In Doctors are not experts on life expectancy1 David Strauss and Robert Shavelle noted that some clinicians testifying as expert witnesses on life expectancy did not have a very good grasp of the subject. As a result some were providing confusing and/or surprising opinions. The authors provided a number of examples. We can confirm that such misstatements continue to be made in this setting today, but the question of whether doctors are experts on life expectancy is not settled by such examples.
One aspect of the science of life expectancy with which many clinicians are not familiar is the life table. Anyone claiming expertise on life expectancy ought to be familiar with standard life tables like those published in the US by the Centers for Disease Control and Prevention.2 The numerical relationships between the numbers in the various columns of these tables ought to be prerequisite knowledge for anyone hoping to assist a court or jury in understanding life expectancy. As Dr Richard B. Singer (a medical doctor familiar with life table methodology and life expectancy) noted,3
“It is important for the risk selection expert (whether a medical director (MD), lay underwriter, actuary, or other specialist) to be able to examine, analyze and manipulate life table data with facility. ...facility will come only if the reader makes use of this knowledge ... in applying the results of all types of mortality studies to his daily work.”
Another important point the authors apparently hoped to impress upon readers of Doctors are not experts1 was that contrary to “doctors” (who their title suggested were not experts on life expectancy) they (Strauss & Shavelle) were. Perhaps they succeeded. The two have opined as experts on life expectancy in tort cases around the world on myriad occasions since their article first appeared in 1998. Both of us once worked for them at Strauss & Shavelle, Inc., and the Life Expectancy Project® and we are familiar with their work in this area. They, like us, continue to provide such expert consulting services on life expectancy and related topics in litigation and in other arenas today.
We are doctors (of philosophy) with PhD degrees in applied statistics (Day SM) and in epidemiology (Reynolds RJ). David Strauss and Robert Shavelle are likewise doctors (of philosophy), with PhD degrees in statistics and applied statistics, respectively. The word doctor, of course, can mean a lot of different things. According to Merriam-Webster online, the origin of the word is:
Middle English doctour teacher, doctor, from Anglo-French & Medieval Latin; Anglo-French, from Medieval Latin doctor, from Latin, teacher, from docēre to teach.
The first two entries for the full definition of doctor at Merriam-Webster are:
a : an eminent theologian declared a sound expounder of doctrine by the Roman Catholic Church —called also doctor of the church
b : a learned or authoritative teacher
c : a person who has earned one of the highest academic degrees (as a PhD) conferred by a university
d : a person awarded an honorary doctorate (as an LLD or Litt D) by a college or university
a : a person skilled or specializing in healing arts; especially : one (as a physician, dentist, or veterinarian) who holds an advanced degree and is licensed to practice
b : medicine man
On view of definition 1c, Drs Strauss and Shavelle would apparently disagree with the claim made in their own article. What may not be as obvious is that they never believed their title literally even on definition 2a. Drs Strauss and Shavelle have long understood that some clinicians or other medical doctors are experts on life expectancy. Many editors of the Journal of Insurance Medicine, and many authors of methodology articles routinely published there, are medical doctors who are also experts on life expectancy.
Drs Strauss and Shavelle have acknowledged that the title of their 1998 article was meant to peak interest and promote discussion rather than to be taken literally. A more accurate (but perhaps less controversial) title might have been:
A medical degree (and/or clinical experience) is neither a necessary nor a sufficient qualification for being an expert on life expectancy.
Clinicians in various fields of expertise have been tested as to their ability to accurately determine the life expectancy of real or hypothetical patients.4-7 They have fared poorly in general. It is worth noting that in these studies,4-7 a statistical analysis of life expectancy has generally been used as the benchmark for comparison. While we are certain that the basic results of these studies are sound (e.g., that “Doctors, nurses and medical students were inconsistent, inaccurate and imprecise in their prediction of life expectancy…”), one should be cautious about how much weight one gives to the statistical analyses used to set benchmarks.
Consider for example Clarke et al.4 who reported that, “Predictions [by doctors, nurses, and medical students] were compared with life expectancy derived from actuarial tables, incorporating a series of numerical mortality ratios used by the life insurance industry.” Herein lies the rub. Knowing that life expectancies were derived from actuarial tables by incorporating a series of numerical mortality ratios leaves much to the imagination, even for those of us intimately familiar with the scientific methods involved. It is beyond the scope of this essay to delve into the possible choices that must be made in such actuarial analyses, including which numerical mortality ratios to use for the various medical conditions, diseases, or lifestyle issues (e.g., smoking, or alcohol or drug abuse) considered. To a great extent, this is immaterial to the findings of Clarke et al. Even in the absence of a reasonable benchmark, the various estimates provided by medical doctors, nurses, and medical students can be compared to each other (within and between groups) to demonstrate a lack of agreement on life expectancy among these medical professionals.
Three “Learning Points” provided by Clarke et al. tend to affirm our hypothesis and proposed title (A medical degree (and/or clinical experience) is neither a necessary nor a sufficient qualification for being an expert on life expectancy):
1. Doctors are as inaccurate as nurses and medical students with a tendency to underestimate when assessing life expectancy.
The meaning of inaccurate here depends in part on the accuracy of the actuarial evaluations of life expectancy used by Clarke et al. as benchmarks for comparison. We do not mean to suggest we have serious doubts about those statistical evaluations, but rather to acknowledge that they can be legitimately questioned. However, even if the true gold standards for life expectancies were known and used, the statement here or a slight modification of it would still be true: Doctors are as accurate or inaccurate as nurses and medical students. If being a medical doctor were a necessary qualification for determining life expectancy, then nurses should not be as accurate as doctors, or doctors as inaccurate as nurses, etc.
2. Inconsistent predictions of life expectancy may lead to patients being managed differently by the same or different doctors despite identical comorbidity.
If being a medical doctor were a sufficient condition for being an expert on life expectancy, then there ought to be a high level of consistency in life expectancy estimates between medical doctors. That this is not the case affirms that being a medical doctor in and of itself does not guarantee expertise in determining life expectancy.
3. Inconsistency does not appear to improve with medical knowledge or experience.
If being a doctor were a sufficient condition for being an expert on life expectancy, presumably being one for a longer period of time would result in a greater level of expertise. Data collected and analyzed by Clarke et al. suggest that this is not the case.
One may wonder whether a similar study might be done with groups of actuaries (or other specialists, see Singer2) and whether the results might not be similar. This would be a worthwhile exercise. Whatever the outcome, however, a careful evidence-based statistical evaluation of life expectancy has a clear advantage over an evaluation by a medical doctor based only on his or her education and experience (which may include very little work related to the science of life expectancy). An evidence-based statistical evaluation, as those used to set benchmarks in Clarke et al., no matter how well or how poorly done, can be reproduced and modified to reflect new or better evidence. The opinions of clinicians are often beyond such examination or evidence-based discussion or modification. For example, it has happened often that two different neurologists in the same case have given vastly differing opinions on the life expectancy of the same individual. What is a court or jury to make of such differing opinions when no explanation as to how the numbers were obtained is provided (other than to say they were determined based on each clinician’s knowledge and experience)? If a clinician’s opinion on life expectancy is not supported by scientific evidence and scientifically sound methods, it may be little more than an opinion (without a scientific basis, and therefore not an expert opinion).
If a medical degree is neither a necessary nor a sufficient qualification for being an expert on life expectancy, then what is? We have personally reviewed non-scientific reports expressing opinions that cannot be supported by any scientific evidence written by people with a variety of qualifications and experience, including clinicians, life insurance underwriters, biostatisticians, and epidemiologists, some with appointments at accredited universities or who had worked for years in the life insurance industry. It would appear to us that none of these qualifications alone is sufficient to bring a level of scientific expertise to a discussion of life expectancy. On the other hand, we have seen a number of reports written by persons with those same qualifications that were scientific, with detail sufficient to allow replication and critical examination of the analyses reported. On what is necessary or sufficient to qualify someone as an expert on life expectancy, a borrowed and slightly modified passage from Modern Epidemiology by Drs Kenneth Rothman and Sander Greenland8 may say it well (with apologies and credit to Drs Rothman and Greenland whose original quote was about being an epidemiologist rather than a life expectancy expert):
Today, notwithstanding the important contributions to the field by many who consider themselves first as statisticians or physicians, life expectancy experts have achieved a separate identity. Being either a physician or a statistician, or even both simultaneously, is neither a necessary nor a sufficient qualification for being an expert on life expectancy. What is necessary is an understanding of the principles of the scientific methods for calculating life expectancy and the experience to apply them.
This is not to say that a clinician or other professional who fails to meet these criteria cannot contribute anything to a discussion of life expectancy. Clinical nuances involved in the health and medical conditions of a given individual may suggest to a clinician that a given individual is at greater or lower mortality risk than average for those who are otherwise comparable and who were the subjects of studies of long-term survival. In this case, the clinician may well suggest that the life expectancy of the individual must be higher, or lower, than life expectancy based on available studies might suggest. In this case, the number of added (or subtracted) years ought to be consistent with available evidence, and the proposed increase or decrease should not be contradicted by available evidence. For example, a scientific analysis of the life expectancy of a 4-year-old child with cerebral palsy of Gross Motor Function Classification System (GMFCS) level V who has a feeding tube may lead to a figure of 20 remaining years (the true life expectancy may be higher or lower than this depending on a number of other factors). A clinician familiar with this child may understand that while he does require and have a gastrostomy tube for feeding, the only reason for this is an aversion to taking food orally, rather dysphagia and a risk of aspiration or gastroesophageal reflux. The clinician may therefore suggest that this particular child’s life expectancy must be greater than 20. As to how much greater, this will be more difficult for many clinicians to determine, but the scientific evidence will provide some bounds on this. For example, the true life expectancy is unlikely to be greater than that of other comparable children who do not require a feeding tube at all.
In our case, we have published extensively on life expectancy and on epidemiologic topics related to life expectancy and we remain active in related fields of research. This year, for example, we will be presenting research findings at the Society for Epidemiologic Research (SER) on the topic of mortality among professional basketball players (NBA) and on certain groups of professional fashion models; we recently lectured on the topic of life expectancy of children with cerebral palsy at the annual meeting of the American Academy for Cerebral Palsy and Developmental Medicine; and every day we apply the results of all types of mortality studies to our daily work. As scientists, we would never suggest that we have the final word on the subject of life expectancy, and we will always welcome scientific input and criticism from other qualified scientists, including medical doctors. There is little question we are equipped to discuss such issues scientifically and to assist a court or jury in their efforts to understand them.
- Strauss DJ, Shavelle RM. Doctors are not experts on life expectancy. The Expert Witness, Summer 1998. Calgary, Alberta: Economica Ltd. http://www.economica.ca/ew03_2p4.htm
- Arias E. United States life tables, 2010. Natl Vital Stat Rep. 2014 Nov;63(7):1-63.
- Singer RB. The application of life table methodology to risk appraisal. In: Brackenridge RDC, Croxon R, Mackenzie BR, editors. Medical Selection of Life Risks, 5th edn. New York: Palgrave Macmillan, 2006.
- Clarke MG, Ewings P, Hanna T, Dunn L, Girling T, Widdison AL. How accurate are doctors, nurses and medical students at predicting life expectancy? Eur J Intern Med. 2009 Oct;20(6):640-4.
- Wilson JR, Clarke MG, Ewings P, Graham JD, MacDonagh R. The assessment of patient life-expectancy: how accurate are urologists and oncologists? BJU Int. 2005 Apr;95(6):794-8.
- Jeldres C, Latouff JB, Saad F. Predicting life expectancy in prostate cancer patients. Curr Opin Support Palliat Care. 2009 Sep;3(3):166-9.
- Chow E, Harth T, Hruby G, Finkelstein J, Wu J, Danjoux C. How accurate are physicians’ clinical predictions of survival and the available prognostic tools in estimating survival times in terminally ill cancer patients? A systematic review. Clin Oncol (R Coll Radiol). 2001;13(3):209-18.
- Rothman KJ, Greenland S. Modern Epidemiology. 1998 Philadelphia: Lippincott-Raven.