Eligibility criteria for the three vaccine trials included being healthy, at low risk for HIV and between the ages of 18 and 50 (18–40 years for one trial, B002). All participants were advised to use condoms. In addition, a long-lasting non-barrier method of contraception, such as Depo-Provera, Norplant, intra-uterine device (IUD) or tubal ligation was required of all female participants of child-bearing potential (oral contraceptives were not allowed). Female participants with an IUD were excluded from cervico-vaginal sampling due to a risk of the IUD being dislodged by the Softcup [10].

Participants were drawn from the following Phase 1 vaccine trials: IAVI B002 (ClinicalTrials.gov Identifier NCT01264445), B003 (NCT01215149), and B004 (NCT01496989), conducted at sites in Eastern Africa, South Africa, and the USA, with the mucosal sub-study conducted at the Kenyan sites only. In Kenya, trials B002 and B003 were conducted in 2011–2012 at KAVI-KNH and KAVI-Kangemi, respectively, and B004 was conducted in 2012–2013 at KAVI-Kangemi. Mucosal specimens were collected at three time points for each trial, with the timing dependent on trial design: in B002 and B003, sampling was at baseline, one month after the final vaccination and at the next vaccine trial visit; in B004 sampling was at baseline, one month after the prime and one month after the boost.

Participants were free to opt out of any collection at any time or to provide samples they had previously refused. Reasons for refusing any sample collection were recorded at each visit. A questionnaire was administered at the final mucosal study visit, asking participants the main reason they agreed to provide mucosal specimens, what mucosal specimens they would agree to in future studies, and any general suggestions for making the procedures more tolerable. Questions were open-ended. If responses fit with a pre-determined list of responses, answers were coded accordingly. If responses did not fit with one of the pre-set answers, they were recorded verbatim. Up to two reasons for refusal were collected at each visit for each sample type not given.

Saliva was collected by placing a Salimetrics Oral Swab (Salimetrics LLC, State College, PA, USA) against the parotid duct for 5 minutes. Oral fluid (transudate) was collected by allowing fluids to pool in the mouth then passed into a Falcon tube [11]. Participants were instructed not to eat or drink anything but water for 2 hours prior to saliva and oral fluid collection.

In female participants, the Instead Softcup (Evofem Inc., San Diego, CA, USA), was inserted by the clinician and kept in place for 5 minutes (10 minutes for B004) to collect cervico-vaginal secretions. The cervix was then accessed with a disposable speculum and two pre-moistened Merocel sponges (Medtronic, Minneapolis, MN, USA) were placed against the cervical mucosa for 5 minutes each, serially. The Softcup and Merocel sponge have been used to collect cervico-vaginal secretions in other research studies [12], [13]. Cervico-vaginal collection was not performed during menstruation. If possible, samples were taken approximately 2 days after bleeding ended, except baseline samples, which were considered missed if the participant was menstruating on the day of vaccination. Male participants provided semen specimens, by masturbation, into a universal container.

Rectal secretions were collected from both male and female participants by accessing the rectal mucosa through a disposable clinician-inserted proctoscope. Rectal secretions were collected using two pre-moistened Merocel sponges placed against the rectal mucosa for 5 minutes each, serially.

B002 and B003 participants were reimbursed a set amount at the end of each visit, regardless of the actual collections performed. In B004, the reimbursement structure was changed so that participants were given a set amount per sample type in order to reimburse participants for the significant additional time involved in providing all specimen types as opposed to just one.

Humoral responses were assessed by anti-HIV specific IgG and IgA ELISAs on frozen samples. Results will be published separately.

Participants' overall acceptance of a mucosal sampling method was calculated as the proportion of participants who provided any specimen for that sampling method during the study. 95% confidence intervals for the observed proportions were estimated using exact (Clopper-Pearson) binomial method in PASS 2008 (NCSS, Kaysville, UT).

Due to limited sample size, the statistical comparisons were primarily exploratory and were conducted for evaluation of any observed site, trial or gender differences in acceptability of mucosal sampling methods. Comparisons of categorical and continuous factors were conducted using the Fisher's exact test and Wilcoxon rank-sum test, respectively. A two-sided p-value of less than 0.05 was considered to be statistically significant. Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC).

Male participants in the vaccine trials outnumbered female participants 2∶1 overall and this disproportion was also reflected in the mucosal sub-study. 27/34 (79%) females and 62/71 (87%) males consented and enrolled in the mucosal sub-study (p = 0.11, Table 1). Participant ages ranged from 18 to 46 years. B002 participants were significantly younger (p<0.0001) than participants in the other two trials due to the stricter age criteria specified in the protocol. Most female participants in all three trials were on an injectable hormonal contraceptive (Depo-Provera), although a few used Norplant or had a tubal ligation. Three female participants had an IUD and were therefore excluded from cervico-vaginal collection.

Of 16 vaccine trial participants that were not enrolled in the mucosal study, seven participants cited discomfort or fear of the sampling methods, two refused without giving a specific reason, for two participants the consent form was not yet available in the local language, and one participant each reported not wanting more procedures, lack of time or parental advice against joining. One participant was excluded for vertebral and pelvic bone deformities and one participant was not enrolled because vaccinations were discontinued following an adverse event not related to vaccination. Participants who declined participation in the mucosal study did not differ significantly in gender and age characteristics from those who participated.

Saliva and oral fluids, the least invasive samples, were collected from all participants in the mucosal sub-study at nearly all visits, while rectal sample collection was the least likely to be accepted (Figure 1). Based on overall acceptance and corresponding 95% confidence intervals, female participants agreed to both types of cervico-vaginal sampling more readily than rectal sampling (Table 2). Similarly, male participants were more likely to provide semen than rectal samples.

Cervico-vaginal sampling by Softcup and Merocel sponge was well-accepted and tolerated by the majority of women. Of 27 female participants, 18 (67%) had cervico-vaginal samples collected with Merocel sponge at baseline and 17 (63%) also had cervico-vaginal samples collected with Softcup (Table 3). Consent for cervico-vaginal sampling remained consistent across the two follow-up visits. Most missed samples were attributable to menstruation or IUD; only three participants refused cervico-vaginal sampling because of physical or emotional discomfort (data not shown). Participants in B004 were given the choice of self-inserting the Softcup or having a clinician place the device. All participants chose to have a clinician insert the Softcup.

Overall, semen was provided by 30 out of 62 (48%) male participants across all studies at baseline, with a similar percentage (29/62, 47%) at the second and third visits (Table 3). There was a significant difference in the proportion of male participants providing any semen specimen in B002, conducted at the KNH clinic (32%) and in B003 and B004, conducted at the Kangemi clinic (68%, p = 0.01). Among participants who did not provide semen specimens, approximately half cited embarrassment/emotional discomfort as the main reason. Another 8 (23%) were specifically uncomfortable about masturbating at the clinic (Table 4).

Rectal secretions were collected with Merocel sponge from 21/89 (24%) male and female participants at each of the three visits, however the male to female ratio differed slightly with each visit (Table 3). Overall, there was no significant difference in the proportion of participants providing any rectal sponge specimens between females (41%) and males (26%) (Table 2; p = 0.21). The proportion of participants who provided any rectal specimen was similar for B002 and B003 (16% combined), but significantly larger for B004 participants (56%) (data not shown; p = 0.0002). A slight downward trend over time in B002 and B003 was reversed in B004, with more participants agreeing to rectal sampling at the second and third visits compared to baseline (Table 3).

When asked at the final mucosal study visit the reason for agreeing to provide specimens, contribution to HIV research was the primary reason given (50–64% for various specimen types), finding out more about one's health ranked second (20–32%), belief that the samples were a requirement ranked third (4–12%), belief that the study would help them access more health care ranked fourth (4–5%), and “easiness of giving” was specified as another reason for providing specimens (3–9%, with the highest rate being for saliva). When asked for suggestions to improve the mucosal sampling experience, 32 out of 89 (36.0%) participants said they had no problems or issues with the sampling methods in this study and 28/89 (31.5%) had no comments. Suggestions to improve semen sampling included quieter rooms, provision of pornographic materials (not used in this study due to Kenyan anti-pornography law) or allowing partners to assist the participant. Rectal sampling was another area of focus, with suggestions to use a smaller proctoscope or to find a method without the need for a proctoscope. Suggestions are summarized in Table S1.

The demographic data collected in the three vaccine trials were limited to age, gender, and race/ethnicity. As a result, other information such as marital status, education, parity for women and socio-economic differences could not be examined. The study participants were selected for their low risk of HIV infection, and their knowledge, attitudes and flexibility may differ from people who are at higher risk, particularly sex workers.

The mucosal sampling time points in each trial were dependent on the vaccination schedules, which varied between protocols. The length of the gap between the second and third visits did not appear to have affected compliance.

The informed consent process was similar between the two sites, but scripts were not used to explain the procedures. Individual differences between counselors could have affected the likelihood of enrollment. Without a strictly standardized method of explaining the procedures, it is possible the consent process had an impact on the initial acceptability of the various sample types. This is an area for further development and research.