A complicated death
Excerpt from the US standard certificate of death (Source: CDC).
A complicated death: A 60-year old male presenting to an emergency room with neck and back pain and an incessant dry cough. A diagnosis of advanced non-small cell lung carcinoma, a form of lung cancer, with metastases to the liver and the spine. The patient is given steroids, and chemotherapy, and is instructed to rest. The steroids reduce his inflammation but increase his chances of infection; the chemotherapy slows the progression of his cancer but irritates his lungs; the bed rest gives his body the space it needs to fight but accelerates the formation of blood clots in his circulation. He comes back to the hospital two months later, with fungus in his lungs, clots in his veins, and a failing heart. His liver and kidneys are failing—a battery of bloodwork sends a simple message: I need oxygen. Even the heart is unable to provide itself with sufficient oxygen to function. Extensive damage to the lungs makes intubation ineffective. And death follows.
Doctors identified many factors—the cancer, the blood clots, the fungus, even the medicine itself—that could have contributed to the patient’s demise. And yet, at a modern medical center, this story is hardly exceptional in its complexity. There are two broad reasons for this: The first is that, as patients live longer, they accumulate more diseases. The second is that, as doctors and scientists make further discoveries about the functions and failings of our bodies, they define more diseases. The International Statistical Classification of Diseases, the world-standard document for defining illness, enumerates some 69,000 diseases with which a person can be afflicted.
Yet despite this dizzying complexity, something needed to be written on the death certificate. A cause. Multiple causes. A chain of events, constrained to fit within four handwritten lines on the standard-issue document that registers a person’s passing. Getting it right matters, because death certificates shape how doctors, policymakers, researchers, and the public understand health. Cause of death data tabulated from death certificates has a vital role to play in identifying disease trends, informing public health measures, and allocating billions of dollars to biomedical research every year.
Getting it right matters. The practice of attesting to a death—of issuing a death certificate—goes hand in hand with the study of death’s cause. A death is only meaningful in a statistical system if it is recorded alongside others, and recorded in the same way. Thus the history of the death certificate is caught up in the ever-evolving search for the cause of death.
* * *
A mound of earth, eroded over millennia by the actions of sun and seasons, with human bones at its core. A sprawling marble structure, topped with a lofty spherical dome and circumscribed by minarets, built on the southern bank of a great river by a king grieving the loss of his wife. A young cherry tree, branches weeping and inauspicious, planted with ceremony in a bright corner of a public garden. We are surrounded by no shortage of stories and edifices reminding us that we die. A newer, more nuanced body of myth and medicine seeks to shed light on the mystery of why we die.
For societies in which human life is lived in the service of gods and goddesses and spirits, the mystery matters little. We die because forces beyond the limits of our control make it so. But to the fifteenth century councilors in the city-states of Northern Italy, for whom the collective memory of a devastating bubonic plague in the previous century was still very much alive, death took on a corporeal dimension. Many city-states set up boards of health that drafted guidelines for containment and quarantine during outbreaks of disease and that mandated the registration of deaths in their region (Moriyama et al., 2011). The information recorded in these Renaissance death registration forms is largely identical to the information recorded in the death certificates of the present day: name, age, date and cause of death, and a loose amalgam of circumstances that are deemed relevant to the patient’s demise. The nature of the information demanded by these Italian boards of health necessitated an understanding of the etiology of death and disease. However, Italian physicians of the period classified deaths according to personal experience and intuition, and do not seem to have relied upon a standardized, systematic classification scheme.
To Enlightenment thinkers—who sought to fragment the world they observed into the tiniest possible pieces—causes of death constituted a fruitful area of study. Foremost among these nascent nosologists was Thomas Sydenham (1624-1689), whom contemporary doctor John Browne christened the “prince of practical medicine, whose character is as beautiful and as genuinely English as his name” (Sydenham, 1676). Motivated by major outbreaks of bubonic plague in London between 1661 and 1665, Sydenham compiled scrupulous notes on the diseases he witnessed and the patients he treated (King, 1970), synthesizing them into a complex network of interrelated etiologies. He studied the natural history of diseases in the same way that his contemporaries at the Royal Society studied the natural history of insects, seeking patterns and relationships that betrayed common origins (Leonard, 1990). To Sydenham, particular sets of symptoms had particular causes, and thus particular treatments (Dewhurst, 2012). His observations formed the basis of his Observationes medicae, a 470-page text first published in 1676 that would become required reading in English medical schools for two hundred years.
Sydenham’s conception of an ecology of disease is a remarkable one, genius in its originality and blatant in its shortcomings. His function-relates-to-form approach to disease speciation, a formula lent to him by his ornithologist and pteridologist colleagues, constitutes a bedrock principle in the search for causal mechanisms of disease and death. Yet Sydenham was an observational empiricist to a fault. He trusted what he could see and doubted what he could not. His system left little room for the possibility of underlying etiologies with disparate physiologic origins. Because he believed that diseases should follow types and lineages, he saw no use in studying symptomatologies that did not conform to his nosological structure (King, 1970). Variation mattered little. Chains of causality mattered a lot.
Sydenham’s Observationes medicae and the investigational methods it popularized set the standard for describing, explaining, and classifying causes of death for European doctors. But long before him, English parishes had been registering deaths and their associated causes using a system that, if a far cry from the sinuous nosology of Sydenham and his followers, played a strong role in communicating causes of death to the public at large. These were the London bills of mortality, issued—sporadically at times, more regularly at others—from the mid-sixteenth to mid-nineteenth century. Recurrent epidemics of the plague motivated their conception, and they are an early example of the collection, analysis, and dissemination of data in near real-time to inform public health actions (Boyce, 2020). The Worshipful Company of Parish Clerks hired “searchers,” usually older women with some experience in domestic nursing or informal medical treatment, to examine dead bodies and determine the reason for their demise. The data were then collated, printed, and sold to the London public as media akin to a weekly periodical (Slauter, 2011; Heitman, 2020).
A Bill of Mortality for the Town of Salem, for the Year 1782
Stakes were high in the searchers’ determination of the cause of death. To have a loved one pronounced dead of the plague was to be sentenced to quarantine, financial peril, and social disdain. Searchers were frequently accused of taking bribes from families who sought to have a death reported as anything other than plague, although archival evidence for such bribes is sparse (Munkhoff, 1999). Other causes of death were possible, and much—insight, wonderment, ridicule—has been made of these. One could die of “Winde” (severe gastrointestinal distress), or “Evil” (mycobacterial cervical lymphadenitis), or “Rising of the Lights” (croup), or “Plannet” (a lamentable misalignment of celestial bodies). The practice of compiling bills of mortality took hold throughout the English world; a 1782 bill from Salem, Massachusetts notes that five people that year simply died “suddenly” (Holyoke, 1783). The informal and provisional nosology of the bills lacked the proclaimed empirical rigor of Sydenham’s system, but it enabled the rapid distribution of data in a way that encouraged analysis and interpretation. A report sold in London in 1665 included burial data from past plague epidemics alongside blank spaces, inviting readers to write in and compare death rates as the present outbreak unfolded (Slauter, 2011). In 1662, John Graunt, a haberdasher by trade, analyzed the bills to discern trends in the severity of disease outbreaks—and thus founded the field of epidemiology, a discipline whose central proposition is that something important about living life can be gleaned by studying death in large numbers (Graunt, 1662).
The bills of mortality registered deaths more or less in aggregate, and the parish system that compiled them did not issue individual certificates for the individual dead. (Why the fifteenth-century Italian practice of registering and issuing individual death certificates never caught on any earlier in northern Europe is its own mystery.) It would take about two centuries after the publication of Sydenham’s Observationes medicae for doctors to link a rigorous disease classification system with the practice of reporting and analyzing deaths at scale—and thus generating a need for the modern death certificate. Key to this development was the establishment of population-wide civil registration programs. England passed the Births and Deaths Registration Act in 1836, establishing the General Register Office for England and Wales (GRO) the following year. The GRO began recording cause-of-death data shortly thereafter and established a program to analyze causes of death for epidemiological purposes (Eyler, 1979). In 1842, Massachusetts became the first state in the United States to establish a civil registration program, with the rest of the nation following suit by 1933.
The civil registration programs of the nineteenth century made reporting a death—and its associated cause—mandatory by law. But statisticians working at the GRO soon realized that the nosology of Sydenham and his followers—based primarily on symptomatic categories and largely ignorant of the pathophysiology of underlying organ systems—made it difficult to identify epidemiological trends that had common root causes. Worse still, the Enlightenment-era impulse to tease apart the subtlest minutiae of a problem resulted in a sprawling system of disease categories that was too detailed for robust statistical inference. The small numbers of death observed in each disease category inhibited the epidemiologists’ ability to generalize, predict, and prevent trends in disease. Simplified but loose categories such as those used in the London bills of mortality were of no use either. As William Farr (1807-1883), chief statistician at the GRO, noted in the organization’s first annual report, the vague causes of death reported on the bills—“Vomiting,” “Rupture,” “Sorelegge”—often confused a complication or symptom of a disease for the primary disease itself (Farr, 1839).
Farr and his colleague Jacob Mark d’Espine (1806-1860), a Swiss epidemiologist, were tasked at the first International Statistical Congress in 1853 to formulate a standardized list of causes of death that doctors throughout the world could use—and could understand (Hirsch et al., 2016). The work was contentious. Two years later, at the second meeting of the Congress, the organizers professed that “As perhaps could easily have been predicted, the two learned Doctors, having exchanged a certain number of written communications between London and Geneva, had to renounce any hope of agreement” (d’Espine, 1856). d’Espine, trained in pathological anatomy, focused his nosology on tracing the progression of a disease through the body. His primary division of diseases as either “chronic” or “acute” reflects a sensitivity to the phenomenology of illness, to the way a disease manifests within an individual (Hirsch et al., 2016). Farr, tasked with deriving insight from the hundreds of thousands death reports that inundated the GRO every year, saw little use in elucidating pathological processes. He instead focused on classifying diseases according to the trends they followed on the scale of entire populations. His primary classification distinguished between diseases that appeared to be “epidemic”—clustered, periodic, relational—and “sporadic”—dispersed, unpredictable, independent. d’Espine looked into bodies to classify disease; Farr looked in between them.
Farr and d’Espine sparred over their nosologies at subsequent meetings of the International Statistical Congress in 1855 and 1857. Fittingly, tragically, a death would decide whose system should form the basis of the disease classification system that is still in use today. d’Espine died suddenly in 1860, and the Congress adopted a list of causes of death based on Farr’s classification in 1864. The International Statistical Institute, which superseded the International Statistical Congress after its dissolution in 1878, revised this list in the last decade of the 1800s (Moriyama et al., 2011). By the turn of the century, statistical offices throughout Europe and the Americas were using this International List of Causes of Death to classify their deceased. The list, revised and expanded periodically throughout the twentieth century, evolved into the International Statistical Classification of Diseases and Related Health Problems (ICD) currently in use by World Health Organization member states.
With a codified list of causes of death came a standardized death certificate. The sixth revision of the ICD, finalized in 1948, introduced a format for reporting causes of death that has remained largely unchanged ever since (Moriyama et al., 2011). Statisticians and physicians at the WHO designed the certificate to address a key challenge in the large-scale tabulation of death data: how to elicit consistent and accurate reporting of the underlying cause of death from physicians working all over the world. The form permits physicians to list up to four ICD codes that forge a causal chain from immediate cause of death (i.e. J96.01, “Acute respiratory failure with hypoxia”) to the underlying cause (i.e. C34.92, “Malignant neoplasm of left lung”), which is recorded by vital statistics systems.
Perhaps Sydenham—empiricist par excellence—would be pleased to see such an approach in wide use today. But the causal, linear format of the modern death certificate complicates the reporting of death with multiple significant contributing factors—the norm among the elderly. Physicians are left with little guidance in these situations. In a handbook accompanying the (most recent) 2003 revision of the U.S. Standard Certificate of Death, the CDC merely states that if two or greater causal sequences contributed to death, the physician should choose to report the chain of events “thought to have the greatest impact” (CDC, 2003). Impact may be difficult to weigh—a narrow but consistent pair of studies has found that risk factors associated with a diversity of diseases are predictive of cause of death regardless of whether that cause is listed as an underlying or contributory factor (Crews et al., 1991; Batty et al., 2019). This suggests that single-cause reporting may underestimate the magnitude of the association between some risk factors (like hypertension) with certain causes of death (like cerebrovascular accident).
The modern death certificate is a chimeric document, both an indifferent piece of statistical data and a dreadful legal form that distresses and dismays. It functions as a shockingly intimate ticket into the lives of the departed. We use death certificates to close our loved ones’ bank accounts, to transfer their hard-earned assets like cars and homes, to erase them from voting registers. We use them to minimize, to the greatest extent possible, the wrinkles that the departed will leave on the society of the living.
Here lies a deep, uneasy tension: important and personal documents are important and personal because they make manifest what is written down for precisely the individual also named in the document. They help to establish someone’s relationship to the world, and they provide an immortal record of that relationship. A passport makes manifest one’s ability to cross borders in peace, a diploma one’s expertise in a particular subject or practical art. Death certificates reverse this logic. They do not make manifest what is already there, but rather what is now absent. They sanction the undoing of the relationships that someone had with the world. The death certificate is mortality immortalized.
The interrelated histories of the death certificate and of death classification show that the cause of death is as much an idea meant to solve practical problems—problems of healing, governing, bookkeeping—as it is a natural fact. Not only do the clinicopathologic correlations of a patient’s history influence the cause of death listed on a certificate, but so does a network of lists, forms, and rules designed by many people over many centuries to fulfill specific needs. A cause of death is to death itself what F = ma is to a falling apple: an immensely useful working model to illuminate something that remains chillingly, deeply, and unendingly mysterious.
Yet, still, getting it right matters.
References
Batty, G.D., Gale, C.R., Kivimäki, M., and Bell, S. (2019). “Assessment of Relative Utility of Underlying vs Contributory Causes of Death.” JAMA Net Open 2, e198024.
Boyce, N. (2020). “Bills of Mortality: tracking disease in early modern London.” The Lancet 395, 1186–1187.
CDC (2003). Physicians’ Handbook on Medical Certification of Death (Hyattsville, MD: Department of Health and Human Services).
Crews, D.E., Stamler, J., and Dyer, A. (1991). “Conditions other than underlying cause of death listed on death certificates provide additional useful information for epidemiologic research.” Epidemiology 2, 271–275.
Dewhurst, K. (1962). “Thomas Sydenham (1624–1689), Reformer of clinical medicine.” Medical History 6, 101–118.
d’Espine, J.M. (1856). Compte rendu de la deuxieme session du Congres International de Statistique (Paris).
Eyler, J. (1979). Victorian Social Medicine: The Ideas and Methods of William Farr (Baltimore: The Johns Hopkins University Press).
Graunt, J. (1662). Natural and Political Observations, Mentioned in a Following Index, and made upon the Bills of Mortality (London).
Heitman, K. (2020). “Authority, autonomy and the first London Bills of Mortality.” Centaurus 62, 275–284.
Hirsch, J.A., Nicola, G., McGinty, G., Liu, R.W., Barr, R.M., Chittle, M.D., and Manchikanti, L. (2016). “ICD-10: History and Context.” American Journal of Neuroradiology 37, 596–599.
Holyoke, E.A. (1783). “A Bill of Mortality for the Town of Salem, for the Year 1782.” Memoirs of the American Academy of Arts and Sciences 1, 546–550.
King, L. (1970). “Empiricism and rationalism in the works of Thomas Sydenham.” Bulletin of the History of Medicine 44, 1–11.
Leonard, A. (1990). “The Theories of Thomas Sydenham (1624–1689).” J R Coll Physicians Lond 24, 141–143.
Moriyama, I., Loy, R., and Robb-Smith, A. (2011). History of the statistical classification of diseases and causes of death (Hyattsville, MD: National Center for Health Statistics).
Munkhoff, R. (1999). “Searchers of the Dead: Authority, Marginality, and the Interpretation of Plague in England, 1574–1665.” Gender & History 11, 1–29.
Slauter, W. (2011). “Write up Your Dead.” Media History 17, 1–15.
Sydenham, T. (1676). Observationes medicae (London: Royal College of Physicians).
(1839). First Annual Report of the Registrar-General on Births, Deaths, and Marriages in England, in 1837-8 (London).