Table 3.

Sequential Models for Predicting Fall-Related Injuries

Model	IRR (95% CI)	AICa	Change in AIC
Base model (age, gender)		2156.4
Age (per additional 5 years)	1.22 (1.10–1.34)b
Female gender	1.11 (0.87–1.43)
1) Base + comorbidity count		2133.7	−22.7c
Elixhauser score (per additional comorbidity)	1.13 (1.08–1.18)b
Age	1.22 (1.10–1.34)b
Female gender	1.16 (0.90–1.48)
2) Base + comorbidity count + prior FRI		2126.1	−7.6d
Claim for FRI in 12 months prior to screening	1.60 (1.20–2.13)b
Elixhauser score	1.11 (1.06–1.16)b
Age	1.22 (1.10–1.34)b
Female gender	1.13 (0.88–1.45)
3) Base + comorbidity count + Screening Question		2115.4	−18.3d
≥2 falls in the past yeare	1.64 (1.32–2.03)b
Elixhauser score	1.12 (1.07–1.17)b
Age	1.22 (1.10–1.34)b
Female gender	1.20 (0.94–1.54)
4) Base + comorbidity count + screening question + prior FRI		2112.3	−3.1f
≥2 falls in the past yeare	1.56 (1.25–1.94)b
Claim for FRI in 12 months prior to screening	1.41 (1.05–1.89)b
Elixhauser score	1.10 (1.05–1.15)b
Age	1.22 (1.10–1.34)b
Female gender	1.18 (0.92–1.51)

AIC indicates Akaike information criterion; FRI, fall-related injury; IRR, incidence rate ratio.

^aAIC is a likelihood measure in which lower values indicate better fit and a penalty is paid for increasing the number of variables in the model.¹⁷

^bP <.05.

^cScreening question, “Have you fallen 2 or more times in the past 12 months?”

^dChange in AIC from base model.

^eChange in AIC from model 1.

^fChange in AIC from model 3.

All models also adjusted with study site, intervention group, and a dichotomous variable for presence of missing claims data in the 12 months prior to screening date.