Franklin University

MATH 810 – PROJECT II

MATH 810 APPLIED STATISTICS

ANALYSIS OF PATIENT OUTCOMES

Introduction:
In the realm of healthcare administration, data-driven decision-making is essential for ensuring efficient
operations and optimal patient outcomes. As a newly appointed executive overseeing various departments
within a regional healthcare system, including an outpatient surgery center, the need for insightful
analysis of operational and financial aspects becomes evident. Despite the presence of limited analytics
on the surgery center's operations and profitability, the acquisition of anonymized patient survey data
offers a promising avenue for comprehensive assessment.
This project delves into an analysis of patient outcomes at the outpatient surgery center, leveraging
anonymized patient survey data sampled over recent weeks. The primary objective is to derive actionable
insights and recommendations based on statistical analyses of the available dataset. By scrutinizing
patient demographics, satisfaction levels, likelihood to recommend, and profitability across different
procedures, this analysis aims to inform strategic decisions aimed at improving operational efficiency and
enhancing patient care.
Through rigorous statistical examination, this project aims to address pertinent questions regarding
demographic distributions, changes in patient satisfaction levels over time, associations between
satisfaction and likelihood to recommend, and disparities in patient profitability across various
procedures. By elucidating these facets, healthcare administrators can gain valuable insights into areas for
improvement, resource allocation, and strategic planning within the outpatient surgery center.

Last Year’s Internal Annual Report.

220818
Franklin University
MATH 810 – PROJECT II

Analyzing Gender and Age Range Distributions: A Comparison between Sample

Data and Annual Report Infographics
The observed (from the sample data) and expected (from last year’s internal annual report) are provided
below:
Observed Expected
13.21 14
15.71 21
37.14 32
27.51 26
6.43 7

Age of patients
40.00% 37.14%
35.00%
30.00% 27.50%
25.00%
20.00% 15.71%
13.21%
15.00%
10.00% 6.43%
5.00%
0.00%
18 to 34 35 to 54 55 to 74 75 to 90 90+

Null Hypothesis (H₀): There is no significant difference between the observed and expected distributions
of age ranges.
Alternative Hypothesis (H₁): There is a significant difference between the observed and expected
distributions of age ranges.

Significance level is 0.05, we reject the null hypothesis if the calculated

2 2
𝜒 2 > 𝜒𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = 𝜒4,0.05 = 9.488

The test statistic:

2
(𝑂𝑖 − 𝐸𝑖 )2
𝜒 =∑
𝐸𝑖

220818
Franklin University
MATH 810 – PROJECT II

(13.21 − 14)2 (15.71 − 21)2 (37.14 − 32)2 (27.51 − 26)2 (6.43 − 7)2
𝜒2 = + + + +
14 21 32 26 7
𝜒 2 = 2.3368

Chi-Square goodness-of-fit results:

N DF Chi-Square P-value
100 4 2.3368777 0.6741

We have that the observed 𝜒 2 is less than the critical value so we fail to reject the null hypothesis and
conclude that There is no significant difference between the observed and expected distributions of age
ranges.
Based on the chi-square test results, we fail to reject the null hypothesis, indicating that there is no
significant difference between the observed and expected distributions of age ranges. Therefore, the age
range distribution in the sample data matches the expectations from the infographics in the annual report.
The chi-square goodness-of-fit test was chosen to compare observed and expected distributions of age
ranges. Its assumptions include categorical data and expected frequencies greater than 5.

Assessing Changes in Patient Satisfaction and Likelihood to Recommend: A

Comparison with the Annual Report.

PATIENT SATISIFACTION
Very Satisfied 28.57%

Would recommend to a friend 37.50%

For very satisfied

𝐻0 : 𝑝 = 0.24
𝐻1 : 𝑝 > 0.24
For would recommend a friend
𝐻0 : 𝑝 = 0.35
𝐻1 : 𝑝 > 0.35

220818
Franklin University
MATH 810 – PROJECT II

The critical value is a Z (0.95) =1.645. So, the rejection Region is such that the calculated Z stat is greater
than the critical Z =1.645
The test statistic is calculated as follows:
𝑝̂ − 𝑝0
𝑍=
√𝑝0 (1 − 𝑝0 )
𝑛

Count of
Row Labels Satisfaction
Somewhat
Dissatisfied 45
Very Dissatisfied 29
Somewhat Satisfied 126
Very Satisfied 80
Grand Total 280

For Very satisfied the test stat is:

80
(280) − 0.24
𝑍= = 1.7911
√0.24(1 − 0.24)
280
1.7911 > 1.645 so we reject the null hypothesis and conclude that the rates of patient being very satisfied
has increased since the annual report.
For would recommend to a friend:
Row Labels Count of WouldRec
N 175
Y 105
Grand Total 280

105
(280) − 0.35
𝑍= = 0.8771
√0.35(1 − 0.35)
280
We fail to reject the null hypothesis and conclude that the likelihood to Recommend has not increased
since the annual report.
The one-sample proportion test was chosen because we're comparing a single sample proportion to a
hypothesized value. Its assumptions include random sampling, independence of observations, and a
sufficiently large sample size. As long as these assumptions hold, the test is appropriate for assessing
whether the observed proportion significantly differs from the hypothesized value.

220818
Franklin University
MATH 810 – PROJECT II

Exploring the Relationship between Patient Satisfaction and Likelihood to

Recommend: Implications for Program Design and Patient Recommendations

Grand
Row Labels N Y Total
Somewhat Dissatisfied 34 11 45
Somewhat Satisfied 75 51 126
Very Dissatisfied 22 7 29
Very Satisfied 44 36 80
Grand Total 175 105 280

Null Hypothesis: There is no association between a patient's satisfaction level and their likelihood to
recommend the surgery center.
Alternative Hypothesis: There is an association between a patient's satisfaction level and their likelihood
to recommend the surgery center.
2 2
The rejection region. 𝜒 2 > 𝜒𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = 𝜒3,0.05 = 7.8147
2
2
(𝑂𝑖𝑗 − 𝐸𝑖𝑗 )
𝜒 =∑
𝐸𝑖𝑗

(row total𝑖 × column total𝑗 )

𝐸𝑖𝑗 =
grand total
45 ∗ 175
𝐸1,1 = = 28.13
280

Cell format
Count
(Expected count)

N Y Total
Somewhat Dissatisfied 34 11 45
(28.13) (16.88)
Somewhat Satisfied 75 51 126
(78.75) (47.25)
Very Dissatisfied 22 7 29
(18.13) (10.87)
Very Satisfied 44 36 80
(50) (30)
Total 175 105 280

220818
Franklin University
MATH 810 – PROJECT II

(34 − 28.13)2 (11 − 16.88)2 (75 − 78.75)2 (51 − 47.25)2 (22 − 18.13)2
𝜒2 = + + + +
28.13 16.88 78.75 47.25 18.13
(7 − 10.87)2 (44 − 50)2 (36 − 30)2
+ + +
10.87 50 30
𝜒 2 = 7.878

Statistic DF Value P-value

Chi-square 3 7.8779785 0.0486

The calculated chi-square statistic is 7.878, which exceeds the critical value of 7.8147 at the 0.05
significance level. Therefore, we reject the null hypothesis, indicating that there is evidence to suggest an
association between satisfaction level and likelihood to recommend.
The calculated p-value is 0.0486, which is less than the significance level of 0.05. This further supports
the rejection of the null hypothesis.
Based on these results, it seems reasonable to expect that a program designed to increase patient
satisfaction would be likely to result in more patient recommendations, as there is evidence of a
significant association between satisfaction level and likelihood to recommend.
I chose the chi-square test for independence because it's suitable for analyzing associations between
categorical variables. The assumptions, including independence and expected cell frequencies, are met
based on the sample data. Thus, the test is appropriate for examining the relationship between satisfaction
level and likelihood to recommend.

Analyzing Patient Profitability across Procedures: Implications for Service Offerings and
Resource Allocation

Null Hypothesis (H0): The average patient profitability (lifetime value) is the same across all different
procedures offered by the center.
Alternative Hypothesis (H1): The average patient profitability (lifetime value) differs among the different
procedures offered by the center.
The test is an F test 𝛼 = 0.05 ,The critical region 𝐹3,276 = 2.637, so we reject the null hypothesis the
observed t test statistic is greater than the critical F.
The Anova results are given below:

SUMMARY
Groups Count Sum Average Variance
Bariatric 46 5145 111.8478 633.5986
Cardiovascular 85 8485 99.82353 589.3375
Oncology 79 8366 105.8987 648.8101
Orthopedic 70 6683 95.47143 631.9629

220818
Franklin University
MATH 810 – PROJECT II

ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 8997.505 3 2999.168 4.806222 0.002798 2.637311
Within Groups 172228.9 276 624.0178

Total 181226.4 279

The calculated F statistic is 4.8062 Which is greater than the critical value 2.6373. We reject the null
hypothesis and conclude that, the average patient profitability (lifetime value) differs among the different
procedures offered by the center.
Multiple comparison test:
Tukey's HSD test, it will provide with confidence intervals for the differences between all pairs of group
means. If the confidence interval includes zero, it means that the difference is not statistically significant.
If it does not include zero, it indicates a significant difference between the group means.

Tukey HSD results (95% level)

Bariatric subtracted from

Difference Lower Upper P-value
Cardiovascular -12.024297 -23.842854 -0.20573933 0.0444
Oncology -5.9490919 -17.924208 6.0260245 0.5739
Orthopedic -16.376398 -28.631537 -4.1212581 0.0036
Cardiovascular subtracted from
Difference Lower Upper P-value
Oncology 6.0752048 -4.0153748 16.165784 0.4055
Orthopedic -4.3521008 -14.773465 6.0692631 0.7024
Oncology subtracted from
Difference Lower Upper P-value

Orthopedic -10.427306 -21.025888 0.17127644 0.0557

The significant pair of means based on the Tukey HSD test results are:
1. Bariatric vs. Cardiovascular
2. Bariatric vs. Orthopedic

The one-way ANOVA test was selected because it allows for comparing means across multiple groups
simultaneously. Its assumptions include independence and normality, which can be verified through

220818
Franklin University
MATH 810 – PROJECT II

visual inspection of Histogram. If these assumptions are met, the ANOVA test is appropriate for
determining if there are significant differences in patient profitability among different procedures.

Normality is achieved. The data is independent and the variances can be assumed to be homogenous.

CONCLUSION
In summary, our examination of patient outcomes at the outpatient surgery center has yielded
significant insights into various operational aspects. Through thorough statistical analysis, we have
addressed crucial inquiries regarding demographic distributions, changes in patient satisfaction levels,
associations between satisfaction and likelihood to recommend, and discrepancies in patient
profitability across different procedures.

Initially, we observed that the gender and age range distributions in the sample data generally coincided
with the expectations from the annual report infographics. However, upon closer inspection, slight
disparities emerged, suggesting potential over- or under-representation in certain demographic groups.

Subsequently, our analysis indicated an apparent increase in rates of patients being Very Satisfied
following the publication of the annual report. However, there was no significant rise in the likelihood to
recommend. This points to nuanced shifts in patient satisfaction levels over time.

Furthermore, we uncovered evidence suggesting a correlation between a patient's satisfaction level and
their likelihood to recommend the surgery center. This underscores the importance of improving patient
satisfaction as a means to generate positive recommendations, emphasizing the potential impact of
targeted programs aimed at enhancing the patient experience

Lastly, our examination of patient profitability across different procedures revealed notable variations.
While the average patient profitability (lifetime value) varied among procedures, further tests identified
specific pairs of procedures with significantly different patient lifetime values. This insight can guide
strategic decisions regarding service offerings and resource allocation within the outpatient surgery
center.

220818