The Contraceptive CHOICE Project is a prospective cohort study seeking to recruit 10,000 women and to promote the use of LARC (17
). Participants are recruited from a university-based clinic, two abortion clinics, and several community-based clinics. Women are eligible for the study if they are between the ages of 14 and 45 years, speak English or Spanish, are seeking a new method of reversible contraception or are not currently using a method, have not had a hysterectomy or tubal sterilization, and are currently sexually active with a male partner or planning to become sexually active in the next six months. Informed consent is obtained for all participants. We obtained approval from the Washington University School of Medicine Human Research Protection Office prior to starting recruitment. After receiving comprehensive counseling on contraceptive options, participants choose a reversible contraceptive method that is provided at no cost for the 3-year study period. Telephone follow-up occurs at 3 months, 6 months, and every 6 months thereafter for the duration of the study - a total of seven contacts.
All women enrolled in the study are offered STI screening at baseline then again at 12, 24 , and 36 months postenrollment. At baseline, all women are instructed on proper technique to self-collect a vaginal swab. Swabs were tested for C. trachomatis and N. gonorrhoeae using the BDProbeTec ET (Becton Dickinson, Sparks, MD) instrument through DNA strand displacement amplification (SDA) technology. These specimens can be stored at room temperature and must be received by the testing laboratory within 16 days of collection.
This sub-study randomized 558 women to home- or clinic-based screening for the 12-month scheduled STI screen. Women were randomized using computer based randomization if they completed the baseline survey and consented to the sub-study at enrollment. For each participant a random number was generated using the uniform distribution with a cut-point of 0.50. Those below this cut-point were assigned to one treatment and above to the control. Staff and participants were blinded to the randomization status until the time of testing (12 months), and randomization was performed by an author who did not have contact with study subjects. At the 12 month survey, participants were required to be living in the United States and using a long-acting reversible method of contraception, defined as levonorgestrel intrauterine system (LNG-IUS), Copper T 380A intrauterine device, or the contraceptive subdermal implant. Participants identified as eligible by these criteria were given instructions to screen for STIs by randomization group. Participants could refuse testing for any reason. Recruitment ended August 1, 2009, after the desired sample size was reached.
Participants randomized to the home-based screening group were mailed a collection kit. The kits were packaged in a plain brown box and could be sent to any address the woman provided (e.g. a friend's house) to address potential privacy concerns. A vaginal swab and collection tube were provided, identical to those used at baseline screening. Detailed, step-by-step instructions with photographs explained how to collect the specimen and return it in a prepaid, pre-addressed mailer (Exakt-Pak, Oklahoma City, OK), which complied with Department of Transportation and United States Postal Service regulations.
Women randomized to clinic-based screening were able to test with their regular health care provider or at four local family planning clinics. Women in the clinic arm were mailed written instructions, which stated that participants could be reimbursed for any expenses of STI testing and treatment, and asked participants to provide medical records of such testing. Medical records release forms for the participant's primary provider were signed at enrollment. However, if the participant planned to test with a different provider, a new release form was mailed to the participant with a postage-paid return envelope.
As another clinic-based option, all women in the clinic-based randomization group were given written instructions to obtain testing at one of four family planning clinics. This was designed to accommodate participants without access to a private provider, though all participants were able to test at these clinics. No appointment was necessary and there was no cost for screening. The clinics were provided with self-collected vaginal swab kits as used in the home-based group.
Specimens from home kits were received by mail daily, and specimens from family planning clinics were returned within five days of specimen collection. Specimens that were not in satisfactory condition for testing (e.g. missing swab or low preservative fluid) were rejected, and the participant was asked to come to the study clinic to collect a new sample. Women with positive tests for C. trachomatis or N. gonorrhoeae were contacted by a research nurse and provided with antibiotic treatment for herself and all partners, at no cost to the participant.
Medical records were requested for all participants in the clinic-based group who had not documented screening by providing medical records or a specimen from the family-planning clinics. A records request was sent to the provider with whom the participant had intended to test. If a valid release for this provider was not available, a release was sent to the participant to sign and return, and requests were sent to any other provider for whom we had a valid medical records release. Two attempts were made to contact each medical provider. We did not request records from the family planning clinics where self-collected screening was available if participants listed this facility as their healthcare provider. Participants who presented there could receive testing through our study at no cost and without an appointment, and were thus unlikely to have been tested by other means at these sites.
Baseline and 12-month follow-up interviews collected detailed information on demographic characteristics, contraceptive use, reproductive history, sexually transmitted infection diagnosis and treatment, and sexual behavior with male and female partners. Women were mailed a written survey addressing satisfaction with the STI screening process fifty-six days after testing was offered. A postage-paid return envelope was provided. At the 18-month telephone interview, the next regularly scheduled follow-up after randomized STI screening, women were asked if they had undergone STI screening. If they had not screened, they were asked to provide one or more reasons why not.
The testing period was defined as 56 days following the 12-month interview date (coinciding with mailing of the satisfaction survey), so that the mailed survey did not serve as a reminder to complete screening. A completed test was defined as a received sample from the home kit or family planning clinic during the testing period; medical record documentation of a test for C. trachomatis and N. gonorrhoeae during the testing period; or self-report of screening on the mailed survey or 18-month telephone survey. A second outcome variable, documented completed tests, considered a participant ‘not screened’ if no medical record documentation could be found confirming her self-report of screening during the testing period.
Baseline characteristics of the home and clinic-based testing groups, and screening completers versus non-completers, were compared using chi-square or Fisher's exact test for categorical variables and Student's t-test for continuous variables. Ordinal data in the satisfaction survey were analyzed by the Mann-Whitney-U test. The hypothesis that the home-based screening group would have a higher proportion of completed tests was tested using the chi-squared test and quantified with a simple relative risk calculation. When the outcome of interest (completed tests) is common (>10%) in prospective studies, logistic regression may produce a biased estimate of the relative risk, with the extent of the bias is related to the prevalence of the outcome (18
). To correct for this bias, we performed a Poisson regression with robust error variance; this method is one methodologic approach that produces an unbiased estimate of both the relative risk and 95% confidence intervals (19
). SPSS 17 (SPSS Inc., Chicago IL) and SAS 9.2 (SAS Institute Inc., Cary, NC) were used for statistical analyses. Based on a testing percentage of 35% in the clinic-based group, a 15% difference testing rates between groups (testing in the home group of 50%), and alpha of 0.05, we needed 242 patients per group (total n = 484) to achieve 90% power.