Bulterys M, Jamieson DJ, O'Sullivan MJ, Cohen MH, Maupin R, Nesheim S, Webber MP, Van Dyke R, Wiener J, Branson BM, for the Mother-Infant Rapid Intervention At Delivery (MIRIAD) Study
Group . Rapid HIV-1 Testing During LaborA Multicenter Study. JAMA. 2004;292(2):219-223. doi:10.1001/jama.292.2.219
Author Affiliations: Division of HIV/AIDS Prevention, Surveillance and Epidemiology, National Center for HIV, STD, and TB Prevention, Centers for Disease Control and Prevention, Atlanta, Ga (Drs Bulterys, Jamieson, and Branson and Mr Wiener); Department of Obstetrics and Gynecology, University of Miami School of Medicine, Miami, Fla (Dr O'Sullivan); Department of Medicine, Cook County Hospital, Chicago, Ill (Dr Cohen); Department of Obstetrics and Gynecology, Louisiana State University School of Medicine, New Orleans (Dr Maupin); Department of Pediatrics, Emory University School of Medicine, Atlanta, Ga (Dr Nesheim); Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY (Dr Webber); and Department of Pediatrics, Tulane University School of Medicine, New Orleans, La (Dr Van Dyke).
Context Timely testing of women in labor with undocumented human immunodeficiency
virus (HIV) status could enable immediate provision of antiretroviral prophylaxis.
Objectives To determine the feasibility and acceptance of rapid HIV testing among
women in labor and to assess rapid HIV assay performance.
Design, Setting, and Patients The Mother-Infant Rapid Intervention At Delivery (MIRIAD) study implemented
24-hour counseling and voluntary rapid HIV testing for women in labor at 16
US hospitals from November 16, 2001, through November 15, 2003. A rapid HIV-1
antibody test for whole blood was used.
Main Outcome Measures Acceptance of HIV testing; sensitivity, specificity, and predictive
value of the rapid test; time from blood collection to patient notification
Results There were 91 707 visits to the labor and delivery units in the
study, 7381 of which were by eligible women without documentation of HIV testing.
Of these, 5744 (78%) women were approached for rapid HIV testing and 4849
(84%) consented. HIV-1 test results were positive for 34 women (prevalence
= 7/1000). Sensitivity and specificity of the rapid test were 100% and 99.9%,
respectively; positive predictive value was 90% compared with 76% for enzyme
immunoassay (EIA). Factors independently associated with higher test acceptance
included younger age, being black or Hispanic, gestational age less than 32
weeks, and having had no prenatal care. Lower acceptance was associated with
being admitted between 4 PM and midnight, particularly on Friday
nights, but this may be explained in part by fewer available personnel. Median
time from blood collection to patient notification of result was 66 minutes
(interquartile range, 45-120 minutes), compared with 28 hours for EIA (P<.001).
Conclusions Rapid HIV testing is feasible and delivers accurate and timely test
results for women in labor. It provides HIV-positive women prompt access to
intrapartum and neonatal antiretroviral prophylaxis, proven to reduce perinatal
HIV transmission, and may be particularly applicable to higher-risk populations.
The Centers for Disease Control and Prevention (CDC) estimates that
between 280 and 370 infants are born infected with human immunodeficiency
virus (HIV) annually in the United States despite recommendations for universal
prenatal HIV screening and widespread use of antiretroviral drugs in pregnant
HIV-infected women.1- 4 Perinatally
acquired HIV infections may result from missed opportunities for prevention,
such as inadequate prenatal care.4,5 Ideally,
all pregnant women should receive early prenatal care with voluntary HIV testing.
However, for those who do not, rapid testing during labor could provide HIV-infected
women with immediate access to antiretroviral prophylaxis.3,6 Most
women in the United States give birth in hospitals, presenting a crucial opportunity
for systematically offering rapid HIV testing and, when indicated, interventions
to decrease perinatal transmission.7
We sought to determine the feasibility of rapid HIV testing during labor,
assess barriers to HIV testing, and facilitate comprehensive care for HIV-infected
mothers and their infants. A US Food and Drug Administration treatment investigational
device exemption permitted the use of a rapid test before its approval in
November 2002.8,9 This test yields
HIV results in 20 minutes, making it ideally suited for point-of-care use.
This report describes the experience of performing rapid HIV testing during
labor and the factors associated with acceptance of rapid testing.
The CDC funded 16 hospitals in 6 US cities (Atlanta, Ga; Baton Rouge,
La; Chicago, Ill; Miami, Fla; New Orleans, La; and New York, NY) to participate
in the MIRIAD (Mother-Infant Rapid Intervention At Delivery) study, which
offered HIV counseling, voluntary rapid testing, and, if indicated, antiretroviral
prophylaxis to women with undocumented HIV status late in pregnancy. Counseling,
voluntary rapid testing, and antiretroviral prophylaxis (as well as study
enrollment) were offered by labor and delivery nurses, midwives, and obstetrics/gynecology
residents in most hospitals. MIRIAD staff performed study interviews (usually
post partum) and medical record reviews for the study at each hospital. Women
were offered enrollment if they were in active labor (defined as having regular
strong contractions and ruptured membranes or cervical dilation >4 cm) at
a minimum of 24 weeks' gestation (potentially viable neonate). Women not in
active labor were enrolled only if they presented at 34 weeks' gestation or
later to the labor and delivery unit (standard HIV testing could be offered
at <34 weeks per hospital protocol).
We developed a standardized procedure for obtaining written informed
consent for rapid testing and study participation during labor that included
use of a flipchart, which pictorially reviews the main study aspects.10 The study was approved by institutional review boards
of the CDC and each participating institution. One MIRIAD study hospital did
not allow collection of variables (cervical dilation, membrane status, frequency
of contractions, gestational age, number of prenatal care visits, age, years
of education, Hispanic ethnicity, and race) on the eligibility form (all other
study hospitals permitted this) from women who were not approached or who
declined rapid testing and study participation (but this hospital did allow
collection of time of admission). Thus, this hospital is not represented by
variables from this group in relevant analyses herein.
Blood was collected for both rapid testing and enzyme immunoassay (EIA)
and, when indicated, Western blot confirmatory testing. Laboratory technicians
and labor and delivery staff performed rapid test proficiency panels for quality
assurance. The EIA, and if needed, Western blot testing, was performed immediately
following rapid HIV testing in the MIRIAD protocol. Nine institutions used
the Abbott HIV-1/HIV-2 EIA (Abbott Laboratories, Abbott Park, Ill) and the
Genetic Systems HIV-1 Western blot (BioRad Laboratories, Hercules, Calif);
7 institutions used the Genetic Systems HIV-1/HIV-2 Peptide EIA (BioRad Laboratories)
and 4 of these used the Genetic Systems HIV-1 Western blot (BioRad Laboratories)
and 3 used the Cambridge Biotech HIV-1 Western blot (Calypte Biomedical, Rockville,
Md). In all institutions, initially reactive EIAs and rapid tests were repeated
in duplicate; specimens with repeatedly reactive EIA or rapid tests were tested
using Western blot. Women identified as HIV-positive by the OraQuick Rapid
HIV-1 Antibody Test (Ora-Sure Technologies Inc, Bethlehem, Pa) or EIA, and
those with discordant rapid test and EIA/Western blot results were followed
up together with their infant for at least 6 months. The infants were tested
using HIV DNA polymerase chain reaction (PCR) at less than 48 hours, 2 weeks,
6 weeks, and 3 months, and if having an indeterminate status, at 6 months.
To determine factors associated with acceptance of HIV testing, odds
ratios (ORs) and 95% confidence intervals (CIs) were estimated using unconditional
logistic regression, adjusting for study site and other covariates. Reported
ORs should not be misinterpreted as relative risks.11 Sensitivity,
specificity, and positive and negative predictive values were determined using
the EIA/Western blot algorithm as the gold standard. For each of these measures,
CIs were estimated using exact binomial methods.12 Median
turnaround times were compared using the Wilcoxon rank-sum test. We used SAS
statistical software version 8 (SAS Institute Inc, Cary, NC) and S-Plus version
6.1 (Seattle, Wash). All P values reported are 2-sided
and P≤.05 was considered statistically significant.
Between November 16, 2001, and November 15, 2003, there were 91 707
visits to the labor and delivery units of the 16 participating hospitals and
7381 women (8% of all visits recorded) were eligible for rapid HIV testing.
Of these, 1637 women (22%) were not approached for rapid HIV testing for reasons
that included no staff member being available or verification for HIV testing
during pregnancy still pending. Every attempt was made to have continuous
labor and delivery coverage but some hospitals were less successful (implementation
issues are being addressed in separate analyses). The remaining 5744 women
were offered rapid HIV testing. Data on frequency of visiting the units were
not collected and some women may have visited the units more than once. Written
informed consent for both rapid testing and study participation was obtained
from 4849 (84%) women.
Thirty-four women tested HIV-1 positive with both rapid test and EIA,
and all were confirmed by Western blot (prevalence = 7/1000). There were 4
false-positive and no false-negative rapid test results. All 4 patients presented
in active labor and were given antiretroviral prophylaxis, which was stopped
when clinicians were notified that the rapid test result was false-positive.
Sensitivity of the rapid test was 100% (95% CI, 90%-100%) and specificity
was 99.9% (95% CI, 99.78%-99.98%). Negative predictive value was 100%; positive
predictive value was 90% (95% CI, 75%-97%). The EIA had 11 false-positive
results: 5 in women with an indeterminate Western blot result (usually a single
p24 band) and 6 others with negative Western blot results. All 11 women had
negative rapid test results. No false-negative EIA results were identified.
The specificity of EIA was estimated to be 99.8% (95% CI, 99.6%-99.9%); positive
predictive value was 76% (95% CI, 61%-87%).
In analyses adjusted for study site, acceptance of HIV testing during
labor was associated with younger age, Hispanic ethnicity, gestation of less
than 32 weeks, time of admission, and no prenatal care (Table 1). In multivariate analysis, black as well as Hispanic women
were more likely than white women to accept testing. Younger age, gestation
less than 32 weeks, and no prenatal care also remained significant (Table 1). Hospital admission between 4 PM and midnight was associated with lower HIV test acceptance (adjusted
odds ratio [AOR], 0.7; 95% CI, 0.5-0.9); acceptance was lowest on Friday nights
(P = .001).
Median time from blood collection to patient notification of rapid test
result was 66 minutes (interquartile range, 45-120 minutes). In contrast,
the median time from blood collection to receipt of EIA results was 28 hours
(P<.001), with more significant delays for specimens
obtained on weekends vs weekdays (39 vs 25 hours; P<.001).
Among women tested during active labor (n = 4073), the factor most strongly
associated with receipt of test results after delivery
was the number of hours the woman spent in the labor and delivery unit before
delivery (Table 2). Women who
arrived 2 or fewer hours before delivery were more likely to receive rapid
test results post partum than women who arrived 12 or more hours before delivery
(AOR, 34.5; 95% CI, 24.9-47.9).
Twenty-seven of the 34 HIV-infected women identified were in active
labor when they first arrived. Of those, 18 received intrapartum zidovudine
(8 of these also received intrapartum single-dose nevirapine). The remaining
9 women who arrived in active labor did not receive intrapartum prophylaxis
because they arrived near the time of delivery and could not receive the rapid
test result in time to start prophylaxis. The median time between receipt
of rapid test result and zidovudine dosing was 33 minutes; zidovudine was
started on average 6 hours prior to delivery (range, 1-18 hours). All HIV-exposed
infants received zidovudine prophylaxis soon after birth (median, 3.8 hours);
17 infants also received single-dose nevirapine (prophylaxis protocols3 varied by institution). Two foreign-born HIV-infected
women presenting late in pregnancy but who were not in active labor could
not be followed up (they may have returned to their home country).
Of the remaining 32 HIV-exposed infants in the study, 17 were delivered
vaginally and 15 by cesarean delivery (none of these 32 infants were breastfed).
Three infants were found to be HIV-infected: 2 were already DNA-PCR positive
at birth and the other infant was negative at birth but positive by 6 weeks
of age. The mother of this infant arrived too late for intrapartum prophylaxis.
Two of the 3 infected infants had a vaginal delivery. The infant who was DNA-PCR
negative at birth but positive by 6 weeks was delivered vaginally. Of the
32 infants, 27 were followed up for 6 months. Many of the women have started
highly active antiretroviral therapy for their own health (follow-up of women
testing positive is continuing and being addressed in separate analyses).
Of the 34 HIV-infected women identified, 30 were black, of whom 21 (70%) were
born in the United States and 9 (30%) were immigrants from Africa and the
Caribbean. Regarding test performance with non-clade B HIV-1, there was 100%
sensitivity in use of the rapid test with whole blood and no indication in
the MIRIAD study that specificity was decreased in African women.
We found that rapid HIV testing yielded accurate and timely results
to women in labor and that implementing rapid testing was acceptable and feasible.
Overall test acceptance was nearly 85%. Lower acceptance during evening shifts
may be explained in part by fewer available personnel. Informed consent was
obtained not only for rapid HIV testing but also for participation in a research
Based on the feasibility of rapid testing demonstrated in MIRIAD, the
CDC now recommends routine rapid HIV testing using an opt-out approach (ie,
a woman is informed that HIV testing will be routinely done during labor if
her HIV status is unknown but she may decline testing).8,13 Each
woman should be informed that a preliminary positive rapid test result means
that she is likely HIV infected but that this result will need to be confirmed.
If her rapid test result is positive, she should be notified that antiretroviral
drugs will be offered to her and to her newborn. If her rapid test result
is negative, she should be notified that she is almost certainly not HIV infected.
One practical approach to implementing routine rapid testing would be for
each hospital to put in place standing orders to immediately inform any woman
in labor whose HIV status is unknown that she will be tested unless she declines.8,13
The rationale for focusing on women in labor is that there is a brief
window of opportunity for interventions to decrease HIV transmission to the
newborn.6,14- 16 This
rationale is related to the pharmacokinetics of the antiretroviral drugs used
for prophylaxis.3 In decision analysis modeling,
rapid HIV testing during labor is cost-saving to the medical system.17 Our study demonstrates that, in general, results
are timely and that antiretroviral prophylaxis can be provided promptly to
HIV-infected women and their infants. In addition, we have previously shown
that point-of-care rapid testing has the potential to save valuable time compared
with sending specimens to the laboratory.18,19
Appropriate training of staff and quality assurance processes are essential
to ensure accurate rapid HIV test results.9,20 Despite
high test performance, there were still instances in which preliminary HIV-1
screening tests (rapid test or EIA) yielded false-positive results. The decision
to recommend antiretroviral prophylaxis on the basis of an unconfirmed test
result will continue to require clinical judgment and knowledge about HIV
prevalence and the performance characteristics of each test. Although EIA
has been the mainstay of HIV screening, the rapid test demonstrated a higher
positive predictive value in the present study.
Several study caveats should be considered. First, clinical interventions
were not standardized but left up to individual practitioners following US
Public Health Service guidelines.3 Second,
the total number of encounters recorded included some women who visited the
labor and delivery unit more than once. Therefore, the percentage of eligibility
reported in this study (8%) is likely an underestimate of the true proportion
of women with undocumented HIV status. Third, our findings about acceptance
rates and the informed consent process may not directly translate to a nonresearch
setting. Fourth, the utility of the program is in part contingent on accurate
and accessible documentation of the HIV status to avoid redundancy of effort.
The MIRIAD findings are important both in the United States and internationally.
In many settings, including in the developing world, pregnant women with unknown
HIV status are often seen by clinicians for the first time during labor.16,21,22 Rapid testing during
labor can enable pregnant women with undocumented HIV status to learn their
HIV infection status so they can receive antiretroviral prophylaxis and be
referred for comprehensive medical care and follow-up.