McKessock 2001.

Methods	Study design: cluster‐RCT Aim: to establish and evaluate a new referral service for women referred to laparoscopic sterilisation, and to report on some methodological issues Unit of allocation: GP practices Unit of analysis: participants Unit of analysis issue: yes, as practices were randomly assigned and participants analysed; unclear also if effects of clustering were taken into account in the analysis Stratification: practices randomly assigned from prestratified lists according to size, location, etc Timing: 1 June 1996 to 31 March 1997 Data collection: prospective data collection through specific questionnaires and hospital records
Participants	Providers: gynaecologists, 230 general practitioners from 57 general practices, and nurses Participants randomly assigned: n = 232; intervention: 75, control: 157 referred for laparoscopic sterilisation Participants withdrawn or lost to follow‐up: intervention: n = 65, control: n = 57 (35 participants later crossed over from control to intervention group again) Baseline characteristics of participants: Age: no information Gender: 100% female Ethnicity: no information Clinical problem: referral for laparoscopic sterilisation Setting: general practices and hospital in the Grampian region, Aberdeen Royal Infirmary Country: Scotland, UK
Interventions	Type of intervention: intervention aimed at restructuring the referral process Intervention: direct referral for laparoscopic sterilisation: GP refers participant directly to clinic ‐ Gynaecology Outpatient Department, GOPD ‐ for laparoscopic sterilisation, bypassing referral to gynaecologist Control: routine referral through GOPD (GP refers to clinic’s gynaecologist, who then refers to sterilisation) GPs randomly assigned to direct referral (intervention) were supplied with a referral pack, which included: Referral criteria (drawn by a multi‐disciplinary team including gynaecologists, GPs and health service researchers, and based on knowledge and expert opinion of those involved and evidence of risk factors for regret following sterilisation from the literature; widespread consultation with local gynaecologists and GPs was carried out before the referral criteria were finalised) Detailed referral sheet Structured referral pro forma (concerning relevant participant history, examination details and counselling provided, which were subsequently sent to the research nurse) Patient information booklet (adapted from Document of Royal College of Obstetricians and Gynaecologists) to be given to the participant by the GP A newsletter kept practices up‐to‐date with the study's progress and encouraged continuing participation; a "theatre list" of two consultants was dedicated to direct referral sterilisation (separate waiting list) Duration of intervention: 10 months
Outcomes	Primary: Participant satisfaction Operative complication rate Participant and NHS costs Secondary:   Short‐term regret GP and gynaecologist satisfaction Waiting time from referral to operation (outcome included in this review) GP adherence to direct referral criteria
Notes	This study was terminated before the expected date after discussions with the funding body, as lower than expected recruitment rates made timely completion impossible. Findings of this trial were reported despite the small numbers, as important lessons were learnt
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Randomization was by referring practice to encourage consistency of referral behaviour and to avoid administrative complications within practices. Participating practices were randomised into intervention and control groups stratified by list size, fund holding status and rural or urban location. The practices in the eight blocks were placed into different coloured envelopes (one colour for each block). A coin was flipped to decide to commence with intervention or control allocation. From the first block of small list practices an envelope was selected and at the same time an envelope from a large list block was selected. These practices were randomised to the same arm of the trial and this process continued, allocating control or intervention alternately until all practices in those two groups were allocated into one of the two arms of the trial: one receiving and implementing the guidelines (intervention arm) and one maintaining the status quo (control arm). This process was carried out for each of the eight blocks" Comment: Method of randomisation is obscure: unclear whether envelopes were opaque and numbered; also unclear whether the envelopes were shuffled
Allocation concealment (selection bias)	Unclear risk	See quote above. Comment: Actual method of allocation is obscure. Unclear whether envelopes were opaque and numbered; also unclear whether the envelopes were shuffled
Blinding of participants and personnel (performance bias) Waiting time and number of visits to GPs before and after operation	Low risk	No masking possible but unlikely to affect results (objective nature of the outcomes of interest of the review)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Unclear whether assessors were blinded, but results unlikely to be affected (objective nature of the outcomes of interest of the review)
Incomplete outcome data (attrition bias) All outcomes	High risk	In the control arm, 57/157 (36.3%) participants withdrew from the trial; among participants randomly assigned to the intervention group, only 10 of 75 received the intervention
Selective reporting (reporting bias)	Unclear risk	Protocol not available; not possible to check whether all prespecified outcomes have been evaluated
Baseline characteristics similar?	Unclear risk	Baseline data only partially reported Quote: "There were no significant differences found in patients' characteristics between control and intervention groups" See Table 1. Referral criteria ‐ suitability for direct referral Comment: This was according to study authors; however, they reported mainly on different clinical criteria, and age was the only non‐clinical characteristic reported
Baseline outcomes similar?	Low risk	Not possible to provide baseline outcome data for the outcome of interest
Free of contamination?	High risk	A large proportion (65/75, 86.7%) of participants assigned to the experimental/intervention group were treated by the standard referral procedure (not eligible for or refused direct referral)
Other bias	Low risk	No obvious other risk of bias identified