Pediatric Health Information System (PHIS): An untapped resource for public health and comparative effectiveness research

As part of 2009's American Recovery and Reinvestment, Congress appropriated $1.1 billion to launch a national comparative effectiveness research (CER) effort. The Institute of Medicine, specified necessary steps to develop and maintain a national CER infrastructure. Key among the steps is the development of CER infrastructure. This includes uncovering existing, large, clinical, prospective, real-world datasets. Given relatively low base rates of: several diseases (especially among children), some treatments, and/or adverse effects, CER frequently must utilize large clinical datasets to achieve adequate sample sizes and to include sufficient clinical and demographic heterogeneity for external validity. But, few datasets have seen description in the literature, leaving few known resources for CER. In this presentation, I address this. I discuss a relatively unknown dataset: the Pediatric Health Information System (PHIS) dataset. I discuss PHIS' nature and methodology, the types of analyses one might conduct using PHIS, and note important analytic considerations relevant to this type of data.

An administrative database, PHIS collects longitudinal, inpatient, emergency department, ambulatory surgery, and observational data from 42 not-for-profit, tertiary care pediatric US hospitals. It collects data regarding pediatric discharges, primarily children aged 0-18 years of age. The dataset allows cross hospital comparisons and includes: patient level discharge data (e.g., demographics, diagnoses, etc.), utilization information (pharmaceuticals, etc.), and physician subspecialties. PHIS offers a model of high quality data collection transferable to other national data collection efforts. PHIS, and datasets like it have the potential to drastically improve the field's ability to conduct CER and public health services research.