To understand the shocking miscarriage of justice meted out to the Dutch licensed paediatric nurse, Lucia de Berk, it’s essential to gain some familiarity with the statistical fallacy known as the ‘Texas Sharpshooter’. This is a type of flawed logic whereby random data points with no relationship to each other are manipulated or analysed until a pattern is identified. The significance of the pattern is assigned retrospectively. To all intents and purposes it means cherry-picking a data cluster to suit a particular argument. The name ‘Texas Sharpshooter’ is a reference to a scenario in which a marksman shoots randomly at the wall of a barn and then, retrospectively, paints a bullseye target around the spot where most of the bullet holes appear, making it appear as if he’s a really good shot. The point is, such clusters appear naturally by chance and don’t necessarily imply there is a causal relationship. By way of example, let us suppose that the manufacturers of Soda Buzz drinks get hold of some research showing that three of the five countries where Soda Buzz sells best appear in a list of the top ten healthiest countries on the planet – they declare therefore that Soda Buzz drinks are healthy.
And now back to our hapless nurse. On 4 September 2001, the unexpected death of a baby girl named Amber in a children’s hospital in The Hague put in train a sequence of events that transformed the life of Lucia de Berk. Toxicology reports suggested the baby’s death may have been caused by digoxin poisoning. This prompted the medical authorities to scrutinize earlier deaths and cardiopulmonary resuscitations at the hospital. They discovered that between September 2000 and September 2001 Lucia de Berk had been on duty at the time of nine such episodes. The crimes were supposed to have taken place in three hospitals where she had worked in The Hague – the Juliana Child Hospital, the Red Cross Hospital and the Leyenburg Hospital. Having identified what they thought was a suspicious trend of deaths and near-deaths, the hospital decided to press charges against her. Although there had been no eye witnesses and no direct incriminating evidence against her, she was sentenced to life imprisonment (with mandatory psychotherapy) in 2003 for four murders and three attempted murders of patients in her care. The following year, appeal court judges upheld her sentence and convicted her on three additional counts of murder and three of attempted murder. Her sentence was unusually heavy by Dutch standards, in part because she had continued to deny wrongdoing and had shown no remorse.
The verdicts depended in large part on testimony by law psychologist and staistician, Henk Elffers, who was used by the courts as an expert witness. According to Elffers, who was also interviewed on Dutch television, the probability of a nurse’s shifts coinciding innocently with so many unexplained deaths and resuscitations was one in 342 million.
Unfortunately, Elffers’ grasp of maths had been seriously flawed. For one thing, he had made a miscalculation involving the p-values (using his computation, any hospital employee who frequently changes hospitals would automatically become a suspect where there are unexplained deaths)! Even if Elffers had been correct about the one-in-342-million chance, his evidence should not have played such a key role in Lucia’s conviction. When someone wins the lottery, we don’t suspect that person of rigging the balls. Crucially, there had been six deaths over three years on one key ward where Lucia had supposedly gone on her killing spree, but in the three preceding years before she had arrived, there had been seven deaths. So, in fact, the arrival of the “serial killer” caused the death rate on this ward to do down!
Lucia de Berk’s diary also played a role in her conviction. On the day one of her patients (an elderly lady in a terminal stage of cancer) died, she wrote that she had “given in to her compulsion”. She wrote on other occasions that she had a “very great secret” and that she was concerned about “her tendency to give in to her compulsion”. These were apparently references to her passion for reading tarot cards. She pursued this interest secretly because she did not believe it appropriate to her employment in a hospital. The court dismissed her explanation that this was a “compulsion” and “perhaps an expression of fatigue” and decided it was evidence that she had euthanised the patients. Lucia’s daughter later stated in a television interview that some of her mother’s notes in the diaries were pure fiction which she had intended to use in writing a thriller.
In October 2008, the case was reopened by the Dutch supreme court, as new facts had been uncovered that undermined the previous verdicts. The deaths were deemed to have been natural, sometimes caused by wrong treatment or bad hospital management, and sometimes unexpected because of faulty medical diagnosis. Lucia was actually credited with saving lives through her prompt and efficient action during a couple of medical crises at the Juliana Child Hospital. The Public Prosecution capitulated, formally requesting the court to deliver a not guilty verdict. Finally, on 14 April 2010, the court delivered the not guilty verdict and Lucia was exonerated.
Lucia De Berk spent over six years in jail. On release from prison, she was penniless, denied unemployment benefits because of her unusual status. She was paralysed down one side following a stroke which she had suffered in 2006 when she had been told that her conviction would be upheld. It is a notorious example of misapplied statistics and it shows how dangerous a little knowledge can be.