Transforming a Patient Registry Into a Customized Data Set for the Advanced Statistical Analysis of Health Risk Factors and for Medication-Related Hospitalization Research: Retrospective Hospital Patient Registry Study

Background Hospital patient registries provide substantial longitudinal data sets describing the clinical and medical health statuses of inpatients their pharmacological prescriptions. Despite multiple advantages routinely collecting multidimensional data, those are rarely suitable for advanced statistical analysis they require customization synthesis. Objective The aim this study was to describe methods used transform synthesize a raw, multidimensional, hospital registry set into an exploitable database further investigation risk profiles predictive survival outcomes among polymorbid, polymedicated, older in relation medicine prescriptions at discharge. Methods A from public extracted CSV (.csv) file imported R package cleaning, customization, Patients fulfilling criteria inclusion were home-dwelling, adults with chronic conditions aged ≥65 who became hospitalized. covered 140 variables 20,422 hospitalizations home-dwelling 2015 2018. Each variable, according type, explored computed distributions, missing values, associations. Different clustering methods, expert opinion, recoding, missing-value techniques customize these sets. Results Sociodemographic showed no values. Average age, length stay, frequency hospitalization computed. Discharge details recoded summarized. Clinical cleaned up best practices managing values applied. Seven clusters diagnoses, surgical interventions, somatic, cognitive, medicines using empirical practices, each presenting status patients included it as accurately possible. Medical, comorbidity, drug Conclusions cleaner, better-structured obtained, combining best-practice approaches. overall strategy delivered exploitable, population-based descriptive, predictive, statistics relating admitted inpatients. More research is needed develop customizing synthesizing large, registries. International Registered Report Identifier (IRRID) RR2-10.1136/bmjopen-2019-030030