TP Hofer, HSR&D Ann Arbor; SJ Bernstein, HSR&D Ann Arbor; S DeMonner, University of Michigan; RA Hayward, HSR&D Ann Arbor

Objectives: Develop a system for efficiently identifying and monitoring preventable hospital complications by using hospital-acquired laboratory abnormalities as potential indicators of poor quality care. Six clinically important hospital-acquired abnormalities were evaluated as generic screens for poor quality: (1) hypercreatinemia (renal failure); (2) hyperkalemia; (3) hypokalemia; (4) hyponatremia; and critical serum elevations of (5) digoxin and (6) aminophylline.

Methods: Laboratory and pharmacy databases at 8 VA hospitals were used to identify cases of hospital acquired laboratory abnormalities. Patient charts were reviewed directly with both explicit and implicit chart review instruments.

Analyses of laboratory and discharge data provided descriptions of the frequency of the hospital laboratory abnormalities. Structured implicit chart review by physicians was used as an exploratory tool to gain clinical insight into the types of preventable causes of these hospital-acquired laboratory abnormalities. An expert panel methodology was used to develop explicit chart review instruments for evaluating quality problems resulting in preventable severe laboratory abnormalities. A nested case-control study was used to determine whether the inappropriate processes of care identified by the expert panel and measured with the explicit review instrument were in fact associated with the development of the acquired hospital abnormalities.

Results: Two of the six potential laboratory abnormalities were eliminated for low frequency (aminophylline toxicity) and inability of our expert panel to determine appropriate processes to prevent or manage the abnormality (hyponatremia). For the remaining four abnormalities, adverse events consisting of severe hospital acquired electrolyte disorders and drug toxicity occur frequently at rates of 0.3 – 2.8% of hospitalizations. Controlling for intensity of treatment, less frequent monitoring of drug or electrolyte levels, and measures suggesting inadequate response to milder levels of electrolyte disorders were highly predictive of developing the adverse events.

In the nested case control study, failing one or more key process of care measures was associated with developing one of the four adverse events (O.R. 2.4 [1.8,3.4]). However, this relationship held principally for hypo and hyperkalemia.

Conclusions: Using these indicators can identify records that have increased rates of problems in common processes of care. The explicit review instrument allows reviewers to efficiently identify, which problems occur most frequently within a given population. This allows potential users of these instruments to design interventions to try to prevent the development of these severe laboratory abnormalities.

Impact: This study presents a general framework for developing and validating quality of care indicators in the majority of settings for which experimental evidence supporting proposed quality indicators is absent.

Adverse events consisting of severe hospital acquired electrolyte disorders and drug toxicity occur frequently at rates of 0.3 – 2.8% of hospitalizations. A case control study showed for two conditions, that developing an adverse event was associated with an increased rate of failure to perform one or more key processes of care. Using this framework, when developing quality indicators, can provide evidence for a link between processes of care and adverse events and support for the assumption that interventions to fix the process problems will reduce the incidence of adverse events events.