Comorbidity Burden Predicts Patient-Important Adverse Outcomes and Informs Care Planning: A Retrospective Study from an Australian Metropolitan Hospital

Background

Mortality is a common clinical outcome indicator but fails to capture patient-important outcomes like functional independence, cognition, and quality of life.

Objective

To assess performance of three validated indices of comorbidity burden-Charlson Comorbidity Index (CCI), AHRQ Elixhauser Index (AHRQ-EI), Van Walraven modification (VW-EI)- for predicting patient-important adverse outcomes and healthcare resource utilisation.

Methods

Retrospective audit of all acute adult admissions to a metropolitan Australian teaching hospital submitted to the Victorian Admitted Episodes Dataset (VAED). Patient-important adverse outcomes were defined as a composite of in-hospital death, discharge to new residential care, or discharge against medical advice (DAMA). Data on ICU admission and hospital length of stay (LOS) was also collected. Receiver operating characteristic (ROC) curves were drawn to evaluate predictive accuracy of each score for composite adverse outcome.

Results

After excluding external transfers, 21,935 unique adult patients accounted for 29,055 separations (mean age 45.3 years; 79.6% female) were included. There were 356 in-hospital deaths (1.6%), with rates increasing by age and differing by gender.

Patients with prolonged hospitalisation (≥10 days, 90th percentile) had higher comorbidity scores than those with shorter stays (AHRQ-EI median 3 vs 0; VW-EI 5 vs 0; age-adjusted CCI 4 vs 0; all p <0.0001). Similarly, patients with repeated admissions (≥3 per year) had greater comorbidity burden compared with those with ≤2 admissions (AHRQ-EI median 0 vs 0; VW-EI 0 vs 0; age-adjusted CCI 2 vs 0; all p <0.0001). Patients admitted to ICU (n=491; 2.2%) showed the same pattern, with substantially greater comorbidity scores than non-ICU patients (AHRQ-EI median 8 vs 0; VW-EI 5 vs 0; age-adjusted CCI 3 vs 0; all p <0.0001). Despite representing only 2.2% of admissions, ICU patients accounted for 5.2% of hospital bed-days and had longer stays (median 5 vs 1 day; p <0.0001).

Among comorbidity indices, an age-adjusted CCI >4 had the strongest predictive performance for adverse outcomes (AUC 0.83; recall 0.70), followed by VW-EI >5 (AUC 0.77) and AHRQ-EI >5 (AUC 0.75). Despite elevated risk, 49% of patients aged ≥65 years and 36.5% of those with high comorbidity burden lacked documented goals-of-care (GOC) discussions.

Conclusion

Comorbidity burden is a strong predictor of patient-important adverse outcomes and increased healthcare resource use. Routine integration into workflows could trigger earlier discussions to better align care with patient values and inform treatment limitations where appropriate.

KEY MESSAGES

What is already known on this topic:

While mortality is a commonly used clinical outcome, it does not fully reflect outcomes that matter most to patients, such as independence and quality of life. Comorbidity indices have been used to predict mortality in healthcare but their application in guiding patient-centered care planning is less established.

What this study adds:

This retrospective audit shows that comorbidity burden, especially the age-adjusted Charlson Comorbidity Index, is a strong predictor of patient-important adverse outcomes and increased healthcare resource use.

How this study might affect research, practice or policy:

Routine use of comorbidity indices may help inform anticipatory care planning and prompt early discussions on goals-of-care discussions to improve alignment of care with patient values.