Recently, one of our stalwart Covers wrote a post Steel, I think you'd be a great Ombudsman, and the Nursing Home world really needs the oversight. My 90 year old mother was in one for 6 weeks a few years back, doing rehab on a fractured knee. The stuff that happened in 6 weeks made both if us have a horror of her ever having to go back. I'm quite sure that she'd commit suicide before she ever went back.It seems to me every day I read another scary story about health care "glitches" throughout the world. Just today there were stories about problems in CanadaInfection Kills 100 Quebec Patients
Aug 5, 1:28 AM (ET)
TORONTO (AP) - A bacterial infection commonly found in hospitals and nursing homes has been blamed in the deaths of 100 patients in the past 18 months in a single Quebec hospital, an infectious disease expert at the facility said on Wednesday.


Dr. Jacques Pepin said cases of Clostridium difficile at University Hospital in Sherbrooke, about 90 miles east of Montreal, have been steadily increasing.

Pepin has called for government action to prevent more outbreaks. The bacteria can cause diarrhea and colon inflammation, and often occurs after a patient has taken antibiotics. Outbreaks of C. difficile have killed almost 90 patients at several other hospitals in Montreal and Calgary, Alberta. And more recently, a patient died of the disease in a hospital in Newmarket, just north of Toronto.

Last week, I readStudy: Hospital errors cause 195,000 deaths
Report doubles earlier Institute of Medicine estimate
Wednesday, July 28, 2004 Posted: 10:08 AM EDT (1408 GMT)

WASHINGTON (Reuters) -- As many as 195,000 people a year could be dying in U.S. hospitals because of easily prevented errors, a company said Tuesday in an estimate that doubles previous figures.
Lakewood, Colorado-based HealthGrades Inc. said its data covers all 50 states and is more up-to-date than a 1999 study from the Institute of Medicine that said 98,000 people a year die from medical errors.
"The HealthGrades study shows that the IOM report may have underestimated the number of deaths due to medical errors, and, moreover, that there is little evidence that patient safety has improved in the last five years," said Dr. Samantha Collier, vice president of medical affairs at the company.Over 30 years ago, there was a fantastic movie satire called
The Hospital
Madness, Murder and Malpractice.
Written By Paddy Chayefsky
Release Year 1971
George C. Scott as Dr. Herbert Herb Bock
Diana Rigg as Miss Barbara Drummond
Barnard Hughes as Edmund Drummond
which capitalized on how each small error could be compounded into an ultimate disaster. At the time, I thought it was a black comedy. Today, it seems to have been the blueprint for healthcare practice throughout the world.

So much for the preliminary setup.
The conditions for debate:

Please consider everything STRICTLY from a Quality Professional viewpoint.
Personal attacks on individuals or institutions are forbidden.
No personal anecdotes - only items from reputable news sources if any "horror stories" are deemed necessary to make a point.
No attacking other posters - if you object, use the report button (little triangle in upper right corner of each post) - let the Moderators sort it out
Now the questions for comment:

Is the seeming crisis in healthcare quality real or merely a matter of perception?
If not real, how should the healthcare industry correct the misperception?
If real, what would be your guess as to root cause or common cause (that is, where would you start your investigation for root cause or common cause?)
Since the situation seems to pervade both socialized and privatized medical systems, is it fair to eliminate the payment system from the primary consideration for cause?
Is Deming right? Is this situation really a management responsibility, not employees?
Finally, are there any public reports of Health Systems (single location or geographic region) where the situation is under control?
Remember, no political diatribes, let's just discuss this from a Quality viewpoint.

I'm wondering how this can be brought into context. For example, in looking at fatal auto accidents we can estimate total miles driven and come up with a relationship. I guess in medical it would be total patients treated.

I'm wondering how this can be brought into context. For example, in looking at fatal auto accidents we can estimate total miles driven and come up with a relationship. I guess in medical it would be total patients treated.Your question raises another point: severity of effect of errors (nonconformances.) Just as we would not equate surface scratches on an automobile paint job with a faulty fuel line connector which drips gasoline on a hot engine, then so, too, we would not equate a one hour wait for an X-ray because of misscheduling with leaving a surgical tool inside a person's body after an operation.

Good question. Good addition to the debate questions, especially #1 "perception."

Another factor would be limitations by specialty. Nursing home vs. emergency room vs. surgery (pre-op, during, and post-op) vs. general care ward (e.g.: diseases) vs. etc.

Another factor would be limitations by specialty. Nursing home vs. emergency room vs. surgery (pre-op, during, and post-op) vs. general care ward (e.g.: diseases) vs. etc.By Jove! I think you've caught the spirit of the debate!
By all means, let's categorize the issues. Ideas for a grid?

Now you're trying to make me think... Let's get some more replies.

Now you're trying to make me think... Let's get some more replies.Here's a sample of the kind of POSITIVE statements I hope our Covers can produce. This outlines some problems and reports/suggests tactics to ameliorate the problems
http://www.intelihealth.com/IH/ihtIH/WSIHW000/8124/29703/350690.html?d=dmtContent

Drug Resource Center
(medical content reviewed by the Faculty of the Harvard Medical School)
At The Hospital
Here are some safety tips to keep in mind when given medications at the hospital ...

There is perhaps no more vulnerable time in your life than when you are ill and hospitalized. Hospitals are alien environments to most people. They are busy, noisy place, populated by people who seem to speak a totally separate language. That is perhaps why the possibility of hospital errors is of such concern to the American public.

Most studies of hospital errors have shown that medication-related errors are the most frequent type. Although it is not obvious, the stepwise process from the time your health-care provider orders a particular medication to the time you receive it requires the actions of many people. This process must be perfect each time, despite the fact that it occurs thousands of times a day in the average hospital.

In 1995, a group of researchers from the Harvard Medical School and the Harvard School of Public Health published a study in the Journal of the American Medical Association highlighting the frequency of such errors, the types of errors committed and, more importantly, the causes of the errors. Contrary to popular opinion, medication errors are not caused by individuals but are caused by failures in what is a complex medication delivery system.

Fortunately, not all errors in medicine translate into a serious reaction or death, and many have little effect at all. Nonetheless, health-care providers, the hospital community and the government are dedicated to instituting systems to prevent and detect errors before they happen, similar to many other industries. Studies show that such measures, when instituted, can significantly reduce the numbers of hospital errors — sometimes by as much as 50 percent.

Here are some examples of how hospitals are working to reduce medication errors specifically:
New computers. Hospitals are installing computerized systems for all medication orders. Computerized ordering systems have been shown to dramatically reduce medication errors. These systems work by eliminating the risk caused by illegible handwriting on a prescription. They also provide your health-care provider with very important information at the exact time it can do the most good; for example, a computerized system can reminding your health-care provider that you have an allergy to a particular medication or that you have a medical condition that can affect the way a medication will work. Unfortunately, these computerized system are very expensive, and only a minority of hospitals actually use them.
More pharmacists. Hospitals are putting more pharmacists on the patient care units. These pharmacists work very closely with other health-care providers. Harvard researchers showed that by increasing the number of pharmacists involved in direct patient care and improving the communication between the pharmacist and other health-care providers, the number of medication errors can be reduced by more than 50 percent.
A review team. Having teams of health-care providers (physicians, nurses, pharmacists and others) examine every reported medication error can help prevent others from occurring. By using this system, hospitals concentrate on the real cause of errors and not on the people who make them. For example, when it was recognized that simple math errors could contribute to errors in administering medication intravenously, one hospital helped design special infusion pumps that tell the health-care provider when the dose of a medication being administered is too high.
What You Can Do
Patients play a very important role in reducing the number of medication errors in the hospital. There are several ways you can participate in making your hospitalization safer. Above all, be an informed and aware patient.
Carry a list of the medications you take. Include the name and number of your health-care provider, as well as your pharmacy. Also include your medication allergies on the list. Update your list each time you change your medications or doses. This list will be important, as you may be meeting many new health-care providers in the hospital.
Ask questions during your hospital stay. It is important that you know what test or procedure you're having and why, as well as what medications you are taking and why. This helps to avoid mistakes such as being given the wrong test or procedure or being given a medication that you have reacted to in the past. Don't assume that your health-care provider knows what is going on, especially when there are many health-care providers consulting on your case. If you didn't hear that you were supposed to have a test that you are scheduled for, ask your health-care provider why there was a change in plans.
Before you are discharged from the hospital, make sure you understand which new medications have been prescribed and which medications have been stopped. Also question why you need each new medication. Sometimes, medications given in the hospital, such as a stool softener or vitamins, are not really necessary when you are home and up and about.
If drug companies wanted to show some effort in this operation, perhaps they might consider offering patient kits to all doctors which include a wallet card for all medications and known allergies, drug interactions, etc. - the allergies part could even be a checklist. This would have greater value for doctor and patient than prescription pads, pens, and junkets to the Bahamas for "continuing education."

How about a similar kind of kit for all patients and "their responsible adults" when they enter hospital?

Anyone know of such kits or efforts by health care providers to supply them to patients?

Dear Wes:

Your posts today (and an earlier one yesterday on PPM Calculation for Bulk Materials) prompted to me prepare the attached flow chart for the Health Care Industry.

What happens when a patient comes to hospital or a health care facility? We can envision a simple four step process as indicated here. First, the patient is admitted. The patient may be healthy, entering the facility, or may be sick. The second step is diagnosis. This may prove to be either correct or incorrect. The next step is treatment. Again errors can occur in the administration of the treatment (wrong prescriptions are filled out, for example, or in extreme cases, as we have heard, the wrong leg is amputated, etc.). The fourth step is the outcome. Even if there is no errors in the diagnosis or treatment steps, the outcome may still not be the desirable one. Many factors affect this outcome (age of the patient for example, severity of the illness, lack of proper medication, etc.)

We could assign or arrive at a probability of the occurrence of an error for each step (by studying many cases where similar illnesses are being treated, with many different patient groups). The overall probability of the occurrence of a medical error would be the product of the probabilities at each step.

Thanks for initiating this thread. I have been interested in this problem but find a real lack of "good" data. Some articles published in medical journals give similar flowcharts but they are quite different from the one given here. A few years ago, there was talk about an "epidemic" of medical errors. I haven't seen much news coverage lately. So, I guess, all is well. May be, I am wrong. I am certainly not too close to this and am not a part of the Healthcare Industry. Anyway, these are just some preliminary thoughts. Now fire away and let's see how we could IMPROVE this further.

Charmed :) :thanx:

P. S. I went back and read all your posts more carefully - I had too focussed on the flow chart idea that hit me. From what you have compiled, it looks like medical errors are back in the news.

Dear Wes:

Are you directly involved with the HealthCare industry now? If so, do you know where more detailed data from the recent study (195,000 death per year due to medical errors) by Lakewood, Colorado-based HealthGrades Inc) could be obtained. I did try to analyze the data generated earlier by the Institute of Medicine (98,000 deaths per year).

Again, I would like to call attention to the fact that the numbers 195,000 deaths per year or 98,000 deaths per year are "ratios" which is not the same as a "rate". The speed of a moving vehicle is the ratio ds/dt where ds is the incremental distance covered in the incremental time interval dt. This ratio is not the same as the ratio s/t which gives the "overall" or average speed of the vehicle, say during a trip when x = 300 miles and t = 5 hours.

The "average" speed s/t = 300/5 = 60 mph. The speeds during the trip could be either higher or lower than this value. Likewise, more generally, the ratio y/x must be distinguished from the slope dy/dx which also a ratio. But, this ratio gives the 'rate' of change. The rate of increase of medical errors dy/dx is not the same as the ratio y/x. We must determine the functional relation y = f(x) where y is the number of deaths caused by medical errors and x is the BIG UNKNOWN, or the variable that is responsible for the death. The problem is similar to that discussed in many quality and Six Sigma books. We are told to determine the X that is responsible for the Y, with Y being the number of defects produced by some process of interest.

Even the very computation of the y/x ratio is problematic in the case of medical errors. I was not satisfied with the data that I found in the earlier study (IIM study, going back to 1999, I believe). Not being a medical field professional, I did not pursue it much further, other than trying to call attention to the difference between y/x and dy/dx with some professionals who I know personally.

If you, or anyone you know, can help me get the data for all the 50 states, from the recent study, I would love to study it. I have studied similar data, that is readily available, for traffic accidents. Unfortunately, I have never been able to obtain the "raw" data on medical errors, although I did contact some of the authors of these studies in the past. With my warmest regards.

Charmed :)

P. S. I am doing this strictly to test some theoretical ideas, such as the wider application of the work function. Such studies and the testing of new theoretical models to understand such data, obviously, have wider societal consequences. I look forward to your response.

http://elsmar.com/Forums/images/smilies/caution.gif Caution - stick to the five basic questions or variations or face "editing"
the questions for comment:

Is the seeming crisis in healthcare quality real or merely a matter of perception?
If not real, how should the healthcare industry correct the misperception?
If real, what would be your guess as to root cause or common cause (that is, where would you start your investigation for root cause or common cause?)
Since the situation seems to pervade both socialized and privatized medical systems, is it fair to eliminate the payment system from the primary consideration for cause?
Is Deming right? Is this situation really a management responsibility, not employees?
Finally, are there any public reports of Health Systems (single location or geographic region) where the situation is under control?

Do not confuse "hospital visits" with "doctor visits"
Do not confuse "long-term care" (nursing homes) with "hospital visits"
Do not confuse "emergency room visits" with "hospital visits"
Do not confuse "public health clinic visits" with any of the others.
Do not confuse "diagnostic lab visits" (phlebotomists, X-rays, CAT scans, MRIs, etc.) with any of the others
Do not confuse "interpretation of diagnostic lab results" (usually done by doctors) with "administering lab tests" (usually done by nurses or technicians.)
Do not confuse "pharmacy fills" of prescriptions (hospital or free-standing pharmacies) with any of the others.

Typically, only doctors "diagnose," but they may delegate some of the preliminary work to nurses and physician's assistants and Emergency Medical Technicians (triage for example.)

In most modern healthcare settings, all treatments of patients are at the ostensible direction of a doctor, regardless of how many layers of delegates may be between the doctor and the patient. Think Deming or ISO (Section 6 - Resource Management!) The manager is responsible for creating and maintaining conditions for processes to proceed according to plan.

If the doctor is "out of the loop" - Why? How can it be remedied?

Hospitals are only buildings. The building may have problems antithetical to patient care (ventilation, cleanliness, heat, cold, light, space), but those conditions are the responsibilities of people. Therefore, "hospitals" don't make mistakes or harm patients, the actions or inactions of people harm patients.

Dear Wes:

Let me try to answer two of the five questions posed (Just my humble opinions.)

Question 1:

I think this medical errors crisis is largely a matter of perception. Not politics, or socioeconomics, but steming simply from the misuse of simple y/x ratios and an imprecise understanding of what x should be used.

Question 2:

More thorough analysis of the "raw" data on deaths due to medical errors is requried. The medical and the health care industry should do what the NHTSA has been doing.

The difference between the ratio y/x and the rate dy/dx, as it applies outside physics must be understood. In the tax problem, for example, the tax rate, also called the marginal tax rate is the slope of the graph of (taxable income) versus taxes owed. The defining equation is y = hx + c where the constant c is negative except for those with the lowest taxable incomes. If you are in the 25% tax bracket h = 0.25. If you are in 33% tax bracket, h = 0.33 and so on. Since c is always negative the ratio y/x = h + (c/x) always seems to be increasing as x increasing with the range for which the constant h applies. I think of the ratio y/x as the "tax burden". This is always increasing as income x increases. The same considerations apply to many problems including traffic fatalities and medical errors that we have started discussing here.

I find the report of Traffic fatalities reports by NHTSA very informative. The same format is used year after year. I can go almost to the same table no. and even to the same page and cross check and compare data for each year. As an example of how new conclusions can be drawn by testing alternative models, I have attached my analysis of the traffic fatality data for the years 1996 to 2002. The Microsoft Excel file should be self-explanatory. If there is interest I would be happy to provide a written text.

I will leave it up to others here to answer the other three questions you have posed. I hope you will be kind enough to entertain my other posts. With my warmest regards.

Charmed :) :thanx:

In Quality,we make a distinction between MAJOR nonconformances and MINOR nonconformances. MAJOR generally means either a life, health, safety issue or the product will not function. MINOR generally refers to issues that do not prevent a product from working.

We require ZERO defects on MAJOR nonconformances. We may track and try to improve ratios on MINOR nonconformances.

In a health care setting, we might consider MAJOR anything that results in death. MINOR might be inconveniences like waiting three hours past a confirmed appointment. Everything else would fall somewhere on that spectrum.

There are times when statistics come into play, but some statisticians are like a kid with a hammer - everything begins to look like a nail.

We don't want to fall into the trap bean counters created at GM when they decided a number of deaths of Corvair drivers and passengers was "statistically acceptable" rather than implementing known remedies to prevent those deaths.

Dear Wes:

To Quote:

We require ZERO defects.

We don't want to fall into the trap bean counters created at GM when they decided a number of deaths of Corvair drivers and passengers as "statistically acceptable" rather than implementing known remedies to prevent those deaths.

I agree. Nothing less than Zero Defects should be acceptable, especially in the Health Care Industry. Every physician would agree.

Is society willing to bear the costs though? That's the "core" of the Health care debate and here, I think, many opinions will be expressed, by employers and their bean counters (since they are paying at least part of the health care costs). In many poor countries (in Africa, in India and China, etc.) quality health care remains a luxury only for the rich. Often, as you know, there is no access to any health care facility, period. May be the last is off topic, if we want to focus on the U.S. Healthcare Industry, in this thread. :topic:

Charmed :) :thanx:

P. S. I guess we cannot forget those who do not have any type of Health Insurance or cannot afford it, and therefore do not have access to quality health care that some others have. (Dick Gephardt, during his brief recent campaign for President, was talking passionately about the superb health care benefits enjoyed by those serving in the Congress and Senate as opposed to health care benefits enjoyed by others in the rest of society.)

The original questions are:
Now the questions for comment:

Is the seeming crisis in healthcare quality real or merely a matter of perception?
If not real, how should the healthcare industry correct the misperception?
If real, what would be your guess as to root cause or common cause (that is, where would you start your investigation for root cause or common cause?)
Since the situation seems to pervade both socialized and privatized medical systems, is it fair to eliminate the payment system from the primary consideration for cause?
Is Deming right? Is this situation really a management responsibility, not employees?
Finally, are there any public reports of Health Systems (single location or geographic region) where the situation is under control?
Remember, no political diatribes, let's just discuss this from a Quality viewpoint.

In another thread, we cover the topic of change management and introduce the concept of organizational culture.

If the seeming crisis (we haven't confirmed that, yet, just anecdotal stories so far) is real, what cultural change would be necessary as a first step?
I heard a speech over the weekend which postulated how much folks could cut down on the spread of colds by just washing their hands more frequently and thoroughly (the speaker suggested singing one verse of "Happy Birthday" for the correct duration of washing.)

Can things really be that simple?

Dear Wes:

After your post, I checked out the Healthgrade Inc. website to see if I could find any hard data to support the assertion that the medical errors crisis is getting worse.

Many news stories seem to be merely repeating what Healthgrades is saying. The 1999 Institute of Medicine (IOM) report estimated 98,000 preventable death per year due to medical errors. Many measures were promised (by IOM, supposedly acting on behalf of the medical community). The new study by Healthgrades now says the situation is much worse. It is now 195,000 preventable deaths per year.

However, so far, I have found no detailed justification yet for these numbers, and how they were arrived at. Still searching and would like to hear more.

Charmed :)

Dear All:

I found the following "data" on medical errors. This is based on a survey conducted by a concerned physician who experienced a near-miss in his own medical practice. The full article may be found at the following link.

http://www.scienceblog.com/community/article3497.html

Notice the estimate of 9 deaths (4 %) based on responses from 466 physicians, with a total of 2500 being surveyed. Of these 210 physicians reported a serious medical error in their own practice within the past six months.

From the persepctive of a "Quality" professional (3.4 defects per million opportunities in Six Sigma methodology), the figure of 4% seems extremely high. Some of the reasons for medical errors are discussed in the article. One of the hospital systems here in the Metro Detroit area has recently announced a Six Sigma initiative (aimed more at cutting costs, reducing waste, and improving general quality, rather than medical errors, specifically).

Charmed :)


*********

45% of specialists report a recent medical errorPosted on Tuesday, August 03, 2004 @ 12:11 PM PDT by bjs


Otolaryngologist Dr. David Roberson has first-hand experience with medical errors. He remembers one near-miss in a patient about to receive a cochlear implant -- and says it typifies the kinds of mistakes he and his colleagues have turned up in a national survey. ''I looked at the CT scan carefully to determine if the cochlea would accept the implant,'' recalls Roberson, from the Department of Otolaryngology and Communication Disorders at Children's Hospital Boston. ''I asked a colleague to look at it also, and he commented that the auditory nerves looked small. I then ordered an MRI which showed the patient had no auditory nerves on either side. I came close to performing surgery and putting a major device in a child's head when there was no possibility of benefit, since she had no auditory nerve. I didn't look carefully enough at the entire scan.''

From Children's Hospital Boston :

In national survey, 45% of specialists report a recent medical error

Authors provide a 'top ten' list of safety recommendations

Otolaryngologist Dr. David Roberson has first-hand experience with medical errors. He remembers one near-miss in a patient about to receive a cochlear implant -- and says it typifies the kinds of mistakes he and his colleagues have turned up in a national survey.

''I looked at the CT scan carefully to determine if the cochlea would accept the implant,'' recalls Roberson, from the Department of Otolaryngology and Communication Disorders at Children's Hospital Boston. ''I asked a colleague to look at it also, and he commented that the auditory nerves looked small. I then ordered an MRI which showed the patient had no auditory nerves on either side. I came close to performing surgery and putting a major device in a child's head when there was no possibility of benefit, since she had no auditory nerve. I didn't look carefully enough at the entire scan.''

Roberson and colleagues sent a brief, anonymous survey to 2,500 members of the American Academy of Otolaryngology-Head and Neck Surgery, and received 466 responses (19 percent). Of these, 210 physicians -- 45 percent -- reported that a medical error had occurred in their practice in the past six months. Errors occurred in all phases of patient care; 78 (37 percent) caused major injury or harm, and 9 (4 percent) were fatal. Both adults and children were affected.

*******************

P. S. The "data" here also highlight the mathematical and philosophical difficulties that I had noted in an earlier post in this thread. The 4% figure is arrived at by converting the ratio y/x = 9/210 = 0.043 to a percentage. The numerator y is the number of errors that led to a fatality. We can be very sure about this. We know how to count the number of preventable deaths. The figure to be used for the denominator x, on the other hand, is far from certain. In the example here, 466 physicians responded to the survey but only 210 reported a medical error in their practice. Perhaps, we should use x = 466 which reduces the percent deaths to just 1.93%.

The situation is similar to that with traffic fatalities, discussed in a separate thread. We drive everyday without getting into an accident. Should the assessment of traffic fatality be based on miles traveled (which is what the National Highway Traffic Safety Administration, NHTSA uses), or should we look at the number of fatalities y that occur in x crashes. This figure tells us something about the chances of surviving a crash, given that you are in a crash. Similarly, given that a medical error has occurred what are the chances that it is fatal. The higher percent should be used, if we are trying to determine the likelihood of a fatality as a result of a medical error.

Unfortunately, (my impression, is that) the data itself might be "faulty". We must, first of all, develop a rigorous methodology for compiling data (NHTSA Traffic Safety Reports provide a very good model). In the recent Healthgrades Inc. report 16 different types of Patient Safety Incidents (PSI) were examined and the x and y values are available for study and analysis. However, I draw somewhat different conclusions from the same data (outside of scope here).

Here's another brick for the quality in health care foundation-building effort.

http://www.usatoday.com/news/health/2006-12-11-sleep-study_x.htm

Study: Long hospital shifts, sleep deprivation can kill
Posted 12/11/2006 10:46 PM ET
By Kathleen Fackelmann, USA TODAY
Medical residents are routinely scheduled to work shifts that last 24 hours or more, yet a study out today suggests that these sleep-deprived doctors are at high risk of making medical mistakes that can harm or even kill patients.
"Working for more than 24 hours is hazardous," says sleep researcher Charles Czeisler at the Harvard Medical School. Scores of studies show that people who stay awake for 18 hours straight can have trouble thinking clearly and can zone out or nod off suddenly.

In July 2003, the Accreditation Council for Graduate Medical Education, a Chicago group that accredits medical residency programs, limited work schedules to no more than 80 hours in a week. But the rules still allow marathon shifts that last up to 30 hours.

Many medical residents must pull these extra-long shifts twice a week — a schedule that leads to extreme fatigue, says Simon Ahtaridis, president of the Committee of Interns and Residents, a 12,000-member union calling for more reform of work hours.

Czeisler and his colleagues had 2,737 first-year medical residents complete a monthly survey that asked detailed questions about their work schedule, sleep and days off. During the year-long study, the residents also were asked to report any medical errors they'd made while on duty.

The researchers found that when residents reported working five marathon shifts in a single month, their risk of making a fatigue-related mistake that harmed a patient increased by 700%. And the risk of making an error that resulted in a patient's death shot up by 300%. The report appears in the journal Public Library of Science (PloS) Medicine.

The residents in this study reported making 156 fatigue-related errors that injured a patient and 31 mistakes that led to a death, Czeisler says. About 100,000 medical residents in the USA routinely work these extended shifts, Czeisler says. "These data suggest there are tens of thousands of preventable injuries to patients annually."

Richard H. Bell Jr., assistant executive director of the American Board of Surgery, contends that 24-hour shifts are sometimes necessary to provide patients with crucial continuity of care. For example, residents might need 24 hours to get a patient through surgery and then to pass on the details of the case to the next resident, he says.