Respondents' recall of injury events: an investigation of recall bias in cross-sectional injury data from the Sudan Household Health Survey 2010.

Recall bias is a well-documented limitation of population-based cross-sectional injury surveys. To fill some gaps in this area, we investigated the extent and nature of recall bias in Sudan Household Health Survey (SHHS 2010) injury data. The extent of incomplete recall was measured by comparing the total reported injuries over 12 months with the annualised number of injuries in the four weeks preceding the survey. Multivariable logistic regression was used to investigate the association of socio-demographic variables, injury attributes and interviewee characteristics with differential recall. Relevant interactions were tested. Overall, reported injuries were 33% of the expected. Injuries among children 1-4 years had lower odds of being reported to have occurred earlier than the four weeks preceding the survey than people aged 65 years and over (OR = 0.24, 95% CI 0.12-0.47). Injuries that received inpatient care in the first week were more likely to be recalled than those that did not receive care (OR = 2.07, 95% CI 1.14-3.75). Respondent's age was associated with differential recall. Differential injury recall should be considered when using SHHS 2010 to compare injury occurrence between children under five and older groups or at the level of health care received.