Objective: To estimate a statistical forecasting model and illustrate its use in projecting the characteristics of Medicare cancer patients who are likely to need hospitalizations. Specific aims: 1) to estimate the probability of requiring hospitalization in a given year among Medicare patients with an incident diagnosis of cancer, and 2) to identify the factors that influence the probability of requiring unnecessary or preventable hospitalizations in a given year. Design: This pilot study is a secondary analysis of a nationally representative sample of older adults with cancer. Sample: Records on cancer hospitalizations that took place over the year 1997 were obtained from the beneficiary-encrypted Medicare claims database, which contains information on 5% of all persons covered by the Medicare Program Part A (97% of the US population aged 65 years and older in 1997). Concepts Studied: The Quality Health Outcome Model that grew out of the 1991 NINR consensus conference on nursing outcomes research was used as the framework to guide the analyses. The variables included in the analyses were: principal and up to four secondary diagnoses, principal and up to two secondary procedures, length of hospitalization, patient age, gender, marital status, ethnicity, hospital region, population density, and hospital code (teaching or community). Methods: Forecasting methods were based on demographic, utilization and comorbidity data. Cancer of four leading types (lung, breast, prostate, colorectal) was identified from claim files having hospital episodes that contain ICD9-CM codes for cancers. All hospital episodes were reviewed in which cancer was coded as primary or secondary. Procedures, length of hospitalization, patient age, gender, marital status, ethnicity, hospital region, and population density were extracted from identified files. Medicare Claim data from inpatient, outpatient and physician/supplier Part B files containing ICD-9 codes and CPT-4 codes were used to identify diagnoses and procedures. Subjects were classified as having no or one or more than one hospitalization during 1997. In addition to this, specific reasons for hospitalization were divided into four general categories: cancer; acute conditions; chronic conditions; and preventable conditions. Multiple logistic regression models were used to examine the associations between demographic characteristics, disease-specific factors and utilization-specific factors and the outcome of hospitalization. Findings: The independent risk factors for hospitalization identified in preliminary analysis were age (AOR 1.8, 95% CI 1.2-2.4), male gender (AOR 1.7, 95% CI 1.1-2.1), COPD (AOR 1. 95% CI 1.4.-2.1), cognitive impairment (AOR 1.6, CI 1.3-2.1). At least one comorbidity was reported in 80% of all beneficiaries, with an average of four conditions reported. Results also showed that there was wide variation in the risk of hospitalization between the leading sites of cancer, with those with lung cancer suffering the highest risks and those with prostate cancer experiencing the lowest risks. There was no evidence of utilization-specific factors influencing risk of hospitalization. Conclusions: Within each site of cancer, preexisting comorbid conditions impacted whether individuals were at risk for hospitalization. The differentials in the risk of hospitalization also varied by several important demographic characteristics, namely age and gender. Future analysis will examine the leading risk factors in various combinations and the probability of requiring unnecessary or preventable hospitalizations. Implications: Development of these models is an important precursory step in approximating hospitalization utilization in elderly cancer patients and for the conceptualization of secondary and tertiary prevention activities for the older population in the future. The use and management of population-based data is one of the challenges future nurse researchers face. Supported by the Oncology Nursing Society Outcomes Research Grant 2001-2002.
Back to Posters
Back to The Advancing Nursing Practice Excellence: State of the Science