Skip to main content
. Author manuscript; available in PMC: 2011 Apr 20.
Published in final edited form as: JAMA. 2011 Jan 5;305(1):50–58. doi: 10.1001/jama.2010.1923

Table 3.

Predictive Accuracy for 5- and 10-Year Survival by Individual Study and Pooled Data Presented as Area Under the Receiver Operating Characteristic Curves

Outcome and Predictors C Statistic (95% Confidence Interval)
CHS8 EPESE15 Health, ABC16,17 Hispanic EPESE8 Invecciare in Chianti18 Osteoporotic Fractures in Men19 NHANES III20 PEP21 Study of Osteoporotic Fractures22 Pooled
5-Year Mortality
Age, sex 0.705 (0.685–0.725) 0.685 (0.658–0.712) 0.606 (0.575–0.637) 0.694 (0.662–0.725) 0.797 (0.754–0.841) 0.700 (0.677–0.723) 0.710 (0.691–0.729) 0.674 (0.616–0.732) 0.646 (0.625–0.667) 0.690 (0.662–0.717)

Age, sex, diseases 0.711 (0.692–0.731) 0.692 (0.665–0.719) 0.616 (0.586–0.647) 0.703 (0.671–0.725) 0.793 (0.747–0.838) 0.704 (0.681–0.727) 0.719 (0.700–0.737) 0.694 (0.737–0.750) 0.662 (0.639–0.684) 0.698 (0.673–0.723)

Age, sex, diseases, BMI, systolic BP, prior hospitalization 0.736 (0.717–0.755)b,c 0.702 (0.676–0.728) 0.650 (0.620–0.680)b,c 0.728 (0.698–0.755)b,c 0.808 (0.765–0.850) 0.728 (0.706–0.749)b 0.744 (0.727–0.762)b,c 0.728 (0.674–0.781)b,c 0.665 (0.643–0.686) 0.719 (0.693–0.745)b

Age, sex, use of mobility aid, functional statusa NA NA NA 0.735 (0.705–0.765) 0.803 (0.756–0.851) NA 0.738 (0.720–0.757) 0.720 (0.663–0.776) NA 0.747 (0.720–0.774)

Age, sex, gait speed 0.734 (0.716–0.753)c 0.711 (0.685–0.737)c 0.642 (0.612–0.673)b,c 0.710 (0.679–0.741) 0.803 (0.760–0.846) 0.729 (0.707–0.751)b,c 0.737 (0.719–0.755)b 0.718 (0.664–0.771)c 0.682 (0.662–0.703)b,c 0.717 (0.694–0.740)
0.741 (0.706–0.775)d

10-Year Mortality
Age, sex 0.721 (0.707–0.734) 0.725 (0.704–0.746) NA 0.700 (0.677–0.724) NA NA 0.741 (0.726–0.757) 0.674 (0.627–0.721) 0.689 (0.676–0.703) 0.712 (0.692–0.731)

Age, sex, diseases 0.728 (0.715–0.742) 0.738 (0.716–0.759) NA 0.709 (0.685–0.733) NA NA 0.749 (0.734–0.764) 0.698 (0.652–0.744) 0.706 (0.692–0.719) 0.724 (0.707–0.740)

Age, sex, diseases, BMI, systolic BP, prior hospitalization 0.745 (0.732–0.759) 0.749 (0.729–0.770) NA 0.733 (0.710–0.756)b NA NA 0.768 (0.754–0.783)b 0.723 (0.678–0.727)b,c 0.709 (0.696–0.722) 0.739 (0.719–0.759)b

Age, sex, functional status, walking aid usea NA NA NA 0.722 (0.699–0.746) NA NA 0.761 (0.746–0.776) 0.702 (0.655–0.748) NA 0.732 (0.698–0.767)

Age, sex, gait speed 0.740 (0.727–0.754) 0.753 (0.733–0.774)b NA 0.709 (0.685–0.732) NA NA 0.766 (0.751–0.780)b 0.723 (0.679–0.768)b,c 0.719 (0.706–0.731)b 0.737 (0.718–0.755)b,c
0.734 (0.692–0.777)d

Abbreviations: ABC, Aging and Body Composition; BMI, body mass index, calculated as weight in kilograms divided by height in meters squared; BP, blood pressure; CHS, Cardiovascular Health Study; CI, confidence interval; EPESE, Established Populations for the Epidemiological Study of the Elderly; NA, not applicable; NHANES III, Third National Health and Nutrition Examination Survey; PEP, Predicting Elderly Performance.

a

Functional status was operationally defined for 3 levels: (1) activities of daily living (ADLs) dependence is defined as report of needing help from another person or being unable to perform any of 6 basic ADLs, 2) Instrumental ADL difficulty is defined as report of no ADL dependence but difficulty performing shopping, meal preparation, or heavy housework, and (3) Independent is defined as no report of ADL dependence or instrumental ADL difficulty.

b

Value is the pooled estimate of the area under the receiver operator characteristic curve for age, sex, and gait speed for the studies that were used in the comparisons of gait speed with use of mobility aids and functional status. Four studies were included in the estimates of 5 y mortality and three in the estimates of 10 y mortality. Values are reported as the C statistic representing area under the receiver operator characteristic curve; values that differ by 0.025 or more are considered substantially different.37

c

C statistic is greater than for age and sex alone.

d

C statistic is greater than for age, sex, and diseases.