ECMT463_FinalDraft
- 1. Datta and Kim 1 Texas A&M University Impact of Amount of Weekly Work Hours for Health Employees on Patient Satisfaction Lalit Datta Minji Kim ECMT 463-906 Haeshin Hwang 29 April 2014
- 2. Datta and Kim 2 Impact of Amount of Weekly Work Hours for Health Employees on Patient Satisfaction I. Introduction Recent legislation over the past few years have indicated that rising health care costs are an important part of expenditure in the United States and around the world, so studying possible factors that may have some effects on health care costs is certainly of major relevance. An interesting debate has arisen that asks if there is any sort of correlation or relationship between the average number of hours that a health employee works each week with the percentage of health patients that are satisfied with the services they receive. Our hypothesis is that when there are more work hours each week for health employees, this could lead to burnout and a negative impact on their well-being, which leads to them either being more likely to leave their job, resulting in more expensive job turnover, or more fatigued during their shifts, leading to a much higher chance of medical malpractice. In the short and the long run, this factor can negatively affect patient care and health care costs. This kind of issue actually has some relation to the theory of diminishing returns, which explains that there is a decrease in marginal output of a process of production as the amount of a single factor is increased, while all other factors remain constant. When an employee works longer hours but is getting tired and being more prone to mistakes, the level of productivity and quality is dropped per unit of labor. This theory implies that longer hours should reduce costs for the employers because fewer employees are needed if each is used for more hours, but that this will lead to a higher chance of medical malpractice that increases chance of patient death, which then increases costs for employers. This theory, however, is not used in the exact same way for this research as explained in the definition above. Our research will try to delve more specifically into the conditions of hospitals in each state rather than focusing on units of labor.
- 3. Datta and Kim 3 Research has already been found from The Center for Health Outcomes and Policy Research at The University of Pennsylvania School of Nursing that shows as the percentage of hospital nurses working shifts over thirteen hours increased, patients' dissatisfaction with care increased (Stimpfel 2012). Another source from the National Nursing Research Unit at the King’s College London showed data that found that patient deaths from pneumonia and myocardial infarction occurred more often where nurses worked longer shifts, as well as in hospitals where there was more often a lack of time off from the job for the nurses (NNRU 2013). Relating to this, a report from the National Health Foundation in 2013 found that a reduction of sepsis mortality greatly reduced healthcare costs, as cost avoidance analysis from 2010-2012 showed that 3,576 lives saved resulted in a reduction of $63, 804, 021 (Kun 2013). Our paper is going to be different from others that are on similar topics primarily because the research that has already been conducted focuses primarily on daily work hours and shift length, while ours will have a different focus that will be primarily on the average total number of weekly hours for health employees and how that affects overall patient satisfaction, so the effects are focused more on long term and less on a specific shift for an employee. For our research, we gathered data from both the Current Employments Statistics survey and the Hospital Consumer Assessment of Healthcare Providers and Systems, both from the year 2008, to create a regression model that would help us determine any kind of correlation. Our method was to utilize the regression component from the Data Analysis tool in Microsoft Excel to minimize the error in the model by using the Ordinary Least squares method. Our results showed that the data did not provide a certain conclusion to whether or not weekly hours for health and hospital employees had any effect on patient satisfaction.
- 4. Datta and Kim 4 II. Model Specification When a hospital employee works longer hours but is getting tired and being more prone to mistakes, the level of productivity and quality is dropped per unit of labor. The theory of diminishing returns implies that longer weekly hours should reduce costs for the employers because fewer of them are needed to be employed if they work longer hours, but that this will lead to an increased level of fatigue in nurses and hospital workers which in turn leads to a higher chance of medical malpractice that increases chance of patient death, which then increases costs for employers. This theory, however, is not used in the exact same way for this research as explained in the definition above. Our research will try to delve more specifically into the health and satisfaction of patients in each state rather than focusing on units of labor. Our main variable in our empirical model is going to be the average number of work hours for each health employee in each state across the United States. We are also going to factor in three control variables that are going to have an effect on both work hours and patient satisfaction. One of these control variables is going to be average hospital cleanliness in each state. The reason that this would be a major control variable is that how clean the important facilities are in a hospital, nursing home, or clinic can play a factor in how satisfied a patient is with the services he or she is getting which is independent from how well the patient is taken care of by the employees in the facility. Another variable that we will control will be salary. We believe salary plays a role in both employee hours and patient satisfaction. Our theory is that if an employee is getting paid higher in his or her state compared to the national average, than he or she is more likely to spend longer hours working resulting in them less likely to need to work a second job to meet their basic needs, while the variable also indirectly affects patient satisfaction because if the employees are getting paid higher, than that most likely means the state has higher
- 5. Datta and Kim 5 funding for health services, leading to better run health facilities which gives higher convenience for the patients. Our last control variable is the level of noise in hospitals at night, where states with hospitals that are more likely to have this issue will mainly be located in major urban areas. This ends up leading to higher likelihood of burnout for the employees as they each have to cater to the needs of more patients, while the noise itself can be a frustrating experience for the patients. In the empirical model, a value of one will be given to states whose average hourly salaries for health employees is above $20, if less than 40% of surveys from the state report loud noises at night, and if over 68% of surveys from the state report the hospitals being very clean. Likewise, a value of zero will be given to states whose average hourly salary for health employees is less than $20, if more than 40% of surveys from the state report loud noises at night, and if less than 68% of surveys from the state report the hospitals being very clean. All these variables should be included in the model to show the overall effect in patient satisfaction. Omitted variables in the model that could affect the outcome that are not taken into effect could be age, which has a correlation with nurse experience, as well as proper technology, which could have an effect on how efficiently the patients are treated with the resources available.
- 6. Datta and Kim 6 Here is our summary table of data and written regression model: Patient Satisfaction Weekly EmployeeHours Cleanliness Salary Quietness Average 68.10417 33.21458333 0.8125 0.541667 0.25 Std Dev 4.534687 1.741238198 0.394443 0.503534 0.437595 Max 79 39.5 1 1 1 Min 58 30.6 0 0 0 # Observations 48 48 48 48 48 Source: The Bureau of Labor Statistics, The Centers for Medicare & Medicaid Services PatientSatisfactioni = ß 0+ ß 1WeeklyEmployeeHoursi + ß 2Cleanlinessi + ß 3Salaryi + ß 4Quietnessi + ui Where WeeklyEmployeeHours represents the average number of hours per week working for each health employee in a respective state in the United States, Cleanliness as a control of whether the sufficient percentage surveyed in each state reported the hospital to be clean, Salary as a control for whether the average salary for health employees in the state is above or below $20, and Quietness as whether the sufficient percentage surveyed in each state reported the hospital to be quiet at night. We obtained data for patient satisfaction, cleanliness, and quietness from the 2008 edition of the Hospital Consumer Assessment of Healthcare Providers and
- 7. Datta and Kim 7 Systems (HCAHPS 2008) while the data for WeeklyEmployeeHours and Salary were found from the Current Employment Statistics Survey(CES) from 2008 (DeAntonio 2010). The null hypothesis is that the number of weekly work hours for health employees has no effect on patient satisfaction, or that ß 1= 0. Our alternative hypothesis that we are testing is that overall, there will be a negative correlation between the hospital’s average weekly hours for an employee, versus patient satisfaction in the hospital, which is ß 1 ≠ 0, or more specifically, with our prediction, ß 1< 0. If this is true, the implication would be that if there is a higher number of hospital employees who, on average, work less hours each week, then the employees are less likely to burn out in their job, and this in turn, would lead to higher patient satisfaction. On the assumption that our hypothesis that longer hours of employees affect negatively on patients’ satisfaction and health, we expect an innovative scheme of human resources and management of hospitals that could be recommended as a great way to manage employees and contribute to providing high-quality nursing and healthcare to patients as well, reducing turnover rates, costs, and other inconveniences for the respective hospital. From the data that we found and mentioned earlier, we used Data Analysis in Microsoft Excel to create a regression model from the data. We predict that ß 1 will be negative because of our alternative hypothesis that more work hours will lead to a decrease in patient satisfaction. We predict that ß 2 will be positive because we understand that whenever a hospital provides a clean environment, the patient will be happier with the services that are being provided. We predict ß 3 to be positive because we believe that higher salaries can result in a desire for higher average hours but that will lead to higher satisfaction for the employees, and indirectly, the patients. Lastly, we predict ß 4 to be positive because we believe that if a hospital is quiet at night, then it is less likely that the hospital is a
- 8. Datta and Kim 8 densely populated area which eases any sort of inconveniences for both the employees and the patients. III. Results Coefficients Standard Error T-Stat P-value Intercept 63.47577027 12.40435485 5.117217 6.89E-06 WeeklyEmployeeHours 0.02720095 0.370143076 0.073488 0.941759 Cleanliness 3.287096424 1.640278108 2.003987 0.051397 Salary 0.462482898 1.332446266 0.347093 0.730214 Quietness 3.214603089 1.492263904 2.154179 0.036873 Observations 48 48 48 48 With degrees of freedom of 43 and a 5% significance level, the critical value we will use is 2.32. Surprisingly, in our results, we find out that ß 1 is positive, but not by much, as the t- statistic is very low. Because of this, as well as predicting the wrong sign, we fail to reject the null hypothesis. A very likely reason for this is the data we found for WeeklyEmployeeHours and PatientSatisfaction were from two different data sources and this may have an effect on accurate estimation of data. Another factor to add is that it is certainly likely that the average weekly hours for health employees makes very little difference in patient satisfaction but it is really actually the average shift length for employees, similarly to research that has already been done on the topic. We predicted correctly that ß 2 was a positive coefficient but the t-statistic is
- 9. Datta and Kim 9 too low, so we fail to reject the null hypothesis. We predicted correctly that ß 3 is positive but again, the t-statistic is too low. We are again in the same situation for ß 4. The unexpected results that were generated may be the result of many factors. One of these include the fact that for average salary and work hours, we used a different data set than the one for patient satisfaction, cleanliness, and quietness. From the data that was the Current Employee Survey, the data for salaries and weekly work hours for health employees also included education employees, which is an unrelated variable that may have greatly skewed the data more than we would have otherwise realized or predicted. Another major factor was choosing an arbitrary percentage that determined assigning values for 1 and 0 for each of our control variables that made our t-statistic and standard error for our variables less accurate than they should have been. However, the case may still be that weekly employee hours actually matter very little to patient satisfaction because there is a huge variation in how much rest that the employees have between shifts, so ultimately weekly hours do not matter to the correlation as much as actually length of each shift for the employees. IV. Conclusion Our motivation for our research was to find out if weekly work hours for hospital and health employees actually play a more important factor in affecting patient satisfaction rather than the average shift length for each employee as research that already exists suggests. We have determined that the data we have found and made a regression model from Microsoft Excel for is not sufficient enough to make a correct conclusion for our research. However, we hope that in the future, we can gather better information and data so we can make a more accurate prediction of this important economic question, as several unused variables that were in the data could have
- 10. Datta and Kim 10 affected the results that we can control for now that we have a better understanding of all the varying effects of patient satisfaction.
- 11. Datta and Kim 11 Works Cited Amy Witkoski Stimpfel, Douglas M. Sloane and Linda H. Aiken. "The Longer the Shifts for Hospital Nurses, the Higher the Levels of Burnout and Patient Dissatisfaction." Nursingworld. Health Affairs, 5 Nov. 2012. Web. 11 Apr. 2014. <http://www.nursingworld.org/MainMenuCategories/WorkplaceSafety/Healthy- Nurse/Longer-Shifts-For-Hospital-Nurses-Higher-Levels-Of-Burnout-And-Patient- Dissatisfaction.pdf>. DeAntonio, Dante. "Program Report." United States Department of Labor. The Bureau of Labor Statistics, 1 Mar. 2010. Web. 28 Apr. 2014. <http://www.bls.gov/opub/mlr/2010/03/art4full.pdf>. Heather Kun, Mia Arias, Jessica Williams and J. Eugene Grigsby. "PATIENT SAFETY FIRST PHASE I RESULTS." National Health Foundation. National Health Foundation, Aug. 2013. Web. 11 Apr. 2014. <http://www.nhfca.org/psf/docs/760.NHF_EndOfYearReport_FINAL.pdf>. "Official Hospital Compare Data Archive." The Centers for Medicare & Medicaid Services. Hospital Consumer Assessment of Healthcare Providers and Systems, 1 Dec. 2008. Web. 28 Apr. 2014. <https://data.medicare.gov/data/archives/hospital-compare>. "What are 12-hour shifts good for?" KING'S college LONDON. National Nursing Research Unit, 1 Mar. 2013. Web. 11 Apr. 2014. <https://www.kcl.ac.uk/nursing/research/nnru/policy/By-Issue-Number/Policy--Issue- 38.pdf>.