Menu
Menu
Menu
Menu
Background: The CORONIS Trial was a 2×2×2×2×2 non-regular, fractional, factorial trial of five pairs of alternative caesarean section surgical techniques on a range of short-term outcomes, the primary outcome being a composite of maternal death or infectious morbidity. The consequences of different surgical techniques on longer term outcomes have not been well assessed in previous studies. Such outcomes include those related to subsequent pregnancy: mode of delivery; abnormal placentation (e.g. accreta); postpartum hysterectomy, as well as longer term pelvic problems: pain, urinary problems, infertility. The Coronis Follow-up Study aims to measure and compare the incidence of these outcomes between the randomised groups at around three years after women participated in the CORONIS Trial.Methods/Design: This study will assess the following null hypotheses: In women who underwent delivery by caesarean section, no differences will be detected with respect to a range of long-term outcomes when comparing the following five pairs of alternative surgical techniques evaluated in the CORONIS Trial:. 1. Blunt versus sharp abdominal entry. 2. Exteriorisation of the uterus for repair versus intra-abdominal repair. 3. Single versus double layer closure of the uterus. 4. Closure versus non-closure of the peritoneum (pelvic and parietal). 5. Chromic catgut versus Polyglactin-910 for uterine repair. The outcomes will include (1) women’s health: pelvic pain; dysmenorrhoea; deep dyspareunia; urinary symptoms; laparoscopy; hysterectomy; tubal/ovarian surgery; abdominal hernias; bowel obstruction; infertility; death. (2) Outcomes of subsequent pregnancies: inter-pregnancy interval; pregnancy outcome; gestation at delivery; mode of delivery; pregnancy complications; surgery during or following delivery.Discussion: The results of this follow-up study will have importance for all pregnant women and for health professionals who provide care for pregnant women. Although the results will have been collected in seven countries with limited health care resources (Argentina, Chile, Ghana, India, Kenya, Pakistan, Sudan) any differences in outcomes associated with different surgical techniques are likely to be generalisable throughout the world.Trial registration: ISRCTN31089967. © 2013 The CORONIS Collaborative Group; licensee BioMed Central Ltd.
The objective of the CORONIS Follow-up Study is to measure and compare the incidence of outcomes at around three years between the randomised groups of women who participated in the CORONIS Trial. This study will assess the following null hypotheses: In women who underwent delivery by caesarean section, no differences will be detected with respect to a range of long-term outcomes when comparing the following five pairs of alternative surgical techniques: 1. Blunt versus sharp abdominal entry 2. Exteriorisation of the uterus for repair versus intra-abdominal repair 3. Single versus double layer closure of the uterus 4. Closure versus non-closure of the peritoneum (pelvic and parietal) 5. Chromic catgut versus Polyglactin-910 for uterine repair All women participating in the CORONIS Trial have been informed that we hope to be able to interview them at least three years after they were included in CORONIS to ask them about their long-term health. The trial entry information leaflet contains the following paragraph: “We are also keen to find out whether the different methods used make any difference to your long-term health. We would like to contact you again after the study has finished. We will ask you how you are and whether you have had any more babies.” In addition, at the ‘six week’ follow-up appointment, women are provided with the following letter: “Thank you for taking part in CORONIS and for letting us know about your health since your caesarean section.” We hope to see all women who take part in the study once more. This will be three years after the caesarean section. At this visit we will also ask about your health since your caesarean section. Because a lot can happen in three years, we will keep in touch with you every 6 months or so to make sure we have your correct phone number and address. If you have any more pregnancies or have any surgery it would be very helpful if you, or your doctor, could write these down on the CORONIS Medical Card. This will make it easier for you to remember everything that has happened since your caesarean section. If you move house or change your phone number, please fill in the Change of Address card and send it back to the Study Office, or phone us. The phone number and address are: [details] “If you do not want to take part in the study any more, you can let us know by phoning or writing to the Study Office. Thank you for your help with this study.” To facilitate long-term follow-up women have been provided with a medical card and a document bag for storing their medical information. This bag has the CORONIS Trial logo printed on the outside and contains information about letting the local Regional Trial Office know if the woman moves house or changes other contact details. The process of maintaining contact with women who participated in CORONIS between their six week follow-up and the three year follow-up is coordinated by each Regional Trial Office to reflect differences in the circumstances between the very different settings for this trial. The list below represents the full range of outcomes relevant to the five intervention pairs being evaluated. 1. Following the CORONIS birth and before any subsequent pregnancy, any new onset or worsening of: i. pelvic pain ii. dysmenorrhoea iii. deep dyspareunia iv. urinary symptoms of poor stream and/or frequency which did not respond to antibiotics 2. Diagnostic laparoscopy or diagnostic laparotomy (not related to pregnancy) 3. Hysterectomy or tubal/ovarian surgery (not related to pregnancy) 4. Bladder or bowel damage in those women who have had surgery, excluding diagnostic laparoscopy and diagnostic laparotomy (not related to pregnancy). 5. Following the CORONIS birth, any new onset of: i. abdominal hernia ii. bowel obstruction 6. Woman’s death 7. Number of women with no subsequent pregnancy i. Voluntary infertility ii. Involuntary infertility 8. Use of fertility treatments 9. Number of women having any subsequent pregnancy and for these women, the following outcomes will be measured: i. Inter-pregnancy interval from the CORONIS birth to the end of the subsequent pregnancy (regardless of loss or birth) ii. Miscarriage of the pregnancy subsequent to the CORONIS birth iii. Ectopic pregnancy For the birth following the CORONIS birth: iv. Gestation at delivery (by best estimate) of the first viable pregnancy (gestational age > 24 or >28 weeks depending on country specific definition) v. Stillbirth vi. Neonatal death vii. Mode of delivery: a. Non-instrumental vaginal b. Instrumental vaginal c. Pre-labour caesarean section d. In labour caesarean section viii. Other pregnancy complications (one or more of the following components: uterine rupture, uterine scar dehiscence, placenta praevia, morbidly adherent placenta, abruption, postpartum haemorrhage requiring transfusion >1 unit of whole blood or packed cells, severe infection within 6 weeks postpartum, hysterectomy up to 6 weeks postpartum, manual removal of placenta) ix. Bladder or bowel damage at the time of subsequent caesarean section 10. Death or serious morbidity of the child who was born at the time of CORONIS participation. Parent report of one or more of the following: child’s blindness, deafness, inability to speak or walk without assistance, or other major morbidity such as the presence of a stoma. Although no difference in death is expected, there is likely to be a time difference between ‘sharp’ and ‘blunt’ abdominal entry and this may, in theory, lead to more neonatal encephalopathy which may lead to a greater risk of later death. All outcomes are potentially relevant for each comparison pair; however, some outcomes are more likely to be influenced by some interventions compared with others. Therefore we have categorised comparisons into (i) main comparisons of interest, and (ii) secondary comparisons of interest (Table 1). This approach has been pre-specified to take account of multiple testing and to simplify reporting of the study. Only the main comparisons of interest will be reported in the main paper, although the remaining comparisons will be presented in appendices to ensure transparency and completeness of the analysis. Comparisons of interest (M denotes main comparisons of interest, s denotes secondary comparisons of interest) Key: Bl vs. Sh = Blunt versus sharp abdominal entry, Ext vs. I-A = Exteriorisation of the uterus for repair versus intra-abdominal repair, Sl vs. Dble = Single versus double layer closure of the uterus, Cl vs. N-Cl = Closure versus non-closure of the peritoneum (pelvic and parietal), Cat vs. P910 = Chromic catgut versus Polyglactin-910 for uterine repair. Most of the information requested will be provided by the woman at face-to-face interview. Many of the outcomes being collected are clinical diagnoses which are made every day by clinicians in these settings. Women will be asked whether they have pelvic pain and pain on intercourse. Whether or not they have sought or received medical care for their pain will be used as a marker of severity. For women who are pregnant at the time of the follow-up, data will be collected once the pregnancy is over. For the relatively small number of women who have one of the outcomes resulting in hospitalisation (including subsequent pregnancy complications, such as uterine rupture or dehiscence, hysterectomy, placenta praevia, or other morbidity including non-pregnancy related hysterectomy and laparoscopy/laparotomy), consent will be requested from the women to access their hospital notes, so that additional information relating to the incident can be collected from the relevant hospital by the assessment doctors. Although medical records in some settings in these countries may be poor, this mechanism is being used to seek confirmation of events. We will not use these data as the primary source of the outcome data – these come from the women. In addition any relevant laboratory reports and histopathology will be used by the clinicians reporting morbidity to support their diagnosis. We will collect these data, but we do not require these for confirmation of the diagnosis. For example a post-partum hysterectomy histopathology report will not be required to support a diagnosis of a morbidly adherent placenta but will be used as supportive evidence, if it is available. A detailed Statistical Analysis Plan (SAP) will be developed and agreed by the Study Steering Committee (SSC) before the analysis is undertaken. For the primary analyses, all maternal and child outcomes will be analysed in the groups into which they were randomly allocated e.g. comparing the outcome of all women allocated “blunt abdominal entry” with all those allocated “sharp abdominal entry”, regardless of technique received. Maternal and child outcomes will be analysed for all eligible women correctly recruited for whom data are available. The outcomes of interest for women who have a subsequent viable pregnancy, such as uterine rupture, will be analysed using two different denominators i.e. women who subsequently have at least one pregnancy and all women randomised. This will take account of the potential for differences in the pregnancy rate between the two arms being compared. The flow of participants through each stage of the trial will be illustrated in a CONSORT flowchart [24,25]. This will include the numbers (with percentages) of losses to follow-up over the period of the follow-up study (including reasons, where known) by pair of interventions. Baseline demographic and clinical characteristics will be described separately for the five intervention pairs in the main trial publication, for those women for whom follow-up data are collected. We will compare the characteristics of such women with those for whom no follow-up assessment was possible using tests of statistical significance, overall and by pair of interventions. The primary analyses will be performed on each of the five pairs of surgical techniques. Outcomes will be summarised by intervention group using appropriate summary statistics (counts and percentages, means and standard deviations, or medians and inter-quartile ranges). Comparisons will be made using relative risks (RR) for dichotomous/categorical outcomes, mean differences for normally distributed continuous outcomes, or median differences for skewed continuous variables (unless the data can be transformed to normality). Analysis of time to event outcomes will employ survival analysis techniques [26]. Multiplicity of statistical testing will be addressed in the detailed statistical analysis plan. With regard to the analysis of child outcomes, multiple births will be treated as independent events in the primary analysis, in order that relative risks may be presented. Adjusted analyses will be performed on all comparisons to investigate the impact of minimsation factors ‘In-labour or not in-labour caesarean section’ and ‘number of previous caesarean sections (none or one)’. These will be agreed by the investigators and the Study Steering Committee and pre-specified in the SAP. Subgroup analyses will be similarly agreed and pre-specified in the statistical analysis plan and results will be presented on forest plots, by intervention pair, wherever appropriate. The robustness of the results will be examined in sensitivity analyses by using multiple imputation techniques to impute missing 3 year follow-up data, and also restricting analyses to a pre-specified follow-up window. Sensitivity of the effect estimates to adjustments for potential clustering of multiple births will also be examined. Interactions will be explored using a structured approach. Plausible interactions are difficult to identify and not all can be assessed due to the nature of the trial design. A strategy for the analysis of interactions for the main trial has been agreed by the Trial Steering Committee to prevent the investigators being misled by spurious interactions, given the multiplicity of possible explorations. This strategy will be employed for the primary analysis of the follow-up study i.e. analyses of interactions will only be performed on the main comparisons; and three-way interactions will not be investigated unless there is strong evidence of a two-way interaction in the presence of main effects. The results of the exploration of interactions will be interpreted cautiously together with other evidence such as biological plausibility and consistency (e.g. across countries). Plausible interactions for each outcome will be agreed by the investigators and the Study Steering Committee and pre-specified in the SAP. The original sample size for the main trial was calculated assuming a 15% primary outcome event rate in the control group (80% power, 5% two-sided significance level, ability to detect a relative risk of 0.85, 15% loss to follow-up) which resulted in a total sample size required of ~ 15,000 women (~ 9,000 in each intervention pair). The actual overall event rate observed during the conduct of the trial was found to be around 9%. The sample size was increased to 15,492 (9,296 per intervention pair) to account for the reduction in the event rate and agreed by the TSC. Assuming an overall response rate of at least 80% for the follow-up [27,28], this will provide us with information from ~12,400 women (or ~7,400 for each intervention pair). If we assume approximately 80% of these women will have had a subsequent pregnancy this will result in ~5,900 women in each intervention pair. This number will be sufficient to detect modest but clinically important differences between any principal comparison for this population. The statistical power available, based on a fixed sample size, for a range of event rates on a selection of outcomes is high (Table 2). These incidence rate estimates are based on two follow-up studies of randomised trials, both of which were from developed countries where the event rates are likely to be lower than those observed in the countries participating in CORONIS. These studies, therefore, are likely to represent a conservative estimate of the anticipated event rates. One study was following a trial of closure versus non-closure of the peritoneum [12] and the other following single versus double layer closure of the uterine incision13. Both studies were small (n = 145 and 144 women) and the estimates of some of the outcomes are therefore imprecise. For example, the estimate of uterine dehiscence is based on 1 event out of 145 women, giving an incidence of 0.69% with a 95% confidence interval of 0.12% to 3.8%. The estimate of dyspareunia is based on 27 events out of 144 women giving an incidence of 19% with a 95% confidence interval of 13% to 26%. power to detect a specified difference for a fixed sample size The power calculations in Table 2 assume equal numbers in each comparison and a 2-5% level of statistical significance.
N/A
