Data quality indicators for daily life chart methodology: prospective self-ratings of bipolar disorder and alcohol use

Background: Self-rating instruments which require a large number of repeated assessments over time are increasingly popular in psychiatry. They are well suited to describing variations in mental states, especially in order to investigate effects of behaviour and events on functioning and mood. For bipolar disorder, the self-rating instrument 'NIMH daily life chart' was developed to assess the course of the illness. This instrument has been validated in the customary ways, yet information about data quality (e.g. completeness, consistency, construct validity, reactivity) was lacking. The goal of this study was to develop several data quality indicators computed from data, in order to be able to detect respondents that provide less valid or reliable data. Methods: During approximately 1 year on average, 137 patients with DSM-IV diagnosed bipolar disorder rated their mood, functioning and number of alcohol units consumed on a daily basis. Three kinds of quality indicators were developed: (1) compliance (i.e. completeness of recording on a daily basis), (2) the association between conceptually related variables - construct validity - and (3) reactivity: any changes in alcohol-drinking behaviour due to the assessments themselves. Relations were measured with Spearman's rho. Results: A relation was found between data quality and illness severity: respondents with lower data quality, according to our operationalisations, were more strongly affected by the illness, as expressed in the number of ill days, than respondents with higher data quality. Conclusion: The more affected patients are by the illness, the lower the data quality to be expected in life chart reports.