Background: Schizophrenia Spectrum Disorder (SSD) patients have repeated symptomatic relapses during their life course. A clear definition of such episodes is required to enable building patient level models describing episode patterns. In current literature, the definitions of relapse and episodes in SSD are mostly based on hospitalisation and occasionally on symptom scales. Hospitalisation rates are affected by deinstitutionalization, hence definitions of episodes should rather depend on healthcare intensity. We present a method for grouping healthcare consumption trajectories into episodes based on individual patterns of healthcare use.

Methods: Administrative data describing the daily healthcare consumption of 13155 SSD patients in the Northern Netherlands is available for research. Data was collected between the years 2000 and 2012. Healthcare use costs are calculated using unit costs from a costing manual.

We use Exponentially Weighted Moving Average (EWMA) control charts to distinguish two states of healthcare consumption based on intensity. Daily healthcare use weighted by cost is used as healthcare intensity. State transitions are determined on a patient level. The chart is restarted after a detected structural change. The approach is validated by determining the association between the Global Assessment of Functioning (GAF) scale and the episode state.

Results: The mean number of episodes was 0.63 per patient per year. For the subsample without hospitalisations this was also 0.63. Average episode durations were similar with 235 and 245 days for the full- and subsample respectively. GAF scores have an inverse relationship with the episode state indicator.

Conclusions: The Repeated Exponentially Weighted Moving Average Control Chart (REWMACC), using a daily healthcare intensity signal, is a feasible and promising tool in quantifying patient level SSD healthcare episodes, useful in health economic models to support prevention based healthcare. The results in the subgroup without hospitalizations show that the method is robust towards changes in health care policy.