Hypothesis Testing (Jane Bolding Case)

Exercise 28: The Jane Bolding Case

This historical exercise is based on an article in Regardie’s, May 1988. The evidence below is the source’s selection, not a complete record of the trial.

“Jane Bolding was a smart, committed, exemplary nurse, but her patients died more often than other nurses’. And when the results of a shocking statistical study were released, the police said she was a killer.”

Raised a Catholic in a poor blue-collar family in Bladensburg, Maryland, she was active in high school (student government, drama club, broadcasting club, and school newspaper), a neat girl, funny, easy to talk to, and a great shoulder to cry on. Her father was an alcoholic, and her mother, a large woman, always had a vicious look on her face and verbally abused her. Bolding was adopted, something she learned at age fourteen during a fight with her parents. She was a bright, serious student who breezed through math and science. She worked as a candy striper at the Prince George’s Hospital Center and aspired to become a doctor, but she was bitterly disappointed that her parents could not afford to send her to college. Two months after graduating from high school in 1975, she got a job as a blood collector in the hospital’s pathology lab.

Three years later she transferred to the Intensive Care Unit (ICU) as a low-level nursing assistant, and later she became a critical care technician. Soon afterward her performance was rated exemplary: intelligent, sound medical judgment, superconscientious, did the best for her patients. Enrolled part-time in a two-year nursing program, she graduated in May 1983 as a registered nurse. Her job was her life. She kept up with nursing research, taught classes in critical care, worked extra shifts, stayed late to write meticulous notes about her patients, rarely charged overtime, and was especially adept at reconciling families to the death or permanent handicap of a loved one.

In May 1988, Bolding was tried before a jury for murdering her patients. The jury acquitted her of the charges.

There are three principal hypotheses that explain the deaths of patients under Bolding’s care:

1. She murdered the patients (either alone or in collusion with another person or persons).
2. Someone else murdered them.
3. They died of natural causes or some unknown phenomenon or force.

Compare these hypotheses using the evidence below. For each item, ask which explanations it supports or contradicts and what additional information would help distinguish them. This is a structured comparison of hypotheses rather than a formal statistical test.

The Evidence

1. Between 1983 and 1985, twenty-eight people in Bolding’s care died in the ICU of the Prince George’s Hospital Center in Maryland. Most of the 194 other nurses who worked in the ICU at some point during that period had only two or three patients die, or none at all.
2. There was a surge in cardiac arrests in the ICU between January 1984 and March 1985. Bolding’s patients suffered fifty-seven fatal or nonfatal arrests. The nurse with the next highest number was present at only five, while most other ICU nurses saw only one or two.
3. During the fifteen-month epidemic, ten patients suffered two or three arrests in an evening, even though they had been fine during the day and night shifts. Eight of those ten were Bolding’s patients. Those in bed, who were least visible from the nurse’s station, were especially prone to unexpected arrests.
4. The epidemic began in January 1984, grew and gathered speed, then stopped cold when she was dismissed.
5. The deaths left no telltale evidence of murder.
6. The probability that pure chance created the epidemic is less than one in 100 million.
7. When Bolding was on duty, the ICU averaged a cardiac arrest every four days. When she was off duty, there was one every fifteen days.
8. By five measures of illness used by the ICU, her patients were no more critically ill than those of the other nurses and had similar prognoses.
9. There had been twenty-one arrests on the evening shift in 1983. Based on that figure, about thirty-one arrests would have been expected over the next fifteen months. In fact, there were eighty-eight, more than twice as many as on the day and night shifts combined. Bolding was the attending nurse in fifty-seven of the eighty-eight. If one subtracts those fifty-seven cases from the total eighty-eight, the result is thirty-one, the number that would have been expected statistically.
10. People who would ordinarily be less likely to have heart failure became, in her presence, more likely to. A woman had a 51 percent chance of an arrest under her care versus 6.7 percent under the average nurse. A man had a 32.4 percent chance under her care versus 8.3 percent under the average nurse. A patient under the age of thirty had a 41 percent chance versus 4.4 percent.
11. Her patients did not have arrests during shifts when other nurses cared for them.
12. No medication, procedure, intravenous fluid, blood product, or other health care worker was as strongly associated with cardiac arrests as was Bolding.
13. There were no witnesses or physical evidence that any crime had been committed.
14. Bolding performed well on “codes” (emergency treatment of cardiac arrests) and seemed to like being part of a cardiac arrest resuscitation team.
15. There was no evidence that Bolding had a lower threshold for calling a “code.”
16. Bolding didn’t work more hours than the other nurses.
17. Right after she was arrested for the murders, she admitted to police interrogators—and later recanted—that she had killed two patients.
18. She failed a polygraph during her police interrogation.
19. The hospital was her first accuser, but later hospital attorneys pooh-poohed the charges. Their turnabout may have sprung from the eight multimillion-dollar civil suits, claiming negligence, that were filed against the hospital over the deaths.
20. Police suspected the deaths were caused by administration of potassium chloride, which in large dosages can cause cardiac arrest. Potassium chloride was one of the most commonly used substances in the ICU. The majority of ICU patients received supplemental doses of potassium chloride through an IV bag over a period of hours. It was never injected all at once undiluted—in a “push,” as doctors call it. Thus, patients seldom experienced a high level of potassium chloride.
21. The source claims that a surviving patient’s excess potassium would clear rapidly and that an overdose would not be revealed by a postmortem examination. These are assertions in the exercise, not independently established findings here.
22. Most ICU patients received medication every hour.
23. It was commonplace to see a nurse at bedside administering medicine into an intravenous bag.
24. About thirteen patients were usually in the ICU, overseen by a revolving crew of six to seven physicians. Medical residents, fellows, respiratory therapists, and other personnel also worked there. But day in, day out, particularly late at night, the ICU was the domain of ten to eleven nurses on duty, each assigned two patients at most.
25. The nurses generally chose the patients they wanted to care for and followed their progress till they were released. Most patients remained in ICU only a week or two.
26. Although many ICU patients were semicomatose and most were so ill that death was a possibility all the time, they were by no means all expected to die.
27. Like many ICU nurses, Bolding preferred more challenging patients, those most grievously ill or injured. Less experienced or less confident nurses were happy to let her take the responsibility for those patients.
28. Medications were left in open areas.
29. One physician’s assistant was on duty during 70 percent of Bolding’s patients’ fatal episodes, yet he also worked during other shifts when there was no epidemic. And the evening-shift epidemic continued when he was off duty and Bolding was on.
30. Three other nurses were on duty during many of the arrests, yet the epidemic continued when they weren’t in the unit.
31. None of the other nurses was present during all of Bolding’s patients’ arrests.

Compare the hypotheses before reaching a conclusion. The stated chance probability depends on a statistical model; it is not the probability that Bolding was innocent. Ask what observations, patient assignments, comparison groups and alternative causes the model accounts for.