The Perfect Crime — Potassium Chloride and the Forensic Detection Gap

Potassium chloride, the forensic detection gap, and why a question asked over dinner kept me talking for an hour.

Someone asked me a question a few months ago, the kind of question that sounds almost playful when it leaves a person’s mouth and then lands somewhere in the brain where it refuses to stay quiet. We were not in a forensic context, we were not at a conference, we were simply in a room with food and the particular kind of conversation that finds its way to uncomfortable places when the wrong people are present. The question was this: George, does the perfect crime exist?

I started talking. After approximately sixty minutes I noticed that the person had stopped eating.

What follows is the condensed version of that conversation, which is to say it is the version with fewer digressions and more citations, but it is the same argument in the same order, and it arrives at the same place. The short answer to the question is yes. The longer answer is that the perfect crime has been committed more times than any forensic investigator would be comfortable estimating, that its mechanism is biochemically elegant, that its primary defense is not concealment but assumption, and that the gap between what forensic pathology can theoretically detect and what it routinely asks for is wide enough that a great many deaths classified in official documents as natural are almost certainly not.

The Ion Nobody Thinks to Look For

The substance in question is potassium chloride, abbreviated KCl, and the first thing to understand about it is that it is not exotic, not controlled in any meaningful sense, and not associated in the public or in most professional minds with the category of lethal weapons. Potassium is the predominant intracellular cation in the human body, essential for cardiac conduction, neuromuscular function, acid-base homeostasis, and the membrane potential of every cell that maintains one. It is kept in meticulous balance by the kidneys, the sodium-potassium ATPase pump in every cell membrane, and the hormonal system of aldosterone and insulin that shifts it between compartments in response to physiological demand. Serum potassium in a healthy person sits in a narrow band between 3.5 and 5.0 millimoles per litre, and when it rises above approximately 6.5 millimoles per litre, it begins to produce the electrocardiographic changes that precede cardiac arrest. Above 8 to 9 millimoles per litre, delivered rapidly, it stops the heart.

Potassium chloride in pharmaceutical form is a ubiquitous clinical agent administered intravenously for hypokalemia and as a critical component of cardioplegia, the deliberate arrest of the heart during cardiac surgery, because a heart that is not beating is considerably easier to work on than one that is. Every intensive care unit stocks it. Every operating theatre uses it. Every veterinary practice has it on the shelf, because the physiology of potassium in a dog or a cat or a horse is essentially identical to its physiology in a human, and the same agent that stops a human heart stops an animal heart, which is why veterinary euthanasia protocols across most of the world include potassium chloride as the final agent, administered after the animal is rendered unconscious by an overdose of barbiturate. Anesthesiologists handle potassium chloride routinely and without particular thought. Nurses handle it routinely and without particular thought. It is monitored, logged, and accounted for in controlled clinical environments, but outside of those environments, potassium chloride as granulated fertilizer is available in any agricultural supply store, on any major online retail platform, without prescription, without identity verification, and without any system designed to track where it goes.

The clinical concentrations used for intravenous administration are a different matter from agricultural grades, but a person with basic chemistry knowledge and access to a laboratory balance can prepare an injectable solution from the agricultural product, and the published literature on self-administered potassium overdose includes cases in which exactly this was done. What is remarkable about these cases from a forensic perspective is not that they occurred but what happened afterward, which in several instances was a death classification of natural causes and a certificate signed by a physician who saw an elderly person with cardiac history and drew the only conclusion that the evidence, examined at the level of routine post-mortem practice, appeared to support.

What Intravenous Potassium Chloride Actually Does

To understand why forensic detection is structurally difficult rather than simply neglected, it is necessary to understand the physiology of hyperkalemic cardiac arrest in some detail, because the mechanism matters for what it leaves behind, which is very close to nothing that routine post-mortem examination will find.

When potassium chloride is injected intravenously in sufficient quantity and at sufficient speed, the plasma potassium concentration rises rapidly, and the consequence of this rise is a progressive disruption of the resting membrane potential of excitable cells, particularly cardiac myocytes. The resting membrane potential of a cardiac cell is determined largely by the ratio of intracellular to extracellular potassium, and when extracellular potassium rises sharply, this ratio compresses, the resting potential becomes less negative, and the cell approaches its threshold potential. As the threshold is reached, the fast sodium channels responsible for the rapid depolarization of the action potential begin to inactivate spontaneously, because a cell sitting near its threshold is a cell that cannot generate a normal action potential and therefore cannot propagate normal conduction. The electrocardiographic consequences are visible in a sequence: first tall peaked T-waves as repolarization is affected, then widening of the QRS complex as intraventricular conduction slows, then loss of P-wave as atrial conduction fails, then a sinusoidal wave pattern as the QRS merges with the T-wave, and finally ventricular fibrillation or asystole.

The cardiac rhythm at the moment of death from acute hyperkalemia can be fibrillation, which is what a pathologist might expect from a sudden arrhythmic death, or asystole, which is what she might expect from a heart that simply stopped in a patient who had been sick for a long time. Neither presentation is specific. Neither provides a finding that, examined at autopsy in the absence of any prior suspicion, would require explanation beyond the cardiac history that most elderly patients carry in their files.

The autopsy findings in potassium-related deaths are similarly non-specific. Congestion of the pulmonary vasculature, a finding present in virtually every death that is not instantaneous. Mild pulmonary edema, again present in the majority of cardiac deaths. Occasionally petechial hemorrhages in the pleura or pericardium, which in an elderly patient with heart disease would be attributed to the cardiac history without a second thought. There is no specific organ finding, no particular color change, no characteristic smell, no histological pattern that says potassium on its own. The potassium that killed the patient is an ion, and ions do not leave the kind of morphological mark that conventional autopsy techniques are designed to identify.

The injection site, if the injection was administered through a peripheral vein in the antecubital fossa or the dorsum of the hand, may show nothing more than a needle puncture mark that is indistinguishable from the puncture marks left by therapeutic intravenous access in any hospitalized patient, or from a subcutaneous injection of insulin, anticoagulant, or any number of other medications commonly given to elderly people. If the injection is administered at the base of the skull in the hairline, a site that is anatomically accessible, that is not examined in the course of routine autopsy unless there is prior reason to look, and that is chosen precisely because it is in a location that routine examination will not reach, the external examination will be normal. The skin will be intact everywhere the pathologist looks. The finding will be the absence of a finding, which is reported as absence of external injury, which is consistent with natural death.

The Postmortem Biochemistry Problem: Why the Laboratory Numbers Are Useless

One might reasonably ask at this point whether a blood test would reveal the elevated potassium. The answer is that the blood test will show elevated potassium, and that this finding is entirely worthless as evidence of ante-mortem hyperkalemia, and that every forensic toxicologist and forensic pathologist knows this, which is why it is rarely ordered or interpreted in deaths where natural cause is assumed.

The problem is postmortem hemolysis. At the moment of death, the sodium-potassium ATPase pumps in every cell membrane stop functioning, because they require ATP, and ATP production ceases with the cessation of oxidative metabolism. Without the pumps, the carefully maintained electrochemical gradient between the intracellular compartment, where potassium concentration is approximately 150 millimoles per litre, and the extracellular compartment, where it is approximately 4 millimoles per litre, begins to collapse. The gradient collapses because potassium leaks outward through channels that remain open after death, flooding the extracellular space and, as erythrocytes begin to lyse under the same conditions, flooding the blood sample with the full contents of every red blood cell. A blood sample drawn from a corpse an hour after death will show potassium levels that are already substantially elevated above ante-mortem levels. A sample drawn six hours after death will show levels that are dramatically elevated. A sample drawn twelve hours after death will show levels that are, in most cases, so far above the physiological range as to be clinically uninterpretable as anything other than a product of postmortem change.

This means that when a clinician or a forensic pathologist orders a potassium level from postmortem blood and receives a result of 12 millimoles per litre, or 18 millimoles per litre, or 25 millimoles per litre, the result provides no information whatsoever about whether the potassium was elevated before death. The number is the noise of cellular death, not the signal of a cause.

Urine is marginally more informative, because the potassium in the bladder was filtered and excreted before death and is therefore a historical record of antemortem renal handling of potassium, and a person with acute severe hyperkalemia will, if they lived long enough for renal compensation to begin, excrete elevated potassium in the urine. But the period between the onset of hyperkalemia severe enough to cause cardiac arrest and actual cardiac arrest may be minutes, and in those minutes the kidneys may not have had time to excrete a detectable excess. The urine finding, when it exists, is supportive of the diagnosis, but its absence does not exclude it.

Vitreous Humor: The Best Available Option and Why It Is Still Not Enough

The compartment that comes closest to offering forensically interpretable potassium measurements is the vitreous humor of the eye, the gelatinous substance that fills the posterior chamber of the eyeball. Vitreous humor is physically separated from the systemic circulation by the blood-retinal barrier, it contains very few cells whose lysis could contribute potassium contamination, it is protected from bacterial degradation in the immediate postmortem period, and it equilibrates with systemic changes in potassium more slowly than blood, because the entry and exit of ions through the blood-retinal barrier is a regulated rather than a passive process.

These properties make vitreous humor the best sample available for postmortem potassium analysis, and forensic pathologists have used it for this purpose since the 1960s, when it was first suggested as a matrix for postmortem biochemistry. The problem with vitreous potassium, and it is a problem that has occupied researchers for decades without being resolved to the satisfaction of anyone who has to stand in a courtroom and defend a number, is that vitreous potassium rises in a reasonably predictable way after death even in the absence of any exogenous potassium administration, because the same process of membrane pump failure and electrolyte redistribution that destroys the interpretability of blood samples eventually affects the vitreous compartment as well, simply more slowly.

The published literature suggests that vitreous potassium rises at approximately 0.14 to 0.17 millimoles per litre per hour in the early postmortem period, a rise that has been used, with mixed success and considerable debate, as the basis for estimating the postmortem interval. The forensic pathologist who finds a vitreous potassium of 10 millimoles per litre in a body that has been dead for approximately twelve hours knows that the expected value for that postmortem interval is roughly 5 to 7 millimoles per litre, and that a value of 10 is significantly above what postmortem change alone would predict, and that this elevation is therefore suspicious. But the confidence interval around any such estimate is wide, the rate of rise is affected by temperature, by antemortem health status, by the ambient conditions in which the body was found, and by the precision with which the postmortem interval can be established, and the result is an elevated number whose elevated status is real but whose magnitude cannot be precisely attributed to exogenous versus endogenous contributions without additional information.

The bilateral asymmetry of vitreous potassium is one additional diagnostic tool: if potassium rises symmetrically in both eyes, as it does in straightforward postmortem change, this is consistent with normal physiology of death. If one eye shows a dramatically elevated value and the other does not, this asymmetry is not what postmortem change alone produces, and requires explanation. This is a finding that has been reported in cases of suspected exogenous potassium administration, but the evidence base is limited to case reports rather than controlled studies, and its diagnostic weight in court remains contested.

What the vitreous humor analysis cannot do, even when it finds an elevated potassium, is prove that the elevation preceded death rather than being a consequence of it, unless the elevation is so far above what any postmortem interval could explain, at any reasonable rate of rise, that only an ante-mortem source makes biological sense. Those cases exist. They are documented in the case report literature. They are rare, partly because truly extreme concentrations require either very large doses or very fast delivery, and partly because most of these deaths are classified as natural before anyone orders the test.

The Injection Site Nobody Examines

There is a forensic angle that the biochemical discussion tends to overshadow, and it is in some ways the most operationally important: the injection site itself, examined histologically, can provide evidence of exogenous potassium administration that biochemical analysis cannot, precisely because the concentration of injected material at the site of administration is orders of magnitude higher than anything detectable in remote compartments after postmortem redistribution.

A peripheral vein injection of concentrated potassium chloride will, if the administration is rapid and into a small vessel, produce local tissue changes that include vascular endothelial disruption, surrounding soft tissue edema, and sometimes localized hemorrhage into the perivascular space. These changes are not specific to potassium, but histological examination of the tissue around the injection site, combined with local fluid analysis for potassium concentration, can reveal a chemical gradient, from very high concentration at the injection site to progressively lower concentrations in tissue further away, that is inconsistent with the diffuse elevation of postmortem change and consistent with a bolus injection at that location.

The problem is that this examination requires that someone suspect the injection site and direct the pathologist to examine it. If the injection is peripheral, in the antecubital fossa, and if the body shows signs of prior therapeutic intravenous access from medical care, there is no particular reason to histologically examine any specific puncture mark rather than simply noting the presence of multiple venipuncture sites in a patient who was hospitalized, or whose medical history includes insulin-dependent diabetes, or anticoagulation therapy, all of which require regular subcutaneous or intravenous injections. If the injection site is at the base of the skull in the hairline, and the autopsy protocol does not include systematic examination of this location, the site is never examined at all.

The question of what standard autopsy protocol includes and excludes is not a trivial one, because most forensic jurisdictions have standard protocols that were written for the detection of conventional causes of death, that have not been systematically updated to address the detection of exogenous potassium administration, and that leave discretion about extended sampling to the individual pathologist, who exercises that discretion under the influence of the initial investigation narrative, which in cases where natural death has been assumed provides no reason to extend the standard protocol.

I should be direct about what this means in operational terms: a person killed by intravenous potassium chloride injection whose body is found in a setting consistent with natural death, who has a medical history that explains sudden cardiac arrest, who shows no external signs of violence, whose initial forensic examination generates no suspicion of foul play, and whose autopsy is conducted according to standard protocol by a pathologist who has no reason to suspect potassium administration, will almost certainly receive a death certificate attributing cause of death to cardiac failure or cardiorespiratory arrest on the background of whatever chronic disease the medical history documents. The injected potassium will have redistributed into the postmortem biochemical noise. The injection site will not have been examined histologically. The vitreous humor potassium may not have been measured at all, because nobody ordered it. And the case will be closed.

The Architecture of the Assumption

The deeper issue, the one that keeps returning regardless of which specific forensic technique is under discussion, is not the technical limitation of any particular measurement but the cognitive architecture that determines whether any measurement is performed at all. The decision to order vitreous humor potassium analysis, to examine injection sites histologically, to collect fluid from the infusion set if one is present at the scene, to treat a death as potentially suspicious rather than presumptively natural: these decisions are made by human beings under conditions of cognitive pressure that strongly favor the simplest available explanation.

The simplest available explanation for the death of a 78-year-old with documented coronary artery disease, heart failure, atrial fibrillation, and a prior myocardial infarction, found unresponsive in bed or in a chair or on the floor of a room that shows no signs of disturbance, is that the 78-year-old’s heart stopped, because a heart with all of those conditions is a heart that has been trying to stop for years. The physician who certifies the death sees the medical history, sees the body, sees nothing that contradicts the history, and certifies. The forensic pathologist, when autopsy is performed, conducts a standard examination, finds the non-specific findings of cardiac disease and organ failure, and issues a cause of death consistent with the certified cause. No one in this sequence has done anything wrong. Every person in it has applied reasonable professional judgment to the information available. The information available was shaped, from the beginning, by an assumption that nobody made explicitly but everybody shared.

This is precisely what makes the potassium chloride scenario forensically extraordinary. It does not require the perpetrator to evade competent investigation. It requires the perpetrator to ensure that competent investigation is never initiated, which is accomplished not by concealment but by plausibility. The body tells a story that is consistent with natural death. The story is good enough that no one asks for another one.

The choice of victim matters enormously in this respect. An otherwise healthy 35-year-old found dead is, in essentially every forensic jurisdiction, a mandatory autopsy case because there is no obvious explanation for the death and the absence of explanation constitutes, in itself, a forensic concern. An 85-year-old with end-stage heart failure, found dead in a care facility, is, in many forensic jurisdictions, eligible for death certification without autopsy, or for a limited autopsy that does not include the extended toxicological sampling that would be required to detect exogenous potassium administration. The gradient of forensic scrutiny that exists across the age spectrum is a direct function of the gradient of clinical plausibility, and the deaths that are most plausibly natural are the deaths that receive the least scrutiny, which makes them the most attractive targets for someone who understands this architecture.

The Cases That Were Found and the Cases That Were Not

The detection of criminal potassium administration in the published literature divides cleanly into two categories: cases where detection was achieved because circumstantial evidence forced a closer look, and cases where detection was achieved by accident or by the persistence of a pathologist who was unwilling to accept the initial attribution. There is, by definition, a third category that is entirely absent from the published literature: the cases where detection did not occur, the deaths that are currently filed under natural cause, the bodies that were cremated before anyone thought to ask a better question.

Niels Högel is the best-documented example of what happens when a healthcare setting allows multiple undetected deaths to accumulate before the pattern becomes visible. Högel was a nurse in cardiac intensive care units in Oldenburg and Delmenhorst between 2000 and 2005, and he injected patients with cardiovascular drugs including potassium, ajmaline, sotalol, amiodarone, lidocaine, and calcium chloride, not with the intention of killing them from the outset but with the intention of inducing a cardiac crisis that would allow him to perform a heroic resuscitation and receive the recognition he craved. Many of the patients he induced crises in died. He was convicted of 85 confirmed murders in 2019, but investigators suspect the actual total may be closer to 300, a number that reflects the impossibility of retrospective forensic investigation in cases where bodies were cremated and death certificates had long since been filed. The operative mechanism that protected him for years was not technical sophistication but institutional plausibility: he worked in a setting where cardiac arrest was expected, where his colleagues were overworked and under-resourced, where the individual deaths were each individually explicable by the patients’ underlying conditions, and where the only signal that something was wrong was a statistical one, an elevated death rate on his shifts, that took years to be recognized as a pattern rather than coincidence.

Orville Lynn Majors, a licensed practical nurse in an Indiana hospital in the 1990s, provides a comparable case on a smaller but equally instructive scale. Majors is believed to have used potassium chloride to kill between seven and as many as 130 patients during his tenure at the Vermillion County Hospital, and the initial detection came not from forensic pathology but from actuarial observation: a state health department survey noted that the death rate at the hospital during the periods when Majors was on duty was six times higher than during periods when he was not. Investigation that followed this statistical anomaly eventually produced the physical evidence: exhumation of patients whose bodies had not yet been cremated, and in some cases tissue analysis that revealed drug concentrations inconsistent with postmortem change and consistent with ante-mortem administration. In a former roommate’s home, investigators found vials of potassium chloride and epinephrine that could be traced to the hospital. Majors was convicted of six murders and sentenced to 360 years. The victims whose bodies had been cremated were beyond retrospective investigation.

The case that I have been studying and that appears in the Instagram post accompanying this article is a different kind of forensic puzzle, and one that illustrates the detection problem not in a healthcare serial killer scenario but in a single, isolated death that was classified as suicide without the investigation that classification should require. A physician was found dead in his own office. Present at the scene were two empty KCl ampules and an infusion needle still in his arm, the plunger of a syringe nearby, and a toxicological picture that showed therapeutic diazepam levels and no detectable potassium chloride in the blood. The official classification was suicide.

The question I keep returning to is not whether the classification was wrong, because I am not the investigating pathologist and the case is not mine to decide, but whether the available evidence actually supports the classification with the confidence that a closed case certificate implies. The toxicology finding of no detectable potassium chloride in blood is consistent with several scenarios: it is consistent with suicide by potassium chloride injection, because postmortem hemolysis renders blood potassium uninterpretable; it is consistent with homicide by potassium chloride injection, for the same reason; and it is consistent with a death from another cause entirely, in which the KCl ampules were present for reasons unrelated to the cause of death. The absence of a positive finding is not, in this context, evidence of anything other than the limit of the investigation that was performed. The vitreous humor potassium, if it was measured, was not documented in the case summary I reviewed. The injection site tissue was not histologically analyzed. The Perikardialflüssigkeit was not collected. The case was closed with the evidence available from a standard protocol, and the standard protocol was not designed to detect what might have been there to find.

What a Complete Investigation Would Actually Require

The multi-compartment approach that I have been working on, and that I intend to publish formally, is not a novel methodology so much as it is a systematic assembly of techniques that already exist in the forensic pathology literature but that are not combined into a standard protocol in any jurisdiction I am aware of. The argument is straightforward: because no single sample provides definitive evidence of exogenous potassium administration, and because the combination of samples from multiple compartments with different kinetics of postmortem potassium rise can triangulate toward or away from the hypothesis, a complete investigation should collect and analyze all of them.

Vitreous humor from both eyes separately, with the bilateral symmetry or asymmetry documented, provides the primary ante-mortem biochemical estimate. The rate of postmortem rise in vitreous potassium is well enough characterized in the literature to allow a probabilistic assessment of whether a given measured value is consistent with the postmortem interval, elevated beyond what the postmortem interval alone predicts, or so dramatically elevated that only an ante-mortem contribution explains it. This assessment requires knowing the postmortem interval as precisely as possible, which requires all of the other postmortem interval estimation tools, rectal temperature, algor mortis, livor mortis, and any available information from witnesses or electronic records, to be applied in combination.

Pericardial fluid provides a second estimate from a compartment that is more protected from the systemic circulation than blood but less protected than vitreous humor, with potassium kinetics that are intermediate between the two. The comparison of pericardial fluid potassium with vitreous humor potassium and with the expected values for each compartment at the estimated postmortem interval provides additional triangulation.

Cerebrospinal fluid, collected from the lumbar cistern rather than the ventricular system to minimize contamination from cerebral decomposition products, provides a third compartment with different kinetics and different potential sources of contamination. Its interpretation requires the same postmortem interval correction as vitreous humor, but the correction curves are less well characterized, which limits its independent diagnostic value while preserving its role as a contributing data point in a multi-compartment analysis.

Urine, when available, should be analyzed for potassium concentration and for the ratio of potassium to creatinine, which provides an estimate of the renal excretory response to potassium load in the minutes before death. A urine potassium-to-creatinine ratio that is substantially elevated above normal suggests that the kidneys were excreting excess potassium antemortem, which is consistent with hyperkalemia in the period before death.

The injection site, when identified or suspected, should be excised for histological examination and for local fluid potassium analysis. The histological changes at an injection site are not specific to potassium, but the combination of histological evidence of a recent intravenous injection, a high local potassium concentration gradient, and a postmortem biochemical picture suggesting ante-mortem hyperkalemia constitutes a triad of evidence that, taken together, is substantially more diagnostic than any single component.

When intravenous equipment is present at the scene, any residual fluid in the infusion set, the drop chamber, or the tubing should be collected and analyzed. This sample, unlike any sample derived from the body, is not subject to postmortem change and will reflect the concentration of the infused material directly. Simon’s 2023 case report, which is the most methodologically complete single-case documentation of this approach in the recent literature, demonstrated that fluid from the drop chamber of the infusion set contained the highest potassium concentration of all the samples collected, and was the most diagnostically unambiguous piece of evidence in a case that would otherwise have been forensically incomplete (Simon, 2023).

The protocol I am proposing is not radical. None of its individual components is novel. The novelty is in the systematic combination and in the argument that this combination should be the standard of care in any death where the cause is unexplained, where the deceased had access to potassium chloride in clinical concentrations, where intravenous equipment was present, or where a statistical pattern of deaths in a particular setting or under the supervision of a particular person has raised any degree of concern. The challenge is that “any degree of concern” is itself a threshold that the cognitive architecture of natural-cause assumption makes very difficult to reach.

The Statistical Dark Matter

There is a question that is impossible to answer with current data and that should haunt every forensic pathologist who has thought about this long enough: how many deaths classified as natural were not?

The published literature cannot answer this question because it documents only the cases that were detected, and the cases that were detected are, by definition, a biased sample of all the cases that occurred. What the literature can offer is some indication of the structural conditions that predict non-detection: advanced age at the time of death, significant pre-existing cardiac disease, a death setting that provides plausibility for natural cause, access by a person with relevant knowledge and means, and the absence of any prior suspicion that would trigger extended investigation. These conditions describe a substantial fraction of all deaths in industrialized societies, because advanced age and significant comorbidities are the modal profile of people who die in hospitals, care facilities, and at home.

The total number of deaths in Germany per year is approximately one million. The fraction of those deaths that receive a full forensic autopsy with extended toxicological sampling is small, estimated at approximately ten to fifteen percent of all deaths in most jurisdictions, and concentrated in the cases with obvious forensic indicators: violent deaths, sudden unexpected deaths in young people, deaths under suspicious circumstances as recognized by the initial investigation. The deaths that are most vulnerable to undetected potassium chloride administration, elderly patients with cardiac history dying in settings consistent with natural cause, are precisely the deaths least likely to receive full forensic investigation.

What this means is not that potassium chloride murder is common. It means that the detection rate for potassium chloride murder is unknown, and that the unknown quantity could be substantially different from zero in ways that the absence of detected cases cannot establish. The cases that come to forensic attention because something in the scene or the history triggers investigation are a tiny and systematically non-representative sample of all the cases in which potassium chloride administration could have caused death. Absence of evidence in this context is not evidence of absence. It is evidence of non-investigation.

The Niels Högel case illustrates the scale of the gap. He worked in cardiac intensive care for five years. He killed at least 85 people and almost certainly substantially more. During those five years, he was noticed by colleagues, discussed at a staff meeting, and allowed to continue. The statistical signal was visible to anyone who looked for it. Nobody looked for it systematically until years had passed. The forensic pathologists who performed autopsies on his victims in the normal course of events signed death certificates attributing cause of death to the cardiac conditions that Högel’s injections had exacerbated or triggered, because those conditions were real and those attributions were plausible, and because nobody presented the pathologists with the question they needed to be asked to look past the plausibility.

The Publication Coming

I want to be transparent about something: this article is, among other things, a preview of a formal publication that is in preparation. The multi-compartment approach I have described, the forensic protocol arguments, and the specific case analysis that forms the core of the evidentiary discussion here will be presented in a peer-reviewed context with the full technical apparatus that a scientific audience requires and that a blog post cannot support. The reason for writing the public version first is precisely the question I was asked at dinner, the same question that thousands of people ask, and the frustration I feel that the answer, when given in the forensic literature, stays in the forensic literature and reaches only people who were already asking it.

The gap I am describing is not merely academic. It is operational, it is present in every country that does not have a systematic multi-compartment protocol for unexplained deaths in vulnerable populations, and it will remain operational until either the forensic community develops and adopts such a protocol or the judiciary begins to require it. Neither of those things happens quickly. They happen through the kind of repeated, uncomfortable, specific argument that eventually accumulates enough mass to change what people think of as normal.

This is one contribution to that accumulation. The conversation at dinner that started it was another. The sixty minutes that followed, which the other person endured with admirable patience, were the draft. This is the version I would like to outlast me.

One Last Thought on the Word Perfect

The question was whether the perfect crime exists. I have spent the preceding pages arguing that it does, and that its perfection is not in the skill of the perpetrator but in the blindness of the system, and I want to complicate that claim slightly before stopping.

The perfect crime, in the sense that nobody is caught and nobody is found, is achievable under the conditions I have described. The perfect crime, in the sense that nobody is harmed, is not achievable by definition. The people who die when forensic investigation fails to ask the right question are dead whether or not they appear in a criminal statistic. The families who are told their parent died naturally, when the truth is that someone decided their parent would die and then ensured they did, carry a grief that forensic silence does not diminish. The injustice is not smaller because it goes unrecorded. It is only less visible.

What forensic pathology owes the people it processes is not a perfect answer, because perfect answers are not available in a discipline that operates on imperfect evidence under real constraints. What it owes them is the best answer that the available science can support, which requires asking the questions that the available science has shown to be worth asking, which requires being willing to look past the assumption that the most plausible story is the only one worth investigating.

The perfect crime remains perfect for as long as nobody looks. The answer to the question asked over dinner is yes. The better question, which takes slightly longer to answer, is: what would it take to make the looking routine?

References

Bertol, E., Politi, L., and Mari, F. (2012). Death by potassium chloride intravenous injection: Evaluation of analytical detectability. Journal of Forensic Sciences, 57(1), 273–275.

Palmiere, C., and Mangin, P. (2016). Fatal intravenous injection of potassium: Is postmortem biochemistry useful for the diagnosis? Forensic Science International, 265, 169–174.

Simon, G. (2023). Detection of fatal potassium overdose: A case report and review of the literature. Diagnostics, 13(7), 1339. doi.org/10.3390/diagnostics13071339

Tarda, L., Sacco, M. A., Gualtieri, S., Mazzuca, W., and Aquila, I. (2024). A case report and literature review of death by potassium chloride: Is it a forensic medicine enigma? Cureus, 16(4), e57776. doi.org/10.7759/cureus.57776

Wetherton, A. R., Corey, T. S., Buchino, J. J., and Burrows, A. M. (2003). Fatal intravenous injection of potassium in hospitalized patients. American Journal of Forensic Medicine and Pathology, 24(2), 128–131.

Zhang, L., Zhao, Q., Wang, Q., Zhang, P., Li, H., and Li, J. (2020). Application of scanning electron microscopy in the auxiliary diagnosis of death caused by potassium chloride intravenous injection: A case report. International Journal of Legal Medicine, 134, 1719–1725.

Zilg, B., Bernard, S., Alkass, K., Berg, S., and Druid, H. (2015). A new model for the estimation of time of death from vitreous potassium levels corrected for age and temperature. Forensic Science International, 254, 158–166.