Table 4.
Hip fracture participants No. Death /Total No.* (%) |
Non-fracture participants No. Death /Total No.* (%) |
Unadjusted OR† |
Multivariable Adjusted OR†† |
|
---|---|---|---|---|
OVERALL | 66 / 192 (34.4) |
214 / 768 (27.9) |
1.4 (1.0, 2.0) |
1.3 (0.9, 2.0) |
0 to <1 YEAR | 29 / 192 (15.1) |
55 / 768 (7.2) |
2.3 (1.4, 3.7) |
2.8 (1.5, 5.2) |
≥1 to 5 YEARS | 18 / 120 (15.0) |
94 / 480 (19.6) |
0.7 (0.4, 1.2) |
0.4 (0.2, 0.8) |
≥5 to10 YEARS | 3 / 53 (5.7%) |
7 / 212 (3.3)‡‡ |
2.2 (0.4, 11.7) |
2.3 (0.3, 19.0) |
Decrease in total number of deaths with each time strata reflects removal of entire quintuplets (case and 4 matching controls) from analysis upon death of any member of quintuplet.
Odds ratios based on conditional logistic regression.
Multivariable model (n=891) includes age, BMI, current smoke status, clinic, history of diabetes, hypertension, stroke, MMSE score, ability to stand from chair at baseline exam, and total hip BMD. If model variables were non-estimable because of small cell size, we removed them.
The participants remaining in the analysis by ≥5 to10 years after their hip fracture were younger, had a higher hip BMD, had higher self rated health, had less diabetes and hypertension, had higher MMSE scores, and were less likely to use their arms to stand from a chair compared to those not remaining the analysis. These differences likely account for the lower absolute death rate in years 5 to 10 compared to years zero to five. This decrease in mortality rate at the ≥5 to10 year time point also occurred when all 5,580 women in the cohort were considered, confirming that cases and controls remained comparable over time.