Universal prenatal HIV screening: are we there yet?

INTRODUCTION

Detection of HIV infection in pregnant women and the increasingly widespread use of antiretroviral therapy have led to a large decline in perinatal HIV transmission in the USA. ¹ Strategies to reduce the risk of mother-to-child transmission include antiretroviral therapy, elective Caesarean section and avoidance of breastfeeding. ^2,3 Thus, determination of HIV serostatus in pregnant women is crucial for efforts to decrease the risk of vertical transmission to the neonate.

In 1999, the American College of Obstetricians and Gynecologists (ACOG) and the American Academy of Pediatrics published recommendations advocating universal prenatal HIV screening to reduce perinatal transmission of HIV. ⁴ Prenatal HIV screening rates in concurrent years were estimated to be 54–69%. ^5–7 Policies on prenatal HIV screening are determined by individual states, with Pennsylvania retaining an opt-in policy of screening. More recent studies have demonstrated that screening rates continue to fall short of universal screening goals. ^8,9 The goal of this study was to explore patterns of prenatal HIV screening in women delivering an infant at a large USA metropolitan obstetric hospital. We determined the prevalence of HIV screening in pregnant women, examined factors associated with increased likelihood of prenatal HIV screening and assessed the availability of HIV test results in medical records at the time of labour.

METHODS

We performed a survey of postpartum women at Magee-Womens Hospital at the University of Pittsburgh Medical Center, a large tertiary care hospital in Pittsburgh, Pennsylvania, where nearly 10,000 deliveries occur annually. Women were enrolled within 72 hours of delivery, on weekdays over a four-week period spanning 25 May to 19 June 2008, using convenience sampling based on the availability of one of the investigators (MRK). Women were not enrolled if they did not speak English (n = 3), they were absent on two interview attempts (n = 2) or if establishment of a confidential environment was not possible (n = 5). All other women who were approached agreed to participate. The primary outcome was prenatal HIV screening. This audit was approved by the Total Quality Control Committee at our institution, and informed consent was not required to complete this quality improvement initiative.

The survey was administered confidentially and consisted of questions related to patient demographics, medical history and risk factors for HIV (e.g. sexual history, illicit drug use). Information about the type of obstetric provider(s) and practice(s) where women received prenatal care was recorded. Women were asked if they recalled being offered HIV screening during pregnancy and the manner in which the test was offered (e.g. available if interested, performed as standard practice unless declined) and reasons for accepting or declining screening. The questionnaire included at least three possible answers for each question, including ‘other’ as an alternate answer if other choices did not match patient response or if the patient preferred to not provide a discrete answer (e.g. illicit drug use). In some instances (e.g. ethnicity, primary obstetric provider type), a specific response was required. Participants were permitted one answer per question item.

After survey completion, each woman's hospital chart was reviewed for documentation of HIV test results and to verify accuracy of self-reported HIV screening. Data recorded during the medical record review included the type of practice where women received prenatal care (private practice, resident/teaching or community outreach clinics) and documentation of HIV screening. Practices were further characterized as physician-based (obstetricians) or certified nurse midwives; care at resident/teaching clinics was performed by residents training in obstetrics and gynaecology. Nurse practitioners and nurse midwives were the primary care-givers at community outreach clinics.

Sample size was calculated based on one of the study objectives to determine whether the proportion of women who received prenatal HIV testing differed by provider setting. In our population, it was expected that 20% of women would receive prenatal care at a resident or community clinic and 80% from a private practice. A previous study found that 70% of women who gave birth at seven teaching hospitals in the USA reported prenatal HIV testing. ⁵ A sample size of 300 women would have 80% power to detect significant differences in the prevalence of prenatal HIV screening, assuming that 70% of women attending a hospital clinic and 50% attending private practices would be screened for HIV, based on a Fisher's exact test evaluated at the two-sided 0.05 significance level. All data were analysed using Stata statistical software, release 11.1 (StataCorp, College Station, TX, USA) and statistical tests were evaluated at the two-sided 0.05 significance level. Certain continuous variables were dichotomized for analysis, such as age (based on sexually transmitted infection [STI] risk of ≤25 versus >25-years-old) and lifetime partners (median numbers of the cohort). Prevalence ratios (PRs) were used to evaluate the association of patient and provider characteristics and provider approach to prenatal HIV screening. Poisson regression with robust variance was used to calculate PRs and their respective 95% confidence intervals (CIs). ¹⁰ Multivariable Poisson regression was used to calculate adjusted PRs for characteristics that were independently associated with prenatal HIV screening. Models were developed using forward regression and variables were retained in the model if the P value from the Wald chi-square test statistic was 0.05 or less.

RESULTS

Three hundred postpartum women participated in this study, of whom 68% (204/300) were married, 16% (48/300) co-habitating, 14% (42/300) single, 1% (4/300) separated or divorced, and two women (<1%) were widowed. Seventy-seven percent (232/300) were white, 15% (45/300) black, 6% (17/300) Asian, 1% (4/300) described their race as ‘other’ and less than 1% (2/300) identified as biracial. Seven women (2%) were of Hispanic ethnicity. The mean age of the cohort was 29 years. The demographics of the study population were similar to the overall obstetrical population at our institution (data not shown). Seventy-five percent (224/300) had at least some college education. Seventy-nine percent (238/300) received prenatal care in a private practice, 16% (49/300) in a resident/teaching clinic or community clinic, and 4% (13/300) received care in multiple settings, in settings unfamiliar to investigators (including out of state), or had no prenatal care. Sixty-five percent (195/300) of women were primiparous. One-third of women (100/300) reported no HIV testing prior to this pregnancy.

Of 300 participants, 195 (65%) reported being screened for HIV during pregnancy, 75 (25%) reported that they were not screened and 30 (10%) were unsure. Of the 75 women not screened, 26 (35%) reported that they were not offered HIV screening and 49 (65%) declined HIV screening. Reasons stated for declining to be screened included previous testing (n = 25), no perceived need for HIV screening (n = 23) and fear of results (n = 1).

We compared subject characteristics with self-reported HIV screening during pregnancy (Table 1). Univariate analysis demonstrated that HIV screening was more common in black women compared with white women and those of other racial backgrounds, in unmarried compared with married women, in younger women, in those with annual household income ≤ US$30,000 and in those with high school education or less. The influence of HIV risk factors on HIV screening in pregnancy was assessed. Women with a prior history of STI (including gonorrhoea, chlamydia, trichomoniasis, human papillomavirus, or genital herpes; no HIV or syphilis infection was reported) were screened more often than those with a negative STI history. Women with five or more lifetime sexual partners were significantly more likely to be screened for HIV (119/163 [73%] versus 75/134 [56%], P = 0.003). History of illicit drug use was not associated with HIV screening in pregnancy.

The impact of the source of prenatal care on HIV screening was examined (Table 2). Women receiving care in a resident or community clinic were significantly more likely to be screened than those receiving care in a private practice (90% versus 58%, P < 0.001). Excluded from this analysis were women who had received prenatal care in multiple settings or sites not easily characterized as a private practice or resident/community clinic (n = 11), and two women who did not receive prenatal care prior to presentation in labour. Women receiving care from a non-obstetrician provider (n = 22) were more likely to undergo HIV screening than those under the care of an obstetrician/gynaecologist on univariate analysis (86% versus 63%, P = 0.001). We next assessed whether an established health-care relationship between the provider and the patient, indicative of the provider's familiarity with the patient and her history, influenced the uptake of prenatal HIV screening. Those patients with an established relationship with their health-care provider predating pregnancy were less likely to undergo screening, but this difference did not reach statistical significance.

A multivariate Poisson model including subject and provider characteristics demonstrated that black race, annual household income ≤US$30,000, and five or more lifetime sexual partners were factors independently associated with increased likelihood of prenatal HIV screening (Table 3). No other subject or health-care provider characteristics were independently associated with HIV screening.

We next evaluated the effect of the health-care provider's approach to screening on the uptake of prenatal HIV screening (Table 4). Of 224 women who recalled the manner in which HIV screening was offered, 164 (73%) reported that screening was offered as standard practice unless declined (similar to an opt-out strategy) and 60 (27%) reported that screening was offered as an available test if desired. Women who reported being informed by their health-care provider that screening was standard practice were significantly more likely to agree to screening. Patient perception of the health-care provider's attitude towards HIV screening during pregnancy was also predictive of HIV screening. Women who perceived that an emphasis was placed on the importance of prenatal HIV screening were more likely to be tested (87% versus 48%, PR 1.81, 95% CI 1.36, 2.42). Similarly, women who reported that the provider encouraged HIV screening were significantly more likely to be screened. Nearly half (89/195, 46%) of the women who underwent HIV screening reported doing so primarily based on the recommendation of the provider.

We reviewed medical records to verify self-reported HIV screening and to determine whether HIV test results were available to health-care providers at the time of parturition. Of 300 study participants, 86% (258/300) had prenatal medical records available for review and 14% (42/300) were missing or unavailable at the time of delivery. Of those records available for review, 115 contained HIV status and an additional five had documentation of screening without results noted; 108 (90%) of these 120 women had reported testing, seven (6%) reported they had not been tested and five (4%) were unsure of prenatal HIV testing status. Forty-five percent (116/258) of available records did not contain documentation of screening and 9% (22/258) contained documentation that screening was declined. Among those women reporting prenatal HIV screening, just over half (53%, 103/195) of the medical records contained HIV results. To examine concordance between patient recall and documentation of HIV screening results, we considered only those participants who recalled HIV screening status with certainty (excluding those who were unsure) and medical charts with confirmed results that were available at the time of parturition (n = 132). This review of records for concordance yielded a kappa value of 0.65 (76% agreement, P < 0.001).

DISCUSSION

Despite recommendations for universal HIV screening of pregnant women, prenatal HIV screening remains suboptimal. Only 65% of participants in this study reported prenatal HIV screening. We and others have found that white women, those with higher socioeconomic status and with fewer sexual partners, are less likely to undergo HIV screening in pregnancy. ^5,8,11,12 The lower proportion of women who take up screening in these populations may be due to health-care providers' perceptions of lower risk of HIV infection. Studies have shown that provider risk assessment for Chlamydia trachomatis infection, depression and substance abuse in pregnant teenagers does not accurately identify those at risk. ^13–15 Health-care provider perception of risk can lead to suboptimal rates of chlamydia screening in at-risk populations. ¹³ In contrast, screening for other infectious diseases in pregnancy (hepatitis B virus, rubella and syphilis) exceeds 96%. ⁷ Disproportionate prenatal screening rates between these infectious agents and HIV highlights the need for all obstetrical providers to adopt universal prenatal HIV screening, as is advocated with opt-out testing strategies. ^3,4,16,17

Although women who received prenatal care in a private practice were less likely to be screened for HIV compared with those receiving care in a resident/teaching clinic setting, this difference did not remain significant on multivariable analysis. Nonetheless, the observed disparity in prenatal HIV screening practices highlights groups of health-care providers who should be targeted for enhanced education to improve implementation of universal prenatal HIV screening. Other studies have corroborated that physicians in academic (teaching) or clinic-type practices demonstrate increased knowledge of STIs, more favourable attitudes towards STI screening and increased C. trachomatis screening rates, compared with physicians in private practice. ^13,18,19 The statewide adoption of an opt-out approach to prenatal HIV screening may minimize disparities in delivery of prenatal care and improve uptake of universal prenatal HIV screening.

The impact of the health-care provider's approach to HIV screening was substantial. Opt-out testing with the tenets of patient notification of testing and right to refusal has been supported by ACOG and the Centers for Disease Control and Prevention. ^3,16 Opt-out screening eliminates risk assessment by the health-care provider and may normalize the HIV screening process. Women who reported that screening was offered by their provider as standard practice, mimicking an opt-out testing strategy, were more likely to report prenatal HIV screening than those who recalled screening offered as available if desired. Previous investigations have shown increased rates of prenatal HIV screening after the adoption of opt-out testing strategies. ^17,20 The state of Pennsylvania requires pretest counselling with informed consent prior to HIV testing, and endorses a policy of opt-in HIV screening of pregnant women. This may represent an additional barrier to universal screening and highlights an area for improvement in delivery of prenatal care.

Patient perception of the health-care provider's attitude regarding the importance of HIV screening had a strong impact on the proportion of women who underwent screening. Women reporting that their health-care provider did not convey a sense of the importance of HIV screening were less likely to be screened. Previous studies similarly support the strongly positive effect of provider encouragement on HIV screening. ^21,22 Thus, greater effort to increase awareness of the importance of prenatal HIV screening among obstetric providers is warranted.

Documentation of prenatal HIV screening in medical records was disappointing. Overall, only 38% of all records contained documentation of HIV screening. Our results are similar to the findings of Schuman et al., ²³ who demonstrated that 37% of women had HIV results available in the obstetric record. The awareness of maternal HIV status at the time of labour enables the provision of intrapartum antiretroviral therapy to HIV-positive women, reducing the risk of vertical transmission of HIV, even among antiretroviral-naïve individuals. ²⁴ While rapid HIV testing for labouring patients may be available in some institutions, and should be performed on those women with unknown HIV serostatus, earlier HIV testing and treatment in pregnancy is of utmost importance. We project that the ongoing implementation of electronic health records at our institution will increase the availability of HIV results at the time of labour.

The main limitation of this study is recall bias, as women were asked to describe events that occurred up to seven months earlier. We explored recall accuracy by examining patient medical records and determining concordance with patient recall, and while the concordance between self-reported HIV screening and medical record documentation was good (κ = 0.65), we believe that recall bias may have played a role nonetheless. This recall bias applies to the main question of the study (HIV screening) and also to the perceptions of their provider's approach to screening earlier in pregnancy, and we suspect that the proportion of those screened for HIV would be overestimated. A second limitation of the study is the limited availability of prenatal records, as 14% were unavailable at the time of labour. This highlights an important need for access to records (particularly for women receiving care outside of the delivery hospital). While it is possible that HIV screening was performed but not entered in the obstetric medical record, this limitation underscores the importance of accurate medical record documentation for the timely availability of results. While our study examined screening approaches in private practices and resident and community clinics associated with one institution, the findings may not be reflective of screening patterns in other obstetric populations and/or health-care settings.

In summary, prenatal HIV screening may not be routinely performed despite longstanding national recommendations, and may be improved with the adoption of an opt-out prenatal HIV screening policy. The role of the health-care provider in offering HIV screening as part of standard practice (similar to an opt-out strategy) and encouraging screening is paramount to optimize the uptake of prenatal HIV screening. Continued education of health-care providers regarding the importance of HIV screening in pregnancy and the elimination of barriers to universal screening is needed. The implementation of electronic medical records may increase availability of HIV results at the time of parturition.