Simulating nursing unit performance with OrgAhead: strengths and challenges.

In this article, we briefly describe our use of a computational modeling tool, OrgAhead, details of which have been reported previously, then discuss several of the challenges computational modeling presented and our solutions. We used OrgAhead to simulate 39 nursing units in 13 Arizona hospitals and then predict changes to improve overall patient quality and safety outcomes. Creating the virtual units required (1) collecting data from managers, staff, patients, and quality and information services on each of the units; (2) mapping specific data elements (eg, control over nursing practice, nursingworkload, patient complexity, turbulence, orientation/tenure, education) to OrgAhead's parameters and variables; and then (3) validating that the newly created virtual units performed functionally like the actual units (eg, actual patient medication errors and fall rates correlated with the accuracy outcome variable in OrgAhead). Validation studies demonstrated acceptable correspondence between actual and virtual units. For all but the highest performing unit, we generated strategies that improved virtual performance and could reasonably be implemented on actual units to improve outcomes. Nurse managers, to whom we reported the results, responded positively to the unit-specific recommendations, which other methods cannot provide. In the end, resolving the modeling challenges we encountered has improved OrgAhead's functionality and usability.

Accurate assessment of clinical nurses' work environments: response rate needed.

Improvement of hospital unit work environments is key to quality patient care, productivity, nurse retention, and job satisfaction. Accurate measurement of such environments is necessary prior to introduction and evaluation of improvement structures and strategies. Characteristics and attributes of work environments are group level phenomena. Accurate assessment of these phenomena requires survey response rates of sufficient size to ensure sample representativeness and data that can reliably be aggregated to group level. What is the sufficient response rate? This question was answered through psychometric testing of five random samples from the population of 23 M.D. Anderson Cancer Center clinical units that had 100% response rates on an environmental survey. Response rates of 40% or more had acceptable psychometric properties for unit-specific scales.

Using OrgAhead, a computational modeling program, to improve patient care unit safety and quality outcomes.

As part of ongoing research to investigate the impact of patient characteristics, organization characteristics and patient unit characteristics on safety and quality outcomes, we used a computational modeling program, OrgAhead, to model patient care units' achievement of patient safety (medication errors and falls) and quality outcomes. We tuned OrgAhead using data we collected from 32 units in 12 hospitals in Arizona. Validation studies demonstrated acceptable levels of correspondence between actual and virtual patient units. In this paper, we report how we used OrgAhead to develop testable hypotheses about the kinds of innovations that nurse managers might realistically implement on their patient care units to improve quality and safety outcomes. Our focus was on unit-level innovations that are likely to be easier for managers to implement. For all but the highest performing unit (for which we encountered a ceiling effect), we were able to generate practical strategies that improved performance of the virtual units that could be implemented by actual units to improve safety and quality outcomes. Nurse managers have responded enthusiastically to the additional decision support for quality improvement.