Original Research
Developing an Efficient Model to Select Emergency Department Patient Satisfaction Improvement Strategies

https://doi.org/10.1016/j.annemergmed.2004.11.023Get rights and content

Study objective

Patient satisfaction is an important performance measure for emergency departments (EDs), but the most efficient ways of improving satisfaction are unclear. This study uses optimization techniques to identify the best possible combination of predictors of overall patient satisfaction to help guide improvement efforts.

Methods

The results of a satisfaction survey from 20,500 patients who visited 123 EDs were used to develop ordinal logistic regression models for overall quality of care, overall medical treatment, willingness to recommend the ED to others, and willingness to return to the same ED. Originally, 68,981 surveys were mailed, and 20,916 were returned, representing an overall response rate of 30.3%. We then incorporated these regressions into an optimization model to select the most efficient combination of predictors necessary to increase the 4 overall satisfaction measures by 5%. A sensitivity analysis was also conducted to explore differences across hospital peer groups and regions.

Results

Results differ slightly for each of the 4 overall satisfaction measures. However, 4 predictors were common to all of these measures: “perceived waiting time to receive treatment,” “courtesy of the nursing staff,” “courtesy of the physicians,” and “thoroughness of the physicians.” The selected predictors were not necessarily the strongest predictors identified through regression models. The optimization model suggests that most of these predictors must be improved by 15% to increase the overall satisfaction measures by 5%.

Conclusion

This study introduces the use of optimization techniques to study ED patient satisfaction and highlights an opportunity to apply this technique to widely collected data to help inform hospitals' improvement strategies. The results suggest that hospitals should focus most of their improvement efforts on the 4 predictors mentioned above.

Introduction

Patient satisfaction has become an increasingly important measure of hospital performance. Thus, predictors of patient satisfaction in the emergency department (ED) are important. Focusing on the right predictors may be critical to retaining current patients and attracting potential new ones. Strategic decisions taken in this arena may affect the financial feasibility of health care institutions, particularly in the United States. Not surprisingly, several studies have identified different predictors of overall patient satisfaction measures using a variety of methodologies.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17

In 2004, Boudreaux and O'Hea18 evaluated 50 studies to identify the strongest predictors of ED patient satisfaction in the United States. Across the multivariate studies, the strongest predictors were patients' interpersonal interactions with physicians and nurses in the ED, perceived technical skills, and perceived waiting times. An earlier review by Trout et al19 concluded similarly that, despite considerable methodologic variability among studies, key themes such as patient information, provider-patient interactions, and perceived waiting times were most strongly associated with overall ED patient satisfaction. In a study published in 2004, Boudreaux et al2 reported that nursing care was consistently the strongest and most significant predictor (P<.05) across multiple assessments over time.

There is substantial literature using multiple regression models to predict overall ED patient satisfaction (eg, overall satisfaction, likelihood to recommend an ED, or intention to return to the same ED).1, 2, 3, 4, 6, 7, 8, 9, 12, 13, 15, 17 These studies are generally followed by the development of hospital quality-improvement initiatives that focus on the strongest predictors from these models. However, in this study, we went 1 step further. We combined the use of regression and optimization techniques to identify the best possible combination of predictors of overall patient satisfaction. It appears that optimization techniques have not been applied to patient satisfaction in ED settings.

Through mathematical modeling, optimization techniques simulate real-world problems using variables, parameters, constraints, and mathematical relationships.20 All of these key elements are defined according to the nature of the problem. Overall, the more information that can be incorporated in the formulation of an optimization model, the more realistic the solution can be. Another key element in an optimization model is the performance criterion, such as the minimization of cost, the maximization of profit, or the yield of a process. In this study, the performance criterion used is the minimization of the predictors' total relative improvement, which may ultimately lead to more effective use of hospitals' resources. The optimization model formulated in this study can be classified as a constrained nonlinear optimization problem.20, 21

The goal of this investigation is to apply optimization techniques to select predictors of overall ED patient satisfaction measures. The results should help hospitals focus their efforts when planning improvement strategies. More concretely, this analysis will identify the best or most efficient group of predictors to drive a preset level of improvement in “overall quality of care,” “overall medical treatment,” “willingness to recommend (the ED to others),” and “willingness to return (to the same ED).”

Section snippets

Setting, Data Collection, and Processing

The database used in this study was obtained from the Hospital Report Project that develops methods and reports on hospital performance in Ontario using the balanced scorecard format.22 The present study was exempted from evaluation by an institutional review board.

The database contains results from a satisfaction survey of 20,500 patients who visited 123 EDs (90 hospital corporations) in the province of Ontario, Canada, from January to March 2002. Originally, 68,981 satisfaction surveys were

Main Results

Results differ slightly for each of the 4 overall satisfaction measures. However, the following 4 predictors were common to all of these measures: “perceived waiting time to receive treatment,” “courtesy of the nursing staff,” “courtesy of the physicians,” and “thoroughness of the physicians.” This efficient selection means that most hospitals should concentrate their improvement actions on these 4 predictors to achieve a 5% increase in the overall ED patient satisfaction measures. For nearly

Limitations

Our model found predictors that, for the least amount of improvement, yielded a preset level of increase in the overall satisfaction measures. These findings are optimal within the context in which they were presented. However, from a cost perspective, they may not be the best possible because we did not include a cost structure in the formulation of the optimization model. In the absence of this information, the model assumes that the cost to increase each of the predictors is equivalent. In

Discussion

The results of this study are not surprising in light of other studies that have identified provider-patient interpersonal interactions, provider technical skills, and perceived waiting times as some of the strongest predictors of overall ED patient satisfaction.18, 19 Therefore, we believe the value of this study rests with the introduction of optimization techniques to select predictors of overall satisfaction. The optimization algorithm selects the best or most efficient combination of

References (34)

  • P.R. Yarnold et al.

    Predicting patient satisfaction: a study of two emergency departments

    J Behav Med

    (1998)
  • M.F. Hall et al.

    Keys to patient satisfaction in the emergency department: results of a multiple facility study

    Hosp Health Serv Adm

    (1996)
  • J.L. Mack et al.

    The effect of urgency on patient satisfaction and future emergency department choice

    Health Care Manage Rev

    (1995)
  • B.A. Davis et al.

    Patient satisfaction of emergency nursing care in the United States, Slovenia, and Australia

    J Nurs Care Qual

    (2003)
  • B. Hutchison et al.

    Patient satisfaction and quality of care in walk-in clinics, family practices and emergency departments: the Ontario Walk-In Clinic Study

    CMAJ

    (2003)
  • B.A. Davis et al.

    Patient satisfaction with nursing care in a rural and an urban emergency department

    Aust J Rural Health

    (1999)
  • O. Carrasquillo et al.

    Impact of language barriers on patient satisfaction in an emergency department

    J Gen Intern Med

    (1999)
  • Cited by (44)

    • Consistency of priorities for quality improvement for nursing homes in Italy and Canada: A comparison of optimization models of resident satisfaction

      2017, Health Policy
      Citation Excerpt :

      Regression coefficients of the predictors from the logistic model were incorporated into the optimization model to select the most efficient combination of predictors necessary to increase the overall WTR measure by up to 15%. The optimization technique was a constrained nonlinear optimization problem selecting the combination of items requiring the lowest total relative improvement to achieve pre-set increases in the dependent variable [9]. The optimization model combined information from the average values of the predictors and the regression estimates in order to identify the predictors (items) that were most strongly related to the dependent variable (WTR) (i.e. those predictors with large regression coefficients) and that had a relatively low current performance (average value in the population).

    • Strengthening quality of acute care through feedback from patients in Ghana

      2015, African Journal of Emergency Medicine
      Citation Excerpt :

      In emergency medicine, the importance of maintaining quality during definitive care in the hospital setting has been extensively stressed. Related benefits include increased patient satisfaction,14–16 intention to seek future care with emergency17,18 and reduced incidence of home management of critical conditions. Poor quality of care and ill preparedness on the part of providers to effectively deliver timely sensitive care lead to fatalities among critical patients.19,20

    View all citing articles on Scopus

    Supervising editor: Robert L. Wears, MD, MS

    Author contributions: ADB and GAS conceived the study and drafted the manuscript. ADB and PB-H supervised the study. GAS analyzed and managed the data and developed the models. CL provided statistical advice on study design and data analysis. PB-H provided advice on data management and editorial assistance. All authors contributed substantially to its revision. ADB takes responsibility for the paper as a whole.

    Funding and support: This work was supported in part by the Hospital Report Project, a joint initiative of the Government of Ontario and the Ontario Hospital Association.

    View full text