BUAD 447 - Project

Fall -2024

Exercise on Quality Management and Customer Focus

Improving Patient Satisfaction

Endocrinology Group Patient Feedback

Background

A regional endocrinology specialty office is experiencing a decrease in the number of patients seen on a weekly basis. The office consists of four doctors, a nurse practitioner, two registered nurses (RNs), four licensed practical nurses (LPNs), and five administrative staff members. Among the clinic staff, there is some concern that this decrease in volume is due to patient dissatisfaction.

A team is assembled from among the staff members to investigate potential causes of decreased patient volume. You are the internal consultant, brought in to assist the team with the analysis.

The Data ComplaintData.xls and PatientFeedback.jmp

        [Both datasets are located at CANVAS --> Files→CASE_STUDIES->Patient Satisfaction]

ComplaintData.xls contains information on 262 customer complaints received over a three-month period. Complaint data is captured through the specialty office website.

Date Date complaint filed

Complaint Type Reason for complaint

PatientFeedback.jmp contains summary information from patient surveys administered over a three- month period.

Week # Week 1, 2, …

Percent Recommend Percentage of responders who would recommend this office

Wait Percent indicating wait time was acceptable

Respect Percent indicating they were treated with respect

Enough Information Percent indicating they received enough information from provider

Problem 1:

Question 1: Which three complaints type makes almost 57% of total complaints? [Hint: Use graphical tool we learn from quality control.]

Problem 2: Rather than speculate on causes of the negative feedback, the team decides to listen to the “voice of the customer” by using a short survey.

“Voice of the customer”, or VOC, is a term used in quality improvement to refer to the process of obtaining feedback on customer likes, dislikes and concerns.

Over a three-month period, patients seen in the office are given a short survey as they finish their appointments. The patients are asked the following questions:

● Was the wait time to be seen today acceptable?

● Were you treated with respect by the provider and office staff?

● Did you receive enough information from the provider?

In addition, all patients are asked:

• Would you recommend this office to others?

A 7-point Likert scale, commonly used in survey research to indicate participant’s agreement with a statement, is used for all responses. A seven-point scale was used where, 1 = Totally Disagree to 7 = Totally Agree. So,

lower ratings indicate the patient is less satisfied, and higher ratings mean the patient is more satisfied.

The team decides to focus on the percentage of satisfied patients. They decide that patients

responding with a 6 or a 7 on the Likert scale will be classified as satisfied, while patients responding  with a 1 – 5 will be classified as not satisfied. At the end of the three-month period, for each question, the percentage of satisfied patients is calculated for each week. These data are found in the file

PatientFeedback.jmp.

Question 2: What is the average, standard deviation, and interquartile range of Percent Would Recommend variable? Percent Would Recommend variable is the primary measure the team will use to assess patient satisfactions.

Question 3: In which week Percent Would Recommend is at the highest level? [Hint: Use run chart from Graph Builder.]

Run charts provide a graphical indication of the central tendency and the variability over time. Run charts also allow time-dependent patterns, such as the decrease in week 5, to be easily detected. (We revisit whether there are patterns over time in the exercises.)

(Graph > Graph Builder; Drag Percent Would Recommend to the Y zone and Drag Week to theX zone. Select

the Line icon at the top of the graph. Right-click on they-axis, choose Axis Settings from menu, and check the    box for Major Gridlines. Note that to apply a 0 to 100% scale (not shown here) Make Minimum = 0, Maximum = 100, Increment = 10, Tick Marks = 1. Click OK to accept axis setting changes, and click Done to close the Graph Builder control panel.)

Run charts for the other three measures, along with Percent Would Recommend, are given below. For ease of comparison, the four measures are plotted on the same graph. The values for these variables tend to move

together, with the exception of Respect. For example, as Percent Would Recommend goes up Enough

Information generally goes up, and as Percent Would Recommend goes down Enough Information generally goes down.

Question 4: What is the relationship between Percent Would Recommend, Wait Time Acceptable,

Respect, and Enough Information? [Hint: Use run chart from Graph Builder. In addition create histogram with dynamic linking]

From the Graph Builder report produced here, select Show Control Panel from the red triangle to reopen the control panel. Then, one at a time, drag each response to the drop zone just inside y-axis. To change the line styles, right-click on the response names in the legend on the right and select Line Style.)

Run charts for the other three measures, along with Percent Would Recommend. For ease of comparison, the four measures are plotted on the same graph. The values for these variables tend to move together, with the

exception of Respect. For example, as Percent Would Recommend goes up Enough Information generally goes up, and as Percent Would Recommend goes down Enough Information generally goes down.

Analyze > Distribution; Drag all 4 variables to Y, Columns, and click OK. Click on red triangle next to Distributions and select Stack. Hold shift key and click on each of the lowest bars for Percent Would Recommend.)

Histograms show the shape, center, and spread of the distributions. Histograms for the four responses, generated from the Distribution platform. The low values for Percent Would

Recommend are shaded, indicating that those rows are selected in the data table. Since the histograms are dynamically linked, the values for the selected rows are highlighted in the other histograms as well.

This confirms that low values for Percent Would Recommend generally correspond to low values for the other variables (except Respect).

Question 5: Create summary statistics table for all four variables and explain each of them in brief Percent Would Recommend, Wait Time Acceptable, Respect, and Enough Information?

Question 6: We use correlations to understand significant relationships between two variables. Generate a correlation plot with graphs for all four variables [Percent Would Recommend, Wait Time Acceptable, Respect, and Enough Information] and explain in brief.

(Analyze > Multivariate Methods > Multivariate; Select all variables of interest as Y, Columns and click OK.)