Resources & Information

Policy Discussions Archive

Clinician obligation to sign death certificates

Discussion Ended on 10/21/2018

NCMB regularly receives calls and emails from funeral directors, families and others who report physician refusals to complete death certificates.

Most instances NCMB has heard about arise from unattended deaths from natural causes. Typically, these decedents had an established relationship with the physician or PA, but for a variety of reasons, the identified provider is reluctant to certify the death. The decedent may not have been seen for several months, or the individual may have been under treatment for stable conditions that posed no apparent immediate threat to his or her life (hypertension, diabetes, etc.). Often, when asked, the physician or PA indicates that he or she feels that they have no idea why the patient died.

The draft position statement under consideration by the Board formally states this position. Access the text of the draft position statement below, under References.

NOTE: It is important to understand that a death certificate is a legal and not a scientific document. Clinicians are not required to establish a specific anatomical reason causing the death. Deaths that are suspected as suicides, homicides or drug overdoses are routinely referred for autopsy to the county coroner. It is simply not feasible from a medical examiner staffing or financial standpoint for every decedent to receive an autopsy. Declining to certify a death will generally NOT result in an autopsy being performed.

The U.S. Centers for Disease Control and Prevention offers free online training in completing death certificates.



I agree with setting the expectation for providers to complete the death certificate. It is an important exercise and learning opportunity to think through possible causes of death in a patient even if we have not seen them in quite some time. The use of “possible” on the death certificate helps acknowledge that we accept the ongoing mysteries related to health, illness and death.

By Harriett Burns on Oct 21st, 2018 at 10:48am

“SUSPECT cause of death is related to progression of and complications from co-morbid conditions of (problem list: i.e htn, DM, hyperlipidemia, COPD)“and including detrimental lifestyle, i.e.smoking, obesity, alcohol,high fat intake .etc.

By Earl Cummings on Oct 20th, 2018 at 11:32pm

Education on how to properly complete the “cause of death” entry will not help when the cause of death is unknown. Listing a guess as to the cause only insures affirmation and dissemination of GIGO (Garbage in, garbage out) data. At best, harmless, but also quite likely to be perpetuated in future studies and statistical analyses, such as those by CDC. That action only disseminates erroneous data and conclusions. Listing “unknown cause, without suspicion of foul play” might be a solution to the entry dilemma.

By Henry F. Gregor, MD on Oct 18th, 2018 at 5:35pm

I’m a forensic pathologist, not practicing in North Carolina right now, but with a NC license.  I’d like to address the questions raised by David Pelzer with respect to his statement:

“Encouraging physicians to “guess” the cause of death serves no purpose.  Giving physicians, as examples, encouragement to list “hypertension” or “complications of diabetes” as causes of death leads us in a path of uncertainty.  If the standard North Carolina death certificate lists “cause of death” as a requirement to be completed, shouldn’t the person completing this form be relatively certain as to the actual cause?  If not, then why is this part of the form??  “

The first thing to remember is that the Medical Examiner has a very limited mandate when it comes to these cases—that, in general, of looking for unnatural death or atypical public health issue.  We do not have the mandate of determining the precise cause of death in natural deaths.  I wish we did—I’m a big fan of autopsies, and of full employment for autopsy pathologists.  But we don’t.  No state or county wants to pay for that.  And Medicare certainly doesn’t want to.

What that means is that in the absence of some suspicious indication of foul play, if a person has significant risk factors for death, it will get signed out that way without an autopsy. 

Is there uncertainty there?  Absolutely.  There have been a number of studies comparing autopsy and external exam accuracy, and it turns out that the error rate for Medical Examiners performing external exams is often between 20 and 30%. 

Why do we accept that?  Because the errors usually involve related events—a PE is called an MI, an AAA is diagnosed as hypertensive heart disease.  From a *forensic* point of view, that’s close enough.  In contrast, our error in manner determination is usually less than 1 percent, usually mistaking accidental for natural or vice versa (e.g. a fall with SDH versus a ruptured berry aneurysm in an older patient).  The rate of missing a homicide or accident that might involve criminal charges is less than 1/10th of one percent.

But, in these situations, the caretaking physician would likely do much better with respect to natural deaths.
The attending or caretaking physician knows the most about the patient—even if he or she has not seen the decedent in some time.  While the attending may not have seen the patient, at least he or she has some familiarity with the case.

In contrast, the Medical Examiner does not have that knowledge.  Instead, he or she must quickly review whatever medical records are immediately available (after all, it can take time to subpoena and receive records and the family waiting), and make some probabilistic estimate of cause of death.

But these diagnoses are anever made with certainty.  They are made probabilistically.  Like any diagnosis—even those made in living patients—diagnosis is a matter of ranking possible diagnoses and choosing the most likely.  Essential;y *no* diagnosis is made with absolute certainty, not even in clinical medicine.  But clinicians treat patients anyway. In general, and knowing that an autopsy is not in the cards, the clinician who knows the patient is more likely than the pathologist to get the right answer.

Nobody likes that uncertainty.  But nobody wants to pay for more autopsies.  Nobody, certainly not clinicians or hospitals, seems to want to ask families for or fund non-forensic autopsies.  So they aren’t going to get done. 

But the family needs closure, even if that closure involves an approximation of the absolute answer, rather than the absolute answer.  They need to get business done—to have the memorial service, say goodbye, get through probate, sell the parent’s house, get on with life.  And when the clinician refuses to accept that he or she is competent to determine the most important risk factor in a natural death, it delays all of this, adds cost to health care and the public health system.  It is a disservice to the family, the state, and to his or her duty as a physician.

Sure, I can drag the body to the morgue and look for a bullet hole in an obese 65-year-old man with a history atrial fib, a 3 vessel CABG, and 10 years of uuncontrolled hypertension and type II diabetes mellitus who suffers a witnessed collapse loading up his plate at the Golden Corral.  But there won’t be one. And it just seems silly to listen to his cardiologist say “I am absolutely incompetent to think of any reason he might drop dead.”

I find it even more odd that clinicians who will not ask for a hospital autopsy because they don’t think it will provide important information will at the same time pretend that suddenly they have no clue about what is going on with their patients.

By William Oliver on Oct 18th, 2018 at 5:01pm

We all feel the impact that generalizations from even partially inaccurate data can have on policies and our daily lives, but we still have to keep some balance between what is a reasonable expenditure of time and resources. A reasonable guess based on history by the physician who knows the deceased best when those on the scene, a coroner or medical examiner find no suspicious circumstances is better than anyone else can do. As long as that deficiency of certainty is held in mind when collating statistics from death certificates. We must always question the assumptions and validity of conclusions drawn from population data.

I’ve seen discharge diagnosis of accidental overdose by a hospitalist in another county when the person had respiratory failure because they had stopped using their non-invasive ventilator. And that same person was discharged home without any inhalers or nebulizer Meds on her discharge medication instructions. I contacted our coroner to share information that might have lead to suicide in an unexpected death of a patient whom I had counseled as he grieved the death of his father after his stepmother’s decision to accept hospice services. His PCP has no knowledge of that. I haven’t heard back from the coroner.

Practice the Art. Do the best you can, where you are, with what you have. No one and no data will ever be perfect.

By Rebecca Love, MD on Oct 17th, 2018 at 11:37pm

I agree with the current position statement. We should be expected to use our best clinical expertise and not hold up the process.

By Dana Chambers MD on Oct 10th, 2018 at 8:35pm

As a hospitalist, I am comfortable in doing death certificates, because I typically have seen the patient within the last day.  And certainly most deaths at home are expected and we usually have a likely diagnosis, so the death certificate can be honestly completed.  However, if a 50 year old with a stable chronic disease is found dead at home months or years after last contact with their primary physician and death was highly unlikely at the time of last visit, I can see how filling out the death certificate could create an ethical conflict.  For cases like that, might the medical society encourage the state to relax the criteria for the medical examiner to accept the case if the primary care provider truly has no idea why they died?  Because when the cause of death is a complete guess, we end up with inaccurate mortality data.  Maybe they died from an opiate overdose rather than an MI?  Or did they have carbon monoxide poisoning?

By Lisa Kaufmann, MD on Oct 10th, 2018 at 7:59pm

Encouraging physicians to “guess” the cause of death serves no purpose.  Giving physicians, as examples, encouragement to list “hypertension” or “complications of diabetes” as causes of death leads us in a path of uncertainty.  If the standard North Carolina death certificate lists “cause of death” as a requirement to be completed, shouldn’t the person completing this form be relatively certain as to the actual cause?  If not, then why is this part of the form??  If you are going to encourage physicians to sign the death certificate and take an educated “guess” as to the cause of death, why don’t you just remove the line “cause of death” from the death certificate all together.  If we are not going to know with relative certainty why the deceased passed away, why do we need to take an educated “guess” and fill out the line stating cause of death anyway. 
I for one, and I believe many other physicians, would be much more agreeable to sign a death certificate if I am not required to state a cause of death on a patient that I may not have seen for months (or perhaps years) and have no knowledge of their recent health status.

By David Peltzer on Oct 10th, 2018 at 7:58pm

Good luck. I’m sometimes asked to sign a certificate for a patient I have not seen for months who was reasonably healthy. I don’t even know that the patient is really dead. A body was found and taken to a funeral home 30 miles away. Did he have an MI or was it a homicide? Drug OD? Suicide It’s asking too much to ask us to make a wild guess when there are many possibilities.
No problem with patients with known potentially fatal disease that I have an ongoing relationship with, but the fact that I am the only doc a patient has seen in the last five years does not qualify me to say why he was found dead in bed at home.

By Charles Davant on Oct 10th, 2018 at 5:36pm

Commenting as an individual PA working in Geriatrics, I support this position and also the continued education of PAs on how to properly complete the ‘cause of death’ notation.

By Samantha Rogers on Oct 10th, 2018 at 1:16pm