0:02

One chance in 342 million. The

0:07

law professor kept staring at the number,

0:10

gaping in shock. There

0:13

had been a string of deaths at

0:15

two local hospitals, and according to his

0:17

calculations, they were almost

0:19

certainly murders. The

0:22

suspect was a 40-year-old nurse

0:24

named Lucia DeBerk. She

0:27

was tall and blonde and worked shifts

0:29

part-time at a children's hospital. There,

0:32

she had been on hand for

0:34

several unexplained deaths. The

0:36

staff quickly grew suspicious. Still,

0:39

the staff knew that everything might just be

0:42

due to chance. Perhaps DeBerk

0:44

was just unlucky in being around

0:46

when the children died. That's

0:48

where the law professor came in. He

0:52

had an undergraduate degree in statistics,

0:54

so he started running numbers. The

0:58

results were damning. According

1:01

to his figures, there was just one chance in

1:03

342 million that DeBerk was a victim of

1:08

bad luck. Put

1:10

another way, it meant with near certainty

1:12

that she was a serial killer. The

1:15

worst serial killer in Holland's

1:17

history. And the

1:19

professor was determined that his math

1:21

would bring her to justice. From

1:30

the Science History Institute, this

1:32

is Sam Keene and the Disappearing

1:34

Spoon, a topsy-turvy

1:36

science-y history podcast, where

1:39

footnotes become the real story. Lucia

1:44

DeBerk had a friendly relationship with her

1:47

fellow nurses, at least on

1:49

the surface. Many

1:52

of them gossiped about her behind her

1:54

back. They made fun of

1:57

her hippie clothing and her obsession with tarot

1:59

cards. In fact,

2:01

De Burke worked at an elderly

2:03

care hospital sometimes, and often read

2:06

patients' fortunes on duty, a big

2:08

no-no. Still, one

2:10

nurse in particular thought De Burke

2:12

was not just flaky, but dangerous.

2:17

Things came to a head with the

2:19

sad case of baby Amber. From

2:22

birth, Amber had suffered from a heart defect

2:24

and had fluid in her lungs. She could

2:26

eat only through a tube. In

2:29

July 2001, at four months

2:31

old, Amber underwent heart surgery.

2:35

She was also prescribed a drug

2:37

called digoxin, which treats congestive heart

2:39

failure. Both treatments seemed to

2:41

help. But on

2:44

September 3, Amber began vomiting and

2:46

had diarrhea. De

2:48

Burke was working that day and brought her

2:50

to pediatricians around 1 a.m. for an examination.

2:54

Despite her bad day, the doctor said

2:56

she looked fine. But

2:59

around 3 a.m., her breathing

3:01

stopped and her heart flatlined.

3:04

The pediatricians tried reviving her, but

3:07

Amber was pronounced dead around 3.30. Her

3:11

sudden death left the medical staff

3:13

baffled and heartbroken. What

3:15

had happened? One

3:18

nurse was especially upset, partly

3:20

because the case reminded her of another

3:22

recent death, that

3:24

of a baby named Okmot. The

3:27

previous January, Okmot had mysteriously

3:30

overdosed on sedatives and died

3:32

after a surgery for stomach

3:34

problems. Just like

3:36

Amber, Okmot was seemingly doing well

3:38

before a sudden crash. And

3:41

Lucia De Burke was working the day he

3:44

died. Soon rumors

3:46

of foul play were flying around

3:48

the ward. The suspicious

3:50

nurse finally blew the whistle and

3:52

told the chief pediatrician. The

3:55

pediatrician had never liked De Burke

3:57

and quickly recalled three other instances

4:00

more unexplained deaths when

4:02

DeBerk was working. Initially,

4:06

doctors had ruled all those

4:08

deaths natural if unexplained, but

4:11

the chief pediatrician now overruled those

4:13

judgments. She reclassified

4:15

the deaths as suspicious and

4:18

told the hospital's director. The

4:21

director started pulling data about DeBerk's

4:23

shifts. What he learned

4:25

sickened him. DeBerk

4:28

had worked at the hospital three years at

4:30

that point, and in her

4:32

134 shifts, there

4:34

were eight unexplained deaths.

4:38

All the other nurses had worked 887 shifts combined, and

4:42

not a single unexplained death had

4:44

taken place. The

4:47

director soon called the elderly care

4:49

hospital where DeBerk worked. He

4:51

asked officials there to comb their records. Five

4:54

more unexplained deaths popped up.

4:58

Upon hearing this, the children's hospital

5:00

director called the police. When

5:03

confronted, DeBerk professed shock. She

5:06

swore that all the deaths were natural. Her

5:09

presence was just random chance, bad

5:11

luck. But the

5:13

police didn't buy it. They

5:16

began investigating and quickly turned up

5:18

more damning evidence. Again,

5:21

DeBerk had worked at the children's hospital

5:23

for three years by then. For

5:25

comparison, the police looked into how many

5:27

unexplained deaths had occurred in the three

5:30

years prior. They found zero.

5:34

Zero unexplained deaths in three

5:36

years. Then DeBerk shows

5:39

up and eight occur? It

5:41

seemed awfully suspicious. The

5:46

police also seized DeBerk's diary. In

5:49

November 1997, she wrote about how she, quote, gave

5:52

in to her compulsion. And

5:55

according to hospital records for that day,

5:57

an elderly cancer patient of hers had

5:59

died. Once again,

6:01

the patient had been stable and her

6:04

doctor recalled being surprised by the sudden

6:06

death. The

6:08

police asked DeBerk about her so-called

6:10

compulsion. She claimed that

6:12

her compulsion was the occult and

6:15

admitted that she had read tarot

6:17

cards to the patient that day

6:19

against hospital rules, but

6:21

the police knew better. Her

6:23

real compulsion was murder.

6:28

Even the most damning evidence of all

6:30

came through. Baby

6:32

Amber had been autopsied, and

6:35

the hospital had kept as evidence the

6:37

gauze used to soak up her bodily

6:39

fluids. Testing the

6:41

fluids revealed suspiciously high levels of

6:44

the drug digoxin, a dozen times

6:46

higher than normal for the amount

6:48

prescribed, and Lucia DeBerk had

6:51

a key to the cupboard where the drug

6:53

was kept. In

6:55

December 2001, the police arrested DeBerk

6:57

and charged her with 13 counts

7:00

of murder. The

7:02

tabloids went wild. A

7:04

nurse, a supposed life saver,

7:06

was instead preying on babies

7:09

and old people. What

7:11

a monster. The

7:13

newspapers dug up dirt on her personal life

7:16

as well. DeBerk had lived in

7:18

Canada as a young woman and had worked as

7:20

an escort there. She had

7:22

also falsified her credentials to get into

7:24

nursing school. During

7:27

all this, DeBerk maintained her innocence,

7:29

but the police soon brought in the law

7:32

professor mentioned before who had a degree in

7:34

statistics. He ran some

7:36

numbers and got the figure of

7:38

one in three hundred and forty-two

7:40

million, a microscopic chance

7:42

that DeBerk was simply the victim

7:44

of bad luck. The

7:47

case shocked the entire Dutch nation.

7:50

He also talked about bringing back public

7:53

hangings or the firing squad. No

7:55

punishment would be too harsh for the likes of

7:57

DeBerk. As

8:01

the case moved to trial, however, something

8:03

funny happened. DeBerk's lawyers

8:06

took a closer look at the eight

8:08

alleged murders at the Children's Hospital. While

8:11

it turned out that DeBerk wasn't actually working

8:13

during two of the deaths, she

8:15

had called in sick once and another time was

8:18

out of town. But

8:20

there was still plenty of evidence to prosecute. The

8:22

diary, the high concentrations of drugs

8:25

in baby Amber and baby Ahmad.

8:29

DeBerk was swiftly convicted on multiple

8:31

counts of murder, and both appeals

8:33

failed. Days

8:35

after the second failed appeal, in

8:37

2006, DeBerk suffered a stroke in

8:39

her prison cell. The

8:42

guards on hand assumed she was faking it

8:44

to win sympathy. They let her

8:46

lie there for 10 hours before calling a

8:48

doctor. As a

8:51

result, DeBerk ended up partially paralyzed

8:53

and unable to speak. But

8:55

honestly, people felt little sympathy for

8:57

a serial killer who preyed on

8:59

children and old people. After

9:03

the stroke, DeBerk was shipped to prison and

9:05

left to rot for the rest of her

9:07

life. Except

9:09

things didn't quite work out that way.

9:12

Less than five years later, the

9:14

most prolific serial killer in Dutch

9:16

history would walk out of

9:18

prison a free woman. Because

9:21

despite everything you've just heard, Lucia

9:24

DeBerk never killed a

9:26

single soul. Lucia

9:35

DeBerk owes her salvation to

9:37

two tenacious siblings. One

9:39

was Mehta D'Nou, a Dutch doctor.

9:42

D'Nou's brother was married to the

9:44

chief pediatrician who had unearthed the

9:46

suspicious cases at the children's hospital.

9:49

D'Nou saw how obsessed her sister-in-law

9:51

was with the case and she

9:53

got interested herself. And

9:55

why not? It involved a

9:58

serial killer. It seemed fascinating. But

10:01

when Dr. Danneux began reviewing the medical

10:03

files in the DeBert case, she felt

10:05

a pit form in her stomach. In

10:08

the majority of the deaths, there was

10:11

no evidence whatsoever of foul play. No

10:14

poison, no violence, nothing. The

10:17

only supposed evidence was the fact that

10:19

DeBert was present. The case

10:21

seemed flimsy. Now

10:24

there was baby Ahmad, who had

10:26

overdosed on sedatives. Danneux

10:28

noticed that he'd been prescribed alarmingly

10:31

high doses by his doctor. There

10:34

were also instructions to give him even

10:36

more sedatives if he appeared restless. From

10:39

what Danneux saw, an overdose was

10:41

almost inevitable. Danneux

10:43

approached her sister-in-law with these findings,

10:46

but her sister-in-law told her to back

10:48

off. She knew DeBert, knew what

10:50

a monster the woman was. A

10:53

huge fight erupted, and Danneux and

10:55

the sister-in-law stopped speaking. Danneux's

10:57

brother cut off contact as well. Luckily,

11:01

Danneux had another brother, Thanh

11:03

Dirksen, a philosophy professor. When

11:06

Danneux mentioned the case to him,

11:09

he started sleuthing around as well,

11:11

focusing on baby Amber. Again,

11:14

Amber had a dozen times more digoxin

11:16

in her blood than she should have.

11:19

But it turns out there's a natural

11:21

molecule in the body that resembles

11:23

digoxin. Indeed, the drug

11:25

mimics this molecule. Unfortunately,

11:28

the police used a cheap test

11:30

that could not distinguish between the

11:32

natural molecule and the drug. Dirksen

11:35

convinced the police to try a more

11:37

sensitive test that could distinguish. And

11:40

according to that test, she had only three and

11:42

a half times more than she should have. And

11:45

even that figure was artificially high. Remember,

11:48

the test involved fluids from the gauze

11:51

left over from Amber's autopsy. That gauze

11:53

had been sitting around for a while,

11:55

and some of the liquid had evaporated.

11:58

That increased the concentration of the liquid. of

12:00

the drug remaining behind, pushing the number

12:02

too high. In

12:05

reality, the concentration in amber was

12:07

normal. Next

12:10

up, Tom Dirksen began dismantling the

12:13

statistical case against De Burgh. During

12:16

his research, Dirksen came across another

12:18

sad death at the children's hospital.

12:21

It involved a baby named Kamal. He

12:24

had overdosed on sedatives. The

12:26

cases looked identical. So

12:28

why wasn't De Burgh charged in Kamal's

12:31

case? Simply because De

12:33

Burgh wasn't working that day. That

12:36

told Dirksen something, that

12:39

the people investigating De Burgh had been

12:41

biased in gathering their data. What

12:43

they should have done is pulled up

12:45

a list of all unexplained deaths, tried

12:48

to determine if each one looked

12:50

suspicious, and then figured out

12:52

whether De Burgh was present. Instead,

12:54

the hospital staff did the

12:56

opposite. They determined when De

12:59

Burgh was present, then started

13:01

fishing for reasons why the unexplained

13:03

deaths looked suspicious. It

13:06

was a textbook example of

13:08

confirmation bias, letting your

13:10

prior beliefs sway your interpretation.

13:15

Dirksen found other problems too. Again,

13:17

the police searched for unexplained deaths in De

13:20

Burgh's ward for three years prior to her

13:22

starting there. They

13:24

found zero. Unfortunately,

13:26

the name of the ward had

13:28

changed right before De Burgh joined.

13:31

So in the database, it just

13:33

looked like there were no unexplained

13:35

deaths, because the police were searching

13:37

under the wrong name. Dirksen

13:41

soon got the records for the

13:43

correct ward and found seven unexplained

13:45

deaths in the three years prior to De Burgh's

13:47

starting. Overall then,

13:49

there were seven unexplained deaths before

13:52

her arrival, eight afterward. And

13:55

to remember, De Burgh was not present for two

13:57

of those eight, so in the third place, it

13:59

was the first time. Three years before a serial

14:01

killer arrives, there are seven unexplained

14:04

deaths. Then the serial killer

14:06

shows up, and there's just

14:08

six? Does that make sense?

14:12

Equally bad, investigators had examined

14:14

data only from the pediatric

14:16

and the elderly care hospitals.

14:19

They ignored three other hospitals

14:21

where Daburk worked, where there

14:23

were no suspicious deaths. And

14:26

think about it. What segments of

14:28

the population are most prone to dying?

14:31

Infants and old people. That

14:34

alone made Daburk's death count seem far

14:37

higher than normal. With

14:39

this better data, Dirksen approached

14:42

some actual, professional statisticians. They

14:45

reran the numbers on the likelihood of

14:47

Daburk being present for so many unexplained

14:49

deaths. And the number

14:51

they got dropped dramatically. From

14:54

one chance in 342 million to one in 1200. Now,

15:00

one in 1200 still might seem

15:02

suspicious, but there were

15:04

a quarter million nurses in Holland. So

15:07

in a normal year, a few hundred of them

15:09

would be present for as many deaths as Daburk.

15:12

It was indeed all just

15:15

bad luck. Dirksen

15:20

and his sister Danu began

15:22

evangelizing about Daburk's innocence. It

15:25

was not easy to get people to listen. Even

15:28

the Dutch Minister of Justice refused to

15:30

listen at first. An

15:32

occult-loving, serial-killing nurse was

15:34

a much juicier story

15:36

anyway. But

15:38

the siblings' doggedness paid off.

15:41

Through media appearances and grassroots organizing,

15:44

they won Daburk a new trial

15:46

in 2010. After

15:48

her stroke, she still struggled to walk,

15:51

but she had recovered enough of her speaking

15:53

ability to testify on her own behalf. During

15:57

the trial, her defense lawyers hammered

15:59

the prosecution. pointing out

16:01

the biased stats and the sloppy drug

16:03

tests. And, incredibly,

16:05

the prosecution agreed. As

16:10

the trial was wrapping up, they

16:12

actually asked the judge to free

16:14

DeBurke. Even the prosecutors

16:16

did not believe in her guilt anymore. As

16:20

a result, DeBurke was acquitted of all

16:22

charges in April 2010. She

16:25

bravely limped out of prison under her

16:27

own power. A free woman,

16:29

at last. Sadly,

16:31

DeBurke's case is far from being

16:33

the only miscarriage of justice due

16:35

to bad statistics. In

16:38

fact, I've put together a

16:40

bonus episode at patreon.com/disappearing spoon

16:43

about other instances. Most

16:46

of the accused are young mothers, who not only

16:48

endured the deaths of their children, but

16:51

were then charged with murder and

16:53

suffered public hatred. They're

16:55

terrifying cases of math gone

16:57

wrong. That's patreon.com/disappearing

17:00

spoon. So

17:03

what lessons should we take from the DeBurke case?

17:07

First, medicine is mysterious.

17:10

Patients doing well one day suddenly die all

17:12

the time. It's

17:14

sad, but it's no one's fault.

17:18

Second, there are statistical lessons.

17:21

Gather your data fairly without

17:23

bias. And when

17:25

examining data, don't let your

17:27

prior beliefs sway your interpretation.

17:30

Third, be careful of

17:32

self-righteousness. DeBurke's accusers thought

17:35

that they had uncovered a serial killer,

17:37

and they were so determined to root

17:39

out the evil that they

17:41

lost all objectivity. Fourth,

17:44

although these were not factors in

17:46

the DeBurke case, there are other

17:48

things to consider with hospital deaths.

17:52

Sometimes nurses are just bad,

17:54

clumsy, or whatever. That

17:56

doesn't mean they're murderers. nurses

18:00

get the toughest cases, the people

18:02

most likely to die anyway. So

18:05

a higher death rate is an artifact, not

18:07

a crime. Even time

18:09

of day can affect things. People

18:12

on overnight shifts make more mistakes

18:14

because they're sleep deprived, and

18:16

specialty doctors usually aren't available overnight,

18:19

which again leads to higher death

18:21

rates. Heck, some

18:23

nurses just work more, double

18:25

shifts. And of course,

18:28

the more you're around the hospital, the more

18:30

deaths there will be on your watch. You

18:33

have to be very careful with statistics.

18:38

But the most important lesson is the

18:40

hardest one to swallow. That

18:42

human beings are simply bad at

18:45

statistics. We have poor

18:47

intuitions for numbers and coincidences.

18:50

We hate randomness, especially

18:52

around death. We

18:55

want someone to blame. So we invent

18:57

patterns and see things that are not

18:59

there. Unfortunately,

19:01

these errors are very hard to

19:03

root out because they're built into

19:05

our brains. Remember,

19:08

the law professor who calculated

19:10

the one in 342 million

19:12

figure was not some bumbling

19:14

fool. He had a degree

19:16

in stats, yet he still

19:18

made ghastly, life-destroying decisions. He

19:22

and others were so convinced that they had

19:24

uncovered a monster that they never

19:26

stopped to consider if they were

19:28

doing something monstrous themselves.

19:43

This is the Disappearing Spoon podcast brought

19:45

to you by the Science History Institute.

19:48

Be sure to check out

19:50

their in-depth science history podcast,

19:52

Distillations. You can find it

19:55

anywhere you get your podcasts

19:57

or on their website, distillations.org.

20:01

You can find more incredible stories

20:03

in my books or at samkeen.com.

20:07

You can also book me as a speaker

20:09

there for your school or event. If

20:12

you like this podcast, please

20:14

spread the word and support

20:17

it at patreon.com/Disappearing Spoon. You

20:20

can also get bonus episodes and

20:22

signed books at Patreon. This

20:25

episode was written by me, Sam

20:27

Keene. It was produced

20:29

by Mariel Carr, Rigoberto Hernandez

20:31

and Sarah Kaplan. Our

20:34

sound design, scoring and mixing is

20:36

by Ro Home Productions. Ro Home's

20:39

creative director is Alex Lewis.

20:42

John Myers is Ro Home's executive producer.

20:46

Thanks for listening. You