Prognostic Index for Mortality

Study
Results
Comment

Knowing the future is impossible. Predicting what is likely to happen based on what we know already is tricky, but possible. A newly-developed prognostic index produces estimates of future four-year mortality in older adults based on age, illness, and function [1].

Study

Community dwelling adults older than 50 years provided information, mainly through telephone interviews. After exclusion of some with incomplete data, the final sample was 19,710 people across the United States. The study began in 1992, and expanded in 1998 to become a representative sample of the older US population. The sample was split into a developmental cohort and a validation cohort, based on geography.

A large amount of information was collected in the interview. As well as age, 18 behavioural and comorbidity variables were examined, including BMI, visual or hearing impairment, and self-reported illnesses. Participants were also asked about 21 functional measures, including activities like bathing, walking several city blocks, or pushing or pulling heavy objects.

Death was assessed using follow up procedures involving cross-referencing with US national death registers. Mortality over four years was then correlated with all the variables, with clever statistical methods to provide a model with 12 predictor variables. These were given points (Table 1), and the predictive accuracy of the developmental cohort was checked against the validation cohort.

Table 1: Points and features in the index

Risk factor	Points
Age
60-64	1
65-69	2
70-74	3
75-79	4
80-84	5
>85	7
Male	2
Diabetes	1
Cancer	2
Lung disease	2
Heart failure	2
BMI<25	1
Current smoker	2
Problems
bathing	2
managing finances	2
walking several blocks	2
pushing/pulling heavy objects	1

Results

The mean age of participants was 67 years; 57% were women, and 10% black in the development cohort, and in the validation cohort mean age was 67 years, 56% were women, and 19% black. Over four years 12% and 13% of people in the cohorts died (over 2,400 deaths in total), and there were 68,000 person years of observation.

Risk stratification by points is shown in Figure 1, with increasing four-year mortality with increasing points in both development and validation cohort. Low risk (0-5 points) had a four year mortality of 3%, Medium risk (6-9 points) 15%, high risk (10-13 points) 41%, and for the highest risk of over 14 points it was 65%.

Figure 1: Mortality with development (square) and validation (circle) cohorts

Comment

It is unusual to find a prognostic tool developed in such a large number of people over such a long period of time, and done so well. The paper comes with an example of a one-page questionnaire with 12 questions, which would allow a professional or patient to calculate their own score.

It is potentially useful, but where, and what for, remains to be seen.

Reference:

SJ Lee et al. Development and validation of a prognostic index for 4-year mortality in older adults. JAMA 2006 295: 801-808.