JMIR Publications

ABSTRACT

Background:

SF-36 is a popular questionnaire for measuring self-perception of quality of life in a given population of interest. Surprisingly, no study compared score values issued from a telephone interview versus an internet-based questionnaire self-completion.

Objective:

This study compared SF-36 score values issued from a telephone interview versus an internet-based questionnaire self-completion.

Methods:

Patients having an Internet connection and returning home after hospital discharge were enrolled in the SENTIPAT multicenter randomized trial the day of discharge. They were randomized to either self-complete a set of questionnaires using a dedicated website (I group) or to provide answers to the same questionnaires administered during a telephone interview (T group). This ancillary study of the trial compared SF-36 data relating to the post-hospitalization period in these two groups. In order to anticipate potential unbalanced characteristics of the respondents in the two groups, the impact of the mode of administration of the questionnaire on score differences was investigated using a matched sample of individuals originating from I and T groups (ratio 1:1), the matching procedure being based on a propensity score approach. SF-36 scores observed in I and T groups were compared with a Wilcoxon-Mann-Whitney test, the score differences between the two groups were also examined according to Cohen's effect size.

Results:

There were 245/840 (29%) and 630/840 (75%) SF-36 questionnaires completed in the I and T group, respectively (P<.001). Globally, score differences between groups before matching were similar to those observed in the matched sample. Mean scores observed in T group were all above the corresponding values observed in the I group. After matching, score differences in six out of the eight SF-36 scales were statistically significant, with a mean difference greater than 5 for four scales and an associated mild effect size ranging from 0.22 to 0.29, and with a mean difference near this threshold for two other scales (4.57 and 4.56) and a low corresponding effect size (0.18 and 0.16, respectively).

Conclusions:

There were 245/840 (29%) and 630/840 (75%) SF-36 questionnaires completed in the I and T group, respectively (P<.001). Globally, score differences between groups before matching were similar to those observed in the matched sample. Mean scores observed in T group were all above the corresponding values observed in the I group. After matching, score differences in six out of the eight SF-36 scales were statistically significant, with a mean difference greater than 5 for four scales and an associated mild effect size ranging from 0.22 to 0.29, and with a mean difference near this threshold for two other scales (4.57 and 4.56) and a low corresponding effect size (0.18 and 0.16, respectively). Clinical Trial: ClinicalTrials.gov NCT01769261, registered January 16, 2013.