteristics at study enrollment that were considered for modelling included: age, education (years completed), parity, number of persons living in the participant’s household, substance use (ever used alcohol, tobacco, marijuana, cocaine/crack, heroin/opiates, or other substances), and mode of HIV infection. Visit-specific (time varying) characteristics included current alcohol use, current tobacco use, duration of ARV use, as well as CDC and WHO clinical classification.17,18 As in our prior analysis, CD4 percent measures and viral load (VL) were considered outcomes rather than predictors of adherence and, therefore, were not included in the modelling. Our prior analysis did not account for psychosocial factors associated with adherence. To address this limitation, an indicator of depression, based on clinical diagnoses and/or use of antidepressants, was included in the modelling; further psycho-social data were not available. Covariates associated with nonperfect adherence in these bivariate analyses with p values of 0.10 or less were considered in the multivariate modelling, together with the categorical time measure used to represent the different study visits. Model selection was performed based on statistical and clinical significance, while study visit was always retained in the models regardless of its significance. To further
examine associations of covariates with trends in adherence measures according to study visit, interactions of study visit with select covariates were included in the modelling. Interaction terms with a p value less than 0.05 were retained in the final model. Lastly, the association between non-perfect adherence and plasma VL, defined as undetectable (
(STD)
[1–10] [PREPRINTER stage]
Int J STD AIDS OnlineFirst, published on April 29, 2015 as doi:10.1177/0956462415584483
Original research article
Long-term postpartum adherence to antiretroviral drugs among women in Latin America
International Journal of STD & AIDS 0(0) 1–10 ! The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0956462415584483 std.sagepub.com
Regis Kreitchmann1, Debora Fernandes Coelho1, Fabiana Maria Kakehasi2, Cristina Barroso Hofer3, Jennifer S Read4, Marcelo Losso5, Jessica E Haberer6, George K Siberry4, D Robert Harris7 and Qilu Yu7
Summary Antiretroviral adherence in the postpartum period is crucial for maternal health and decreasing risks of mother-to-child HIV transmission and transmission to sexual partners. Self-reported antiretroviral adherence was examined between 6to 12-weeks and 30 months postpartum among 270 HIV-infected women enrolled in a prospective cohort study from 2008 to 2010 at multiple sites in Latin America. Adherence data were collected at each study visit to quantify the proportion of prescribed antiretrovirals taken during the previous three days, assess the timing of the last missed dose, and identify predictors of adherence. Mean adherence rates were 89.5% at 6–12 weeks and 92.4% at 30 months; the proportions with perfect adherence were 80.3% and 83.6%, respectively. The overall trend for perfect adherence was not significant (p ¼ 0.71). In adjusted regression modelling, younger age was associated with an increased probability of non-perfect adherence at 18 and 24 months postpartum. Other factors associated with increased probability of non-perfect adherence were higher parity, current use of alcohol and tobacco, and more advanced HIV disease. Women with perfect adherence had lower viral loads. Interventions for alcohol and tobacco use cessation, and support for young women and those with advanced HIV disease should be considered to improve postpartum adherence.
Keywords HIV, antiretroviral, adherence, postpartum Date received: 22 August 2014; accepted: 2 April 2015
Introduction
1
Globally, it is estimated that 35.3 million people are living with HIV, with 2.3 million new infections occurring each year.1 Among young individuals, half of these infections occur in women. As of December 2012, over 900,000 pregnant women living with HIV used antiretroviral (ARV) prophylaxis or treatment. Coverage of ARV programs for prevention of mother-to-child transmission increased from 57% (51–64%) in 2011 to 62% (57–70%) in 2012.1 For women who are eligible for treatment, highly active antiretroviral therapy (HAART) is continued during the postpartum period and is critical for preserving maternal health and increasing child survival. A recent multinational randomised trial, HPTN 052, found that initiation of ARV treatment of HIV-infected individuals with CD4 cell counts between 350 and 550
Irmandade da Santa Casa de Misericordia de Porto Alegre and Universidade Federal de Ciencias da Saude de Porto Alegre (UFCSPA), Porto Alegre, Brazil 2 Universidade Federal de Minas Gerais, Belo Horizonte, Brazil 3 Instituto de Puericultura e Pediatria Martaga˜o Gesteira and Department of Preventive Medicine Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil 4 Maternal and Pediatric Infectious Disease Branch; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Bethesda, MD, USA; Current affiliation for Dr. Read: National Institute of Allergy and Infectious Diseases; National Institutes of Health, Bethesda, MD, USA 5 Hospital General de Agudos Jose Maria Ramos Mejia, Buenos Aires, Argentina 6 Massachusetts General Hospital, Boston, MA, USA 7 Westat, Rockville, MD, USA Corresponding author: Regis Kreitchmann, Irmandade da Santa Casa de Misericordia de Porto Alegre, Prof. Annes Dias 285, 90020-090 Porto Alegre, Brazil. Email: [email protected]
Downloaded from std.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on July 12, 2015
XML Template (2015) [24.4.2015–5:41pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150065/APPFile/SG-STDJ150065.3d
(STD)
[1–10] [PREPRINTER stage]
2
International Journal of STD & AIDS 0(0)
cells/mm3 significantly reduced transmission to their uninfected sexual partners, compared to those in whom treatment was delayed.2 Given that rates of HIV serodiscordance reach 40 to 49% among couples affected by HIV in many settings, there is increasing impetus to continue ARVs (usually a HAART regimen) in HIV-infected women after delivery, even if they do not meet current indications for ARV treatment (ART).3,4 Indeed there are increased efforts to keep these women on HAART after delivery, with many countries having implemented this as policy (WHO Option Bþ).5 Continued adherence to HAART in the postpartum period is also critical for prevention of HIV transmission to sexual partners.6 The effectiveness of ART is a function of the extent and duration of viral suppression with high levels of adherence considered necessary to achieve optimal treatment outcomes.7 Factors associated with imperfect adherence to ARVs include patient-level factors (e.g. education about and commitment to ARV treatment, psychiatric illness, stigma, use of alcohol, or other drugs), treatment-level factors (e.g. adverse events, pill burden, and number of daily doses), and healthcare system-level factors (e.g. accessibility and confidentiality). Any intervention to improve adherence must consider the patient holistically, taking into account behavioral, cognitive, emotional, and social characteristics.8 However, relatively few studies of ARV adherence have included postpartum women,9–15 and fewer still have been conducted in Latin America, although an estimated 1.5 million people were living with HIV infection in Latin America at the end of 2012, and slightly more than half of them were women.1 In a prior analysis, we examined self-reported adherence to ARVs during pregnancy and the early postpartum period (6 months postpartum) in a population of HIV-infected women enrolled at multiple sites in Latin America in the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) International Site Development Initiative (NISDI) LILAC (Longitudinal Study in Latin American Countries) study.12 We found that selfreported adherence to ARVs was significantly higher during pregnancy compared to the early postpartum period (p < 0.01) but did not differ between the 6- to 12-week and 6-month postpartum visits (p > 0.4). Substance use (alcohol and tobacco) and older age were found to be important predictors of increased risk of ARV non-adherence in multivariable modelling at selected visits. The current analysis extends the former analysis to examine long-term adherence through the inclusion of time-points beyond the 6-month postpartum visit that were not previously available or had been completed by too few study participants to support formal analyses. This analysis is intended to provide long-term data on ARV adherence
during the postpartum period, which are currently very limited, and assist in identifying HIV-infected women at greatest risk for non-perfect adherence to ARVs to better inform the targeting of resources for adherence counselling and support.
Methods LILAC LILAC was a multi-center prospective cohort study of HIV-infected pregnant women and HIV-exposed, uninfected infants conducted at participating clinical sites in Latin America.16 In this study, women enrolled at 22 weeks of pregnancy and were followed through delivery and postpartum. Maternal study visits were conducted at hospital discharge after delivery, at 6–12 weeks and six months postpartum, and every six months thereafter, for up to five years after delivery. The medical history and laboratory samples were collected and a physical examination was performed at six-month intervals. ARVs were prescribed according to local guidelines and adherence data were collected at each study visit to quantify the proportion of prescribed ARVs actually taken during the previous three days, to assess the timing of the last missed dose, and to identify adherence difficulties experienced by study subjects. The protocol was approved by the ethical review board at each participating site, as well as the institutional review boards at the sponsoring institution (NICHD) and at the data management and statistical center (Westat). Prior to enrollment, all participants provided written informed consent for study participation.
Study population and definitions for this analysis The study population for this analysis comprised women enrolled in the LILAC protocol from June 2008 to June 2010 during their first on-study pregnancy, who were prescribed ARVs for at least one of the six targeted postpartum visits (6–12 weeks and 6, 12, 18, 24, and 30 months postpartum), and who completed the selfreport adherence form for the visit. Self-reported adherence to ARVs was derived based on all ARVs included in the subject’s regimen at a particular study visit and considered doses missed yesterday, two days ago and three days ago. For the three-day time period the total number of missed and expected doses were used to calculate percent adherence as follows: Percent adherence ¼ 100 1 P
P
number of doses missed expected number of doses prescribed
Downloaded from std.sagepub.com at Kungl Tekniska Hogskolan / Royal Institute of Technology on July 12, 2015
XML Template (2015) [24.4.2015–5:41pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150065/APPFile/SG-STDJ150065.3d
(STD)
[1–10] [PREPRINTER stage]
Kreitchmann et al.
3
For each of the six study visits, an indicator of nonperfect adherence was derived from percent adherence to distinguish those who were less than 100% adherent over the three-day period from those who were 100% adherent.
Statistical analysis Summary statistics, such as means, standard deviations, and medians were used to describe continuous-scaled characteristics of the study population and to describe percent adherence at each study visit; frequencies and proportions were used to describe categorical-scaled characteristics of the study population and perfect adherence at each study visit. The trend according to study visit in the proportion with non-perfect adherence was modelled using a generalised estimating equation (GEE) method to account for within-subject correlation and to provide population-averaged estimates of parameter effects. In order to allow for a possible non-linear trend in adherence according to study visit and allow for pair-wise comparisons between visits, a categorical time variable was used to model study visits. All subjects with at least one adherence measure were included in the analyses. Associations between subject characteristics and non-perfect adherence were examined in bivariate GEE models that examined the effect of one characteristic at a time across the six postpartum visits. Based on our prior analyses of these data,12 characteristics at study enrollment that were considered for modelling included: age, education (years completed), parity, number of persons living in the participant’s household, substance use (ever used alcohol, tobacco, marijuana, cocaine/crack, heroin/opiates, or other substances), and mode of HIV infection. Visit-specific (time varying) characteristics included current alcohol use, current tobacco use, duration of ARV use, as well as CDC and WHO clinical classification.17,18 As in our prior analysis, CD4 percent measures and viral load (VL) were considered outcomes rather than predictors of adherence and, therefore, were not included in the modelling. Our prior analysis did not account for psychosocial factors associated with adherence. To address this limitation, an indicator of depression, based on clinical diagnoses and/or use of antidepressants, was included in the modelling; further psycho-social data were not available. Covariates associated with nonperfect adherence in these bivariate analyses with p values of 0.10 or less were considered in the multivariate modelling, together with the categorical time measure used to represent the different study visits. Model selection was performed based on statistical and clinical significance, while study visit was always retained in the models regardless of its significance. To further
examine associations of covariates with trends in adherence measures according to study visit, interactions of study visit with select covariates were included in the modelling. Interaction terms with a p value less than 0.05 were retained in the final model. Lastly, the association between non-perfect adherence and plasma VL, defined as undetectable (
