You are working as a statistician with the Boston Police department. You were given this case to work on to see if there is enough data to support an arrest warrant.
Kristen Gilbert started working at the VAMC as a nurse in 1989. A proficiency report obtained by the Boston Police Department, which described her as “highly skillful”, calm, and compassionate. She organized charity drives, collections for the needy, and organized a memorial service for a colleague who died of cancer (CBS News, Killer Nurse Gets Life, 2009).
However, Gilbert did tend to be nearby when patients died. Coworkers sometimes referred to her as the “angel of death”. Eventually her fellow nurses grew suspicious. Gilbert’s coworkers stated to raise concerns about an increase in medical emergencies (codes) and deaths in Ward C between August 1995 and February 1996. A criminal investigation was launched in February and Gilbert left her job shortly after learning she was the target of the investigation.
The search warrant of hospital records showed that on Ward C between January 1, 1995 and February 19, 1996 Gilbert was present or on duty for 40 codes which ended in death.
As the statistician for this case answer the following question
What information would you need to see if there is an association between Gilbert being on duty and code/death rates?
What other explanations might there be for why Gilbert had more deaths on her shifts?
What type of statistical test would you use for this case to see if there is an association?
What are your null hypotheses and alternative hypotheses?
The following is an analysis of 1641 eight-hour shifts. Use this data to decide if there is enough statically support for an arrest warrant. (Use 95% confidence interval)
Deaths during shift?
Gilbert present Yes No
Yes 40 217
No 34 1350
