Sunday, May 22, 2016

Medical Treatment Is Never A Zero Risk Decision

Harvey W. Cushing, MD

Ever since Medical Errors - the Third Leading Cause Of Death came out the media and bloggers have been abuzz with the headline.  It is only a matter of time before we hear about how many full 747s crashing would equal the number of patients that are killed by physician errors each year.  There are any number of articles in the press each day about how there are algorithms or checklists from some committee or government agency that will eliminate risk in medical diagnosis and procedures.  There are an equal number of stories about how machine intelligence or will eliminate all of this human error.  There was a story a few weeks ago about a robot that could suture up operative wounds from surgery.  After seeing that robot in action, I would not want it anywhere near me or anyone I cared about.  There is often palpable conflict of interest involved in the news stories.  Stories designed to generate interest or influence politicians.  Stories to express a political viewpoint.  Even in the professional literature, much of what is written is a political viewpoint often to develop leverage against physicians and force them to do things in a certain way.  The selling point to the public and their elected officials is that medicine, particularly physicians would perform better and more uniformly if they would adhere to certain protocols and that these protocols need to include the input of people who have never been trained as physicians - most notably business managers.          

I can illustrate why these analyses are absurd based on an experience I had about 6 years ago.  At that time my wife was struggling with intense ear pain and the provider she was seeing wanted to start her on another course of antibiotics for ear pain.  I had her see an ENT surgeon instead, who performed an examination that only an ENT physician can do and then read the CT scan that he ordered.  That CT scan was read as normal by the radiologist.  This ENT physician not only read it as abnormal, but got additional imaging studies to demonstrate that my wife had a pituitary tumor.  Referral to an endocrinologist confirmed that it was a growth hormone secreting tumor, further explaining additional symptoms that she had described to various physicians over the years.  I went to work at that point and found the neurosurgeon who was one of the pioneers in transsphenoidal adenoma resections in the world.  Although not malignant, the prognosis of these tumors is not benign with disfigurement from acromegaly, progressive endocrine abnormalities, and eventual congestive heart failure all from the effects of excessive growth hormone secretion.  There were also complications of inadequate excision and secondary procedures like gamma knife irradiation in an attempt to obliterate the tumor.  We were sitting in the consulting room of the neurosurgeon as soon as we could arrange it.

He had pulled up a coronal section of the MRI scan showing the tumor wrapped around my wife's right internal carotid artery.  He made the following statement:  "This is the tumor and this is the artery (pointing to areas on an MRI coronal view of the pituitary gland and surrounding anatomy).  I am going to try to remove as much of the tumor as possible.  If I accidentally nick the carotid artery in this area, there is nothing that can be done about it.  That complication happens and I know good neurosurgeons who have had that happen to them.  I can assure you I will remove only as much tumor as possible.  I will remove only as much as I would remove if you were one of my family members."

This is the unspoken truth about medical errors and whether they are preventable or not.  In this case doing nothing results in an inevitable slow death and severely compromised quality of life.  The procedure is not without risk.  In this case the risk was minimized by selecting the most skilled neurosurgeon in the area, but he openly acknowledges that nicking the carotid artery in a place that cannot be repaired is not only a possibility, but it happens to the best neurosurgeons.  Any cursory analysis should illustrate why.  We are talking about an operation that occurs in about a 1 1/2 inch space at the tip of an endoscope on a tumor that extends away from that tip into a small crevice between the carotid artery and sphenoid bone.  Anyone really interested in this can go to YouTube and view several videos of the procedure from the view of the surgeon.  In this case our neurosurgeon explicitly talked about the issue of how much tumor could be removed.  He was acutely aware that the risk involved in removing too much tumor may be unacceptably high.  Incompletely removing the tumor involves the risk of continued exposure to excessive growth hormone and the secondary gamma knife procedure.     The fact that this surgery can be done at all seems like is a miracle to me.  Early in my  career, I had treated a patient who underwent a transtemporal approach to the same kind of tumor and that had resulted in significant postoperative disability.
In the decision to proceed my wife clearly found the risk acceptable because acromegaly and a slow death from endocrine complications and further procedures were not.  The operative procedure went perfectly and 6 years later her growth hormone, IGH levels, and serial MRI scans show no tumor recurrence.  I don't have to speculate about what might have gone wrong in my wife's case because Henry Marsh (6) writes about it in his compilation of neurosurgical complications Do No Harm.  In the chapter "Pituitary Adenoma" he did a transphenoidal procedure on a man with acromegaly.  His description of the relevant anatomy: "There are however, two major arteries next to the pituitary gland that can, if the surgeon is exceptionally unlucky, be damaged during the operation."  His patient did well until post-operative day number 3, when suddenly his right arm was paralyzed and he could not communicate.  Brain imaging confirmed a major left hemispheric stroke.  Having no actual operative complications Marsh concludes: "This must have been caused in some unknowable way by the operation."

But let's take a look at what would have happened in my wife's case if there had been a complication or series of complications using the methodology cited by the authors in the studies.  First off, in the case of a complication where does the chain of errors begin.  The misdiagnosis of otitis media and two antibiotic prescriptions?  The misread CT scan of the sphenoid bone?  It seems like those are two preliminary errors right there even though neither was immediately threatening.  Combined with a surgical error that would be three medical errors had a complication occurred.  And what about the self correcting aspects of this process?  Does the ENT surgeon get any credit here for correcting the misdiagnosis of otitis media or the misreading of the CT scan?  Any focus on medical errors never looks at the self correcting aspects - how many times they are caught and how many times standard second opinions from colleagues or trusted referrals modify the treatment plan.  How much morbidity and mortality would occur without this level of self correction?  How is it estimated?  The best example I can recall of the problems of error determination was in a NEJM editorial to address the first report on medical errors by the Institute of Medicine.

In that editorial (5), the lead author of some of the most quoted studies discusses the issue of the definition of errors, the definition of preventability, and the subjective nature of these determinations.  His preliminary analysis is that the IOM conclusion that 44,000 to 98,000 deaths caused by medical errors "create an impression that is not warranted by the scientific work underlying the IOM report."  That number has been inflated to more than 250,000 according to the recent quoted report (1).   He also expresses an opinion completely consistent with Marsh's observations and goes on to point out that the observation of error can not cannot be reliably made by third party observers:

"Even with the best surgical technique and proper precautions, however, a hemorrhage can occur. We classified most postoperative hemorrhages resulting in the transfer of patients back to the operating room after simple procedures (such as hysterectomy or appendectomy) as preventable, even though in most cases there was no apparent blunder or slip-up by the surgeon. The IOM report refers to these cases as medical errors, which to some observers may seem inappropriate." (6)

He goes on to elaborate on four important aspects of the IOM report, 2 of which have to do with politics and public perception.  The first is the idea that either nothing is being done about safety or that things are getting more dangerous.  In many ways this is analogous to the public perception that violent crime is high when it is at a 30 year low.  He gives an even better example of Harvey Cushing's accomplishments during his neurosurgical career.  In 1913, the mortality rate for craniotomies for brain tumors was 80%.  In twenty years it was reduced to 13%.  Currently it is 1%.  All of that without the IOM or any of the current federal or state regulators.  The whole idea that physicians are motivated more by adhering to what politicians or regulators want rather than what is in the best interest of their patients has always been an incredible one to me.  I guess it makes sense only if you are a politician or a regulator and you really believe that your naively designed incentives and disincentives have meaning.   The second complication is increased reporting requirements and all of the complications that involves including the impact on confidential peer reviewed complications.  Every physician has participated in some form of complications conferences or what surgeons used to call morbidity and mortality conferences.  The modern goal of these conferences is to try to identify procedures and interventions to improve patients safety.  Opening that system up to increased reporting and subjecting it to the "dead weight of the litigation system" is something the author cautions against, but is often explicit in the news headlines.

Makary thought the IOM missed the boat on the issue of how malpractice litigation is a risk against internally driven safety initiatives.  I think they and subsequent authors missed the boat on the issue of biological variability.  To some degree is it addressed by the subjective determination of preventability but not entirely.  If a neurosurgeon is observed to have performed an errorless operation and the patient sustains significant complications, what are all of the factors that go into calling that an error or not?  A lot of it may have to due  with the baseline state of the patient.  Are they already compromised by injury, illness or congenital variations.  Neurosurgeons after all are not operating on healthy people.  The example also applies to medicine and psychiatry.  As an example, I usually tell people up front these days that there is about a 40% chance that they will noticeably improve with an antidepressant if their symptoms are significant.  There is another 20 or 30% chance that they will feel somewhat better but not get back to their baseline state.  There is a 10-20% chance that they will get significant side  effects and not want to continue the medication and in a small number of cases (1-2%) there is a chance of really severe side effects from the medication.  Pharmacogenomics suggests that it can be invaluable in this process, but there is variability between predictions using the available analyses from the same manufacturer and between manufacturers.  Even with that warning there are a number of people who will consider it an "error" that they received a medication that caused them to get a rash or in the worst case serotonin syndrome.  I think that there is a natural tendency for some to see any medical procedure that does not go well to be attributable to error on the part of the physician.  I think that the personal experience of most physicians will bear that out.        

Another less well known intervention in psychiatry is the care of the chronically suicidal person.  Many of these persons have a history of serious suicide attempts and remain at significant risk.  The usual risk factor analysis of their suicide potential does not add very much.  There are frequently inadequate resources to treat them or they refuse to access what is available.  Frequent short term admissions to inpatient units adds nothing to their care.  The psychiatrist who accepts their care usually is seeing them more intensely and accepting more crisis calls from them than other patients.  Until a sufficient therapeutic alliance develops, there may be many sessions with a lot of depression and anger and very little objective information on what the patient is thinking.  The psychiatrist and patient in this case need to accept that the risk involved in this situation from suicide or a suicide attempt - is necessary to make progress and enhance the patient's ability to function and enjoy life.

My main point here is that people currently need to assume some risks in order to get better.  There is nobody more than physicians who would want medicine to be a no risk endeavor, but the reality is that is not going to happen anytime soon.   All patients realize this at some level, but that may do little to mitigate the anger or disappointment when it occurs.  Every family has a tale to tell about a medical miracle or a medical mistake.   In the situations like my wife faced with my support, people are clearly willing to take the risk in face of undesirable consequences.  The physicians involved see this as very serious work.  Should the classification of medical errors be refined and analyzed?  Of course they should - but the approach being used in the press invoking planes full of patients being sent to their death by physicians being equated to terrorists is not accurate or helpful.  The backlash is significant and impairs the ongoing error analysis and correction process.  Equating a heterogenous collection of complications classified by different methods as errors and listing them as a standard cause of death is also not accurate or helpful to the science of medical error analysis and correction.

George Dawson, MD, DFAPA


1:   Makary MA, Daniel M. Medical error-the third leading cause of death in the US. BMJ. 2016 May 3;353:i2139. doi: 10.1136/bmj.i2139. PubMed PMID: 27143499.

2: Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, Newhouse JP, Weiler PC, Hiatt HH. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med. 1991 Feb 7;324(6):370-6. PubMed PMID: 1987460.

3: Leape LL, Brennan TA, Laird N, Lawthers AG, Localio AR, Barnes BA, Hebert L, Newhouse JP, Weiler PC, Hiatt H. The nature of adverse events in hospitalized patients. Results of the Harvard Medical Practice Study II. N Engl J Med. 1991 Feb 7;324(6):377-84. PubMed PMID: 1824793. 

4: Localio AR, Lawthers AG, Brennan TA, Laird NM, Hebert LE, Peterson LM, Newhouse JP, Weiler PC, Hiatt HH. Relation between malpractice claims and adverse events due to negligence. Results of the Harvard Medical Practice Study III. N Engl J Med. 1991 Jul 25;325(4):245-51. PubMed PMID: 2057025. 

5: Brennan TA. The Institute of Medicine report on medical errors--could it do harm? N Engl J Med. 2000 Apr 13;342(15):1123-5. PubMed PMID: 10760315.

6:  Marsh H.  Do No Harm: Stories Of Life, Death, And Brain Surgery.  Thomas Dunne Books/ St. Martin's Press; New York,  NY; 2014.


Portrait of Harvey Cushing, MD by Edmund C. Tarbell [Public domain], via Wikimedia Commons at accessed on May 22, 2016.


Supplementary 1:  Another excellent example of biological variability is joint replacement surgery.  I see patients, friends and family members who have mixed experiences.  The majority turn out very well, in some cases extremely well.  In the case of extremely good joint replacement surgery, according to all of these people I talk to, the surgeon seems to discourage them talking to other people about how good the results were, generally by telling them that they had an extraordinary result in terms of how the replaced joint functions.

Supplementary 2:  A good example from a recent post about the aggressive treatment of pain and the risk of addiction and overdose deaths.  20 years ago when physicians were being criticized by various factions for not aggressive enough use of painkiller prescriptions and the use of opioids for chronic noncancer pain - the risk of death by overdose was minimized by those proponents.  Now that there is a clear cause of death from overuse of opioids and the rates of addiction have increased - physicians are faulted for not knowing how to prescribe opioids.  It seems like they had a better idea 20 years ago before yielding to the critics.  From the perspective of this post - there is always risk either way.  Only politicians and regulators can deny that.

Supplementary 3:  The Twitter graphic ( and yes I am serious).  The IOM did not start until about 1970.  None of these entities has a track record remotely close to Harvey Cushing in the 1920s and 1930s.


No comments:

Post a Comment