Assessment of appropriateness of hospitalisations in Ukraine: Analytical framework, method and findings

Abstract

Objectives This article reviews the applicability of a customised version of the Appropriateness Evaluation Protocol (AEP) to evaluate the magnitude of inappropriate hospitalisations in two regions of Ukraine. Data and methods The original AEP was modified to develop a customised tool, which included criteria for the appropriateness of hospitalisation and duration of inpatient stay. The customisation of the tool followed the Delphi procedure. We randomly selected 381 medical records to test the feasibility and reliability of the method and 800 medical records to evaluate the scope of inappropriate hospitalisations. We used descriptive and analytical statistics, receiver operating characteristic curve analysis and Cohen's kappa to check the consistency between the findings of primary reviewers and experts. Result We observed high levels of agreement in conclusions of primary reviewers (reference standard) and experts during testing of the reliability and validity of the method. The external validity check showed that the use of the tool by different experts provided high accuracy: 95.1 sensitivity, 76.6 specificity and area under ROC-curve (AUC)=0.948 (<0.001) for analysis of the appropriateness of admissions; 95.3 sensitivity, 84.7 specificity and AUC=0.900 (=0.001) for the duration of hospitalisations. Cohen's kappa coefficient (κ) indicated agreement in expert evaluations of 0.915 (95% Crossed D sign Crossed D sign† 0.799 to 1.000) and 0.812 (95% Crossed D sign Crossed D sign† 0.749 to 0.875), respectively. We found that over one-third of admissions (38.1%; 95% Crossed D sign Crossed D sign† 33.9 to 43.5) and over half of total bed-days were unnecessary (57.4%; 95% Crossed D sign Crossed D sign† 56.4 to 58.5). The highest levels of stay were observed in hospitals' general medicine departments (64.6%; 95% Crossed D sign Crossed D sign† 63.0 to 66.3)compared with other departments included in the analysis. Conclusion The proposed method is robust in assessing the appropriateness of hospitalisations and duration of inpatient stays. The quantified levels of unnecessary hospital care indicate the need for improving efficiency and quality of care and optimising the excessive hospital capacities in Ukraine.