Diagnostic value and reliability of the present-on-admission indicator in different diagnosis groups: pilot study at a Swiss tertiary care center

• Main Authors: Karen Triep (Author), Thomas Beck (Author), Jacques Donzé (Author), Olga Endrich (Author)
• Format: Book
• Published: BMC, 2019-01-01T00:00:00Z.
• Subjects: article
• Online Access: Connect to this object online.
• Notes: 10.1186/s12913-018-3858-3; 1472-6963
• Keywords: Present on admission; Routinely collected health data; Diagnostic value; Quality indicator; Complications and comorbidities; Quality of inpatient care; Public aspects of medicine; RA1-1270
• Source: BMC Health Services Research, Vol 19, Iss 1, Pp 1-10 (2019)

Abstract Background With few exceptions the International Statistical Classification of Diseases (ICD) codes for diagnoses and official coding guidelines do not distinguish pre-existing conditions from complications or comorbidities which occur during hospitalization. However, information on diagnosis timing is relevant with regard to the case's severity, resource consumption and quality of care. In this study we analyzed the diagnostic value and reliability of the present-on-admission (POA) indicator using routinely collected health data. Methods We included all inpatient cases of the department of medicine during 2016 with a diagnosis of deep vein thrombosis, decubitus ulcer or delirium. Swiss coding guidelines of 2016 and the definitions of the Swiss medical statistics of hospitals were analyzed to evaluate the potential to encode information on diagnosis timing. The diagnoses were revised by applying the information present-on-admission by a coding specialist and by a medical expert, serving as Gold Standard. The diagnostic value and reliability were evaluated. Results The inter-rater reliability for POA of all diagnoses was 0.7133 (Cohen's kappa), but differed between diagnosis groups (0.558-0.7164). The rate of POA positive of the total applied by the coding specialist versus the expert was similar, but differed between diagnoses. In group "thrombosis" SEN was 0.95, SPE 0.75, PPV 0.97 and NPV 0.60, in group "decubitus ulcer" SEN 0.89, SPE 0.82, PPV 0.89 and NPV 0.82, in group "delirium" SEN 0.91, SPE 0.65, PPV 0.71 and NPV 0.88 For all diagnoses SEN 0.92, SPE 0.73, PPV 0.87, NPV 0.82, summing up the cases of all diagnosis groups. Conclusions Coding the POA indicator identified diagnoses which were pre-existent with insufficient reliability on individual patient's level. The overall fair to sufficient diagnostic quality is appropriate for screening and benchmarking performance on population level. As the medical statistics of hospitals carries no variable on pre-existing conditions, the novel approach to apply the POA indicator to diagnoses gives more information on quality of hospital care and complexity of cases. By preparing documentation for POA reporting diagnostic quality must be increased before implementation for risk-assessment or reimbursement on the individual patient's level.