Taylor 2003.

Study characteristics
Methods	RCT. Compared written self‐disclosure intervention to standard care (wait‐list control condition). Hypothesis: patients who completed the written self‐disclosure procedure would evidence improvement in health care utilization, disease severity and subjective health status 3 month following the intervention.
Participants	Population of interest: N = 81 (met eligibility). Number randomised: 70. Number of participants received the intervention: 70. Dropouts: 14. Excluded from data analysis because of unusual health care utilization patterns: 17. Final sample size: 39, intervention group (WSD) n = 18, control group (SMC) n = 21. Recruited from 2 CF centres located at 2 children’s hospitals in the southeastern part of the USA. Initial eligibility criteria included: (1) a diagnosis of CF; (2) age at least 15 years; (3) enrolled in the CF clinic for at least 9 months; (4) physical and mental, ability to complete the research protocol as judged by, the project staff and CF physician or nurse; (5) willingness, to write for 20 min on 3 separate occasions, over a 5‐day period and completion of self‐report, measures on two occasions over the next 3 months; and (6) access to a telephone on the days that the participants were to write at home.
Interventions	Written self‐disclosure intervention: 3‐sessions written self‐disclosure intervention, adapted from Pennebaker's laboratory‐based protocol First session in a private room within the CF clinic, two home‐writing sessions prompted by a telephone call (each 20 minutes). Instructions were in accord with those of Pennebaker (1989, 1993, 1997) and required that participants write about their "deepest thoughts and feelings about the most distressing experience of their entire life for a period of 20 minutes" (Pennebaker 1997).
Outcomes	Health care utilization: number of (1) outpatient visits; and (2) inpatient hospitalizations days. Physiological disease severity: (1) FEV1; (2) BMI. Subjective health status: (1) 'The Patient Health Questionnaire' (PHQ; Spitzer Taylor, Wallander, Anderson, Beasley, and Brown et al 1994; Spitzer, Kronke, Williams et al 1999): symptoms of (a) depression; (b) anxiety; (c) somatic complaints; and (d) psychological distress. (2) 'The Stressful Life Events Scale': 10 psychosocial complaints common among health care seeking populations (e.g., difficulties with family support, problems with significant others, and financial concerns); (3) SF‐12: severity and frequency of 12 physical and mental health problems, as well as of their impact on the patients' overall perceived health status. Feasibility and acceptability: modified version of the 'Visit Specific Satisfaction Questionnaire' (VSQ; Ware & Hays 1988).
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The authors provide insufficient information about the sequence generation. After contacting the authors they provided the information that they followed a published random numbers sequence and assigned the patient to one of the two treatments based on the next number in the sequence being odd or even.
Allocation concealment (selection bias)	Unclear risk	The authors provide insufficient information on the allocation concealment.
Blinding (performance bias and detection bias) All outcomes	Low risk	The physician and the clinic staff were not informed as to whether individual patients were in the WSD or the SMC condition. Patients and providers of the intervention knew who was in the WSD or the SMC condition (Taylor 2003).
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Number of drop‐outs in each group is not reported, demographic comparison data between those excluded and included in the final sample, using T‐tests and chi‐square analyses, revealed no significant differences between the two groups.
Selective reporting (reporting bias)	Low risk	All of the study’s pre‐specified outcomes have been reported.