Monday, February 13, 2012

Death at Birth: A Tragedy Caused by Overwork and Undertraining - and Technology

As a result of my recent post "Plaintiff's Trial Lawyers Are to Blame for EHRs That 'Tattle' on Doctors - And Harm Patients", a malpractice attorney, Patrick Malone in Washington, D.C., sent me a link to a tragic summary of a case his firm handled.

The story illustrates the toxic effects that health IT and other technology, especially when not "done well", can have on patient care.

I present it below with just a bit of comment, with permission of attorney Malone. Emphases also mine:

The Patrick Malone firm represented the family of Keren Kipoliongo, a baby girl who died in her mother’s womb just minutes before the successful Cesarean section delivery of her twin sister Gabrielle.
This tragedy occurred at Virginia Hospital Center in Arlington, Virginia in July 2008. At midnight one evening, Mrs. Kipoliongo and her husband presented to the labor and delivery unit at the hospital, as scheduled per their obstetrician’s instructions. The plan was to induce labor with the drug Pitocin (oxytocin) and monitor both babies during the labor with fetal heart monitors, to make sure they could withstand the contractions and be safely delivered by the normal vaginal route.
This was the first pregnancy for the couple, and they were excited about the prospects of having twins. They had set up their home in northern Virginia with a nursery with two cribs and two of everything else they would need.
The obstetrician whom Mrs. Kipoliongo had seen regularly throughout her pregnancy checked them into the L&D unit and then went to sleep in the doctors’ lounge. He instructed the nurse and a resident physician (doctor in training) to call him if needed.
Mrs. Kipoliongo was left alone for large stretches through the night. It wasn’t until a nursing shift change at 7 a.m. that she started to get close attention from the nurses. By that time, baby Keren was in bad shape as shown by fetal monitor patterns in which her heart beat was slowing ominously with the mother’s contractions. She had already shown throughout the overnight period a flattening or lack of variability from beat to beat. (The flattening of the heart rate often means the baby is not getting enough oxygen.) [Why was this not noted and acted upon promptly? Read on - ed.]
They finally brought the obstetrician in, and he decided around 8:15 a.m. to do a Cesarean section. Baby Keren was still alive at that point. Unfortunately she died some time over the next hour, because when the doctor finally cut into the mother’s womb after 9 a.m., he found that she was dead, with the umbilical cord wrapped tightly around her neck.
When our law firm investigated what had happened on behalf of the family, we found several disturbing facts. These included:
  • Despite the lack of reassuring signs on baby Keren’s heartbeat, the nurses failed to do any of a number of standard things that can improve the baby’s condition, such as turning the mother, giving the mother oxygen, and giving extra fluids.
  • The hospital overworked the nurse assigned to Mrs. Kipoliongo’s labor, giving her a second patient for whom she was also primarily responsible. Standards require one-on-one nursing during a labor that is as involved as hers was. The stimulating drug Pitocin requires close monitoring. In addition, Keren was under-sized (a condition called “intrauterine growth restriction”), and that is why they were proceeding with a scheduled birth rather than wait for labor to start on its own. (At autopsy, Keren was found to have completely normal body organs; her only issue was being small.) [I note that the overwork might have included having to interact with an EHR presenting a mission hostile user experience, as I've seen in critical care environments and have worked to specifically avoid when I was in charge as here - ed.]
  • Inspection of the computerized hospital record showed that this nurse, Julie Bates Gilpin, made most of her entries in the record hours after the purported observation time, including a whole series of entries after her shift ended. [I have seen this phenomenon myself; some EHR's allow entries even without a time stamp, and the only way to ascertain when entries were made is via discovery of metadata (automatically generated audit trails and other data that shows who manipulated data, and when. Hospitals often resist providing full metadata, despite Federal Rules of Civil Procedure (FRCP) governing e-discovery enacted almost 6 years ago - ed.]
  • We asked the nurse at a deposition why her entries showed such regular times of observation, at exactly 15 minutes after the hour, 30 minutes, 45 minutes, etc. She told us that was because she “rounded” on her patients every 15 minutes. But later we discovered – after winning a “motion to compel” in which the judge required the hospital to turn over records of the other patients Ms. Gilpin was monitoring that night (with names blacked out) – that she was also recording the same exact pattern of observations on her other patient, at 15 past the hour, 30, 45, and so on. So when we compared our client’s record with the other patient, it was clear that Nurse Gilpin was making computer entries showing that she was in two places at once. Again, most of these entries were made hours later and back-timed. [I also wonder, during the actual clinical encounter time was the nurse treating the actual patients or the e-Patient, i.e., the 'virtual patient' represented by the computer screens? - ed.]
  • We also wanted to know why the alarm on the fetal monitor went off 12 times during the night but was answered only one of those times by Nurse Gilpin. We discovered that these monitors have alarm reset buttons like a snooze button on an alarm clock. You can hit the button and turn off the alarm without doing anything else for the patient. Most of the time, the “snooze button” was hit by a nurse at the nurse’s station. What did this mean? The monitor was supposed to be tracking two separate twin babies. [Not exactly what I would call a "state of the art" alert design - ed.] The alarm would go off when the signal was lost for one baby. The nurses are supposed to reposition the patient so they can pick up the signal again. They didn’t do that for most of the times that the signal was lost. So it turned out that hours went by during the night when nothing was tracked on baby Keren’s heartbeat.
  • Even worse, we found out that when we took Nurse Gilpin’s testimony, she no longer remembered how to do her job on the labor and delivery unit even though she had worked there up until a few months before the time of her deposition. We asked her how a nurse reads the fetal monitor and what are bad signs for the baby, and she said she didn’t know.
  • And even worse than that (the wrongdoing kept piling up in this case as we continued our investigation), the resident doctor, whose name was Dr. Sahardat Nurudeen, and who was the only physician monitoring our client’s labor progress for nearly eight hours, also claimed at her deposition that she couldn’t read a fetal monitor tracing even though she was in training to become an obstetrician. (In fact, she had completed her full obstetric residency and was preparing for her board certification test at the time of her deposition.) We say “claimed,” because it could be that she just wanted to avoid answering tough questions at the deposition about individual segments of Keren’s monitor strips. Either way, it wasn’t good for high quality, accountable patient care.
We finally concluded that there were many chances during that night to save Keren’s life, but the understaffing and overwork and apparently poor training (how could a nurse or a young doctor forget how to read a fetal monitor, a basic part of their training?) [and the toxic effects of ill-conceived and ill-implemented technology - ed.] led to an unnecessary tragedy.

This story should give pause to technophiles in medicine who ignore technology's possible downsides, and the clinicians who are compelled to center their clinical lives around such technology (while the purveyors of that technology are often held harmless by hospital executives in violation of the latter's fiduciary responsibilities).

-- SS

4 comments:

Afraid said...

Oh my goodness, what a sad, sad story. I have seen similar, nurses using entries in the medical record recording events that never occurred. This keeps the nurse manager from giving the nurse heck based on some algorithmic quality indicator triggered if the EMR decides. And keeps the administrator from giving the nurse manager heck....

So the 15, 30, 45 ... minute entries are done even if it was simply impossible to do that much work on a given shift.

So the nurse knows better, knows she is over worked, but has no support from her admin. Same for the house doc. They can't fox the problem and they can't keep their jobs if they don't essentially lie about it.

When the crap hits the fan, the lower downs are whacked, not those that caused the problem.

Typical. Seems almost OK until someone dies.

InformaticsMD said...

Afraid February 13, 2012 8:53:00 PM EST writes:

I have seen similar, nurses using entries in the medical record recording events that never occurred. This keeps the nurse manager from giving the nurse heck based on some algorithmic quality indicator triggered if the EMR decides.

This 'misbehavior' is one of the unexpected adverse consequences of new information/communication technologies. as predicted through empirical research in the field of Social Informatics.

Except, it's unexpected only to health IT amateurs, i.e., the people who usually are in charge of designing and implementing these systems and the toxic policies surrounding the data that comes from them.

-- SS

Anonymous said...

The above post's legal case demonstrates what appears to be a classic example of a health IT workaround. Not sure it follows logically that the company that sold the EMR software used in the above case intended that nurses using the EMR improperly, in a manner that partially falsifies when tasks are actually done, be caught doing so in court. But I do agree that these nonsensical workarounds should be addressed for usability by the EMR company and the hospital. Forcing nurses to enter 'bad data' into EMRs for quality control purposes makes no logical sense. The total clinical workflow with EMR interactions should facilitate accurate and easy tracking of a healthcare practitioner's tasks. This, in my opinion, is a requirement for hospital process improvement efforts based on EMR data, and for proper use of EMR data in litigation. Whether this requirement is articulated properly to EMR companies, or discovered as such, is another big question.

Afraid said...

Absolutely SS, these unintended consequences are totally predictable if you know the environment of a hospital floor in today's world. In the world of woulda, coulda, shoulda, it is malfeasance not just nonfeasance.

They do a failure mode effectivity analysis in the airplane and automotive industries, why not HIT?

To look for and miss something is understandable, to not look at all for potential problems is beyond the pall of innocent behavior. Only very inexperienced employees would do that, and it would be incumbent upon their management to ensure that the inexperienced are consgtrainted.

In HIT, it seems the opposite, hire the young and inexperienced because they wont excercise due dilligence. It is a systemic oversight problem which cant really be attributed to an innocent mistake because other software/systems industries are required to perform and document potential failures and their mitigation.