Session number: 1037
Abstract title: Use of Computer Telephony for the Administration of the Veterans' SF-36 and SF-12
Author(s):
JA Rothendler - Center for Health Quality, Outcomes and Economic Research
LE Kazis - Center for Health Quality, Outcomes and Economic Research
DR Miller - Center for Health Quality, Outcomes and Economic Research
A Lee - Center for Health Quality, Outcomes and Economic Research
BG Fincke - Center for Health Quality, Outcomes and Economic Research
S Gaehde - Center for Health Quality, Outcomes and Economic Research
Objectives: The Veterans' SF-36 and SF-12 are important instruments for evaluating health status. Large-scale administration using mailed forms involves considerable expense and delays in acquiring data. We evaluated administration of these surveys using "computer telephony" (CT) with questions asked via computer over the telephone in an interactive, automated fashion, and responses entered using telephone keypads.
Methods: Utilizing a randomized controlled trial involving 1,998 veterans, we evaluated 3 administration modes for each survey instrument: 1) Mailed surveys (“M-arms"), 2) CT without supplying subjects with the questions beforehand (“T-arms"), and 3) CT in which the questions were supplied beforehand (“Tq-arms"). To reduce possible effects of response bias on inter-arm comparison of scores, we calculated “difference” scores by subtracting expected scores (based on age, gender, marital status and diagnoses) from observed scores.
Results: The CT arms had significantly lower response rates compared to the M-arms (24% vs. 71%, p<.05), and Tq-arms had a higher response rate compared to the T-arms (27% vs. 21%, p<.05). Among responders, there were no significant differences among arms in age, marital status or disease burden (as measured by expected Physical Component Summary (PCS) or Mental Component Summary (MCS) scores). For responders, completeness of responses, as measured by calculable summary scores, was not statistically different among arms. For the Veterans’ SF-36, the MCS difference score and four scale difference scores for the T-arm were significantly greater compared to the M-arm (p<.05). There were no significant differences among modes in PCS or MCS difference scores for the Veterans’ SF-12 or in PCS difference scores for the Veterans’ SF-36.
Conclusions: While response rates for our specific implementation of CT were low compared to mailed surveys, there did not appear to be additional barriers to response based on age or overall disease burden, and completeness of responses was similar. Certain survey scores acquired by CT differed from mailed surveys and should be interpreted with caution.
Impact statement: This study improves our understanding of the use of CT for administration of the Veterans' SF-36 and SF-12. Further research is needed to explore methods of improving response rates and to better understand observed effects of administration mode on scores.