Response to a pilot single-centre randomized trial: the PATRASTOM trial

Charlotte Murkin, Jane Blazeby, Natalie Blencowe, Leila Rooshenas, Barnaby Reeves, Ian Daniels, Thomas Pinkney, Neil Smart

Research output: Contribution to journalLetter (Academic Journal)peer-review

1 Citation (Scopus)

Abstract

Dear Editor,

We read the paper on the PATRASTOM pilot randomized controlled trial (RCT) [1] with great interest. Prevention of parastomal hernia (PSH) is an area that needs high-quality evidence. Good pretrial work, such as pilot studies, has an important role in the design of definitive multicentre trials. We have several methodological concerns, however, that limit the findings and interpretation of this study.

Pilot studies aim to test the feasibility, reliability and validity of the proposed design of the main RCT [2]. As such, pilot RCTs are designed to establish whether, for example, it is possible to recruit and randomize patients, and whether the preferred end-points in the main trial can be collected with acceptable completeness [3]. They provide the opportunity to ‘test drive’ the components of the main RCT; published reports are largely descriptive and contain detailed methodological information. Such studies may describe information needed to calculate the sample size required for the main trial (e.g. standard deviation of an outcome; frequency of missing outcome data) but should not estimate the effect size (any estimate would be very imprecise).

In view of these important considerations, what can be taken from the PATRASTOM study? The available recruitment data show that randomization was acceptable to patients and to surgeons, with 75 of 92 eligible patients (91%) recruited. A substantial proportion of participants (n = 15, 20%), however, were later excluded because intra-operative findings made the participant ineligible. This observation suggests that in the main trial it might be preferable to randomize participants in the operating room when eligibility can be known definitively; excluding randomized participants runs the risk of introducing bias. Secondly, PATRASTOM used the incidence of PSH as a primary end-point, which was assessed intra-operatively at the time of ileostomy closure. It appears, however, that the proposed main trial will investigate end-stomas. Therefore, piloting this former outcome during closure of a loop ileostomy is not relevant to the main trial. Furthermore, whilst PATRASTROM assessed quality of life as a secondary end-point using two established measures, neither measure has been validated in the patient population of interest. The main study will need to implement methods for detecting PSH (e.g. computed tomography and/or symptom assessment), yet these methods have not been piloted in the current study.

To conclude, PATRASTOM produced early data on safety of the interventions, but we believe that there were missed opportunities to obtain important information in this ‘pilot’ study. We also note that the original trial registration did not refer to the ‘pilot’ nature of the study [4].

The CIPHER (Cohort study to Investigate the prevention of Parastomal HERnia) study will soon open in the UK, and aims to address some of these issues. It includes the development of a measure for assessing symptoms of PSH and will describe key components of stoma formation. This nonrandomized cohort study will collect information about these components to investigate which surgical factors are prognostic for the development of PSH. It is hoped that the CIPHER study will eventually inform the optimal design of a trial evaluating interventions to prevent PSH.
Original languageEnglish
Article number10.1111/codi.13338
Pages (from-to)622–623
Number of pages2
JournalColorectal Disease
Volume18
Issue number6
DOIs
Publication statusPublished - 8 Jun 2016

Structured keywords

  • ConDuCT-II
  • Centre for Surgical Research
  • BTC (Bristol Trials Centre)

Fingerprint Dive into the research topics of 'Response to a pilot single-centre randomized trial: the PATRASTOM trial'. Together they form a unique fingerprint.

Cite this