HW 02: Statistical inference

due Wed, June 30 at 11:59p

Clone assignment repo + start new project

Data

Postoperative sore throat is an annoying complication of intubation after surgery, particularly with wider gauge double-lumen tubes. Reutzler et al. (2013) performed an experimental study in Germany among patients having elective surgery who required intubation with a double-lumen tube. Prior to anesthesia, patients were randomly assigned to gargle either a licorice-based solution or sugar water (as placebo).

Sore throat was evaluated 30 minutes, 90 minutes, and 4 hours after conclusion of the surgery, evaluated using a numeric scale from 0 to 10, where 0 = no pain and 10 = worst pain. For the purposes of this assignment, we will treat these pain scales as numeric.

The data are available in your assignment repository as a .csv file. Some relevant variables of interest are:

Exercises

Overall hint: When performing a hypothesis test, you must provide the significance level of your test, the null and alternative hypotheses, the p-value, your decision, and an interpretation of the p-value in context of the original research question. If you are using a non-simulation-based approach, you must also provide the value of your test statistic and the distribution of that test statistic assuming the null hypothesis is true.

Overall hint: To ensure reproducibility, for all exercises requiring a simulation-based approach, set a seed of your choice. Additionally, ensure that the number of repetitions is sufficiently large.

Hint: Be careful with missing values of the variables you’re analyzing in each question!

  1. Construct and interpret a 95% confidence interval for the mean sore throat pain score 30 minutes after arrival in the PACU among all patients using both a simulation-based approach and a CLT-based approach. Compare these two intervals.

  2. Suppose that these patients are representative of German patients undergoing surgeries that require intubation. Is there evidence that the mean BMI among such patients differs from the mean BMI among all German adults of 26 kg/m\(^2\)? Assess this hypothesis using a simulation-based approach. Provide a visualization of your simulated null distribution and observed data (sample statistic).

  3. It may be possible that ASA classification may be associated with throat pain after surgery. Create a new binary variable that corresponds to whether a patient experienced any throat pain at all, 30 minutes after surgery (i.e., if they had a non-zero pain score at that time). Then assess whether there is a relationship between ASA classification and having any throat pain after surgery among all patients undergoing surgeries that require intubation.

Now, let’s examine any potential effects of licorice solution on reducing throat pain after surgery.

  1. Assess whether there was a lower mean throat pain score 30 minutes after surgery among patients who received licorice compared to patients who received sugar solution placebo. Use a CLT-based approach.
  2. Comprehensively assess whether a lower proportion of patients who received licorice solution reported having any pain 30 minutes after surgery compared to sugar solution. Use a simulation-based approach.
  3. Based on your analyses, do you think that licorice gargle prior to surgery is effective in reducing post-intubation sore throat? Explain your answer, referencing any data, formal statistical tests, or study design as necessary.

In Exercises 7 - 11, determine whether the statements are TRUE or FALSE. If the statement is FALSE, explain why it is FALSE.

The mean BMI among patients receiving licorice solution was 25.6 kg/m\(^2\) and the mean BMI among patients receiving sugar solution placebo was 25.6 kg/m\(^2\). In assessing whether there is a difference in mean BMI between the two treatment groups using a CLT-based approach, the researchers obtained a p-value of 0.925.

  1. If there is truly no difference in mean BMI between these two groups, then the probability of seeing a difference in BMI as large as our observed difference or even larger is approximately 0.925.
  2. Assuming \(\alpha = 0.05\), then our p-value of 0.925 would be strong evidence that there is no difference in the mean BMI between the two treatment groups.
  3. The probability that we have made a Type 2 error is less than 10%.
  4. If we were to repeatedly construct 95% confidence intervals for the difference in mean BMI in the same way from the original population, then we know that 95% of those intervals would truly contain the true population difference in means.
  5. If we instead found a p-value of 0.021, then at the \(\alpha = 0.05\) level, we would have enough evidence to conclude that there is a difference in mean BMI between the two treatment groups.

Acknowledgements

Today’s dataset was made available by the Lerner Research Institute and Dr. Amy S. Nowacki of the Cleveland Clinic. This dataset is representative of a study by Ruetzler et al. (2013).