The cause of death on a death certificate is the sequence of diseases, injuries, or conditions that led to a person’s death, written in a specific chain that traces backward from the final event to the condition that started the lethal process. This chain is not a single line but a structured medical statement that distinguishes between the immediate cause (the event that directly killed the person), the underlying cause (the disease or injury that set the fatal sequence in motion), and any contributing conditions that played a role but were not directly in the chain. Getting this right matters more than most people realize, because the data from death certificates drive everything from disease surveillance to life-insurance claims.
How the Cause of Death Section Works
The medical certification portion of a U.S. death certificate uses a two-part format established by the World Health Organization and adopted internationally. Part I asks the certifier to list a chain of events leading to death, working from the immediate cause on the top line down to the underlying cause on the bottom line. Each line also has a space for a time interval, indicating roughly how long each condition was present before death. Part II provides room for other conditions that contributed to death but were not part of the direct causal sequence.
To make this concrete: if someone with longstanding diabetes developed kidney failure, which led to a fatal heart attack, the certificate might read “Acute myocardial infarction” on the top line (immediate cause), “Chronic renal failure” on the next line (due to), and “Type 2 diabetes mellitus” on the bottom line (underlying cause). The underlying cause is the one that public health agencies use to compile mortality statistics. It is the answer to the question “what started the chain of events that killed this person?”
Separate from the cause of death is the manner of death, which classifies the circumstances into one of five categories: natural, accident, suicide, homicide, or undetermined. The manner is not a medical diagnosis but a legal determination about the context of the death. A death certificate also distinguishes between a cause and a mechanism. “Cardiac arrest” and “respiratory failure” are mechanisms, meaning they describe how the body shut down. They are not causes of death on their own, because virtually everyone’s heart and breathing stop at death. Writing “cardiac arrest” as a standalone cause without an underlying disease is one of the most common errors certifiers make.1JAMA. Cause of Death: Proper Completion of the Death Certificate
Who Fills Out the Certificate
For deaths that occur from natural disease, the attending physician, the physician who last treated the patient, or sometimes a hospice medical director is responsible for completing the cause-of-death statement. These are the majority of deaths. But when a death is sudden, unexpected, violent, or suspicious, a different system kicks in. In the United States, jurisdiction falls to either a medical examiner or a coroner, depending on the state or county. Medical examiners are physicians, often board-certified forensic pathologists, who are appointed to the role and can perform autopsies. Coroners, by contrast, are elected officials who typically are not physicians and may have limited formal training in death investigation.2American Journal of Public Health. Death Certification in the United States
This patchwork system means the quality and rigor of death certification varies widely depending on where someone dies. Some jurisdictions have centralized medical examiner offices staffed by forensic pathologists; others rely on coroners who may be funeral directors, sheriffs, or other local officials with no medical background. The practical impact is that a cause of death determined in one county may have been arrived at through a very different process than one determined in a neighboring county.
How Often Death Certificates Get It Wrong
Errors on death certificates are not rare edge cases. A systematic review and meta-analysis that pooled data from studies worldwide found a long list of recurring problems. The most common was a missing time interval, which appeared in roughly four out of five certificates. But errors that directly affect the cause of death were common too: an incorrect underlying cause of death appeared in about 39% of certificates, an improper causal sequence in about 36%, and listing a mechanism (like cardiac arrest) where an actual cause should be was found in about a quarter of cases.3PubMed. Common errors in reporting cause-of-death statement on death certificates: A systematic review and meta-analysis
Some of these errors are clerical (abbreviations, illegible handwriting, blank fields). Others are substantive and can change the statistical picture of how people die. When a certifier writes “sepsis” without tracing it back to the pneumonia, surgical wound, or urinary tract infection that caused the sepsis, the death gets coded to a vague category instead of the specific disease that public health officials need to track. When two competing causes appear on the same line, automated coding systems cannot determine which one to count.
Why are the error rates so high? Focus group research with physicians who certify deaths found several recurring themes. Many said they had received little to no formal training in how to fill out the form. Others described time pressure, uncertainty about the patient’s full medical history (especially when the patient was not their own), and the difficulty of choosing a single underlying cause when several chronic diseases were all advancing at the same time.4PLOS ONE. Provider reported challenges with completing death certificates: A focus group study demonstrating potential sources of error
What Autopsies Reveal About Accuracy
Autopsy studies consistently show a gap between what the death certificate says and what a thorough postmortem examination finds. A landmark study compared death certificates with autopsy findings and reported that in 29% of cases, a major disagreement led to reclassification of the death into an entirely different major disease category. In an additional 26%, the certificate and autopsy agreed on the broad disease category but pointed to a different specific disease within it.5PubMed. The autopsy as a measure of accuracy of the death certificate That means more than half the certificates in that study had a cause of death that was either flatly wrong or wrong in its specifics.
More recent forensic autopsy data reinforce this picture. In an analysis of 952 forensic cases, about 40% had a change from the initial assessment after the full autopsy workup was complete. Of those, roughly 17% of total cases represented a true, unexpected change in diagnosis, meaning the preliminary cause of death was substantively wrong, not just refined.6PubMed Central. Accuracy and validity of determined cause of death and manner of death following forensic autopsy prosection
Autopsy rates have fallen sharply over the past several decades in most high-income countries, which means the built-in quality check that used to catch these errors is applied far less often. The result is that the inaccuracies identified in autopsy studies likely persist in the broader population of death certificates, mostly uncorrected.
Why Older Adults and Complex Cases Are Especially Hard
Choosing a single underlying cause of death becomes genuinely difficult when someone had multiple serious chronic diseases. An 85-year-old with heart failure, diabetes, chronic kidney disease, and dementia who develops pneumonia and dies presents a real challenge: any of those conditions could plausibly be the underlying cause, and the “chain” the certifier constructs depends heavily on clinical judgment. Research on older adults with cognitive impairment shows that the distribution of listed causes of death shifts meaningfully depending on which combination of chronic conditions the person had.7PubMed Central. The Influence of Multimorbidity on Leading Causes of Death in Older Adults With Cognitive Impairment Two people who died in very similar ways might end up with different underlying causes on their certificates simply because their certifiers made different judgment calls about which disease started the chain.
This ambiguity is not a sign of negligence. In clinical trials involving older participants, determining cause of death is so difficult that researchers typically convene entire adjudication committees, teams of physicians who review the case records and try to reach consensus. Even with access to full medical records and dedicated expert time, these panels find the task challenging.8PubMed Central. Adjudication of cause of death in older adults: Learnings for death certification from the ASPirin in Reducing Events in the Elderly study A busy attending physician doing this alone, sometimes days after the death, with incomplete chart access, faces a harder version of the same problem.
Drug Overdoses and the Literal Text Problem
Drug overdose deaths illustrate a different kind of certification complexity. When someone dies of a drug overdose, the certifier needs to list the specific substances involved, often based on toxicology results that may take weeks to come back. Modern overdose deaths frequently involve multiple drugs, which makes the literal text on the certificate a dense pharmacological statement. Analysis of overdose deaths in North Carolina found that about 75% involved more than one drug class by 2017 to 2019, with an average of 2.4 drug mentions per death in 2019. Among fentanyl-related deaths that year, the most common co-occurring substances were cocaine, heroin, alcohol, and methamphetamine.9PubMed Central. Literal text analysis of poly-class and polydrug overdose deaths in North Carolina, 2015–2019
This matters for public health surveillance because national coding systems have to pick one underlying cause from that list of substances. Which drug gets counted as “the” cause can change how an epidemic looks in the statistics. If fentanyl is listed as the underlying cause but cocaine and heroin are also present, the death counts toward fentanyl in national tallies, potentially underrepresenting the role of stimulants or other opioids. Understanding what the literal text actually says, rather than just the single coded cause, gives a more accurate picture of what is killing people.
Sensitive Causes and Family Pressure
Death certificates are legal documents, and the cause of death listed on them can have real consequences for surviving family members. Life insurance policies may exclude certain causes of death. Military survivors’ benefits may depend on whether a death is linked to service-related exposures. The stigma around causes like suicide, overdose, or HIV can weigh on families. Physicians who certify deaths have described the uncomfortable position of being pressured, sometimes subtly, to shade the cause of death in a way that protects the family’s financial or emotional interests. While providers in focus group research did not report actually changing their certifications based on family wishes, they acknowledged that these conversations were difficult and created a source of tension around the process.4PLOS ONE. Provider reported challenges with completing death certificates: A focus group study demonstrating potential sources of error
The legal weight of the document also means that errors can be hard to fix. In most U.S. states, amending a death certificate after it has been filed requires a formal process involving the vital records office. If a family believes the cause of death is wrong, perhaps because later test results or autopsy findings pointed to a different cause, correcting it is not as simple as calling the doctor’s office. This procedural friction means that some inaccurate certificates remain in the record permanently.
Why Accuracy Matters Beyond the Family
Cause-of-death data from death certificates are the backbone of public health statistics in most countries. These data are used for disease surveillance, for tracking emerging health threats, for directing prevention programs, and for modeling life expectancy.10PubMed Central. Public Health Impact: How Medicolegal Death Investigation Data Help the Living When death certificates systematically undercount a particular disease, that disease gets less funding, less research attention, and less public awareness. Diabetes is a well-known example: it is frequently listed as a contributing condition rather than the underlying cause, which means national statistics underrepresent how many people die because of it.
During the COVID-19 pandemic, the accuracy of cause-of-death certification became a public controversy. Guidelines were issued specifically for how to document COVID-19 on death certificates, including whether it should be listed as the underlying cause or as a contributing condition depending on the clinical circumstances.11PubMed Central. Cause of Death Certification in COVID-19 Deaths The debate over whether COVID deaths were being “overcounted” or “undercounted” was fundamentally a debate about how cause-of-death chains were being constructed and coded. The same underlying ambiguity that has always existed in death certification was suddenly visible to the general public.
Electronic Certificates and Automated Coding
One of the most promising developments in death certification is the shift from paper to electronic forms. Paper certificates are plagued by illegible handwriting, blank fields, and formatting errors that make automated coding difficult. Electronic systems can build in real-time validation: prompting the certifier when they list a mechanism without an underlying cause, flagging missing time intervals, or rejecting vague terms that are not useful for coding.
Norway’s experience offers a concrete example. The country moved rapidly from zero electronic death certificate submissions in 2017 to 95% by 2022. Looking specifically at diabetes-related deaths, the shift to electronic filing was associated with dramatically less use of unspecified diabetes diagnoses, fewer certificates that needed to be corrected after automated processing, and less use of vague or ill-defined diagnosis codes.12PubMed Central. Introduction of electronic death notification in Norway-Impact on diabetes mortality registration In other words, the electronic forms nudged doctors toward more specific and accurate cause-of-death statements.
On the coding side, automated systems like IRIS (used internationally) can process death certificates and assign underlying cause-of-death codes without manual intervention for a substantial portion of cases. In one implementation study, the system handled about 69% of certificates without any human input, and among those, 78% received exactly the same four-digit disease code that a human coder would have assigned.13PubMed. The implementation of an automated coding system for cause-of-death statistics That leaves room for improvement, but it also means the coding process can be made faster and more consistent, freeing human coders to focus on the genuinely ambiguous cases.
France has also studied the quality difference between electronic and paper certificates, evaluating them on the amount of information provided, the internal consistency of the causal chain, and the precision of the underlying cause selected.14PubMed Central. Quality comparison of electronic versus paper death certificates in France, 2010 These kinds of head-to-head comparisons help health systems build the case for full digital adoption, which remains uneven globally.
Forensic Imaging as a Complement to Autopsy
For suspected unnatural deaths, the traditional gold standard has been a full forensic autopsy. But postmortem CT scanning is increasingly used as either a complement or a preliminary step. A study comparing forensic CT findings with autopsy results in suspected unnatural deaths found strong agreement on the primary cause of death, with concordance reaching about 86% before toxicology and histology results were considered and roughly 95% when those results were factored in. For specific injury types like gunshot wounds, intracranial bleeding, and strangulation, CT correctly identified the cause in every case examined.15PubMed. Forensic postmortem computed tomography in suspected unnatural adult deaths
Postmortem imaging does not replace autopsy entirely. It cannot detect many types of poisoning, subtle organ-level disease, or causes that require tissue samples under a microscope. But it offers a noninvasive option that can be performed quickly and may satisfy the needs of some death investigations without the time and resources of a full dissection. In jurisdictions that lack forensic pathologists, or for families with religious or personal objections to autopsy, imaging can fill part of the gap.
Molecular Autopsy for Unexplained Sudden Death
Sometimes even a full autopsy cannot explain why someone died. This happens most strikingly in cases of sudden cardiac death in young, apparently healthy people. The heart looks structurally normal, toxicology is negative, and the forensic pathologist is left without a diagnosis. The death certificate in these cases may read “sudden unexplained death” or “sudden arrhythmic death syndrome,” which is essentially an admission that no cause was found.
Molecular autopsy offers a way forward. By extracting DNA from postmortem blood and sequencing genes associated with inherited heart rhythm disorders, pathologists can sometimes identify a genetic mutation that explains the death. Studies have found that molecular autopsy identifies a likely genetic cause in up to about 30% of sudden unexplained death cases in young people.16European Heart Journal. Sudden cardiac death in the young: the molecular autopsy and a practical approach to surviving relatives Earlier research focused specifically on cardiac ion channel genes found pathogenic mutations in over a third of sudden unexplained death cases tested.17PubMed. The role of molecular autopsy in unexplained sudden cardiac death
The value extends beyond the person who died. If the molecular autopsy reveals a mutation linked to a condition like long QT syndrome or catecholaminergic polymorphic ventricular tachycardia, surviving family members can be tested for the same mutation. Those who carry it can be monitored and treated before the same thing happens to them. In cases of sudden infant death, similar genetic testing has identified mutations in roughly 5 to 10% of cases, potentially reframing what had been classified as unexplained crib death into a diagnosable and, for surviving siblings, preventable condition.17PubMed. The role of molecular autopsy in unexplained sudden cardiac death The molecular autopsy does not change the fact that someone has died, but it can change what happens next for the family they left behind.18PubMed Central. Molecular autopsy in sudden cardiac death