Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Prognostic index for mortality of older people after hospital admission


Older people are more likely to be admitted to hospital, and for many of them who may not previously suffered a major illness this can be a major event that defines the future course of their life. Knowing their expected mortality in the year after discharge may be helpful in planning ongoing care with them and their families, and with the primary care physicians. A new study [1] gives a simple bedside method for doing this, and also forms an exemplar of how diagnostic or prognostic tests can be developed.


Patients aged 70 years and older admitted for two days or more to general medical services of two hospitals in Ohio, and taking part in a randomised trial, were the subjects of the study. Excluded were elective admissions, patients admitted to intensive care units, or admissions to subspecialty services. About 4% of enrolled patients had missing data, and were excluded. About a quarter of all admissions between 1993 and 1997 were used.

Patients from one hospital were used as a derivation cohort, from which risk categories and scoring system were devised. Patients from the other hospital were used as a validation cohort to test the rules.


Predictors of mortality were obtained from standardised interviews with patients, or from surrogates if patients were too ill. As well as demographic information, independence in five activities of daily living was assessed. These were bathing, dressing, using the toilet, transferring from bed to chair, and eating. A person who required assistance to perform one of these activities was classified as dependent for that activity. As well as the interview, comorbid conditions and admission laboratory values were obtained from medical records by trained medical abstractors.


The outcome was death within one year of discharge from hospital. This information was obtained from follow up interviews with patients or families, or from a death index. This information was available for all patients.


The relationship between the one-year mortality and each of many variables was found by logistic regression analysis. Those factors that were independently significantly associated with mortality were given points.


Six risk factors were independently associated with mortality - male sex, dependency for activities of daily living, congestive heart failure, cancer, and admission creatinine and albumin values. The risk factors, the strengths of association, and points given for each factor are shown in Table 1. In the derivation and the validation cohort, increased scores were associated with an increased risk of death in the year after hospital discharge.

Table 1: Independent risk factors

Risk factor Odds ratio (95%CI) Points
Male sex 1.4 (1.1 to 1.8) 1
Dependent in 1-4 activities of daily living 2.1 (1.6 to 2.8) 2
Dependent in all 5 activities of daily living 5.7 (4.2 to 7.7) 5
Congestive heart failure 2.0 (1.5 to 2.5) 2
Solitary cancer 2.6 (1.7 to 3.9) 3
Metastatic cancer 13.4 (6.2 to 29) 8
Admission creatinine >265 µmol/L 1.7 (1.2 to 2.5) 2
Admission albumin 30-34 g/L 1.7 (1.2 to 2.3) 1
Admission albumin less than 30 g/L 2.1 (1.4 to 3.0) 2
The adjusted odds ratio was from multivariate logistic regression using the derivation cohort

Figure 1: Risk and scores for derivation and validation cohorts


This is an ideal paper to read for detailed methods about developing clinical scoring systems. It has the same sort of strength as found in the development of the Ottawa ankle and knee rules ( Bandolier 21 and 49 ). It is interesting that a number of these clinical scoring systems are now appearing, predominantly from north America.

Knowing that the risk of dying in the next year is low or high may be useful to professionals planning care, or lay carers or patients. Individual cases will differ. Perhaps the real strength of this type of evidence, and scoring systems in general, is they remind us about what is important and become an ongoing quality check.


  1. LC Walter et al. Development and validation of a prognostic index for 1-year mortality in older adults after hospitalization. JAMA 2001 285: 2987-2994.
previous or next story in this issue