Multiple Imputation to Account for Missing Data in a Survey: Estimating the Prevalence of Osteoporosis

Abstract

Epidemiologists should consider using multiple imputation more often than is current practice.

Menopausal hormone therapy and risk of cholecystectomy: a prospective study based on the French E3N cohort

“…Missing data for adjustment variables were imputed by means of a multiple imputation procedure. 20 We explored potential interactions between use of menopausal hormone therapy and covariates by including an interaction term in the models.…”

Geographic variation of bone mineral density and selected risk factors for prediction of incident fracture among Canadians 50 and older

“…Recruitment rates varied by center and this may introduce selection bias; however, previous analysis suggests little bias for most of the age range [38]. Fractures outcomes were based on self-reported fractures, and this implies potential mechanisms of additional ascertainment bias provided there were between-center differences in fracture questionnaire response rates.…”

Prospectively measured 10-year changes in health-related quality of life and comparison with cross-sectional estimates in a population-based cohort of adult women and men

“…Multiple imputation was initially considered and would have merit for correcting at least some of the bias [21,22]. However, it was not utilized since it too has limitations within the context of non-ignorable data.…”