Serious clinical
computing problems in the worst of places: an ICU
(Note:
This is a case account from several years ago when I was Director of
Medical Informatics. It was one of the major experiences that led to the realization
that clinical IT difficulties were not being adequately addressed anywhere, and
that Director of Medical Informatics positions were poorly structured and
inadequately empowered.)
In an East Coast USA tertiary hospital’s
medical intensive care unit, an electronic patient record and physiologic
monitoring system was desired by the medical and nursing staff to save time and
improve care. The MIS department was put in charge of software and hardware
selection and configuration. The
implementation of the ICU system was near complete upon the arrival of a
medical informatics expert, hired as “Director of Medical Informatics” largely
at the prompting of the Senior Vice President for Medical Affairs to resolve
major clinical IT problems and political infighting in the hospital.
The MIS department by policy only used
Compaq computers. Any other computer manufacturer's computer was deemed
"risky." There were stories of a bug in a model of Macintosh causing
packet storms and taking down a network. Other computer brands might also have
compatibility problems and require special drivers (not that this was ever
tested). Therefore, Compaq PC's were the "one
shoe" that would "fit all" needs in the organization.
The ICU rooms were very small. In order to
fit the standard-issue Compaq desktop computer into such rooms, along with a
standard CRT monitor, custom (expensive) cages were designed and ordered so
that the machines could be bolted to the ceiling of each room. Special
custom poles and cabling harnesses were also designed, ordered and installed,
custom-made to fit each room at great expense – over five thousand dollars per ICU room. On each pole was mounted a
standard computer CRT monitor, keyboard and ordinary computer mouse.
Issues such as air filtration, maintenance,
and contaminant circulation from the power supply fans of each machine, heat
generation from the CRT's, dirt accumulating on the mouse and keyboard, and
other ergonomic and medical issues were not considered. The informaticist,
asked to evaluate the system, was rather shocked by this, especially with the
rapid rise of resistant strains of pathogenic bacteria in the
The informaticist
was a physician who had excellent training in infectious disease (his chairs of
postgraduate medical education were infectious disease experts, one of national
standing), as well as a thorough understanding of electronic equipment. He had a clerkship in biomedical engineering,
was a microcomputer hardware expert, as well as a licensed radio
telecommunications specialist by examination, the amateur extra class, the
highest certification. He had
constructed complex electronic equipment from kits as well as from raw parts
and blank circuit board stock, including an infrared sensing heart monitor,
high powered shortwave transceivers, and computer equipment (when individual
parts required soldering onto blank circuit boards), as examples. This was known to the hospital staff.
His concern – a concern in his judgment not
just possible but highly likely –
was that the dust balls and dirt that accumulate in any actively fan-cooled
electronic equipment operated in contaminated or dirty environments might serve
as an excellent medium for colonization by pathogenic organisms, and that the
organisms could be spread by the air circulated by the fan. These concerns were ignored by MIS staff, the
infectious disease department and the head of the ICU (and by rumor also
ridiculed).
The informaticist
was shocked even more by an apparently cavalier attitude of the senior hospital
management towards these concerns. Even
at his recommendations, industrial form factor computers (small,
convection-cooled or low-air-circulation, and mountable to a wall), flat-screen
LCD monitors (compact, low-power-consumption, low-heat-producing),
easily-cleanable track pads, and other ergonomically better-suited solutions
were dismissed with the refrain "we don't support them". No such technology was purchased for
evaluation. In fact, as it turned out, the MIS system architects had actually never
before seen or even heard of a combination keyboard and track pad that had been
available off-the-shelf for a number of years from nearby CompUSA outlets - for
the grand sum of $50 retail.
Worse, the ICU system had ergonomic and
technical problems. A clinician hit his head on one ceiling-mounted computer. A
monitor nearly toppled and caused an injury. Redesign and relocation of the
hardware mounting cages and poles had to be performed at more expense. The ICU
room crowding and tight workstation ergonomics were not appreciated by busy ICU
personnel. The informaticist noted the ergonomic
problems and recommended solutions, but was resented since the technology being
suggested (e.g., industrial/clinical computing form factors, track pads instead
of mice, and flat screens from a number of non-Compaq vendors) were "not
supported by I.S." Recommendations
to periodically culture the dirt and dust inside the ceiling mounted computers
were also ignored.
The informaticist
sensed a motivation of the MIS department might be to protect the investment
they’d made for the custom installation of desktop computers on the ceilings of
ICU rooms, requiring the Rube Goldberg-style mounting cage, swivel pole,
interconnecting cables, and other hardware that had been acquired at great
expense to “force-fit” inappropriate computers into the ICU rooms for the
convenience of the MIS department. The informaticist also sensed a motivation of MIS to meet
remaining project deadlines and budgets, perhaps related to bonuses or to
advance their careers – at the expense of proper due diligence regarding
seriously ill ICU patients.
Unbelievable, thought the informaticist. The MIS culture - and the culture of
indifference to that culture by clinical leadership – might be endangering
patients, and his advice was being ignored and marginalized.
As an engineering specialist later commented
when reading this story:
If there is a specific application (example: a PC
requirement for an ICU room), then the specifications for what should be in the
room should be established first. If the requirements are that the equipment be
compact (small, out of the way somehow), this should be part of the
specification. MIS personnel and the users (clinicians who operate in the room)
should define what is required, including any "hardening" requirements
(CPU shielding from magnetics or radiation),
ventilation requirements, and display requirements (touch sensitive LCD
screens?).
Obviously, there needs to be consideration as to how
to engineer this system (I'm still thinking get the CPU out of the room, wire
in, that way there's only the monitor and input device/keyboard/mouse), but the
system should be designed and not just slapped in as the same CPU as what's in
the accountants office. More importantly, if the MIS group insists on a
standard CPU (Compaq, HP, IBM), then they are responsible (and in my view
liable) for guaranteeing that the system meets the standards required.
In pharmaceutical plants, there are areas which are
designed to be low or no particulate generating (Class 100 areas) where product
might be exposed to the environment. In these areas, all the equipment is
required to be of clean room design (low generation, no aerosols, etc.). When
equipment such as computers are required for the area, the engineer MUST find a
way to keep the system from putting material in the area. This means either
locating the CPU (with its fan and heat generating components) outside of the
room (with sealed cables coming into the room) or putting the CPU under an
exhaust hood which removes the exhaust air from the room. Either way, the
system must be designed to the needs of the area.
The system was also repeatedly
crashing. The informaticist
noted portable x-ray machines in frequent use in patient rooms, as happens in
ICUs, and realized x-ray scatter could be a cause of PC crashes. MIS personnel,
on the other hand, thought insufficient RAM might be causing the crashes
(empirically - there was no actual evidence for this) and were set to spend a
large amount of money to upgrade each machine.
When the informaticist
inquired if the computers used parity-checked RAM as a precaution against
memory errors caused by x-ray radiation or other factors (a reasonable question
for an ICU setting), it became apparent to him that the MIS personnel,
including the MIS director, did not know if the computers were so-equipped, and
worse, did not know what parity checking memory was or why it might be
needed in such a setting.
After this basic question and a lack of
response, the informaticist reported the potential
problem to the head of the ICU. The response of the MIS personnel was to shun
the informaticist and say non-complimentary things to
administration about the informaticist's role and
beliefs. The MIS personnel assured administration that all computers had the
parity feature, and that is was not very important anyway since only
"satellites in Earth orbit" needed protection from x-rays.
The informaticist
pulled a memory module from a machine and found it was an 8 bit, not a 9 bit,
module and therefore did not support parity checking. However, administration
did not believe the informaticist's concerns about
ergonomics or technical issues such as this, after the assurances and spin from
MIS.
It later turned out that neither x-ray
scatter nor memory quantity was causing these particular crashes. In fact, a
vendor software design deficiency was found to be causing the crashes. The
vendor had apparently designed the system so that each individual client
workstation was responsible for initiating printing of periodic reports on the
patient in its room, rather than centralizing this function at the server. This
introduced a manifold increase in potential points of failure and was found to
be the major source of the trouble. Significant modifications to make the
reporting function server-based solved the problem.
This was a design deficiency that the informaticist, who had developed significant software in
the past, would have recognized quickly as suboptimal for a critical care
setting such as an ICU in the first place. Also, the informaticist
still recommended parity memory for any critical care setting as a relatively
inexpensive legal due diligence. This advice was not heeded.
The informaticist's
credibility with administration had been tarnished by MIS. The informaticist's concerns about the technical abilities of
the MIS department to support equipment so closely involved in this critical
patient care setting were also resented by administration. Regarding organizational changes recommended by
the informaticist on clinical IT leadership, the chief
medical officer seemed more concerned with the possible effects on IT personnel’s
careers than with the effects of faulty computing on patient well being. The informaticist
found such priorities simply stunning.
Meanwhile, the system proved more costly to
support than MIS had predicted, requiring extensive development and
customization (over and above the inflated costs of the fancy mounting
accouterments), since it was immature, not entirely reliable, and
user-unfriendly. One very valuable system feature, the severity scoring system,
was never enabled. That feature might have allowed patients to be transferred
out of the ICU earlier, saving a significant amount of money.
The system struggled with proving a return
on investment, was nearly canceled after a year, and was given a
"try-it-for-one-more-year-but-prove-the-ROI" reprieve only after a
large degree of pleading and politicking by key personnel (including the informaticist). Its future became uncertain, plans to
spread the technology to other ICU's in the organization were canceled, and
administration had been needlessly "turned off" to this type of
technology.
On the whole, the principal actors in the
case described above were petty, small-minded, self-serving, and narcissistic at
the expense of patient safety.
The dysfunction in this malignant hospital environment (madness,
actually) was one factor in the informaticist later resigning this role and
taking a position in industry.
**************************************
Addendum: the informaticist’s fears have since been well-confirmed. It would have been helpful to patients if the
hospital had actually taken a scientific approach and investigated the concerns
of its own expert.
Dirty PCs
fuel hospital super bugs
Hospital
Computer Keyboards May Spread Danger
Computer equipment
could help MRSA spread
The informaticist can
only wonder if the iatrogenic infection, morbidity, and mortality rates had
increased in patients who’d spent time in MICU when the computerized
physiologic monitoring system that is the subject of this story was in
operation.
**************************************
It is worth reviewing the first linked story
above, from The Register, a
Dirty PCs fuel hospital super bugs
by Peter Warren, The Register
Published Thursday
26th May 2005 10:23 GMT
Scientists in the
Researchers working
in hospitals have found that the deadly
bacteria clings to the keys of the computer keyboards used to update
patient records and therefore can re-infect the hands of staff even after they
had washed their hands.
There were 55 deaths
from MRSA in
The
The researchers also
found that cleaning IT equipment with soap and water was not enough to remove
the bacteria.
The only way to
clear the infection from the keyboards, according to Dr Gary Norskin from the
"A computer keyboard is like any other
surface in a hospital and has to be sterilised,"
said Norskin,
The
Computers quickly become magnets for airborne dust and
bacteria-harbouring dirt, which builds up on their
internal cooling fans. The fans represent
a further health hazard because of their potential to blow that same dust
around a ward. "Anything that can
put bacteria into the air is a risk," said Norskin.
"If you bang into a computer and disturb that dust you can effectively
create a dust cloud."
Doctors at the
"We have
computers everywhere because our goal it to have a paperless hospital and to
have computers everywhere a patient goes,” said Jeff Jones,
"Computers harbouring bacteria is a very big concern of ours because
we have computers only three feet away from patients in our operating rooms and
we can't have dirt in places like that," he said, adding that tuberculosis
is another potential risk from technology as it is the world's number one
airborne disease.
"We did
experiment with waterproof keyboards that you can wipe clean but found out that
they were generating a lot of keystroke errors that could have been just as
dangerous for patients," said Jones.
A spokesman for the
Dr Paul Grime, the
British Medical Association's spokesman on MRSA commented: "If computers
and keyboards are going to be next to people's beds then this is something that
we have to be aware of because this equipment is no different from any other
hard surface in a hospital but the key to this is hand hygiene and staff have
to get used to washing their hands before and after touching a patient.”
Such health risks
have not gone unnoticed by the computer industry which has moved quickly to
respond to the threat created by technology in hospital.
.