Yudin, Mark H. MD, MSc1; Moravac, Catherine BSc1; Shah, Rajiv R. MD1
In the past two decades, the number of individuals with the acquired immunodeficiency syndrome (AIDS) and infected with the human immunodeficiency virus (HIV) has steadily increased in Canada and worldwide. There were 60,160 positive HIV test results in Canada up to the end of 2005.1 Women have accounted for an increasing proportion of positive HIV test reports (10.6% before 1996 and 25.4% in 2005) and AIDS cases (6.6% before 1996 and 21.7% in 2005).1 Among these women, at least 70% are of child bearing age.1
Mother-to-child (vertical) HIV transmission remains an important but preventable mode of transmission. With current obstetric practices and the use of highly active antiretroviral therapy, the rates of transmission can be decreased from a baseline rate of 25% to approximately 1%.2,3 This can be accomplished, however, only if all pregnant HIV positive women are identified and offered treatment. Currently, universal screening of pregnant women is recommended by the Society of Obstetricians and Gynecologists of Canada and the American College of Obstetricians and Gynecologists (ACOG) to identify all pregnant women with HIV infection.4,5 Targeted testing of pregnant women considered at high risk is no longer recommended, because it fails to identify a substantial proportion of HIV positive women.6 With the adoption of universal screening, the Canadian Perinatal HIV Surveillance Program has identified an increasing number of HIV-positive pregnant women (87 in 1993 and 173 in 2005) with a concomitant decrease in the proportion of infected infants (39.5% in 1994 and 4% in 2005)1
There are two different approaches to universal antenatal HIV screening: opt-in and opt-out. Under an opt-in strategy, HIV testing is offered to all pregnant women, but can be done only after informed consent has been obtained. With the opt-out approach, HIV testing is done as part of routine screening, along with other baseline prenatal blood work. The patient must be informed that the test is being ordered, and it is only omitted if the patient opts out of testing. Jurisdictions adopting an opt-out approach to screening have higher testing rates than those using an opt-in strategy, and rates have increased when the strategy was switched from opt-in to opt-out.7,8 In Canada, one half the provinces use an opt-in approach and the other half use an opt-out approach, with Ontario recommending the opt-in strategy. Data published in 2002 revealed that the rate of prenatal HIV testing in Ontario before 2001 was approximately 54%.9
The primary objective of this study was to estimate the rate of accepting HIV testing using a universal opt-out approach and to compare these rates to the provincial average. Our hypothesis was that by using this policy, pregnant women would be more likely to accept antenatal HIV testing and that our rates would therefore be significantly higher than the provincial average. The secondary objective of the study was to explore patient characteristics that may play a role in accepting the test.
MATERIALS AND METHODS
Before initiation of the study, approval was obtained from the hospital Research Ethics Board. The study was performed at the Women’s Health Care Centre at St. Michael’s Hospital in Toronto, Canada. This is a women’s health ambulatory care clinic in downtown Toronto serving a multiethnic patient population that is of varied socioeconomic status. In our clinic, all pregnant patients have a first intake visit with a trained obstetric nurse. This is a 1-hour visit devoted to counseling, antenatal testing, and the dissemination of information. Pre-test counseling occurs and HIV testing is offered using an opt-out approach, with the nurse available to answer any questions that might arise. If patients refuse testing, the reasons for refusal are explored.
The study was performed in two parts, the first being a chart review and the second being a prospective cohort. In part one, charts of women initiating care from September 1, 2003, to January 31, 2004, were reviewed. In part two, a prospective cohort was followed from February 1, 2004, to March 29, 2005, with a data sheet used by the prenatal nurse at the first visit to document acceptance of HIV testing or reasons for declining. There was no difference in the way that testing was offered in either part of the study. For both portions, the following information was collected: presence of an HIV test; results of the test; if testing was not performed, presence or absence of documentation of a reason for refusal; and if testing was refused by the patient, the reason for refusal. Demographic information was also collected to explore factors that may have played a role in accepting or declining the test, including age, racial background (determined by asking what racial background the woman identified with), first language (English or other), fluency in English, country of origin (Canada or other), and insurance status. All obstetric patients cared for during the study period were included, with known HIV positivity before the first visit being the only exclusion criterion.
All statistical analyses were performed using SAS 8.2 (SAS Institute Inc., Cary, NC). Analyses were separately performed for the retrospective and prospective portions of the study. Rates of accepting HIV testing were defined as the proportion of women who accepted testing at their first visit. Among women who refused, the reasons for refusal were tabulated to determine which were the most common.
To determine the association between patient characteristics (racial group, using English as a first language, fluency in English, born in Canada, and insurance coverage) and likelihood of acceptance of testing we used logistic regression models. Women were categorized as Caucasian, black, Hispanic, Asian, or other. We explored several models, comparing each racial group to each other group (eg, Asians used as the reference group compared with Caucasians, blacks, Hispanics, and other races), and also comparing each group with all others (eg, Asians compared with non-Asians).
For comparisons between our clinic testing acceptance rates and the provincial average, the overall proportions of acceptance and 95% confidence intervals (CIs) in the retrospective and prospective portions of the study were calculated. These were then compared with the provincial average screening rate of 54% at the time of study initiation and to the rate of 84% while the study was ongoing.
For the sample size calculation we used the provincial average screening rate of 54% at the time of study initiation and assumed that our rate was at least 60%. A total of 537 subjects were required to test whether the prevalence in our study was statistically different from the provincial value, using a two-tailed t test with 80% power at the .05 level of significance. We therefore planned to enroll at least 537 women in each phase (retrospective and prospective) of the study.
In the retrospective part of the study, 545 pregnant women had first intake visits, and 462 (84.8%; 95% CI 0.82–0.88) accepted HIV testing. During the prospective portion, 1233 women had first visits, and 1,140 (92.5%; 95% CI 0.91–0.94) accepted testing. Table 1 presents the racial background of the women for both study periods, and the proportion of women tested from each race. Most Asian women were Korean, with the remainder originating from China or Japan. Most Hispanic women were from Mexico or South America. Among the black women, one half were of African descent and one half were from the Caribbean. Table 2 presents data on other patient characteristics. While only one half of the women reported English as a first language, the majority were fluent in English. Almost all women had some form of insurance coverage.
With women who refused testing, we attempted to estimate the reason for refusal. In the retrospective portion of the study, 83 (15.2%) of the women were not tested. Of these, 55 (66%) had no reason documented. During the prospective portion, 93 (7.5%) of the women were not tested. In this phase, a data sheet was used by the prenatal nurse to try to increase the likelihood of having documentation of reasons for refusal. Nonetheless, in 27 (29%) cases there was no reason recorded. For women with documented reasons, the most common were that the pregnancy was found to be nonviable at the first visit and therefore testing was not needed, a recent test had been done and the patient felt a repeat test was unnecessary, or the feeling by the patient that she was not at risk.
Logistic regression modeling with multiple pair-wise comparisons was performed to estimate whether there were differences between racial groups with respect to accepting HIV testing. During the retrospective phase, Asians were significantly less likely than Hispanics to accept testing (odds ratio [OR] 0.4; 95% CI 0.2–0.9; P=.02) and less likely than non-Asians as a group to accept testing (OR 0.6; 95% CI 0.4–0.9; P=.03). Hispanics were significantly more likely than Asians to accept testing (OR 2.6; 95% CI 1.2–5.6; P=.02). During the prospective phase, Asians were significantly less likely than non-Asians to accept testing (OR 0.4; 95% CI 0.3–0.6; P<.001), and Hispanics were significantly more likely than non-Hispanics to accept testing (OR 6.9; 95% CI 2.2–22.0; P=.001). Table 3 presents the results of the logistic regression model comparing Asians to each other race. Caucasians and blacks were both significantly more likely than Asians to accept testing. Table 4 presents the results of the model comparing Hispanics to each other race and shows that each racial group was significantly less likely than Hispanics to accept testing. In other models, Hispanics were more likely to accept testing than Caucasians (OR 4.8; 95% CI 1.5–15.8; P=.01), blacks (OR 7.6; 95% CI 2.1–27.3; P=.002), and other races (OR 6.3; 95% CI 1.5–25.7; P=.01).
In addition to race, other patient characteristics were examined to determine whether they were predictive of accepting HIV testing. Neither English as a first language nor nationality (country of birth) were significantly associated with accepting HIV testing. However, being fluent in English was predictive in the retrospective phase, with those fluent being more likely to be tested (χ2, P<.01). Using logistic regression, women who were fluent in English were twice as likely to be tested as those who were not fluent (OR 2.0; 95% CI 1.2–3.3; P=.01). Among all women who were not fluent in English, Asians were significantly less likely to accept testing than the other racial groups (OR 0.1; 95% CI 0.03–0.2; P<.001).
Using binomial proportions with 95% confidence intervals for our testing rates, we determined that there was a significant difference between our rate of 84.8% in the retrospective phase and the 54% provincial average at the time of study initiation (P<.001), but no significant difference between our rate of 84.8% and the more recent rate of 84%. For the prospective phase, there was a significant difference between our rate of 92.5% and both 54% (P<.001) and 84% (P<.001). Finally, there was a significant difference between the rate of 84.8% during the retrospective phase and 92.5% in the prospective phase (P<.001).
In this study, by having a first visit with a trained obstetric nurse using an opt-out strategy, the rate of accepting HIV testing in our clinic of 92.5% was significantly higher than the provincial average. Rates were affected by patient race, with Asian women significantly less likely and Hispanic women significantly more likely to accept testing. Rates were also affected by language, with women who were fluent in English significantly more likely to be tested. Rates were not affected by nationality (country of origin) or insurance status.
Although universal voluntary HIV screening is advocated for pregnant women by both Society of Obstetricians and Gynecologists of Canada and ACOG,4,5 there is great variation in how this is actually put into clinical practice. Different jurisdictions use different approaches for universal screening, with both opt-in and opt-out strategies acceptable. In the United States, the opt-out approach is recommended by the Centers for Disease Control and Prevention10 and ACOG.5 In Canada, different provinces have different policies, with Ontario recommending an opt-in approach.
Although both the opt-in and opt-out strategies are effective, testing rates are higher when using an opt-out approach to screening. In a surveillance report of testing rates for American states and Canadian provinces, rates were consistently higher in jurisdictions using the opt-out strategy, with the published rate for Ontario being 54%.9 In a study from Denver, the use of a verbal opt-out system resulted in a testing rate of 98.2%.11 Studies have also shown that rates increase when switching from an opt-in to an opt-out approach. In a study performed in Alabama, after adopting an opt-out strategy in 1999, testing rates increased from 75% to 88% (P<.001).7 In Alberta, after implementing opt-out testing in 1998, rates increased by 28%, and only 3.3% of women declined testing over the first year with the new approach.8 In our clinic, we were able to document a testing rate with the opt-out approach that was significantly higher than the provincial average in a province that recommends using opt-in testing. The testing rates in Ontario have risen steadily over the past 5 years, with the rate for 2005 at 91.2%.12 With provincial opt-out testing this rate might be increased even further.
There may be several reasons why an opt-in approach results in lower testing rates, including women choosing not to opt in because of a perceived lack of risk, previous testing, or a lack of endorsement by the health care provider. Other attractive attributes of opt-out testing include less anxiety, no fear of being “singled out” for testing, and less time required on pre-test counseling.13 However, some critics of opt-out testing have raised concerns that it may compromise autonomy or be unethical because of the absence of traditional informed consent. It is therefore paramount that opt-out testing be done correctly, with each woman informed that the test will be performed as part of the routine prenatal blood work and given the opportunity to refuse.
In addition to examining our testing rates with an opt-out strategy, we also wanted to explore patient characteristics that may be associated with accepting HIV testing. We found that race was a predictor of accepting, with Asian women less likely and Hispanic women more likely to be tested. Race has been shown in other studies to be a predictor of prenatal HIV testing.7,14–17 This may be due to ethnic or cultural diversity with respect to how HIV is perceived. Different racial groups may view themselves to be more or less at risk for HIV, and may have different levels of acceptance toward the relevance of antenatal HIV testing. Counseling may also differ in different parts of the world, and women born outside of North America may have been counseled differently in their home countries regarding the importance of prenatal testing. The finding that acceptance rates varied by race is provocative, and further study may elucidate specific factors within racial groups that motivate women to agree to or refuse testing during pregnancy.
Although having English as a first language did not predict likelihood of testing, being fluent in English did, with women who were fluent being significantly more likely to accept HIV testing. There was also an interaction between race and fluency in English, with non–English-speaking Asian women less likely to accept testing than non–English-speaking women of other races. This may be related to the lower rates of testing among Asian women in the study overall. We do not believe that fluency in English was a surrogate marker for understanding testing in this study, because our prenatal nurses are multilingual, with fluency in both Korean and Spanish available. However, it is possible that there were subtle differences in the way the testing options were presented to non–English-speaking women which may have influenced their decision to accept or refuse testing. It is interesting to note that fluency in English was a significant predictor of acceptance of testing but having English as a first language and nationality were not. Perhaps fluency in English reflects integration into Canadian society, and these women therefore more readily accepted testing.
Previous studies have identified other factors that affect the likelihood of accepting HIV testing, including patient age, education level, marital status, income, insurance status, and provider age and gender.14–18 In this study, insurance status was not related to accepting HIV testing, although with universal health care in Canada the majority of our patients had insurance coverage. There was a relationship between physician and acceptance of testing, but this relationship was confounded by patient race, because certain physicians have large patient populations from specific ethnic backgrounds. We were unable to document a relationship between physician gender and rates of accepting HIV testing.
Some limitations of this study should be acknowledged. The circumstances under which prenatal HIV counseling and testing are performed in our clinic may not be representative of what occurs in other offices. Many prenatal care providers do not have a system such as ours in place, and may feel that they don’t have the time to devote to pretest and posttest counseling. The use of the study data sheet coupled with the length of time devoted to counseling may have positively affected the acceptance rates we demonstrated. The study was performed at an inner city Canadian prenatal clinic with a patient population that varied widely with respect to racial background and socioeconomic status and in which the majority of women were insured. Therefore, our results may not be applicable to all practice settings.
The findings of our study lend further support to the existing literature that an opt-out strategy for HIV screening in pregnancy results in higher testing rates than an opt-in approach. By applying opt-out testing, we were able to demonstrate higher testing rates than the provincial average. An opt-out strategy for HIV screening in pregnancy should be considered in all jurisdictions to maximize testing rates and therefore identify HIV positive women, providing the opportunity for treatment and prevention of vertical transmission.
2. Connor EM, Sperling RS, Gelber R, Kiselev P, Scott G, O’Sullivan MJ, et al. Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment. Pediatric AIDS Clinical Trials Group Protocol 076 Study Group. N Engl J Med 1994;331:1173–80.
3. Cooper ER, Charurat M, Mofenson L, Hanson EC, Pitt J, Diaz C, et al. Combination antiretroviral strategies for the treatment of pregnant HIV-1-infected women and prevention of perinatal HIV-1 transmission. J Acquir Immune Defic Syndr 2002;29:484–94.
4. Society of Obstetricians and Gynecologists of Canada. Clinical Practice Guidelines, Policy Statement: HIV testing in pregnancy. J Obstet Gynaecol Canada 1997;19:767–8.
5. ACOG Committee on Obstetric Practice. ACOG committee opinion number 304, November 2004. Prenatal and perinatal human immunodeficiency virus testing: expanded recommendations. Obstet Gynecol 2004;104:1119–24.
6. Samson L, King S. Evidence-based guidelines for universal counselling and offering of HIV testing in pregnancy in Canada [published erratum appears in CMAJ 1998;159:22]. CMAJ 1998;158:1449–57.
7. Stringer EM, Stringer JS, Cliver SP, Goldenberg RL, Goepfert AR. Evaluation of a new testing policy for human immunodeficiency virus to improve screening rates. Obstet Gynecol 2001;98:1104–8.
8. Jayaraman GC, Preiksaitis JK, Larke B. Mandatory reporting of HIV infection and opt-out prenatal screening for HIV infection: effect on testing rates. CMAJ 2003;168:679–82.
9. Centers for Disease Control and Prevention (CDC). HIV testing among pregnant women—United States and Canada, 1998–2001. MMWR Morb Mortal Wkly Rep 2002;51:1013–6.
10. Centers for Disease Control and Prevention. Revised recommendations for HIV screening of pregnant women. MMWR Recomm Rep 2001;50:63–85.
11. Breese P, Burman W, Shlay J, Guinn D. The effectiveness of a verbal opt-out system for human immunodeficiency virus screening during pregnancy. Obstet Gynecol 2004;104:134–7.
13. Simpson WM, Johnstone FD, Goldberg DJ, Gormley SM, Hart GJ. Antenatal HIV testing: assessment of a routine voluntary approach. BMJ 1999;318:1660–1.
14. Wang F, Larke B, Gabos S, Hanrahan A, Schopflocher D. Potential factors that may affect acceptance of routine prenatal HIV testing. Can J Public Health 2005;96:60–4.
15. Ethier KA, Fox-Tierney R, Nicholas WC, Salisbury KM, Ickovics JR. Organizational predictors of prenatal HIV counseling and testing. Am J Public Health 2000;90:1448–51.
16. Lansky A, Jones JL, Frey RL, Lindegren ML. Trends in HIV testing among pregnant women: United States, 1994-1999. Am J Public Health 2001;91:1291–3.
17. Lindsay MK, Adefris W, Peterson HB, Williams H, Johnson J, Klein L. Determinants of acceptance of routine voluntary human immunodeficiency virus testing in an inner-city prenatal population. Obstet Gynecol 1991;78:678–80.
18. Ogilvie G, Adsett S, Macdonald G. Do physicians discuss HIV testing during prenatal care? Can Fam Physician 1997;43:1376–81.
Figure. No caption available.