Setting
Kabul, the capital of Afghanistan, had a population of approximately 2.5 million in 2005, with an estimated facility-based delivery rate for urban areas of 85.2% [9, 12]. There are four public maternity hospitals in the city, providing free obstetric care locally and serving as regional referral centers for other facility- or home-based deliveries. At the time of the study, the three hospitals with highest delivery volumes in the previous quarter (Malalai, Rabia Balkhi, and Khair Khana) were included. Prior to enrollment, approval was obtained from the Afghan Ministry of Public Health and University of California, San Diego (Protocol #060083) institutional review boards. All study procedures were followed in accordance with the ethical standards of these review boards.
Participants
Eligibility for this cross-sectional study was confined to pregnant women admitted in stable medical condition; women with conditions requiring urgent medical intervention (e.g. eclampsia, hemorrhage), imminently delivering (cervical dilation ≥ 8 cm), or unable to provide informed consent were excluded. Potential participants were consecutively evaluated from June to September, 2006. Potential study participants either declining participation or ineligible for participation were enumerated and assessed by age or parity for comparison to enrolled participants.
Questionnaire Development
Questionnaires were developed with input from a local advisory board of reproductive health specialists in Kabul. Following pilot-testing and board review, the final questionnaire was edited, translated, and back-translated for accuracy. The instrument assessed demographics, socioeconomic status, and time spent outside the country, including prior refugee status. Past medical and obstetric history was queried, including number, site, year, and outcome of prior deliveries and prenatal care for the current pregnancy. Risk factors for sexually transmitted and blood-borne infections, such as drug use or number of therapeutic injections, were also assessed.
Enrollment Procedures
Study representatives were female medical providers (midwives and physicians) who were not employees of the study hospitals. Study representatives were trained in human subjects' research, questionnaire administration, rapid testing, and pre and post-test counseling. Trained representatives were present in shifts to ensure 24 hour enrollment coverage. All women presenting for admission were assessed for study entry. Following medical evaluation and admission, eligible women were invited to accompany the study representative to a private room for informed consent; all women provided written informed consent prior to enrollment. Consented participants were assigned a study number, unlinked to any hospital identifiers, that was the sole identifier on all study materials.
Participants completed an interviewer-administered questionnaire. Pre-test counseling, whole blood rapid testing with finger stick specimen, and post-test counseling with provision of results were then performed. Following these activities, participants were escorted back to the labor ward. Participants who were too uncomfortable for interview or testing following consent were excused from the study (n = 3); those who were somewhat uncomfortable during interview had the interview suspended and testing performed, with the questionnaire completed after delivery. Free testing was the only compensation provided for participation.
Patient Testing and Prophylaxis
Rapid testing was performed with Abbott Determine® HIV 1/2, Syphilis TP, and HBsAg (Abbott Diagnostics Japan, Tokyo, Japan) and with Standard Diagnostics Anti-HCV (Standard Diagnostics Laboratories, Kyonggi-do, Korea). A second rapid test (OraSure OraQuick HIV 1/2®, OraSure Technologies, Bethlehem, PA) was available for tandem HIV testing, with Western Blot (HIV BLOT 2.2®, GeneLabs Diagnostics, Singapore) available for confirmation. Treponema pallidum particle agglutination (TPPA) assay (Fujirebio Diagnostics, Malvern, PA, USA) and rapid plasma reagin (RPR) (Spectra Diagnostics, Cairo, Egypt) were available for syphilis confirmation and clinical monitoring. All participants with positive rapid tests underwent intravenous sampling, from which serum was prepared and stored for confirmatory testing.
HBV was confirmed with a second, serum-based rapid test (Standard Diagnostics HBV, Standard Diagnostics Laboratories, Korea), as nucleic acid testing was not available. Use of the Determine® and Standard Diagnostics® rapid tests with sensitivities of 99.0% and 99.0% and specificities of 99.0% and 100.0%, respectively, have a positive predictive value (PPV) of 99.988%, assuming a baseline prevalence of HBsAg of 5.0% [13, 14]. HCV antibody was confirmed with RIBA (RIBA 3.0 SIA®, Chiron Corporation, Emeryville, CA, USA). All confirmatory testing was performed at the VCT Center in Kabul by trained laboratory personnel.
Participants with any positive rapid test (or two for HIV) were treated as presumptive positives and received intra- or post-partum prophylaxis. Protocols were in place for management of prevention of vertical transmission of HIV or HBV and maternal and neonatal syphilis, following international guidelines [15–17].
Analysis
Student's T-test and the binomial probability test were used to compare participants and those ineligible or declining entry. Descriptive statistics were generated to further characterize the study population. Prevalence for each infection was calculated with confidence intervals based on the Poisson distribution for anti-HCV and HBsAg. The sample size was sufficient to determine at least a 28% difference in demographic and risk variables between HBsAg-positive and HBsAg-negative participants (P = 80%, two-sided alpha = 0.05). Correlates of HBsAg were assessed with univariable logistic regression. Data were entered and cleaned with EpiData 3.1 (EpiData Association, Odense, Denmark) and analyzed with Stata 8.0 (Stata Corp, College Station, Texas).
Hepatitis B vaccination modeling
The Hepatitis B Disease Burden and Vaccination Model Version 1.2 from the Centers for Disease Control and Prevention was used to estimate hepatitis cases averted for several vaccination strategies, given measured HBsAg seroprevalence [18]. The model requires input of prevalence of maternal hepatitis B e antigen (HBeAg) and child and adult hepatitis B core antibody (anti-HBc) among those positive for HBsAg, for which there is no data from Kabul. A small study done among Afghan refugees in Iran reported that, of women testing positive for HBsAg, no women had HBeAg [11]. However, due to the small sample size, we elected to use data from other sources for this calculation. India was the country of closest geographic proximity with published HBeAg prevalence among antenatal populations positive for HBsAg, ranging between 7.8% and 18.0%; [19–21] we used the estimate of 12.5% for presumed HBeAg prevalence among women testing positive for HBsAg in the model. The model population was the birth cohort for the maternity hospitals of Kabul for 2006 less the probable neonatal and infant deaths (based on national infant mortality rate, resultant n = 45,090). Estimates of anti-HBc of 30% for children and 60% for adults were used to complete the model, which are identical to those used for a model created by the World Health Organization in development of a HBV vaccination strategy for Afghanistan [22].