Background

The Centre for the AIDS Programme of Research in South Africa (CAPRISA) 004 trial demonstrated 39% reduction in acquisition of HIV infection, with 54% reduction in women who used tenofovir gel consistently.2 The CAPRISA 008 trial which followed, was an open-label randomised controlled trial to assess the implementation, effectiveness and safety of 1% tenofovir gel provision through family planning services in KwaZulu-Natal, South Africa.19 This trial had three main goals:

1. to provide post-trial access to tenofovir gel,

2. to develop and assess an implementation model for pre-exposure prophylaxis (PrEP) provision through family planning services and

3. to collect additional safety data.

The South African Medicines Control Council approved CAPRISA 008 only for consenting, research-experienced CAPRISA 004 participants, who were sexually active, aged 18 years and older, and remained HIV uninfected. The study was conducted at the CAPRISA Vulindlela (rural) and eThekwini (urban) Research Clinics in KwaZulu-Natal, South Africa and their neighbouring family planning clinics. CAPRISA 004 participants consented to be contacted for other trials at the study exit visit and contact numbers were readily available to reach out to volunteers for CAPRISA 008. Of the 786 eligible for CAPRISA 008, 73(9.3%) were not contactable and 716(91.1%) were prescreened; 268(37.4%) were excluded either for reasons of relocation, lack of interest, work/study commitments, HIV seroconversion, failure to return for enrolment, untraceability at last known address, pregnancy or planned pregnancy, death and other reasons. Only 448 (56.9%) were screened with 382 (48.6%) enrolled, with the dropout due to failure to return within the screening window, HIV seroconversion, lack of sexually activity, pregnancy and co-enrolment in another trial. The screening-to-enrolment ratio was 1.7:1. Retention varied between 90% and 100% for all visits throughout the trial.20 This trial was registered with the South African Department of Health (reference: DOH-27-0812-4129) and ClinicalTrials.gov (reference: NCT01691768) on 5 July 2012. The trial was also approved by the University of KwaZulu-Natal's Biomedical REC (BFC273/010).

CAPRISA 008 was nurse-driven to replicate service in the public sector, with a clinician available for consultation as needed. Inclusion criteria included a negative urine pregnancy test and written agreement to initiate non-barrier contraception at baseline. Urine pregnancy testing was conducted as in CAPRISA 004, with fresh urine samples collected on the day of the visit for point-of-care testing in the clinic. A positive urine pregnancy test during follow-up resulted in cessation of study product until the outcome was confirmed and the urine pregnancy test reverted to negative. Participants remained in follow-up while pregnant. This criterion was the same for the efficacy study,2 was part of the contraceptive counselling and was agreed to by all participants through the informed consent process. Pelvic examination, including collection of genital samples, was done at enrolment and every 6 months, or more frequently if clinically indicated. Genital samples were stored to assess for markers of safety, risk exposure, product adherence, potential post-trial assessments of activity against sexually transmitted infections (STIs) and tenofovir resistance.19 Individual tenofovir levels indicative of product adherence were not analysed in real time due to prohibitive costs. Adherence during the study was monitored by clinic and pharmacy staff on an individual participant basis, matching sexual acts to returns of used and unused product applicators.19

This case study is of an urban CAPRISA 008 participant, who apparently, driven by personal circumstances, concealed vital clinical information, substituted urine samples for pregnancy testing, underwent trial procedures, received contraception and study drug, potentially compromising her safety and that of her fetus. Although product safety had been established in the 004 study, safety had not been demonstrated in pregnancy.

Case study summary

A woman aged 25 years with one previous pregnancy (5 years earlier) was screened early in 2013. At baseline, she perceived herself at low risk of HIV acquisition, based on her in-depth knowledge of her partner and the length of their relationship. The partner was 7 years older than her, and known to be HIV uninfected. She self-reported her date of birth, as her identity document was illegible, and supported this with an affidavit which stated that she was unable to afford a copy of her identity document, due to her indigent status. She reported that she was financially dependent on a child grant, intermittent informal work and variable support from her partner. She reported irregular menses secondary to depot medroxyprogesterone acetate (DMPA) in 2012, with the last menses prior to screening in March 2013. She also reported inconsistent condom use since cessation of DMPA and agreed to restart contraception. A stat dose of DMPA was administered and 20 male condoms dispensed. Clinically, her weight was 54 kg, body mass index was 23.7 and her waist circumference was 79 cm. Urine pregnancy test was negative, with no symptoms or signs on medical history and general physical examination. In the absence of gynaecological symptoms, a pelvic examination was not mandated at the screening visit.

At enrolment she reported a white, inoffensive discharge. Physical examination was repeated and was normal; pelvic examination was reported as ‘difficult’, with a poorly visualised cervix lying posterolaterally and an offensive, frothy white discharge evident. A urine pregnancy test was not repeated, as the screening test was done less than the protocol mandated 21 days previously. Genital samples were collected as per protocol. Following a review of the visit notes and laboratory results, the clinician provided a script for study product, based on the reported sexual frequency, which amounted to 10 gels for 1 month (a maximum of two gels to be used within 24 hours, coitally based).2 The discharge was managed syndromically with cefixime, metronidazole and doxycycline as per the then approved South African STI treatment guidelines for non-pregnant patients.21

During follow-up the participant was seen monthly at four consecutive visits (relevant findings are reflected in table 1) until visit 4, when marked abdominal distension was noted by the research nurse.

The clinician confirmed the pregnancy on abdominal palpation, noting marked foetal movement, which the participant claimed not to have felt. The initial and repeat urine pregnancy tests were both negative and a blood pregnancy test was requested. Study product was held pending referral for further investigations. The participant denied knowledge of the pregnancy at this stage. An ultrasound and the blood pregnancy test confirmed pregnancy. The participant accepted the diagnosis but continued to deny prior knowledge of the pregnancy. A few months later, she delivered a healthy live male infant via caesarean section, conducted because of a previous caesarean section and for foetal distress secondary to transient neonatal tachypnoea. A bilateral tubal ligation was performed simultaneously at her request. Paediatric assessment at birth revealed no congenital abnormalities. At the study exit, the participant's baby was noted to be normal with no developmental problems.

In tandem, an extensive root cause analysis was conducted by the protocol team, to exclude problems regarding pregnancy test kit integrity and user error. No problems were identified. However, it was noted that research nurses in HIV prevention were relatively inexperienced in conducting obstetric examinations, did not always conduct bimanual palpation of the uterus and that the method of conducting waist measurement involved only partial disrobing and thus did not facilitate detection of a distended abdomen, especially during winter when bulky clothing is worn.

The clinician requested a private meeting with the participant at the request of the protocol team to discuss the participant's motivation for her decisions and actions and to review the overall health of the baby. The clinician discussed the safety concerns of the protocol team (unknown risks of product use in pregnancy) and invited the participant to describe the thinking behind her decisions. The participant appeared embarrassed and anxious during the discussion and went to great lengths to express her remorse, apologising profusely and disclosing that she knew she was pregnant before she was screened. She then also disclosed substituting her urine samples at clinic visits with samples she had obtained earlier in the day from a person at her home that she knew would be negative. She described taking the urine sample container to the ablution facilities and decanting the sample she brought from home into the container and presenting this to the staff for testing. She said that the pregnancy had been unplanned, and that her partner was angry and threatened infidelity. Consequently, she perceived her risk for HIV acquisition as high and judged ongoing gel usage to be essential to her remaining HIV negative. She resorted to hiding her knowledge of her pregnancy from trial staff, denying symptoms and substituting urine samples. She reported adherence to the gel regimen and that she had not shared gel with anyone else. She was aware of all the risks related to procedures and STI medication but said that her desire to prevent HIV acquisition had overruled these considerations.

Based on the root cause analysis and her disclosure, the participant was counselled and excluded from further participation in the trial, in consultation with CAPRISA's bioethics head, on the grounds that she no longer met inclusion criteria for the study. The case was also reported to the local REC which raised concerns about the site's failure to detect a pregnancy of >12 weeks at screening. The case was also discussed with all clinical project staff to raise awareness, as well as with the Community Research Support Group, who supported the decision to exit the participant from the trial.

Recommendations

The case reviewed in this paper supports the following pertinent recommendations:

General:

1. Increase awareness among researchers and staff, RECs and society, of the multiple motivations for participation in post-trial access trials or substudies, and the limitations of REC review.

2. Study benefits may be perceived differently among participants and that focussing on paid inducements and risk/benefit factors is required.

Specific to trial design:

1. Pregnancy testing methods should not rely only on participant sample collection, for example, serum pregnancy testing.

2. Clinician-driven baseline assessments in post-trial access studies should be used to determine eligibility.

3. Weight and height measurements should be conducted in procedure rooms where privacy is guaranteed, rather than in group areas where participants are less inclined to disrobe appropriately.

4. Collection of real-time genital samples should be done to determine exposure to study product when safety concerns arise.

5. Study nurses require advanced training in pregnancy evaluation where pregnancy is an exclusion criterion.

These recommendations should be considered on an individual study level taking into account human and financial resource constraints, length of study visits, study design and the likelihood of the occurrence of similar cases.