Prognosticating with the hospitalized-patient one-year mortality risk score using information abstracted from the medical record

There were 43,883 overnight hospitalizations at our hospital in 2011, and 38,886 hospitalizations were excluded: 1883 hospitalizations were in the same-day surgery or the bone-marrow transplant unit; 2485 did not have a valid health card number; 34,515 were not randomly selected; the records of 3 randomly selected patients had been blocked by our hospital’s privacy department; and 1 patient could not be linked with the population-based administrative datasets.

The 4996 study patients were middle-aged and predominantly female (Table 2). The extensive majority of patients was admitted from the community, was independent for ADL, had a family member as the principal contact, and had no admissions by ambulance in the previous year. Most people had no significant comorbidities or ED visits in the year prior to their admission. The mean chart-HOMR score was 22 (standard deviation [SD], 12), which is associated with a 1.2% expected risk of death within 1 year of hospital admission (Appendix 1).³

A total of 563 patients (11.3%) died within 1 year of admission to hospital (Table 2). In the study cohort, each chart-HOMR component was associated with death status. People who died were older, more likely to be male, had a greater number of important comorbidities, had more ED visits and admissions by ambulance in the previous year, and were more likely to have been discharged in the previous 30 days, and were admitted urgently, directly to the intensive care unit, or with complicated diagnoses. The mean chart-HOMR score differed extensively by survival status (37.4 [SD, 7.5] in those who died vs. 19.9 [SD, 12.2] in those who survived). Three of the socio-functional variables (delirium and falls noted on admission documents, and dependent for any ADL) also varied with death status.

The chart-HOMR score was strongly associated with the likelihood of death within 1 year of admission. When included in a logistic regression model having 1-year death as the outcome, a 1-point increase in the chart-HOMR score was associated with a 19% increase in the odds of death (P < 0.0001). This model (with only the chart-HOMR score) was highly discriminative (C statistic, 0.888) and well calibrated (Hosmer-Lemeshow test, 12.9 [8 df, P = 0.11]).

Observed and expected death risks by chart-HOMR score were similar (Figure 1). The observed total number of deaths (n = 563; 11.3%) exceeded the expected number of deaths (n = 437, 8.7%). When we regressed observed death risks on expected death risks for chart-HOMR scores (clustered into 22 groups), the Hosmer-Lemeshow test was significant, indicating that differences between observed and expected risks were beyond that expected by chance (Hosmer-Lemeshow test, 141.9, 21 df, P < 0.0001). The intercept of this model (0.035; 95% CI, 0.01-0.06) was statistically significant (P = 0.01), indicating that the observed number of cases significantly exceeded the expected; however, its calibration slope (1.02; 95% CI, 0.89-1.16) did not deviate significantly from unity, indicating that the relationship between the observed and expected death risk (based on the chart-HOMR score) remained intact (Figure 1).

The deviations between observed and expected death risks reflected deviations between the c chart-HOMR score and the administrative-HOMR score, with the former being significantly lower than the latter (Figure 2). Overall, the chart-HOMR score was 0.96 points lower (95% CI, 0.81-1.12) than the administrative-HOMR score. The HOMR score components that were notably underestimated using chart data included those for the age-Charlson Comorbidity Index interaction, living status, and admit points. Points for only 2 components (admitting service and admission urgency) were higher when calculated using chart data.

Four additional socio-functional variables collected from medical record review were significantly associated with 1-year death risk independent of the chart-HOMR score (Table 3). Admission documentation noting either delirium or falls were both associated with a significantly increased death risk (adjusted odds ratio [OR], 1.92 [95% CI, 1.24-2.96] and OR 1.96 [95% CI, 1.29-2.99], respectively). An independently increased death risk was also noted in patients who were dependent for any ADL (adjusted OR, 1.99 [95% CI, 1.24-3.19]). The presence of an ED geriatrics consultation within the previous 6 months was associated with a significantly decreased death risk of 60% (adjusted OR, 0.40 [95% CI, 0.20-0.81]). Adding these covariates to the logistic model with the chart-HOMR score significantly improved predictions (likelihood ratio statistic = 33.569, 4df, P < 0.00001).

In a large random sample of patients from our hospital, we found that the HOMR score using data abstracted from the medical record was significantly associated with 1-year death risk. The expected death risk based on the chart-HOMR score underestimated observed death risk but the relationship between the chart-HOMR score and death risk was similar to that in studies using administrative data. The HOMR score calculated using data from the chart was lower than that calculated using data from population-based administrative datasets; additional variables indicating patient frailty were significantly associated with 1-year death risk independent of the chart-HOMR score. Since the HOMR score was derived and initially validated using health administrative data, this study using data abstracted from the health record shows that the HOMR score has methodological generalizability.⁸