On January 31, 2000, Dr. Harold Shipman was convicted at Preston, England, of murdering 15 of his patients by administering lethal doses of diamorphine (pharmaceutical heroin). Investigations indicate that, during his working life, he killed about 220 to 240 of his patients. The bodies of many victims were cremated. Twelve victims were exhumed, and 9 of these deaths were included in the indictment. Most victims were elderly and had histories of natural disease. Autopsies confirmed known natural disease but showed no evidence of acute lethal events. Analysis of skeletal muscle disclosed significant quantities of morphine, to which the deaths were attributed. Circumstantial evidence was strong, as illustrated by the convictions in 6 deaths without autopsy or toxicology, because the bodies had been cremated. Organic compounds are remarkably stable in buried bodies. Even so, detection and quantitation of morphine in exhumed bodies may become problematic after burial for 4 years or more. Morphine glucuronide is slowly converted back to free morphine in the buried corpse.
On January 31, 2000, Dr. Harold Frederick Shipman was convicted at Preston, England, of murdering 15 of his patients and of forging a will of 1 of them. The evidence was that he had killed them by administering lethal doses of diamorphine (pharmaceutical heroin). He was sentenced to 15 concurrent terms of life imprisonment and was told by the judge that in his case life imprisonment would mean that he would remain in prison until his death. Until his arrest for murder in September 1998, Dr. Shipman had been a well-respected family physician in Hyde, near Manchester, where all 15 of his victims had lived. His life chronology is set out in Table 1. The police investigation of Dr. Shipman was initiated following his crude and hopelessly incompetent attempt at forging the will of 1 victim (Table 2).
Shipman had a conviction for dishonestly obtaining drugs to feed an addiction during 1975, when he was 29 (Table 1). The police considered that he had obtained diamorphine for his killings in the same way that he had obtained pethidine for his addiction in 1975. At that time, he issued prescriptions for patients who required the drug and for those who did not, retaining some or all the drug himself. The police discovered discrepancies in the records for diamorphine. These discrepancies included prescriptions dispensed but not listed in the medical records; prescriptions dispensed for recently dead patients; the amount of drugs, recorded as delivered to the patients’ homes by nurses, being less than that dispensed; and unused diamorphine being taken from patients’ homes for destruction by Dr. Shipman. From 1993 to the time of his arrest in 1998, the police estimate that some 20,000 mg of prescribed diamorphine was available to Dr. Shipman.
In prioritizing cases for investigation, the police targeted incidents in which Dr. Shipman was with the deceased hours before death, as corroborated by documents or witnesses. They looked also for some of the following elements: an alteration of the medical records, fabricating a relevant clinical history to support the stated cause of death; inconsistencies between the stated cause of death and the medical history; an unsolicited visit by Dr. Shipman on the day of death; inconsistencies between the cremation certificate and the investigative facts; mistruths told by Dr. Shipman to family and friends of the deceased; Dr. Shipman either finding the body or being present at the time of death; and death in the physician’s rooms, of which there were 5 throughout a 2-year period.
By the time of the start of the trial in 1999, the police were investigating 137 deaths. At the conclusion of their investigations in May 2000, after the trial, they had investigated nearly 200 deaths. In addition to the 15 cases forming the basis of the conviction, the police believed there was sufficient evidence of murder in a further 24. In another 2 cases, prosecution was regarded as not in the public interest, because the patients were already terminally ill. In 86 cases, the evidence was regarded as insufficient for prosecution and in 69 gave no reason for suspicion.
After the arrest of Dr. Shipman, the local health authority undertook reviews of the deaths of his patients by using the data available. 1 They found that the death rate among women aged 65 years or older within that area was 2.7 per 100, but for Dr. Shipman’s patients, the rate was 26.0 per 100. In addition, 66.7% of Dr. Shipman’s patients died in their own homes, but the proportion for all deaths in the area was only 19.2%. For patients aged 65 years and older (both men and women), the rates of death among Dr. Shipman’s patients were higher than expected between 1987 and 1989, below the expected rate for 1990 to 1992, and higher than expected for all years from 1993 to 1998.
The Department of Health commissioned a more elaborate statistical analysis 2 of deaths in Dr. Shipman’s clinical practice (Table 3). The review examined 3 main sources of information, namely, surviving clinical records, cremation certificates, and death registration data. Two hundred eighty-two clinical records were available, including those of the 15 patients he had been convicted of murdering. The records were reviewed in light of what were considered to be the typical features of the 15 murders: all were older women; Dr. Shipman was present at death or shortly before death; death occurred suddenly at home and, in 14 of the 15 cases, in the afternoon; and there was only a weak association between the clinical history and the cause of death as certified. Of the 267 clinical records reviewed, 180 involved cases in which Dr. Shipman himself had issued the medical certificate of cause of death (MCCD). Of these, 102 (57%) were classified as highly suspicious, 39 (21.8%) as moderately suspicious, and 38 (21.2%) as not suspicious.
United Kingdom cremation certificates contain information such as details of persons present at the death and the mode, duration, and time of death. Cremation certificates were available for 292 of Dr. Shipman’s patients and 475 from comparison general practitioners. The key findings were that Dr. Shipman’s patients were more likely to be reported as having died in the afternoon (55% between 1300 and 1900 HR in comparison with 25% for the other general practitioners), with Dr. Shipman present (19.5% contrasted with 0.8% of the comparison practitioners) or with no one present (40.4% versus 19.0%) and relatives or caregivers less likely to be present (40.1% versus 80.2%), with death occurring in a short time (in 29 minutes or less for 60.4% of Dr. Shipman’s patients contrasted with 22.7% for the other general practitioners).
The third method of review involved death registrations. The Office of National Statistics provided information on death notifications where the medical certificate of cause of death had been completed by Dr. Shipman, 1 of a matched sample of 6 Hyde general practitioners for 1977 to 1998, or 4 comparison practitioners in Todmorden for 1973 to 1976. The estimated excess of deaths from 1974 to 1998 among patients dying at home or on practice premises was 236 (95% confidence interval, 198-277). There were 6 deaths on practice premises. The excess number of deaths among all patients for whom Dr. Shipman had issued a medical certificate of cause of death was 297 (95% confidence interval, 254-345). The numbers of excess deaths distinctly diverged from the expected level for women aged 65 to 74 years from 1975 onward, for women younger than 65 years from 1986 onward, and for men aged 65 to 74 years from 1989 onward. For men younger than 65 years, there was no accumulation of excess deaths. The excess numbers of deaths were highest from 1995 onward, but this trend was not a sudden new feature of Dr. Shipman’s clinical practice: it was an exacerbation of the trend traceable to his earliest years as a family practitioner. There was a marked decline in the annual number of deaths between 1990 and 1992, and the numbers classified as suspicious in these years were also low. This period represents the 2 years leading up to, and the year after, Dr. Shipman’s departure from the group family practice to set up himself in practice alone (Table 1).
The evidence that emerged at the trial raised general concerns about the regulations applying to controlled drugs, the effectiveness of the death investigative system, and a range of other problems. One year after the conviction, on January 31, 2001, Parliament appointed a judge to conduct a public inquiry. The terms of reference for the judicial inquiry encompassed suspicious deaths that occurred at any time during Dr. Shipman’s professional career. Before the establishment of the inquiry, the local coroner conducted inquests into the deaths of 27 former patients of Dr. Shipman and reached a verdict of unlawful killing in 25 with an open verdict in 2. The government had granted permission to the coroner to open inquests in about 260 deaths in all, but a decision was later made to adjourn those inquests until the findings of the public inquiry were available.
The judicial review 3 considered 526 deaths in which the medical certificate of cause of death was issued by Dr. Shipman and classified them as unlawfully killed, probably unlawfully killed, a real suspicion of unlawful killing, probably natural, natural, and insufficient evidence. The inquiry conducted an exhaustive case-by-case investigation, reaching an individual determination for all of the cases. Two hundred twelve deaths were classified as unlawful killings or probable unlawful killings, and a further 43 cases as suspicious of unlawful killing, giving a total of 255 cases. Two hundred forty-four of these 255 deaths occurred in the patient’s home or on the physician’s practice premises. In reaching these conclusions, the inquiry called on evidence of relatives, friends, and health professionals, and used other investigatory information, including additional documentary evidence that had not been previously available. Transcripts, statements, and documents can be found on the inquiry Web site (http://www.the-shipman-inquiry.org.uk).
The result of all of the investigations conducted thus far indicate that it is possible to have confidence in the general conclusion that Dr. Shipman, during his working life, murdered about 220 to 240 of his patients.
Of the 15 deaths forming part of the criminal indictment, only 9 were subject to exhumation, autopsy, and toxicological analyses. The other 6 bodies had been cremated. The key circumstantial evidence surrounding these 6 deaths is set out in the appendix.
In the United Kingdom, the proportion of deaths followed by cremation has been gradually increasing, from 55% in 1970, to 65% in 1980, to 72% in 1997. As a consequence, the retrospective investigation of the deaths of Dr. Shipman’s patients presented inevitable difficulties.
The medical histories, certified causes of death, and autopsy findings in the 9 exhumations forming part of the criminal indictment are set out in Table 4. Highlighted in the Table are those aspects of the medical history for which there was computer evidence of falsification. The evidence for this falsification was a discrepancy between the true entry date and time according to the internal computer clock and the stated entry date and time. The interval between the death and the autopsy after exhumation ranged from 38 to 852 days. The data on date of death, burial, exhumation, and autopsy are set out in Table 5. Samples available for toxicology analyses were limited, and a decision was made to analyze skeletal muscle obtained from the anterior midpart of the thigh. In the first few cases, liver was analyzed also. The analytical results are set out in Table 6. Some indication of the relative preservation of the tissues can be gleaned from the weights of the liver and the water content of the muscle. In addition, the weights of the heart and lungs and their water content are set out in Table 7. No toxicological analyses were performed on the heart and lungs. Three additional cases subject to exhumation, autopsy, and toxicology did not form part of the criminal indictment but were the subject of inquests after the trial. The relevant data in these cases are set out in Tables 8 and 9.
The prosecution relied heavily on the circumstantial evidence, which was extremely strong, as indicated by the 6 convictions in deaths in which the bodies had been cremated and no pathology or toxicology was available (see the appendix). At trial, the prosecution took the position that the morphine concentrations in the 9 exhumed bodies indicated a substantial dose of either morphine or diamorphine, which must have been the cause of death, notwithstanding any natural disease present. The accuracy and precision of the morphine analyses (immunoassay and gas chromotography-mass spectroscopy) were undisputed. The morphine concentrations in skeletal muscle were comparable with those of a previously published case of exhumation after homicidal poisoning 4 (table 10) and with published levels of morphine in skeletal muscle in morphine fatalities generally. 6–8 Theoretically loss of tissue water content over time might lead to a relative increase in the concentration of the morphine detected. 4 This was addressed by the prosecution analyst through the measurement of the water content of the muscle samples, which proved comparable to the expected normal approximately 74% hydration (Table 6). Similarly, the validity of heart weights was also assessed by means of tissue hydration (Table 7). The latter showed considerable intersubject variability, which proved of little or no consequence to the arguments presented at trial.
Over-the-counter medications containing morphine are available in the United Kingdom. Tablets may contain up to 0.4 mg morphine, and liquids may contain up to 0.02%. These products would be insufficient, the prosecution successfully argued, to account for the substantial levels of morphine found in the bodies, implying that the morphine originated from either morphine or diamorphine, which was either pharmaceutical or illicit. Dr. Shipman claimed in the case of K.G., whose will he had forged and whose death was the first investigated by police, that for some time he had suspected she might be abusing drugs. This remained his position at trial. With respect to the other deaths, the defense was in the difficult position of having no plausible explanation for the substantial levels of morphine found in the bodies. Head hair analysis in 8 of the 9 exhumations, including K.G.’s, disclosed only trace amounts of morphine, consistent with contamination in the postmortem period. The low levels were incompatible with established opiate abuse, and the absence of detectable acetylmorphine excluded the regular use of heroin. In one case (case 6, I.L. in Tables 4–7) a high morphine concentration of up to 7.0 ng/mg was found in the hair. It was speculated that this amount might be the result of degradation of prescribed pholcodine.
There was an exchange of expert reports between prosecution and defense before trial. As a possible result, the prosecution confined its scientific argument to the position that the substantial quantities of morphine found in the bodies must represent the cause of death and that the circumstantial evidence indicated that Dr. Shipman had administered the drug. They did not seek to establish by scientific means whether the drug administered was morphine or diamorphine, the estimated dosage, the route of administration, or an estimated time from administration to collapse and death.
This series of fatalities from diamorphine poisoning reinforces the established wisdom that organic compounds are remarkably stable in buried bodies and that a range of drugs may be detectable in exhumed corpses. 4 The present cases (Tables 8 and 9) suggest that detection and quantitation of morphine in exhumed bodies may become problematic after burial for 4 years or so. It is also apparent (Table 6) that morphine glucuronide is slowly converted back to free morphine, so that the longer a body is buried, the more likely it is that all the morphine detected will be free morphine.