We performed a cross-sectional study of women undergoing either D&E or second trimester medical induction with misoprostol alone. We investigated access to and acceptability of D&E and medical induction by interviewing women after completion of their abortion. Length of procedure, medications used, safety and efficacy were compared using clinical information collected during or after the procedure.
The study was approved by the University of Cape Town Research Ethics Committee, Stellenbosch University Committee for Human Research and Allendale Investigational Review Board. Permission to perform the research was obtained from the Provincial Department of Health, Western Cape. All participants provided written informed consent, and confidentiality and anonymity were ensured. Study procedures were carried out in accordance with the revised version of the Helsinki Declaration and with the standards of Good Clinical Practice.
The study was conducted between February and July 2008 at five public sector hospitals in Western Cape Province, South Africa. These sites were five of the nine public hospitals in the province that provided second trimester abortion services, either by D&E or medical induction, during the study period. The remaining four hospitals performing second trimester abortion were not included in the study due to logistical reasons which included distance, low volume and permission being denied at one facility. Women generally were referred to these hospitals based on regional proximity, and clients were not offered a choice of which facility to attend. No hospital offered both D&E and medical induction, and therefore women did not have a choice of abortion method after presenting to the facility.
Three of the study hospitals offered outpatient D&E services, and the other two offered medical induction on an inpatient basis. Of the five sites, two D&E sites and one induction site served a primarily urban population, while the other sites served a more peri-urban and rural population. We collected data at each site on the total number of second trimester abortions performed during the study period by reviewing logs of abortion procedures maintained at the facilities.
All women attending these hospitals for second trimester abortion on days when study interviewers were present were invited to participate in the study. Eligibility was assessed prior to the abortion, and criteria included age 18 years or older, gestational age between 12.1 weeks and 20.9 weeks, and ability to communicate in English or Xhosa. All participants provided written informed consent before enrollment to both an interview and hospital record review.
Bilingual interviewers administered structured questionnaires in a private location at the hospital after the woman recovered from the abortion. Data were obtained on age, spoken language, education, employment, housing conditions and reproductive history. Information was also obtained about the number and timing of clinic visits leading up to admission for the current abortion. After initial data collection, the long delay between date of the first clinic visit and date of admission for the abortion as reported by the participant became apparent. To substantiate this finding, additional information was collected on the date and result of the ultrasound scan from participants' charts. Although the ultrasound may not have been performed at the first clinic visit, it was the only documented record of participants' access to abortion-related care.
For details relating to the abortion experience itself, clients were asked to rate the following on a 5-point Likert scale: overall satisfaction with the abortion, pain and emotional discomfort.
For participants undergoing medical induction, a procedure form was completed by the attending doctor or by the study investigators following discharge of the participant. Data were collected on gestational age by ultrasound, date of initial ultrasound, duration of procedure (from first misoprostol dose to placental expulsion or surgical procedure, if performed), duration of hospital stay, timing and dosage of misoprostol and any other medications, and details of additional surgery, if performed. When forms were incomplete, participants' medical records were reviewed. For participants undergoing D&E, the procedure form was completed by the doctor performing the abortion immediately after the procedure. Gestational age by ultrasound, duration of surgical procedure, details of the procedure, and whether fetal expulsion occurred prior to surgery were noted. Because most of the relevant information on the D&E procedure was not documented in the medical record, it was not possible to collect this information later if the form was incomplete.
The primary outcome of the study was procedure efficacy, defined as complete abortion performed on site, with no need for repeat surgical procedure in the case of D&E, or no need for any surgical intervention in the case of induction. Secondary outcomes included safety and acceptability. Safety was assessed by measuring the prevalence of immediate major complications (death, seizure, injury to organ requiring abdominal surgery, or hemorrhage requiring transfusion) or minor complications (hemorrhage without transfusion, fever, cervico-vaginal trauma, transfer to another facility because of inability to complete procedure, unplanned surgical procedure (for induction) and fetal expulsion prior to procedure (for D&E)) [16]. Because it was not standard at these sites to have women return for a follow-up visit, information on delayed complications was not available. Acceptability was determined by overall satisfaction, pain and emotional discomfort, and likelihood of recommending the abortion method to others.
Data were analyzed using Stata (version 10, College Station, TX). We summarized data on participant characteristics, abortion-related hospital and clinic visits, procedure details, efficacy, complications and acceptability by abortion method using proportions for categorical variables and medians for continuous variables. 95% confidence intervals (CI) were calculated for proportions, and proportions were compared using chi-square tests or Fisher's exact tests when appropriate. Where medians were compared, Wilcoxon rank sum tests were used. Two-sided significance tests were used throughout. Missing values were excluded from the analysis, and valid percentages are reported for all results. Statistical comparisons were performed by procedure rather than by site as numbers were often too small to allow for analysis by site.
The number of subjects recruited was sufficient to detect a difference in efficacy between 85% for induction and 98% for D&E, with 90% power and two-sided alpha of 0.05.