Skip to main content

Table 4 Effect of email intervention on questionnaire completion, number of questionnaire reminders, and questionnaire completion rate

From: Increasing follow-up questionnaire response rates in a randomized controlled trial of telehealth for depression: three embedded controlled studies

  Control Urgent email Unadjusted odds ratio, b or hazard ratio 95 % confidence interval; P Adjusted odds ratio, b or hazard ratioa 95 % confidence interval; P
Primary analysisb
 Completed questionnaire 109/116 (94) 104/115 (90) 0.61 0.23, 1.63; 0.32 0.53 0.19, 1.49; 0.23
Sensitivity analysis
 Only those who required a reminder: Number of reminders = 0 54/61 (89) 69/80 (86) 0.81 0.30, 2.24; 0.69 0.66 0.22, 1.95; 0.45
Secondary analysisc
 Number of reminders = 0 55/116 (47) 35/115 (30)     
 Number of reminders = 1 26/116 (22) 35/115 (30)     
 Number of reminders = 2 10/116 (9) 16/115 (14)     
 Number of reminders = 3 7/116 (6) 12/115 (10)     
 Number of reminders = 4 7/116 (6) 8/115 (7)     
 Number of reminders = 5 11/116 (9) 9/115 (8) 0.27 −0.15, 0.69; 0.20 0.25 −0.18, 0.68; 0.26
Sensitivity analysisc
 Only those who required a reminder: number of reminders = 0 0/61 (0) 1/80 (1)     
 Number of reminders = 1 26/61 (43) 34/80 (43)     
 Number of reminders = 2 10/61 (16) 16/80 (20)     
 Number of reminders = 3 7/61 (11) 12/80 (15)     
 Number of reminders = 4 7/61 (11) 8/80 (10)     
 Number of reminders = 5 11/61 (18) 9/80 (11) −0.22 −0.72, 0.28; 0.38 −0.15 −0.65, 0.36; 0.56
Secondary analysisd
 Time to complete questionnaire, median days 9.0 11.0 0.84 0.64, 1.10; 0.22 0.88 0.67, 1.15; 0.34
Sensitivity analysisd
 Only those who required a reminder: time to complete questionnaire, median days 22.5 19.0 1.12 0.78, 1.60; 0.54 1.08 0.75, 1.55; 0.67
  1. Data are reported as number/total (percent), unless otherwise indicated. All primary analyses are intention-to-treat analyses. Sensitivity analyses included only those participants who required a reminder (excluded those who responded before a reminder was due and those who opted to complete the Patient Health Questionnaire (PHQ-9) over the phone before being sent any form of questionnaire) and, in addition to this, removing two outliers in terms of length of time it took to return the questionnaire (>80 days)
  2. aAnalyses adjusted by participant sex, baseline age and Patient Health Questionnaire (PHQ-9) score, main trial allocation arm, questionnaire method sent (paper or online)
  3. bLogistic regression analysis of pre-call group allocation on whether the 8-month questionnaire was completed or not (odds ratio)
  4. cLinear regression of pre-call group allocation on total number of reminders (linear regression coefficient, b). In the sensitivity analysis that includes only those who required an email reminder (i.e., excluded those who responded before a reminder was due and those who opted to complete the Patient Health Questionnaire (PHQ-9) over the phone before being sent a questionnaire), there was one participant who required an email reminder, but it was not sent due to administrative error
  5. dCox regression of pre-call group allocation on time to complete the questionnaire in days since the questionnaire was sent to the participant (hazard ratio). Zero values are not permitted in the time-to-event analysis, and so 0.5 days were allocated for participants who completed the questionnaire the day it was sent to them, while 91 days were allocated to the 18 participants that did not complete the questionnaire. For median: control n = 109 (sensitivity n = 54); urgent email n = 104 (sensitivity n = 69)