Patient satisfaction: how patient health conditions influence their satisfaction.
Article Type: Report
Subject: Patient satisfaction (Analysis)
Hospitals (Management)
Hospitals (Missouri)
Medical research
Medicine, Experimental
Authors: Otani, Koichiro
Waterman, Brian
Dunagan, W. Claiborne
Pub Date: 07/01/2012
Publication: Name: Journal of Healthcare Management Publisher: American College of Healthcare Executives Audience: Trade Format: Magazine/Journal Subject: Business; Health care industry Copyright: COPYRIGHT 2012 American College of Healthcare Executives ISSN: 1096-9012
Issue: Date: July-August, 2012 Source Volume: 57 Source Issue: 4
Topic: Event Code: 200 Management dynamics Computer Subject: Company business management
Product: Product Code: 8000200 Medical Research; 9105220 Health Research Programs; 8000240 Epilepsy & Muscle Disease R&D; 8060000 Hospitals NAICS Code: 54171 Research and Development in the Physical, Engineering, and Life Sciences; 92312 Administration of Public Health Programs; 622 Hospitals SIC Code: 8062 General medical & surgical hospitals; 8063 Psychiatric hospitals; 8069 Specialty hospitals exc. psychiatric
Geographic: Geographic Scope: Missouri Geographic Code: 1U4MO Missouri
Accession Number: 299343690
Full Text: EXECUTIVE SUMMARY

With increasing emphasis in healthcare on patient satisfaction, many patient satisfaction studies have been administered. Most assume that all patients combine their healthcare experiences (such as nursing care, physician care, etc.) in the same way to arrive at their satisfaction; however, no research has been conducted prior to the present study to investigate how patients' health conditions influence the way they combine their healthcare experiences. This study aims to determine how seriously ill patients differ from less seriously ill patients during their combining process.

Data were collected from five large hospitals in the St. Louis area by administering a patient satisfaction questionnaire. Multiple linear regression analyses with a scatter term, a severity measure, and interaction effects of the severity measure were conducted while controlling for age, gender, and race.

Two models (overall quality of care and willingness to recommend to others) were analyzed, and the severity of illness variable revealed interaction effects with physician care, staff care, food, and scatter term variables in the willingness to recommend model (six attributes were analyzed: admission process, nursing care, physician care, staff care, food, and room). With more seriously ill patients, physician care becomes more important and staff care becomes less important, and seriously ill patients are proportionately more likely to combine their attribute reactions only in the willingness to recommend model.

All six attributes are not equally influential. Nursing care and staff care show consistent influence in both models. These findings show that if healthcare managers want to increase their patient satisfaction, they should enhance nursing care and staff care first to experience the most improvement.

INTRODUCTION

Healthcare environments have been changing rapidly and drastically in the United States. Not only must healthcare managers contend with current changes, they must also anticipate future changes. One shift that has occurred in healthcare is an emphasis on patient satisfaction. Traditionally, the quality of healthcare has been measured by healthcare professionals using objective measures. But patient satisfaction is not an objective measure and may be influenced by the so-called bedside manner of caregivers. Nevertheless, patient satisfaction is gaining attention for several reasons. First, patient satisfaction is considered customer satisfaction. In any field, customer satisfaction is a key determinant of an organization's survival; healthcare is no exception. Some observers recommended that hospitals and healthcare systems go so far as to regard patients as guests (Fottler, Ford, and Heaton 2011) who are seeking not only positive clinical outcomes but also quality service experiences. Satisfied customers or patients bring business. With increasing empowerment of patients who can decide their choices of providers, patients are no longer passive (Fottler et al. 2006; Scotti, Harmon, and Behson 2007; Ford et al. 2006).

Second, managed care organizations use the patient satisfaction data to choose, negotiate with, and decide the level of bonus to pay to providers (Zimmerman, Zimmerman, and Lund 1997). In response, many studies have investigated possible measures to improve patient satisfaction at the organizational level (Powers and Jack 2008; Pink, Murray, and McKillop 2003; Marley, Collier, and Goldstein 2004; Hotchkiss, Fottler, and Unruh 2009; Fottler et al. 2000; Ford et al. 2006). Healthcare managers need to pay particular attention to their patient satisfaction data and develop the organization's competitive advantage with the positive feedback they receive.

Third, satisfied patients are less likely to engage in doctor shopping, and doctor shopping increases duplication of testing. Finally, satisfied patients tend to comply with the treatments prescribed (Ford, Bach, and Fottler 1997; Parente, Pinto, and Barber 2005; Zandbelt et al. 2007). This factor is particularly important regarding patients with chronic diseases, the population of which will very likely increase with the increasing number of elderly.

Patient satisfaction research has shifted its focus from developing patient satisfaction measures to identifying which patients are more or less satisfied and to finding which healthcare attributes (such as nursing care, physician care, food, etc.) are more or less influential to improving patient satisfaction levels. These studies attempt to determine how patients combine their healthcare attribute reactions (healthcare experiences) to arrive at their overall satisfaction or their intention to recommend or return (Dansky and Brannon 1996; Oswald et al. 1998; Ross, Steward, and Sinacore 1993; Otani et al. 2009). They provide useful findings to help healthcare managers improve their patient satisfaction. However, most of these studies assume that all patients combine their healthcare attribute reactions in the same way to arrive at their level of satisfaction. A few studies argue that not all patients combine their healthcare attribute reactions in the same way; they found gender and racial differences (Otani and Harris 2004; Otani, Herrmann, and Kurz 2010b). However, research investigating how patients' health conditions influence the combining process is lacking. For example, no studies have asked the question, "How do patients with more or less serious health conditions differ in their combining process of healthcare attributes?" This study aims to determine how seriously ill patients differ from less seriously ill patients in this process. It uses a new severity measure to classify patients' illness levels, from which it is possible to identify more seriously ill and less seriously ill patients.

THEORY AND CONCEPTUAL FRAMEWORK

After patients are discharged from a hospital, they often receive a patient satisfaction questionnaire. In it they are asked to recall many types of experiences during their stay. Patients combine these experiences to arrive at their overall evaluation of the care they received. Because some hospital experiences are inevitably good and others not so good, and because some experiences are more influential than others, how do patients combine these experiences? One of the most well-known customer satisfaction models relevant to patient satisfaction studies is the Fishbein model (Fishbein and Ajzen 1975). According to the model, patients combine salient attribute reactions (experiences) with different weights and arrive at their overall evaluation. Thus, attribute reactions with larger weights influence the overall evaluation more than attribute reactions with smaller weights. In addition, positive attribute reactions (excellent experience in some areas) can compensate for weak attribute reactions (fair experience). In this regard, the Fishbein model is referred to as a compensatory model.

Later, patient satisfaction researchers began to use a noncompensatory model. The noncompensatory model was originally developed in psychology and later operationalized as a mathematical model for applied research (Ganzach 1993; Kahneman and Tversky 1979). The noncompensatory model does not always admit trade-offs among attribute reactions. Thus, positive attribute reactions do not always compensate for weak attribute reactions to arrive at the overall evaluation. Several patient satisfaction studies that use the noncompensatory model found the model valuable in that it provides more information to healthcare managers (Otani 2006; Otani and Harris 2004; Otani, Hen-mann, and Kurz 2010a; Otani et al. 2010).

The mathematical noncompensatory model in this study employs a scatter term. The scatter term, essentially a standard deviation of all attribute reactions, was developed by Brannick and Brannick in 1989 and indicates which of three options (compensatory, risk averse, or risk taking) patients employ to arrive at their overall evaluation. The compensatory option assumes that patients average out their attribute reactions (positives and/or negatives), and thus, positive attribute reactions can compensate for negative attribute reactions proportionately (Fishbein and Ajzen 1975). The noncompensatory options (risk averse and risk taking) assume that patients may be disproportionately influenced by either highly positive or highly negative reactions, and thus, little trade-off occurs among attribute reactions (Ganzach 1993; Kahneman and Tversky 1979).

To show how patients use these options, a simplified hypothetical situation is useful. Suppose that there are two hospitals, and for simplicity, there are three salient healthcare attributes: nursing care, physician care, and environment. Each attribute is equally influential to patients, meaning that their weights or coefficients in a model are the same. Now assume that, based on patient questionnaire results, one hospital is awarded 5 (excellent) on nursing care, 4 (very good) on physician care, and 3 (good) on environment. The other hospital receives 4 (very good) on three of the attributes. In this hypothetical case, the average is 4 for both hospitals because we assume that the three attributes are equally influential.

Which of these two hospitals would patients choose? Those patients who choose the first hospital are assumed to use the risk-taking option. Even though the influences of the three attributes are the same, patients are disproportionately influenced by the highly positive reaction (5 on nursing care) to the first hospital. Patients who choose the second hospital are assumed to use the risk-averse option. In this case, patients are disproportionately influenced by the mediocre reaction (3 on environment) to the second hospital. Patients who do not show a specific preference are assumed to use the compensatory option. Thus, when little difference is seen between competing hospitals regarding the average scores, it is particularly critical for the hospitals' healthcare managers to know which of the three options (combining processes) patients use.

Another variable used in this study is a severity of illness measure of patients derived from diagnosis-related groups (DRGs). The DRG patient classification system was introduced to facilitate development of the prospective payment system for hospital care. The purpose of the DRGs, used for reimbursement purposes by the Centers for Medicare & Medicaid Services (CMS), was to categorize patients into different resource-use intensity levels. A need later arose to develop a measure that is applicable not only to resource use and payment but also to quality improvement and internal management of hospitals. Many severity of illness measures were developed and tested by numerous researchers (Iezzoni et al. 1996a; Iezzoni et al. 1996b; Lindenauer et al. 2002; Meurer et al. 1998). These studies examined the usefulness of the measures for case-mix purposes to compare quality (mortality), length of stay, or cost levels among hospitals. One widely used severity measure is the all patient refined diagnosis-related group (APRDRG) system, which was developed by 3M Health Information Systems (2003).

Many studies use the APR-DRGs in areas such as comparing hospitals regarding resource and outcome measures, implementing critical pathways, and demonstrating case-mix complexity (e.g., Davis et al. 2002). Among these studies, some used the APR-DRGs for case-mix purposes to compare the length of stay of Veterans Administration patients, community-acquired pneumonia patients, and pediatric asthma patients (Schein et al. 2008; Sedman et al. 2004; Shen 2003). Others used the APR-DRGs for case-mix purposes to compare the cost of acute inpatient palliative care, total joint arthroplasty, pediatric asthma care, and community-acquired pneumonia care (Davis et al. 2005; Lavernia et al. 2009). Another study used the APR-DRGs to identify levels of high-risk patient falls to initiate an intervention program to prevent falls (McAlister 2009). Many studies that used or investigated the APR-DRGs concluded that they are a useful measure (Kuo et al. 2004; Lagman et al. 2007; Lavernia et al. 2009; McAlister 2009; Romano and Chan 2000; Sedman et al. 2004; Shen 2003). The effects of the severity of illness measure on patient satisfaction were assessed by using the APR-DRGs in this study. The APR-DRGs were grouped and assigned based on discharge diagnoses and provided four categories--minor, moderate, major, and extreme--with severity of illness.

METHODS

Data Source

BJC Healthcare in St. Louis, Missouri, provided the data for this study. It is a regional, 12-hospital, integrated healthcare delivery and financing system serving the St. Louis metropolitan area, mid-Missouri, and southern Illinois. This study gathered data from five of the system's large hospitals. It excluded the Children's Hospital, rural community hospitals, and small community hospitals, either because they were notably different from the five hospitals chosen in size and location or they did not maintain the patient demographic data necessary for consistent analysis. In addition, the Children's Hospital, a pediatric hospital whose patients are typically younger than 20, was excluded because the focus of this study was adult patient satisfaction.

The five hospitals included in this study consist of one academic hospital and four large community hospitals. We used a telephone-based survey of discharged patients, which was conducted by a national telephone survey company specializing in patient satisfaction measurement. We used a probability sampling method with stratification by department and hospital. Patients were initially contacted 7 to 14 days post-discharge. Those who did not respond were contacted until they completed the survey, indicated refusal to complete the survey, or were unable to be reached over the course of two weeks. The study analyzed discharged patients 20 years old or older who were discharged between January 2005 and June 2010. There are 32,053 cases in the data for analysis.

Measures

The two dependent variables for this study include the following one-item questions:

1. Overall, would you rate the quality of care and services received during this hospital stay as excellent, very good, good, fair, or poor?

2. Would you rate your willingness to recommend this hospital to family and friends as excellent (5), very good (4), good (3), fair (2), or poor (1)?

The six independent variables related to attribute reactions were admission process, nursing care, physician care, staff care, food, and room.

Each of the six construct variables includes multiple questionnaire items in the survey that measure the same construct variable with the 5-point, Likert-type scale. The reliability and validity of the instrument used in this study has been evaluated and found to be strong in numerous studies using a combination of principal component analysis, confirmatory factor analysis, and structural equation analysis (Burroughs et al. 1999; Burroughs et al. 2001).

A composite index was created for each of the six independent variables as the arithmetic mean of all items measuring the attribute. The Cronbach's coefficient alpha was estimated to test the internal consistency of the items for each attribute (Exhibit 1). All computed Cronbach's alpha values for this data set were larger than .80, indicating good internal consistency, except for the food attribute, which includes only two items. To maintain stability of the attribute, it was decided to keep the two items in the food attribute. Regarding the APR-DRGs, a new variable with two levels of severity of illness was created by combining the first two categories, minor (1) and moderate (2) and the latter two categories major (3) and extreme (4). The control variables considered for analysis included age, gender, and race.

Analysis

This study analyzed how patients combined the six attribute reactions to arrive at their evaluation of overall quality of care and willingness to recommend. Multiple linear regression analyses with a scatter term and interaction effects of the severity of illness measure were conducted while controlling for age, gender, and race.

The general multiple linear regression model used in this study is shown

together with a scatter term and interaction effects of the severity measure:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where Y is the dependent variable, a is the intercept, b is a coefficient, X is an attribute reaction, [bar.X] is the average of the [X.sub.i] s, S is a severity level, and e is the error term. The first two terms of the right-hand side of the equation are always used in a linear regression model, and the third one is the scatter term. The fourth one is a severity measure, and the fifth one is an interaction effect. The coefficient of the scatter term, [b.sub.k+1], indicates which option patients employ:

If [b.sub.k+1] is zero or not statistically significant, patients are assumed to have used the compensatory model.

If [b.sub.k+1] is statistically significant, patients are assumed to have used the noncompensatory model.

And then:

If [b.sub.k+1] is negative and statistically significant, patients are assumed to have used the risk-averse option.

If [b.sub.k+1] is positive and statistically significant, patients are assumed to have used the risk-taking option.

RESULTS

Of the 32,053 cases for analysis, female patients accounted for 18,556 (57.9 percent) and male patients accounted for 13,496 (42.1 percent). The average age was 56.7 years old, and the standard deviation was 18.8 years. Regarding race or ethnicity, white patients accounted for 10,702 (77.6 percent), African American patients accounted for 5,416 (20.3 percent), Hispanic patients accounted for 74 (0.3 percent), Asian patients accounted for 101 (0.4 percent), and others accounted for 388 (1.4 percent). For descriptive statistics and summary descriptions of the survey items and composite indexes, see Exhibit 1. The mean score for each item and attribute (composite index) indicates generally positive reactions ranging from excellent (5) to very good (4), except for reactions to the food. Note that the number of cases for any composite indexes is always larger than the numbers of any items in the same attribute, because the composite index score was computed as long as at least one item in the attribute was responded to.

Results of multiple linear regression analyses are shown in Exhibits 2 and 3. These analyses included six attributes, the severity indicator, the scatter term, the interaction effect variables, and the control variables. Exhibit 2 shows the overall quality of care model, which accounts for 57.4 percent of the variance. Among the six attributes, five are statistically significant and are positively related to the overall quality of care at the [alpha] = 0.05 level. (Throughout this study, [alpha] -0.05 is used to test the significance level.) Nursing care is most influential, followed by staff care, admission process, physician care, and room. Food is not statistically significant. The severity indicator is negatively related but is not statistically significant. The scatter term is negative and is statistically significant. It indicates the risk-averse option. None of the interaction effect terms are statistically significant. With control variables, age and sex are not statistically significant. Among race variables, only African American is statistically significant and is negatively related, indicating that African American patients are less satisfied with the overall quality of care than white patients (reference group) are.

Exhibit 3 shows the willingness to recommend model, which accounts for 57.8 percent of the variance. Among the six attributes, five are statistically significant and are positively related to the willingness to recommend. Staff care is most influential, followed by nursing care, physician care, room, and admission process. The order of the five attributes with the willingness to recommend model is different from that with the overall quality of care model. Food is not statistically significant. Note, however, that food is statistically significant and is positively related to both overall quality of care and willingness to recommend when food alone was analyzed separately (not shown here). The severity indicator is negatively related but is not statistically significant. The scatter term is negative and is statistically significant, again indicating the risk-averse option. Among interaction effect terms, physician care, staff care, food, and the scatter term show statistically significant interaction effects. Physician care, food, and the scatter term are positively related and staff care is negatively related. These results indicate that when patients' conditions are serious, physician care, food, and the scatter term increase their influence positively. On the other hand, staff care's influence decreases when patients' conditions are serious. With control variables, age and sex are not statistically significant. Among race variables, African American and others are statistically significant and are negatively related, indicating that African Americans and others are less willing to recommend than white patients (reference group) are.

DISCUSSION

The purpose of this study was to investigate how the severity of illness level influences patients' attribute reaction combining process to arrive at their conclusions of overall quality of care and willingness to recommend to others. The severity of illness variable itself was not statistically significant in either model; however, it revealed the interaction effects with physician care, staff care, food, and scatter term variables in the willingness to recommend to others model but not in the overall quality of care model. Exhibit 4 shows the comparisons of only significant independent variables for the two models. The directions of the interaction effects are different among attributes. For physician care and food, the interaction effects are positive, indicating the influence levels of physician care and food attributes increase among seriously ill patients (physician care and food become more important for seriously ill patients than for less seriously ill patients). On the other hand, for staff care, the interaction effect is negative, indicating the influence level of staff care decreases among seriously ill patients. For the scatter term, the interaction effect is positive, indicating seriously ill patients become slightly less risk averse and move toward compensatory in integrating their attribute reactions (seriously ill patients are less likely to be disproportionately influenced by their negative reactions than less seriously ill patients are).

The severity of illness measure alone was also tested against overall quality of care and willingness to recommend, and it was statistically significant and negatively related for both models (not shown here). This negative relation indicates that seriously ill patients are less likely to rate overall quality of care better or are less likely to recommend to others. These results are the same for the original four-category severity measure and the severity of illness measure created for this study used in the analyses. Thus, the severity of illness variable itself lost its significance after other attribute reaction variables were included in the models. This weak predictive power of the APR-DRG system's severity of illness rating is consistent with a study by Woodbury, Tracy, and McKnight (1998), who surveyed discharged patients from a tertiary care hospital.

The scatter term is statistically significant and is negatively related to both overall quality of care and willingness to recommend to others. This finding is consistent with other patient satisfaction studies (Otani and Harris 2004; Otani, Harris, and Tierney 2003; Otani, Herrmann, and Kurz, 2010a, 2010b; Otani et al. 2003). Patients are risk averse when they combine their attribute reactions to arrive at their evaluation of overall quality of care and willingness to recommend to others. In other words, they combine their attribute reactions by giving disproportionate weight to a negative attribute. The interaction effect with the severity of illness measure was found only in the willingness to recommend to others model, as previously mentioned.

LIMITATIONS AND SUGGESTIONS

This study adds new findings to the existing knowledge by including the severity of illness measure in the combining process of patient satisfaction. This study also confirms that certain attributes of hospital care are more or less influential on the evaluation of overall quality of care and willingness to recommend to others. However, it has some limitations. First, this is a cross-sectional study. With this study design, it is possible to establish an association, but it is not appropriate to claim a cause-and-effect relationship. To address this point, we referred to well-established theories and literature that support the assumption that patients combine healthcare attribute reactions to arrive at their evaluation of overall quality of care and willingness to recommend to others. Second, although in this study we analyzed a large data set, the data were collected from large hospitals in one geographic area. Patients in other areas or from smaller hospitals may react differently. Third, although the response rate fluctuates slightly each time the survey is conducted, the overall response rate to the survey on which our study is based was 36.7 percent. This percentage is acceptable compared to the rates in other patient satisfaction studies, but it is possible that nonrespondents may combine their attribute reactions differently from respondents. One study, although not in the healthcare field, examined respondents, active nonrespondents, and passive nonrespondents. The active nonrespondents were purposeful nonrespondents, and passive nonrespondents simply forgot to respond. The study's authors found that active nonrespondents were less satisfied than respondents were, and passive nonrespondents were attitudinally similar to respondents. They conclude that nonresponse bias does not appear to be a substantive concern for satisfaction-type variables (Rogelberg et al. 2003). In our study, the respondents were slightly older than nonrespondents. Male patients responded in greater percentages than female patients did. (Male patients tended to respond rather than decline compared to female patients when they were asked to participate in the survey.) White patients responded in greater percentages than African American patients did. However, these differences are minimal, and thus, respondents and nonrespondents are similar regarding age, sex, and race.

CONCLUSION AND PRACTICE IMPLICATIONS

The results of this study revealed that the severity of illness measure is a significant factor for patients as they combine their attribute reactions to arrive at their willingness to recommend to others. The severity of illness level interacted with physician care, staff care, food, and the scatter term. With more seriously ill patients, physician care became more important and staff care became less important, which reflects a rational patient logic--it is reasonable that a patient would think this way. The scatter term interacted with the severity of illness measure only with the willingness to recommend to others model. It indicated that seriously ill patients were proportionately more likely to combine their attribute reactions. The seriously ill patients employed more of the compensatory model while less seriously ill patients employed more of the risk-averse model. However, when they evaluated their overall quality of care, there was no interaction effect with the scatter term. Therefore, when patients evaluate overall quality of care, they might be consistently subjective regardless of the severity level of illness. Thus, they might be consistently and disproportionately influenced by a weak attribute regardless of the health conditions from which they suffer. However, when they recommend, they might be less subjective. It is probably because the willingness to recommend may be a more objective mental activity.

The results of this study also revealed practical implications for healthcare managers. First, all six attributes (admission process, nursing care, physician care, staff care, food, and room) were not equally influential. As Exhibit 5 shows, some attributes were clearly more influential than others. Nursing care and staff care showed their consistent influence in both models. Admission process, food, and room showed much less influence than nursing care or staff care. These findings indicate that if healthcare managers wish to improve their patient satisfaction scores--in either overall quality of care or willingness to recommend--they should improve nursing care and staff care first to take the most effective path to improvement. Interaction effects of the severity of illness measure were seen with physician care and staff care in the willingness to recommend model, but even after considering these effects, staff care (0.448-0.050 = 0.398) and nursing care (0.206) are more influential than physician care (0.141 + 0.039 = 0.18). Considering nurses and staff members are under the control of healthcare managers much more than physicians who have hospital privileges and are typically not employees of the hospital, these findings may be good news for healthcare managers who make every effort to improve patient satisfaction. Patients also showed that they use the risk-averse model. In this study, they were disproportionately influenced by poorly performing attributes, and thus, strongly performing attributes might not compensate for poorly performing attributes. Therefore, it is important that healthcare managers discover which attributes are poor performers by administering satisfaction surveys and improve those attributes.

[GRAPHIC OMITTED]

Recently, CMS announced the Hospital Value-Based Purchasing program. This program rewards hospitals for the quality of care provided to Medicare patients. The quality measures include several areas, among them patients' experience of care (Ferman 2011; McHugh and Joshi 2010). Thus, hospitals that perform better on patient satisfaction will receive more payment from CMS than hospitals that do not, directly affecting hospitals' bottom line. Hospital managers now have one more reason to pay more attention than ever to their patients' satisfaction.

ACKNOWLEDGMENTS

The authors thank Emily Ostmann for assistance with data collection, Suzanne Rumsey for assistance with an earlier version of this manuscript preparation, and Dawn Adams for assistance with editing.

This study has been approved by the Institutional Review Board at Purdue University, Ref. #0710005884.

No authors have any conflicts of interest, and this study was not funded by any organization.

REFERENCES

Brannick, M., and I. P. Brannick. 1989. "Nonlinear and Noncompensatory Processes in Performance Evaluation." Organizational Behavior and Human Decision Processes 44: 97-122.

Burroughs, T. E., A. R. Davies, J. Cira, and W. C. Dunagan. 1999. "Understanding Patient Willingness to Recommend and Return: A Strategy for Prioritizing Improvement Opportunities." Joint Commission Journal on Quality Improvement 25 (6): 271-87.

Burroughs, T. E., B. M. Waterman, J. C. Cira, R. Desikan, and W. C. Dunagan. 2001. "Patient Satisfaction Measurement Strategies: A Comparison of Phone and Mail Methods." Joint Commission Journal on Quality Improvement 27 (7): 349-61.

Dansky, K. H., and D. Brannon. 1996. "Discriminant Analysis: A Technique for Adding Value to Patient Satisfaction Surveys." Hospital & Health Services Administration 41: 503-13.

Davis, M. P., D. Walsh, S. B. LeGrand, R. L. Lagman, B. Harrison, and L. Rybicki. 2005. "The Financial Benefits of Acute Inpatient Palliative Medicine: An Inter-institutional Comparative Analysis by All Patient Refined-Diagnosis Related Group and Case Mix Index." Journal of Supportive Oncology 3 (4): 313-16.

Davis, M. P., D. Walsh, K. A. Nelson, D. Konrad, S. B. LeGrand, and L. Rybicki. 2002. "The Business of Palliative Medicine--Part 2: The Economics of Acute Inpatient Palliative Medicine." American Journal of Hospice & Palliative Medicine 19 (2): 89-95.

Ferman, I. H. 2011. "Value-Based Purchasing Program Here to Stay: Payments Will Be Based on Performance." Healthcare Executive 26 (3): 76-78.

Fishbein, M., and I. Ajzen. 1975. Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Philippines: Addison-Wesley.

Ford, R. C., S. A. Bach, and M. D. Fottler. 1997. "Methods of Measuring Patient Satisfaction in Health Care Organizations." Health Care Management Review 22 (2): 74-89.

Ford, R. C., S. A. Sivo, M. D. Fottler, D. Dickson, K. Bradley, and L. Johnson. 2006. "Aligning Internal Organizational Factors with a Service Excellence Mission: An Exploratory Investigation in Health Care." Health Care Management Review 31 (4): 259-69.

Fottler, M. D., D. Dickson, R. C. Ford, K. Bradley, and L. Johnson. 2006. "Comparing Hospital Staff and Patient Perceptions of Customer Service: A Pilot Study Utilizing Survey and Focus Group Data." Health Services Management Research 19 (1): 52-66.

Fottler, M. D., R. C. Ford, and C. P. Heaton. 2011. Achieving Service Excellence: Strategies for Healthcare, 2nd edition. Chicago: Health Administration Press.

Fottler, M. D., R. C. Ford, V. Roberts, and E. W. Ford. 2000. "Creating a Healing Environment: The Importance of the Service Setting in the New Consumer-Oriented Healthcare System." Journal of Healthcare Management 45 (2): 91-107.

Ganzach, Y. 1993. "Goals as Determinants of Nonlinear Noncompensatory Judgment Strategies: Leniency vs. Strictness." Organizational Behavior and Human Decision Processes 65: 422-40.

Hotchkiss, R. B., M. D. Fottler, and L. Unruh. 2009. "Valuing Volunteers: The Impact of Volunteerism on Hospital Performance." Health Care Management Review 34 (2): 119-28.

Iezzoni, L. I., A. S. Ash, M. Shwartz, J. Daley, J. S. Hughes, and Y. D. Mackieman. 1996a. "Judging Hospital by Severity-Adjusted Mortality Rates: The Influence of the Severity-Adjustment Method." American Journal of Public Health 86 (10): 1379-87.

Iezzoni, L. I., M. Shwartz, A. S. Ash, J. S. Hughes, J. Daley, and Y. D. Mackiernan. 1996b. "Severity Measurement Methods and Judging Hospital Death Rates for Pneumonia." Medical Care 34 (1): 11-28.

Kahneman, D., and A. Tversky. 1979. "Prospect Theory: An Analysis of Decisions Under Risk." Econometrica 47: 263-91.

Kuo, P. C., A. R. Douglas, D. Oleski, D. O. Jacobs, and R. A. Schroeder. 2004. "Determining Benchmarks for Evaluation and Management Coding in an Academic Division of General Surgery." Journal of the American College of Surgeons 199 (1): 124-30.

Lagman, R. L., D. Walsh, M. P. Davis, and B. Young. 2007. "All Patient Refined-Diagnostic Related Group and Case Mix Index in Acute Care Palliative Medicine." Journal of Supportive Oncology 5 (3): 145-49.

Lavernia, C. J., A. Laoruengthana, I. S. Contreras, and M. D. Rossi. 2009. "All Patient Refined Diagnosis-Related Groups (APRDRGs) in Primary Arthroplasty." Journal of Arthroplasty 24 (6, Suppl.): 19-23.

Lindenauer, P. K., R. Chehabeddine, P. Pekow, I. Fitzgerald, and E. M. Benjamin. 2002. "Quality of Care for Patients Hospitalized with Heart Failure." Archives of Internal Medicine 162: 1251-56.

Marley, K. A., D. A. Collier, and S. M. Goldstein. 2004. "The Role of Clinical Process Quality in Achieving Patient Satisfaction in Hospitals." Decision Sciences 35 (3): 349-69.

McAlister, S. 2009. "APR DRG Weights and the Relationship to Patient Falls." Nursing Economics 27 (2): 119-23.

McHugh, M., and M. Joshi. 2010. "Improving Evaluations of Value-Based Purchasing Programs." Health Services Research 45 (5, Pt. 2): 1559-69.

Meurer, J. R., E. M. Kuhn, V. George, L S. Yauck, and P. M. Layde. 1998. "Charges for Childhood Asthma by Hospital Characteristics." Pediatrics 102 (6): E70.

Oswald, S. L., D. E. Turner, R. L. Snipes, and D. Butler. 1998. "Quality Determinants and Hospital Satisfaction. Perceptions of the Facility and Staff Might Be Key Influencing Factors." Marketing Health Services 1891: 18-22.

Otani, K. 2006. "Enrollees' Global Rating Process of Health Care with the National CAHPS[R] Benchmarking Database." Health Care Management Review 31 (3): 205-12.

Otani, K., and L. E. Harris. 2004. "Different Integration Processes of Patient Satisfaction Among Four Groups." Health Care Management Review 29 (3): 188-95.

Otani, K., L. E. Harris, and W. M. Tierney. 2003. "A Paradigm Shift in Patient Satisfaction Assessment." Medical Care Research and Review 60 (3): 347-65.

Otani, K., P. A. Herrmann, and R. S. Kurz. 2010a. "Improving Patient Evaluation of Hospital Care and Increasing Their Intention to Recommend: Are They the Same or Different Constructs?" Health Services Management Research 23 (2): 60-65.

Otani, K., P. A. Herrmann, and R. S. Kurz. 2010b. "Patient Satisfaction Integration Process: Are There Any Racial Differences?" Health Care Management Review 35 (2): 116-23.

Otani, K., R. S. Kurz, T. E. Burroughs, and B. Waterman. 2003. "Reconsidering Models of Patient Satisfaction and Behavioral Intentions." Health Care Management Review 28 (1): 7-20.

Otani, K., B. Waterman, K. M. Faulkner, S. Boslaugh, T. E. Burroughs, and W. C. Dunagan. 2009. "Patient Satisfaction: Focusing on 'Excellent.'" Journal of Healthcare Management 54 (2): 93-103.

Otani, K., B. Waterman, K. M. Faulkner, S. Boslaugh, and W. C. Dunagan. 2010. "How Patient Reactions to Hospital Care Attributes Affect the Evaluation of Overall Quality of Care, Willingness to Recommend, and Willingness to Return." Journal of Healthcare Management 55 (1): 25-38.

Parente, D. H., M. B. Pinto, and I. C. Barber. 2005. "A Pre-post Comparison of Service Operational Efficiency and Patient Satisfaction Under Open Access Scheduling." Health Care Management Review 30 (3): 220-28.

Pink G. H., M. A. Murray, and I. McKillop. 2003. "Hospital Efficiency and Patient Satisfaction." Health Services Management Research 16 (1): 24-38.

Powers, T. L., and E. P. lack. 2008. "Using Volume Flexible Strategies to Improve Customer Satisfaction and Performance in Health Care Services." Journal of Services Marketing 22 (3): 188-97.

Rogelberg S. G., I. M. Conway, M. E. Sederburg, C. Spitzmiiller, S. Aziz, and W. E. Knight. 2003. "Profiling Active and Passive Nonrespondents to an Organizational Survey." Journal of Applied Psychology 88 (6): 1104-14.

Romano, P. S., and B. K. Chan. 2000. "Risk-Adjusting Acute Myocardial Infarction Mortality: Are APR-DRGs the Right Tool?" Health Services Research 34 (7): 1469-89.

Ross, C. K., C. A. Steward, and J. M. Sinacore. 1993. "The Importance of Patient Preferences in the Measurement of Health Care Satisfaction." Medical Care 31 (12): 1138-49.

Schein, J., C. Janagap-Benson, R. Grant, V. Sikirica, D. Doshi, and W. Olson. 2008. "A Comparison of Levofloxacin and Moxifloxacin Use in Hospitalized Community-Acquired Pneumonia (CAP) Patients in the US: Focus on Length of Stay." Current Medical Research & Opinion 24 (3): 895-906.

Scotti, D. l., I. Harmon, and S. J. Behson. 2007. "Links Among High-Performance Work Environment, Service Quality, and Customer Satisfaction: An Extension to the Healthcare Sector." Journal of Healthcare Management 52 (2): 109-24.

Sedman, A. B., V. Bahl, E. Bunting, K. Bandy, S. Jones, S.Z. Nasr, K. Schulz, and D. A. Campbell. 2004. "Clinical Redesign Using All Patient Refined Diagnosis Related Groups." Pediatrics 1147(4): 965-69.

Shen, Y. 2003. "Applying the 3M All Patient Refined Diagnosis Related Groups Grouper to Measure Inpatient Severity in the VA." Medical Care 41 (6, Suppl. II): 103-10.

3M Health Information Systems. 2003. All Patient Refined Diagnosis Related Groups (APR-DRG) Version 20.0. Wallingford, CF: 3M Health Information Systems.

Woodbury, D., D. Tracy, and E. McKnight. 1998. "Does Considering Severity of Illness Improve Interpretation of Patient Satisfaction Data?" 1ournal for Healthcare Quality 20 (4): 33-40.

Zandbelt, L. C., E. M. Smets, E J. Oort, M. H. Godfried, and H. C. de Haes. 2007. "Medical Specialists' Patient-Centered Communication and Patient-Reported Outcomes." Medical Care 45 (4): 330-39.

Zimmerman, D., P. Zimmerman, and C. Lund. 1997. "Customer Service: The New Battlefield for Market Share." Healthcare Financial Management 51 (10): 51-53.

PRACTITIONER APPLICATION

Sue Ehinger, PhD, president, Parkview Regional Medical Center and Affiliates, Fort Wayne, Indiana

Healthcare organizations are in an environment of continuous change at a rapid pace. The patient experience has become as important as key healthcare metrics such as mortality and readmission rate. It is now common to hear of organizations purchasing consulting services to improve their customer perceptions, sometimes spending significant resources to change the satisfaction level experienced. Whether through outside consulting or a combination of the numerous webinars and books available, organizations are spending significant time and seeking to improve the patient experience--their sustainability depends on it.

Why has it become so important to create the ideal patient experience? It is ironic that healthcare organizations are seeking assistance to improve the patient experience because healthcare itself is an environment of service. There was a time when those who entered the career of healthcare truly wanted to care for individuals or serve those who entered their doors. That the government will be measuring healthcare organizations' performance brings a much-needed opportunity for them to become recommitted to service. The Centers for Medicare & Medicaid Services has issued a final rule that established the Hospital Value-Based Purchasing program, which will tie acute care Medicare payments to quality of care. It is important to note that the patient experience is a significant element of the Value-Based Purchasing program. Even the government realizes the impact of the patient experience on the quality of care delivered.

This tremendous attention to patient satisfaction has already improved the delivery of care. A simple initiative such as having nurses conduct hourly rounding has been found to improve the patient fall rate. Explaining what you are doing and why you are doing it improves medication and care compliance and other aspects of care. Hospitals are purchasing inpatient portal systems that assist in a better delivery of education to the patient and family. The words we say maximize the impact of the care delivered.

Organizations must seek to understand the attributes that influence the patient experience. Logically, nursing care, physician care, and staff care would be most influential, followed by quality of food and room environment. It is interesting to find that the more seriously ill the patient, the more important physician care becomes. For the less seriously ill, nursing and staff care are most influential. The individuals with whom the patient interacts most frequently drive that patient's willingness to recommend the organization to others for care.

The patient experience has now become a differentiator in highly competitive marketplaces. It is a key determinant of sustainability, as satisfied customers exhibit loyalty and give repeat business. Patients today have choices and shop for the organization that exceeds their expectations. It will be organizations that serve that will survive.

For more information about the concepts in this article, contact Dr. Otani at otanik@ipfw.edu.

Koichiro Otani, PhD, Indiana University-Purdue University Fort Wayne, Fort Wayne, Indiana; Brian Waterman, Thomson Reuters Healthcare, Chicago; and W. Claiborne Dunagan, MD, BJC Healthcare, St. Louis, Missouri
EXHIBIT 1

Descriptive Statistics of Survey Items and Composite Indexes

         N      Mean    sd                  Description

Admission Process

1      24,540   4.10   1.00   Promptness and efficiency of the
                              admission or registration

2      24,680   4.27   0.88   Courtesy and helpfulness at admission
                              or registration

C.I.   24,954   4.19   0.87   Composite Index: Cronbach's [alpha] =
                              0.818

Nursing Care

3      24,713   4.00   1.00   Responsiveness of the nurses when you
                              called

4      24,405   4.25   0.98   Helpfulness of the nurses to reduce or
                              eliminate any pain

5      25,269   4.23   0.97   Nurses' ability to communicate with you

6      25,149   4.18   0.98   Nurses' ability to provide adequate
                              instructions or explanations of your
                              treatment or tests

C.I.   25392    4.16   0.90   Composite Index: Cronbach's [alpha] =
                              0.908

Physician Care

7      24,536   4.16   1.03   Availability of your doctor when needed

8      25,183   4.31   0.95   Doctor's ability to communicate with you

9      25,112   4.30   0.94   Doctors ability to provide adequate
                              instructions or explanations of your
                              treatment or tests

10     24,962   4.31   0.94   Doctors involvement of you in decisions
                              about your care

C.I.   25,359   4.26   0.88   Composite Index: Cronbach's [alpha] =
                              0.930

Staff Care

11     25,093   4.21   0.95   Staffs willingness to help if you had a
                              question or concern

12     25,095   4.11   1.01   Responsiveness of the staff to your
                              requests

13     25,283   4.25   0.92   Courtesy and helpfulness of the staff

14     25,287   4.29   0.92   Amount of dignity and respect shown by
                              the staff

15     24,561   4.12   1.02   Clear and complete explanation provided
                              by the staff about your medications and
                              their side effects

16     24,828   4.15   0.98   Clear and complete explanation provided
                              by the staff about how to care for
                              yourself at home

C.1.   25,400   4.18   0.83   Composite Index: Cronbach's [alpha] =
                              0.928

Food

17     23,745   3.40   1.19   Rate the food that was delivered to
                              your room

18     23,796   4.15   0.93   Rate the courtesy and helpfulness of
                              the staff serving the food

C.I.   24,074   3.78   0.91   Composite Index: Cronbach's [alpha] =
                              0.609

Room

19     25,293   4.01   1.04   Rate the cleanliness of your room

20     23,263   4.12   0.96   Rate the courtesy and helpfulness of
                              the staff who cleaned your room

C.I.   25,344   4.05   0.96   Composite Index: Cronbach's [alpha] =
                              0.863

Dependent Variables

       14,831   4.17   0.98   Overall, rate the quality of care and
                              services received during this hospital
                              stay

       25,267   4.34   0.96   Rate your willingness to recommend this
                              hospital to family and friends

Note: The response can vary between 1 and 5, with 5 being excellent.

EXHIBIT 2

Parameter Estimates of Attributes and Control Variables with Overall
Quality of Care

                                Parameter   Standard
Independent Variables           Estimate     Error     p-Value

Intercept                         0.527      0.056      0.00
Admission Process                 0.126      0.010      0.00
Admission Process X Severity     -0.003      0.020      0.89
Nursing Care                      0.398      0.013      0.00
Nursing Care X Severity           0.006      0.027      0.83
Physician Care                    0.067      0.010      0.00
Physician Care X Severity         0.040      0.021      0.06
Staff Care                        0.303      0.017      0.00
Staff Care x Severity            -0.036      0.034      0.29
Food                             -0.011      0.009      0.24
Food X Severity                   0.005      0.019      0.79
Room                              0.023      0.009      0.01
Room x Severity                  -0.005      0.019      0.78
Severity                         -0.001      0.099      0.99
Scatter Term                     -0.091      0.010      0.00
Scatter Term X Severity          -0.023      0.021      0.27
Age                               0.000      0.000      0.98
Sex, Female                      -0.016      0.011      0.18
African American                 -0.075      0.014      0.00
Hispanic                         -0.252      0.130      0.05
Asian                            -0.002      0.094      0.98
Others                            0.023      0.051      0.65

Note: n = 25,402, [R.sup.2] = 0.5743. Gender: Male is a reference
group. Race: White is a reference group.

EXHIBIT 3

Parameter Estimates of Attributes and Control Variables with
Willingness to Recommend to Others

                                Parameter   Standard
Independent Variables           Estimate      Error    p-Value

Intercept                         0.505      0.042      0.00
Admission Process                 0.055      0.007      0.00
Admission Process x Severity      0.016      0.014      0.26
Nursing Care                      0.206      0.010      0.00
Nursing Care X Severity          -0.022      0.019      0.24
Physician Care                    0.141      0.007      0.00
Physician Care X Severity         0.039      0.015      0.01
Staff Care                        0.448      0.012      0.00
Staff Care X Severity            -0.050      0.024      0.04
Food                              0.009      0.007      0.17
Food x Severity                   0.049      0.013      0.00
Room                              0.074      0.007      0.00
Room X Severity                  -0.016      0.013      0.24
Severity                         -0.061      0.071      0.39
Scatter Term                     -0.054      0.007      0.00
Scatter Term x Severity           0.045      0.014      0.00
Age                               0.000      0.000      0.06
Sex, Female                      -0.004      0.008      0.62
African American                 -0.138      0.010      0.00
Hispanic                         -0.015      0.084      0.86
Asian                            -0.040      0.065      0.53
Others                           -0.076      0.035      0.03

Note: n = 25,402, [R.sup.2] = 0.5781. Gender: Male is a reference
group. Race: White is a reference group.

EXHIBIT 4

Comparisons of Only Significant Variables for the Two Models

   Overall Quality of Care        Willingness to Recommend to Others

1  Nursing care (positive)        Staff care (positive)
2  Staff care (positive)          Nursing care (positive)
3  Admission process (positive)   Physician care (positive)
4  Physician care (positive)      Room (positive)
5  Room (positive)                Admission process (positive)

   Scatter term (negative)        Scatter term (negative)
                                  Physician care x severity (positive)
                                  Staff care x severity (negative)
                                  Food x severity (positive)
                                  Scatter term x severity (positive)

   African American (negative)    African American (negative)
                                  Others (negative)
Gale Copyright: Copyright 2012 Gale, Cengage Learning. All rights reserved.