The analysis of creatinine level
given in this case shows a situation where a serious violation of
a one-sample t-test assumption occurred causing a need for an alternative
method. This emphasizes the importance of testing assumptions. This
case also illustrated the importance of visualizing each variable
as a necessary first step prior to applying a statistical method.
This revealed the left-skewed creatinine distribution. Not only
was the one sample t-test inappropriate in this situation, but the
median was a better measure of centrality than the mean due to the
left-skewed creatinine distribution. While both the one sample t-test
and the test of proportion shown here produced the same conclusion,
in other circumstances, two different tests may lead to contradictory
conclusions.
For this data, it is
estimated that 63.7% of patients have creatinine levels indicative
of Stage 1 kidney insufficiency. The 95% confidence interval tells
us that the true proportion could be as low as 58.7% and as high as
68.4%. This suggests that the hospital should be prepared to handle
patients with kidney insufficiency.
Bear in mind that the
data was simulated and it is not known the extent to which it represents
real patient data on kidney insufficiency. It is always prudent to
review results with subject matter experts who can lend valuable insight
into the interpretation of the analysis in the problem context. The
data set provides additional information on other diagnoses such as
diabetes and coronary artery disease. As a next step, relationships
between these co-morbidities and kidney function should be examined.