Challenges in validating quality of care data in a schizophrenia registry: experience from the Danish National Indicator Project

Charlotte Gjørup Pedersen,¹ Jaimie L Gradus,² Søren Paaske Johnsen,³ Jan Mainz<sup>1,4

1</sup>Department South, Aalborg Psychiatric Hospital, Aalborg, Denmark; ²VA Boston Healthcare System, National Center for PTSD and Boston University, Boston, MA, USA; ³Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark; ⁴Department of Health Services Research, University of Southern Denmark, Odense, Denmark

Background: Improvement of quality of care for psychiatric patients is a key objective of health care systems worldwide. Consequently, there is an increasing interest in documenting quality of care; however, little is known about the validity of the available data on psychiatric care.
Objective: To assess the validity of process of care data recorded in the Danish National Indicator Project (DNIP), a national population-based registry containing quality of care data of patients diagnosed with schizophrenia in Denmark.
Methods: A random sample of 1% of patients with schizophrenia registered in the DNIP between 2004 and 2009 (111 inpatient and 85 outpatient) was identified for validation. Medical records for these patients, which were used as the gold standard, were retrieved and reviewed for information on the processes of care received. Agreement between the data in the DNIP and the medical records were assessed by computing sensitivity, specificity, and positive and negative predictive values.
Results: The agreement between the recorded processes of care in the DNIP and in the medical records varied substantially across the individual process of care variables. However, a collection of the processes of care demonstrated a high agreement (80% or more) between data in the DNIP and the medical records, according to all examined aspects of data validity (sensitivity, specificity, and positive and negative predictive values). The medical records contained varying levels of missing information regarding the processes of care, from 1% for antipsychotic medication prescription to 54% for psychoeducation.
Conclusion: Current documentation practices in Danish psychiatric hospitals appear to be inconsistent and may preclude the use of psychiatric medical records as the gold standard when validating registry data.

Keywords: validation, quality of care, registry, schizophrenia, The Danish National Indicator Project