OBJECTIVES: It has been found that respondents are more likely to report health problems on the EQ-5D-5L as the day goes on. A possible cause may be due to the EQ-5D recall period (today). Respondents may report no problems with pain or discomfort in the morning because the day is still young and they have yet to go to work. In this study, we test whether the EQ-5D-5L is more susceptible to this bias than health instruments with longer recall period (SF-36, AQoL).

METHODS: The Multi Instrument Comparison (MIC) project examined the health of 7 patient groups as well as the general population using a variety of instruments. This study re-examined U.S. and German responses (N=1,456 and 1,269, respectively) on the EQ-5D and other health instruments by time of day to assess the potential effects of recall period.

RESULTS: Results from the EQ-5D responses confirm a previous finding that the likelihood of reporting and severity of health problem increases as the day goes on (p-value<0.01). Controlling for demographic differences between respondents, the utility value predictions for late-night response are .04 lower than to those from other parts of the day. However, time of day is also associated with the likelihood of reporting and severity of problems on other instruments, including those with four-week recall periods (p-value=0.01). A comparison of ordinal logit models for similar questions found no significant differences in the time of day bias between instruments.

CONCLUSIONS : I find no evidence that the short recall period of the EQ-5D caused its association with time of day. The time of day bias was found across a variety of instruments and appears to be unrelated to recall period and more related to the health problem itself.