Factors Associated with Human Immunodeficiency Virus Screening of Women During Pregnancy, Labor and Delivery, United States, 2005–2006
The purpose of this study was to estimate prenatal human immunodeficiency virus (HIV) screening rates prior to and on admission to labor and delivery (L&D) and to examine factors associated with HIV screening, including hospital policies, with a comparison of HIV and hepatitis B prenatal screening practices and hospital policies. In March 2006, a survey of hospitals (n = 190) and review of paired maternal and infant medical records (n = 4,762) were conducted in 50 US states, DC, and Puerto Rico. Data from the survey and medical record review were analyzed using SAS software v9.2 (SAS Institute, Cary, NC). HIV testing before delivery occurred among 3,438 women (73.9 %); African American and Hispanic women were more likely to be tested than white women [aOR 2.22, 95 % CI (1.6–3.1) and aOR 1.55, 95 % CI (1.1–2.2), respectively]. Among women without previous HIV testing, 138 (16.6 %) were tested after admission to labor and delivery. Policies to test women with undocumented HIV status in at delivery were present in 65 (36.3 %) hospitals. HIV testing after admission to L&D was more likely in hospitals with policies to test women with undocumented HIV status [aOR 5.91, 95 % CI (2.0–17.8)]. Overall, policies and screening practices for HIV were consistently less prevalent than those for hepatitis B. Many women are not being routinely screened for HIV before or at delivery. Women with unknown HIV status were more likely to be tested in L&D in hospitals with testing policies.
Since the mid-1990s, the number of perinatal HIV infections in the United States has decreased dramatically. CDC estimates that 1,650 HIV-infected children were born in 1991, decreasing to 480 HIV-infected children in 1996 . In 2011, there were an estimated 127 children diagnosed with HIV in the United States through perinatal transmission . While these trends demonstrate substantial progress towards reducing perinatal HIV transmission in the United States, recent estimates show an increase in the number of HIV-infected pregnant women giving birth each year, from 6,000 to 7,000 in 2000 to approximately 8,700 in 2006 [3, 4]. This increase indicates a growing need for interventions to prevent perinatal HIV transmission.
Early first-trimester testing provides a timely window to offer ARV prophylaxis and other interventions to prevent mother-to-child transmission of HIV. Avoidance of breastfeeding, scheduled cesarean section when appropriate, and provision of antiretroviral medications to the mother during pregnancy and to the exposed infant can reduce transmission rates to <1 % ; however, a woman must first be tested and identified as having HIV infection in order for these time-sensitive interventions and treatments to have optimal benefit. The United States Public Health Service first recommended routine voluntary counseling and HIV testing for all pregnant women in 1995 . The Institute of Medicine, the American College of Obstetricians and Gynecologists, and the American Academy of Pediatrics have similarly recommended universal HIV screening of pregnant women since the late 1990s [7–9]. To streamline incorporation of HIV testing into routine prenatal care, CDC released updated recommendations in 2006, advising providers to screen pregnant women as part of the standard battery of prenatal tests unless a woman refuses (“opt-out” testing). Screening women in labor and delivery (L&D) in the absence of a documented prenatal HIV test result is also included in the testing recommendations . Since 2006, many states have changed their routine HIV testing laws and statutes to align with the revised CDC testing recommendations [11, 12]. However, there continues to be some confusion among obstetricians on their state’s testing requirements and some providers in states with “opt-out” laws do not always follow that approach .
Human immunodeficiency virus-infected women who receive insufficient prenatal care miss the opportunity for screening and interventions during pregnancy, thereby increasing the risk of transmitting HIV to the infant. Conducting rapid HIV testing for women with an unknown HIV status in L&D settings has been found to be an accurate and feasible way to identify HIV infection in women and to provide short-term prophylaxis to both mother and infant, which can reduce the risk of perinatal transmission . Thus, it is vital for hospitals to routinely screen women who are admitted to L&D without a documented HIV test result. In 2008, 3 % of women with known HIV infection in 15 high HIV prevalence areas were diagnosed in L&D, while 71 % were diagnosed before pregnancy and 24 % during pregnancy .
In 2002, a US national survey using self-reported data found that 69.2 % of women were tested for HIV during prenatal care . Since 2002, no prenatal HIV testing estimate using data from all states have been published. The purpose of this study was to: (1) identify factors associated with HIV screening of pregnant women in the United States, (2) estimate HIV screening rates of pregnant women during pregnancy and at L&D using objective methods, (3) review hospital policies in L&D units as they relate to HIV testing, (4) compare hospital practices and policies for HIV screening to those for maternal hepatitis B screening and (5) provide a baseline estimate of HIV testing practices and policies before CDC’s 2006 HIV testing recommendations.
Materials and Methods
The data for the study came from a nationwide hospital practice survey with medical record reviews. The survey was primarily designed to assess practices around perinatal hepatitis B testing  but also collected data regarding HIV testing and practices. This paper will focus on questions regarding HIV from both the policy survey and the record review. The CDC National Center for Immunization and Respiratory Diseases Human Research Protection Contact reviewed the protocol and determined it to be a non-research assessment of public health practice.
Hospitals were eligible for inclusion in the sample if they (1) had an L&D unit in 2005, (2) delivered more than 100 live births per year, and (3) were located in the 50 states, DC, and Puerto Rico. Based on the 2003 American Hospital Association (AHA) database, 3,102 hospitals (52 %) were eligible for survey selection. The sampling strategy incorporated a two-stage stratified design. The first stage of the survey consisted of a simple random sample of birthing hospitals in each state. The sampling weight was the inverse of the selection probability of the hospital and was calculated using the following formula: Hospital weight = numbers of total eligible hospitals in each state/numbers of selected hospitals in each state. The second stage consisted of a sample of approximately 25 consecutive paired mother and neonatal records from each hospital in the sample. The actual number of paired records ranged from 21 to 29. Thus: Final weight = hospital weight x (numbers of total live births per year in each hospital/numbers of selected paired mothers and neonates in each hospital).
Epi Info 2002 was used to calculate the appropriate sample size. Allowing for an 80 % response rate, a target sample size of 246 hospitals was required to estimate the proportion of hospitals reporting a given policy with a 95 % CI of not more than ±7 % points wide. A random sample of 254 hospitals was drawn from 51 strata (50 states plus one stratum for DC and Puerto Rico). All hospitals were called to verify delivery services and to identify a contact person. Twelve hospitals did not provide delivery services and were excluded.
The hospital policy survey was sent during March 2006, by mail or in person, to either a nursing supervisor or a clinical nurse manager at the selected hospitals. The hospital survey examined hospital policies (1) to review records in L&D units for a prenatal HIV test and (2) to test women with an undocumented HIV status as soon as possible after admission to L&D.
Medical Record Review
a documented HIV test performed prior to admission and a test date (maternal record)
an HIV test performed on admission to L&D (maternal record)
a recorded maternal HIV test result (neonatal record)
SAS software v9.2 (SAS Institute, Cary, NC) was used to analyze data from the hospital survey and medical record review. SUDAAN v10 was used to account for the complex stratified sampling design. In the tables, all percentages of outcome variables and their associated odds ratios are weighted to represent all births in the United States for 2006. The calculation of weights was associated with the basic sample design described above.
Multiple logistic regression using weights described above was employed to explore relationships between HIV testing prior to admission and the following variables, while accounting for potential confounders: age, race, insurance status. In addition, associations between odds of testing on admission to L&D and hospital-level characteristics including live births/year, written policy to review prenatal HIV test results and written policy to test women with an undocumented HIV test in L&D were analyzed. Geographic location of the hospital, categorized as urban or rural, was omitted from this analysis due to collinearity with the number of live births/year. Testing rates from the medical record review and hospital policies for HIV and hepatitis B were compared using a matched pairs analysis. McNemar’s test was used to test for differences between the discordant pairs, and 95 % confidence intervals for the proportions were calculated. All statistical tests were performed at the 5 % significance level.
Characteristics of hospitals with a written policy to review medical records for a prenatal HIV test result and to test women with an undocumented HIV test result at L&D, Hospital Survey and Medical Record Review, United States 2005–2006 N = 190
Written policy to review medical record
Written policy to test for HIV
95 % CIc
95 % CIc
Live births (annual)
p = 0.07
p = 0.03
Level of neonatal care
Level 1: basic
p = 0.01
p = 0.12
Level 2: specialty care
Level 3: neonatal ICd
p = 0.05
p = 0.14
p = 0.48
p = 0.14
p = 0.19
p = 0.52
Not for profit
p = 0.01
p = 0.01
Among the hospitals surveyed, 55.2 % had a written policy to review medical records at L&D for a prenatal HIV test, and 36.3 % had a written policy to test for HIV as soon as possible after admission for women admitted to L&D who did not have a documented HIV test result (Table 1). Hospital characteristics significantly associated with the presence of a written policy to review medical records included higher level of neonatal care (Chi squared p = 0.01), OB as attending provider (Chi squared p = 0.05) and urban location of the hospital (Chi squared p = 0.01). The number of live births/year was significantly associated with hospitals having a written policy to test women in L&D with an undocumented HIV status (p = 0.03).
Patient characteristics and documentation of testing for HIV before admission to L&D by selected demographic characteristics, Medical Record Review and Hospital Survey 2005–2006, United States N = 4,762
No. with documented test (wt%)
No. without documented test (wt%)a
95 % CI
95 % CI
Overall, 3,438 (73.9 %) of women were tested for HIV prior to admission (Table 2). African American and Hispanic women were more likely to have been tested for HIV prior to admission than white women [Adjusted OR (aOR) 2.22, 95 % CI 1.6–3.1 and aOR 1.55, 95 % CI 1.1–2.2, respectively]. Age and insurance status were not found to be statistically significant for HIV testing prior to admission in the adjusted analysis. A maternal HIV test result was documented in 64.2 % of neonatal records.
Documentation of testing for HIV at L&D among women with no previous HIV testinga, by selected demographic and delivery hospital characteristics, Medical Record Review and Hospital Survey, United States 2005–2006 N = 1,324
No. with documented test (wt%)
No. without documented test (wt%)
95 % CI
95 % CI
Live births (annual)b
Delivered at hospital with written policy to test at L&D
Additionally, women were more likely to be tested in L&D in hospitals with ≥2,000 live births/year (aOR 5.20, 95 % CI 1.6–17.3) than in smaller hospitals. Women were more likely to be tested for HIV on admission to L&D if they delivered in a hospital with a written policy to test women with an undocumented HIV status (aOR 5.91, 95 % CI 2.0–17.8). Among women screened for HIV, either prior to or on admission to L&D, a maternal HIV test result was documented in 64.2 % of neonatal records.
HIV and HBV Policies and Testing
HIV and hepatitis B hospital policy and testing practices comparison. Hospital survey and medical record review, United States 2005–2006
HIV and hep B
Hep B only
Neither HIV or hep B
95 % CI
95 % CI
95 % CI
95 % CI
Hospital policy n = 190
Written policy to review medical records
p < 0.01
Written policy to testb
p < 0.01
Medical Record n = 4,762
Tested prior to admission
p < 0.01
Tested on admission
p < 0.01
Test recorded in neonatal record
p < 0.01
Tested prior to and/or on admission
p < 0.01
A review of maternal medical records showed that prior to or on admission to L&D, 77.0 % of women received both an HBsAg test and an HIV test, whereas 1.3 % received only an HIV test, 19.7 % received only an HBsAg test, and 2.0 % received neither test. Furthermore, neonatal medical records documented that 63.1 % women received both an HBsAg test and an HIV test, whereas 1.1 % received only an HIV test, 27.2 % received only an HBsAg test, and 8.5 % received neither test. In both the maternal and neonatal records, women were more likely to receive an HBsAg test only than an HIV test only (Chi squared all p < 0.01). Overall, 91.1 % of women were screened for hepatitis B prior to admission to L&D and 74 % were screened for HIV prior to admission to L&D.
This study found that HIV testing of pregnant women fell short of meeting universal testing recommendations, particularly in L&D settings. Approximately one out of four pregnant women was not screened for HIV before admission to L&D, and few were subsequently screened in L&D. These findings represent missed opportunities to test women and to identify HIV infection, to provide interventions to prevent perinatal transmission and to link the mother and infant to care and support services.
At the time of the study, only 55.2 % of hospitals had a policy to review records for HIV test results, and 36.3 % had a policy to test for HIV on admission to L&D if no prenatal result was documented. The presence of a policy to test women with an undocumented HIV status was found to be significantly associated with testing for HIV on admission to L&D. Other variables related to HIV testing include race/ethnicity and age. Testing was more likely among African American and Hispanic women, as well as younger women, which suggests that some providers may test their patients based on perceived risk, despite recommendations to test all pregnant women.
In comparing HIV and HBsAg testing practices and policies, a higher percentage of women were tested prenatally for hepatitis B (91.1 %) than for HIV (74 %), and a greater proportion of hospitals had policies related to hepatitis B than for HIV. Additionally, 19.5 % of women were tested for hepatitis B prior to admission to L&D but did not receive an HIV test. Other studies have found higher screening practices for hepatitis B than HIV [18, 19], further highlighting this missed opportunity. Because screening for HIV and hepatitis, in addition to other infections such as syphilis, is recommended for all pregnant women during the first prenatal care visit and in L&D for women with no documented test [10, 20], streamlining policies and screening practices to include all recommended tests may help close this gap between prenatal screening rates.
Data from this study also show there was an incomplete transfer of HIV testing information from maternal records to neonatal records. A positive maternal test result recorded in the neonatal record may help facilitate appropriate diagnostic testing of the infant and prompt necessary treatment. Adverse outcomes for infants have been related to errors in test result transcription and documentation [21, 22]. Transferring test result information accurately and completely should be routine practice in hospital settings.
Although current prenatal HIV testing rates and related policies are unknown, findings from this study can guide efforts to minimize missed opportunities and maximize positive birth outcomes. Hospitals and prenatal care providers should actively promote routine HIV screening early in pregnancy and on admission to L&D for women with an unknown HIV status. Provider endorsement of HIV testing is found to be a factor in a woman’s decision to accept HIV testing ; therefore, it is essential to work with prenatal care providers to assure HIV screening is recognized as an essential component of the standard of care for all pregnant women. Education and training programs targeting prenatal and hospital providers should take into account each state’s current law or regulation regarding testing pregnant women for HIV. Other studies have found a higher prevalence of HIV testing in opt-out areas and when potential barriers, such as written informed consent, are removed [24–26]. State testing laws were not included in this analysis, but are likely a factor in screening practices of pregnant women. Further data are needed to assess how prenatal HIV screening rates are influenced by state laws, specifically laws regarding screening pregnant women.
Additional activities should include working with hospitals to develop and implement testing policies and incorporate them into routine practice in L&D settings. Small hospitals with basic level of care and those located in rural areas should be included in these initiatives. Finally, strengthening mechanisms to transfer information between providers and medical records should be a key component of education efforts.
This study had several limitations. First, the response rate to the survey was 78.5 %. Thus, the characteristics of the sample may not reflect the characteristics of all hospitals with >100 annual live births. Second, it is unknown what proportion of untested women were offered an HIV test and declined. Additionally, using consecutive paired sampling of medical records rather than a random sampling method may have introduced some bias in the second stage sampling. Also, some record reviews were completed by hospital staff and not trained abstractors, so the quality of the abstracted data may vary. Furthermore, if an L&D record did not include the prenatal care record or if prenatal test results were not recorded in the hospital record, then the prenatal estimate may be biased. Lastly, the study was conducted before CDC published its revised 2006 HIV testing recommendations. Although data were not collected for more recent years, the data presented in this paper are the most up to date national estimates of prenatal HIV testing using objective clinical data and can be used as a baseline to assess the impact of the “opt-out” testing strategy on prenatal testing practices and policies.
The findings from this study demonstrate that, during the study period, HIV screening for pregnant women was not a universal practice in prenatal care or L&D settings. The institution of written policies in hospital settings is one strategy to identify HIV infection among pregnant women. Further efforts are needed to promote HIV screening for all pregnant women during each pregnancy in order to reduce, and ultimately eliminate, mother to child HIV transmission in the United States.
We extend great appreciation to all delivery hospitals and perinatal hepatitis B coordinators for participating in the survey and record review without which this data would not be available and Pascale Wortley, MD MPH (CDC National Center for Immunization and Respiratory Diseases, Immunization Services Division, and Respiratory Diseases, Immunization Services Division, Atlanta, GA). Health Research and Evaluation Branch and Immunization Services Division, CDC.
Conflict of interest
None of the authors have a conflict of interest.