Aim
to conduct a randomised controlled trial to answer the clinical question
In women with asymptomatic vaginal candidiasis early in pregnancy does treatment with clotrimazole prevent spontaneous preterm birth (less than 37 weeks gestation)?
Hypothesis
that treatment of asymptomatic vaginal candidiasis with clotrimazole is associated with a 40% reduction in spontaneous preterm birth <37 weeks gestation.
Study Design
We will use a prospective, randomised, open-label, blinded-endpoint (PROBE) study design with 2 arms [44, 45]:
Treatment
6-day treatment with vaginal clotrimazole
Usual care
screening result is not revealed, no treatment, routine antenatal care
Although, double-blind, placebo-controlled randomised trials are regarded as the 'gold standard' of clinical trials, they do have some disadvantages. For our hypothesis, a placebo arm would not represent current clinical practice (no screening and no treatment). Knowledge of vaginal colonisation with Candida may change participants behaviour such that they seek active therapy (clotrimazole vaginal preparations are available over the counter in Australia) rather than, or in addition to, the study medication [45]. Further, the time, expense and feasibility of obtaining a placebo preparation should not be under-estimated [46]. And finally, as a vaginally administered placebo will necessarily contain an alcohol preservative, it may be biologically active and the possibility of an independent affect on vaginal flora cannot be excluded.
These disadvantages can be overcome by a PROBE study design [44, 45]. As in our pilot study, the PROBE study design maintains strict randomisation and allocation concealment procedures but asymptomatic women will be randomised to open-label treatment (Canestan®), or no treatment and the screening result is not revealed (consistent with current clinical practice) [43–45]. The primary endpoint (preterm birth) is unambiguously defined and those assessing the end-point will be blinded to the screening results and treatment allocation. Another advantage of using open-labelled medication is that patient management adheres more closely to routine clinical practice than in double-blind, placebo-controlled trials, making the results more generalisable to the pragmatic management of patients [44].
Setting
The study will be carried out in at least five participating maternity hospitals. A range of antenatal care options are provided in these hospitals including standard antenatal clinic care, midwifery-led care, birth centre care, GP-shared care and private obstetric care.
Participants/Eligibility criteria
Pregnant women presenting for antenatal care prior to 20 weeks gestation with singleton pregnancies will be eligible for inclusion into the study. Treatment by 20 weeks of gestation is likely to be necessary to prevent preterm birth. Exclusions: women who present beyond 20 weeks gestation or have a history of hypersensitivity to clotrimazole will be ineligible. Women with symptomatic vaginal infection due to Candida will be treated and are therefore ineligible for the study.
Sample size
A sample size of 3,208 women with Candida colonisation (1,604 per arm of the trial) is required to detect a 40% reduction in spontaneous preterm births (compared with term birth) among women with asymptomatic candidiasis from 5.0% in the control group to 3.0% for those women treated with clotrimazole (significance 0.05, power 0.8). This is a more conservative risk reduction than that found in the trial by Kiss [14], where analysis of this subgroup suggested treatment of asymptomatic candidiasis was associated with a 65% reduction in spontaneous preterm birth but the numbers were small and the subgroup analysis was post-hoc.
Recruitment of 3,208 women will require 16,040 women to be screened based on a Candida carriage rate of 20%. In our pilot study the Candida carriage rate was 19.6% and the participation rate among eligible women was 64%. If a participation rate of 64% could be maintained across all participating centres, ~25,000 would need to be approached to achieve screening of 16,040 women. A lower participation rate means more women would need to be approached. We expect that a participation rate of at least 60% could be maintained across all centres and this would mean approaching ~26,000 women.
Recruitment and collection of baseline data
The study procedure is feasible as it draws on usual antenatal care in pregnancy, and is similar to the Kiss trial (see Table 1). The research nurse will ask eligible women to participate, explain the trial and obtain informed consent, collect baseline data and ask women to self-collect a vaginal swab. This is only a minor departure from current practice.
As women of child-bearing age are known to be very mobile [47], participants will be asked to provide their parents and other alternative contact details (eg friend or relative) to enhance subsequent follow-up. Women will also be asked to specify their preferred method of contact - women in our pilot study have shown a strong preference for email contact. Self-collection of vaginal swabs is commonly used for screening in pregnancy [48]. We have also successfully used this method of specimen collection in a study of vaginal Ureaplasma urealyticum colonisation in pregnancy[49] and in our pilot study [43]. The detection rate of Candida using self-collection is not different from techniques that are more invasive, requiring the use of a speculum [50].
Randomisation
Eligible women who are culture positive for Candida will be randomised to receive active treatment with clotrimazole or usual care. The randomisation schedule will be prepared and centrally administered by a researcher not involved in patient care. A computer random number generator will be used to prepare the randomisation schedule with permuted blocks and 1:1 randomisation. The women randomised to open-label clotrimazole treatment will be notified by phone by a research nurse. A central pharmacy will be responsible for dispensing the study medication which will be sent to participants by mail from the coordinating centre. This approach has been acceptable and successful in the pilot study.
Specimen collection and transport
Surveillance for Candida isolates will be conducted at 12-19 weeks of pregnancy. A vaginal swab will be self-collected and inoculated immediately into Amies Transport Media, stored at room temperature and transported to Royal North Shore Hospital (Sydney) within 24 hours. All swabs will be identified with date of collection, patient's study number, packaged according to IATA guidelines, and transported to Pacific Laboratory Medicine Service (PaLMs), Royal North Shore Hospital.
Isolation and identification of Candida from vaginal swabs
The detection and identification of Candida from vaginal swabs is readily achieved. The swab will be inoculated onto full-plates of chromogenic agar (CHROMagar™Candida, CHROMagar, Paris, France) to isolate, provisionally speciate and semi-quantitatively enumerate (scant, moderate, or heavy) Candida species and to obtain colonies for further work such as susceptibility testing by broth microdilution methods. This medium is chosen to ensure detection of mixed infections. Cultures will be incubated for 72 hours at 35°C in 5% CO2. Identification to species level will be based on colony morphology, and colour.
Intervention
The treatment group will receive a 6-day course of commercially available vaginal clotrimazole pessaries (Canestan®). Women will be advised to insert one pessary as gently and deeply as possible into the vagina while lying on her back, preferably at night, for 6 nights. We have chosen 6 days of treatment because this is: supported by the Cochrane Systematic Review of treatment for Candida eradication in pregnancy [36], this is how commercially available vaginal clotrimazole is marketed[40] and this was the regimen used in the Kiss trial [14]. Women who miss one or more daily doses will be advised to insert a single pessary as soon as they remember or the next evening, and continue using the pessaries until the course of treatment is finished.
Follow-up
A brief (< 5 minutes) follow-up questionnaire will collect information about symptoms of candidiasis, use of antibiotics and/or corticosteroids (risk factors for symptomatic candidiasis), treatment of candidiasis (including study medication, doctor prescribed and over-the-counter medications), treatment side-effects and compliance with the treatment regimen. The questionnaire will be web-based with email notification or mailed with a reply-paid envelope according to the participants stated preference at trial entry. These secondary outcomes will be collected at 28-32 weeks, a gestation when the potential for unmasking the 'no treatment' arm will be irrelevant.
Care of all women and their babies will be otherwise managed according to the standard practices of the model of maternity care she has chosen. Women who develop symptoms of Candida infection will be offered appropriate treatment. The results of the vaginal swab culture taken at the time of trial entry will not be disclosed nor be available in the pregnancy record.
Outcomes and Explanatory Factors
Baseline data
Brief baseline data will be collected to assess comparability of the study groups. The baseline assessment will include age, socio-demographic data, smoking and alcohol use, medical conditions, history of candidiasis, history of prior pregnancies including preterm birth, gestational age at enrolment and method of gestational age assessment (ultrasound or dates). This information will be obtained from existing databases (as outlined below) or where this is not available from the participants at study entry.
Primary Outcome
birth <37 weeks gestation following spontaneous onset of labour or following preterm prelabour rupture of membranes.
The primary and secondary outcomes will be obtained (with informed consent) from existing computerised obstetric and hospital databases. This is a novel and cost-efficient approach [51]. We have previously demonstrated that the outcomes for this study are reliably collected in routinely collected obstetric and hospital databases [52–56]. It also allows identification of transfers to another hospital. The latter are uncommon and we will utilise health record linkage to determine outcomes for these few women who would otherwise be lost to follow-up. Consequently, we anticipate negligible loss to follow-up in this trial. Because this approach is novel, we will validate the primary outcome (preterm birth) by reviewing the medical records of all women identified in the computerised databases as delivering preterm, those where gestational age is missing (usually <1%) and a random sample of those recorded as term deliveries. We estimate that approximately 300 records will need to be reviewed to validate preterm birth.
Secondary outcomes
Include any preterm birth <37 weeks, medically indicated preterm birth <37 weeks (including indication), preterm birth <32 weeks, preterm prelabour rupture of the membranes, spontaneous pregnancy loss <20 weeks gestation, fetal growth restriction (< 10th birthweight for gestational age percentile), perinatal mortality (stillbirth or neonatal death), admission to neonatal intensive care unit, duration of stay in hospital (maternal and infant), birth weight, Apgar score at 5 minutes, and a composite neonatal morbidity indicator including respiratory distress, assisted ventilation, intraventricular haemorrhage, necrotising enterocolitis, retinopathy and pneumonia [57].
Other maternal and infant explanatory factors
Other conditions may arise in pregnancy that influence preterm birth risk including pregnancy hypertension, gestational diabetes, antepartum haemorrhage, placental abnormalities (placenta praevia, placental abruption), congenital anomalies, onset of labour (spontaneous, induced or no labour) and mode of delivery.
Analyses
Analyses will be by intention to treat. The primary analysis will compare the proportions of spontaneous preterm births in the two groups, using a chi-squared test. The relative risk and 95% confidence interval of preterm birth and other outcomes in women with active treatment compared with those with usual care will be calculated. In the secondary analysis of the primary outcome, logistic regression will be used to adjust for centre and any differences in important baseline characteristics. Secondary outcomes that are binary will be analysed in a similar manner, while those that are quantitative will be compared between groups using t-tests or the distribution-free equivalent, as appropriate.
Sub-group analyses will be used to explore associations between preterm birth and clotrimazole treatment by degree of Candida colonisation (light, moderate, heavy), Candida albicans and non-albicans species, gestational age at enrolment, and among medically indicated preterm births by indication (hypertension, growth restriction, fetal malformation, placental abnormalities). To test for differences in the effect of clotrimazole among sub-groups, a chi-squared test for interaction will be used, using the trend version for the ordered variable, degree of colonisation.
Interim analyses: the Data Monitoring Committee
An independent Data Monitoring Committee (DMC) with established terms of reference and stopping rules will be appointed to review interim data and other evidence (including updated overviews of relevant randomised controlled trials). The DMC will:
-
Advise on protocol modifications suggested by investigators or collaborators (e.g. sample size or trial endpoint modifications)
-
Assess the impact and relevance of new external evidence
-
Assess data quality, including completeness
-
Monitor recruitment figures, balance in recruitment by centre and losses to follow-up
-
Monitor evidence for treatment harm
-
Monitor planned sample size assumptions
-
Monitor treatment side-effects and compliance with the treatment regimen
-
Recommend that the trial continues to recruit participants or that recruitment should be terminated
An interim analysis will be conducted when outcomes are known for the first 50% of women recruited. The O'Brien-Fleming method[58] will be used to preserve the overall P-value at 0.05; for the interim analysis we will test at P = 0.0031 (z = 2.95430) and for the final analysis at P = 0.0497 (z = 1.96257) (these calculations were performed using PASS).
The DMC will inform the Steering Group if in their view an arm of the study is either clearly indicated or contra-indicated. The final decision on whether to stop the trial will rest with the Steering Group.
Ethics Approval
Ethics approval has been granted by the Northern Sydney Central Coast Area Health Human Research Ethics Committee (Protocol 0907-165M)