Patterns of unit and item nonresponse in the CAHPS Hospital Survey

Area of Science:

• Health Services Research
• Survey Methodology
• Health Care Quality Measurement

Background:

• The Consumer Assessment of Health Care Providers and Systems (CAHPS) Hospital Survey is crucial for evaluating healthcare quality.
• Understanding and mitigating nonresponse bias is essential for accurate survey results.

Purpose of the Study:

• To identify predictors of unit and item nonresponse in the CAHPS Hospital Survey.
• To assess the impact of nonresponse bias on survey ratings and reports.
• To determine the necessity and utility of nonresponse weights.

Main Methods:

• Multivariate models were employed to predict unit and item nonresponse using 11 administrative variables.
• Descriptive statistics were utilized to analyze the influence of nonresponse on CAHPS Hospital Survey data.
• Correlations between nonresponse weights and survey ratings were examined.

Main Results:

• Unit nonresponse was higher among younger patients and non-Hispanic white patients.
• Item nonresponse showed a positive correlation with patient age.
• Nonresponse weights demonstrated small negative correlations with most care ratings and did not significantly improve precision for sample sizes below 1,000.
• Case-mix adjustment effectively reduced nonresponse bias in certain scenarios.

Conclusions:

• Nonresponse weights are not advised for between-hospital comparisons using the CAHPS Hospital Survey.
• Nonresponse weights may offer minor benefits for overall estimates or demographic comparisons, particularly when case-mix adjustment is absent.