CEMACH

Chapter 15: Pathology

Harry Millward Sadler

Introduction

In this triennium, 2003-05, the pathology assessor reviewed 353 case report forms. Virtually all of the report forms for the Direct and Indirect deaths were assessed: 120 out of 132 and 161 out of 163 respectively. The remaining assessed deaths were from Coincidental or Late causes.

For 102 of the 353 deaths there was either no postmortem, or the postmortem report was not available or not released by the coroner. Some of these were documented biopsy-confirmed cases of progressive malignancy where an autopsy would not have added value, in others the reports could not be released as the deaths were sub judice. In a few cases the coroner refused to release the autopsy report. However, in at least 17 of the assessed cases there were clinical issues that should have been addressed by a properly conducted autopsy. 

The quality of the autopsy reports

As in previous Reports, the quality of the autopsy reports fall into five categories: excellent, good, adequate, poor and appalling and these groups are used again here. It must be commented that although some reports are categorised as poor because they did not sufficiently address pertinent clinical issues or differential diagnoses, they were adequate for the purposes of a coronial autopsy as they identified the ultimate cause of death. The numbers of reports in each category are shown in Table 15.1.

Table 15.1 Quality of autopsy reports assessed; United Kingdom: 2003-05.

Excellent
Good
Adequate
Poor
Appalling
Total
  n (%) n(%) n(%) n(%) n(%) n(%)
Direct 28 (27) 27 (26) 25 (24) 19 (18) 3 (3) 103 (100)
Indirect 21 (18) 46 (40) 19 (17) 25 (22) 4 (3) 115 (100)
Coincidental 4 (12) 10 (30) 13 (39) 6 (18) 0 (0) 33 (100)
Total 53 (21) 83 (33) 57 (23) 50 (20) 7 (3) 251 (100)

 

Whilst it is gratifying that the proportion of “excellent” and “good” autopsy reports is increasing, it is still disappointing that there were seven appalling reports. Although the number was higher than in the previous triennium, because the overall numbers had risen, the proportion remained similar, as Table 15.2 shows.

Table 15.2 Quality of autopsy reports of Direct and Indirect deaths assessed; United Kingdom: 1997-2005.

 
Excellent
Good
Adequate
Poor
Appalling
Total
  n (%) n(%) n(%) n(%) n(%) n(%)
1997-99 13 (16) 29 (36) 21 (26) 11 (14) 7 (9) 81 (100)
2000-02 28 (21) 44 (32) 40 (29) 21 (15) 3 (2) 136 (100)
2003-05 49 (22) 73 (33) 44 (20) 44 (20) 7 (3) 218 (100)

 

The most consistent failings were a lack of correlation of the clinical history with the pathological findings, and to be satisfied with just establishing the immediate cause of death. Her Majesty’s Coroners are empowered to establish how, where and when a death occurred. However, coronial practice seems to vary: most seem to accept that an understanding of the processes resulting in an untimely and unexpected death is necessary but other coroners will only accept investigations adequate to establish the immediate cause of death. In this review there have been several instances where the postmortem has given a final cause of death without determining the processes leading up to death. This is unsatisfactory not only for the purposes of critical clinical review of standards of care but also because the next of kin, not unnaturally, wish to know not only the cause of death but also to understand why it occurred. There are also instances cited in this Report where the available information suggests that the stated cause of death is wrong.

The fundamental point is that maternal deaths are often complicated and an accurate, detailed autopsy is required; a poor autopsy is at best unhelpful and at worst misleading.  A poor autopsy satisfies no one despite fulfilling the technical legal requirements. As high quality autopsies can be conducted under the auspices of the coroner, it is difficult to know whether poor quality autopsies are created by the restrictions placed on the pathologist by individual coroners or whether the restrictions are a convenient smoke screen for poor performance. The assessors were also concerned about some cases where there was no autopsy at all:

A new mother who had had an uneventful pregnancy and birth telephoned a relative complaining of a very severe headache. She deteriorated rapidly and died before she arrived in hospital. A close family relative had had a subarachnoid haemorrhage in the past and because of this, and the clinical picture, the Procurator Fiscal allowed a certificate of death from subarachnoid haemorrhage to be issued without an autopsy.

There were many other possible causes of her death, for example this could have been a direct death from cerebral vein thrombosis. It is not uncommon for deaths from cerebral haemorrhage diagnosed on imaging not to have an autopsy so unfortunately the underlying cause is not established. There are similar examples involving other disease categories in this triennium. Particular and detailed recommendations concerning specific types of maternal death have been made in previous triennial Reports2 and publications3. Though still relevant they are not repeated here unless there is a particular need for emphasis from the current review.

Direct deaths

Deaths from pulmonary embolism were still the largest category of Direct deaths, but in this triennium there have been apparent increases in deaths from amniotic fluid embolism and from sepsis.

Pulmonary embolus 

Death from pulmonary embolus was given as the cause in twenty-three autopsy reports, of which eight were excellent or good, nine were adequate but six were poor or appalling. Again there was a common failing not to search for evidence of previous episodes of pulmonary embolus:

A woman who had a termination of pregnancy presented to the Emergency Department (ED) with chest pain few weeks later. She had tachycardia and was apyrexial but, despite this, she was diagnosed as having pelvic inflammatory disease. She died one week later. 

Her autopsy report was less than one side of a sheet of A4. It identified impacted emboli in the pulmonary arteries of both lungs and stated there was no evidence of embolism (sic) in the pelvic veins. There was no histology or attempt, even macroscopically, to demonstrate previous pulmonary emboli. There was also no attempt to confirm or refute the clinical diagnosis of pelvic inflammatory disease. Pregnancy was not stated in the cause of death although in the opinion of the assessors it was a contributing factor of major significance.

This case contrasts dramatically with those who had excellent reports: in one, thrombi in the internal iliac vein and tributaries were histologically confirmed as recent in some veins and organising in others. Many small pulmonary vessels were undergoing extensive recanalisation but the major pulmonary emboli were confirmed as fresh thrombi with no organisation.

Box 15.1

Pathology learning points: pulmonary embolism

At autopsy:

Predisposing causes for thromboembolism should be identified.

The source of the emboli should be stated.

Evidence for episodes of thromboembolism preceding death should be sought.

Pre-eclampsia/ eclampsia

The case reports of 16 of the 18 women who died from pre-eclampsia or eclampsia were reviewed. Eleven were associated with cerebral haemorrhage or infarction, three with the HELLP syndrome and one was an anaesthetic death. The remaining woman had pre-eclampsia but her immediate cause of death was necrotising fasciitis involving the cervix and uterus. Three women who died from cerebral haemorrhage did not have an autopsy although they were referred to the coroner.  Nine of the remaining 13 reports were good or excellent and there was only one poor autopsy:

A non-English-speaking woman was admitted for induction of labour. On admission her blood pressure was 136/83 mm/Hg and the diastolic remained at this level through most of her labour, which lasted around ten hours. Her raised blood pressure was not treated. During the second stage of labour she had a convulsion, at which point her blood pressure had risen and she had marked proteinuria. A CT scan revealed a cerebral haemorrhage and she died in intensive care. The postmortem report identified cerebral haemorrhage but there was no clinical resume, the macroscopic description of all organs was exceedingly brief and there was no attempt to perform any histology.

In four other cases there was a very rapid rise in blood pressure during labour, leading directly to death. The rapid escalation and/or onset of fulminating hypertension in pre-eclampsia has been noted in previous Reports. No deaths from fulminating pre-eclampsia arising between antenatal reviews occurred in this triennium, but this is a possibility that may present as a sudden community death to be investigated by a pathologist. In such instances histology may be the only evidence for the sequence of events.

Haemorrhage

Thirteen of the seventeen maternal deaths from haemorrhage have been reviewed. Two were associated with placenta previa and two with abruption. The remaining nine cases were all postpartum haemorrhage. Of these, three women had refused blood products, one was associated with postoperative infection of the cervix and uterus, and one occurred several days postpartum.

No autopsy was performed in two cases although both were referred to the coroner. One of these cases was postpartum haemorrhage of unknown cause where amniotic fluid embolus could not be excluded from the differential diagnosis. The autopsy reports were good or excellent in seven, adequate in one and poor or appalling in three cases. This failure to address the clinical issues is apparent in all the poor reports and sometimes the quality of the autopsy report is such that the diagnosis still seems in doubt:

A known drug user concealed her pregnancy until admitted in labour. Shortly after her delivery she left hospital. Lignocaine was then found missing from the ward trolley. Three days later she was found dead and the lignocaine was found beside her. 

Her postmortem took place some days after death. The autopsy report contained no clinical history. In the report her liver was described as showing acute shock due to blood loss but the features were not identified. There was no description of blood loss on the external examination of the body and no source of bleeding was identified in the report. No  histology reported. Toxicology was taken but lignocaine was not assayed. Nonetheless the cause of death was given as postpartum haemorrhage.

A non-English-speaking woman had recurrent bleeding and was known to have a cervical fibroid. Heavy bleeding occurred in her second trimester and, although her husband said he would take her to hospital, he didn't. She collapsed at home the next day and required emergency admission. At this point she was unconscious, her wrists had been hurt and were bandaged, her haemoglobin was 7g and there was intrauterine death. At hysterotomy a partially organised retroplacental clot was found. She had severe disseminated intravascular coagulation (DIC) and a CT scan showed contusions associated with small undisplaced fractures of her temporal and frontal bones. She died despite neurosurgical intervention for subdural haematoma. 

In the autopsy report there was no mention of her wrist injuries, no reference to the DIC or its possible causes, and no examination of her skull. The report then identified a swollen brain with no signs of raised intracranial pressure.

Amniotic fluid embolus

Seventeen Direct deaths plus two Late Direct deaths were attributed to amniotic fluid embolism this triennium, a considerable, though not statistically significant, increase. The possible reasons for this have already been discussed in Chapter 5 - Amniotic Fluid Embolus (AFE), and include variation by chance, increased case ascertainment, less stringent diagnostic criteria and increased use of prostins. Two of the deaths were Late deaths with mothers surviving in intensive care for prolonged periods after the precipitating event so that the diagnosis is of necessity based on the clinical features without pathological correlation. One other death was diagnosed clinically as AFE where autopsy was not performed. There were three other deaths in which the diagnosis was also made clinically but not confirmed at autopsy, for example:

A woman was admitted in her third trimester because of hypertension and proteinuria; labour was induced. She had a rapid delivery but was transferred to theatre because it was thought she had delivered an incomplete placenta. At this stage her observations were stable but she suddenly collapsed with very low blood pressure. She was resuscitated but arrested again and died shortly afterwards.

At autopsy there were numerous petechial haemorrhages on serosal surfaces and pallor of the renal cortex. Extensive histology demonstrated the features of severe pre-eclampsia. There were no lacerations in her genital tract and no retained products were found, neither was any evidence of amniotic fluid embolism identified within the pulmonary vessels. Despite this, death was given as an anaphylactic reaction against amniotic fluid.

In 11 of the remaining deaths there was good clinicopathological correlation. In two there was an atypical clinical presentation and amniotic fluid embolus was not suspected clinically but was demonstrated at autopsy. For example:

A mother presented at term in spontaneous labour with fetal distress. Arrangements were being made for an emergency caesarean section when she became unrousable. She was resuscitated and at operation there was an excess of blood in her abdomen. An exploratory laparotomy revealed a two centimetre tear on the inferior aspect of the left lobe of the liver. Despite transfusion and blood products she continued to bleed from the site, though not elsewhere, and despite transfer to a liver unit, she died.

A thorough and detailed autopsy confirmed the liver rupture with no underlying liver pathology on histology. Cytokeratin positive squames were found in her lung capillaries on immunochemistry. The cause of death was attributed to spontaneous rupture of the liver with the amniotic fluid emboli thought to be resuscitation ‘artefact’ but it is also conceivable that the vigorous resuscitation caused rupture of the liver in a patient with a bleeding diathesis from amniotic fluid embolus.

These cases illustrate our poor understanding of the underlying mechanisms surrounding AFE and the consequent difficulties for making an accurate diagnosis. There is evidence that foetal squames can be found in the maternal pulmonary circulation without initiating the clinical syndrome4 suggesting that there is either an idiosyncratic response or a threshold effect. In the majority of cases the diagnosis will be obvious both clinically and pathologically and frequently foetal squames will be visible in the maternal pulmonary circulation on routine histological stains. There are obviously cases at both ends of the clinical and pathological spectrum when very careful clinicopathological correlation is a necessity. In these situations immunocytochemistry for foetal squames is mandatory and will usually resolve the dilemma. It is possible that foetal squames are a surrogate marker for other components of amniotic fluid that trigger an immunological reaction in a susceptible individual5. Consequently, in situations where the clinical and pathological findings lack concordance further investigations are indicated. An immunological reaction to the foetal sialyl Tn antigen has been found and the essence of this antigen can be demonstrated by immunocytochemistry6,7. In the light of our current knowledge, this would be an appropriate further investigation for such diagnostic conundrums. 

Box 15.2

Pathology learning points: AFE

The diagnosis of amniotic fluid embolism needs careful and critical pathological evaluation.

Difficult cases require detailed immunocytochemical studies for fetal squames and possibly for fetal mucins.

Sepsis

Puerperal sepsis is increasing in incidence. The Health Protection Agency's enhanced surveillance of all Group A streptococcal infections in 2003 identified a total of 2085 cases with an overall mortality of 27%. Three per cent of the total cases were women with puerperal sepsis8. In this Report the deaths of 19 deaths women have been directly attributed to sepsis. Overall the most common predisposing factor was an association with obesity as eight of the deaths were in women with a Body Mass Index (BMI) over 30. Review of the reports suggested that there was increased difficulty in diagnosing severe infections in such women.

The most common organisms were eight cases of streptococci and six cases of coliforms, often admixed with enterococci. There were also three exotic infections by citrobacter koseri and listeria monocytogenes as well as aeromonas subrea in an HIV positive patient. Two women had predisposing immunodeficiency states, one of whom was known to be HIV positive. The other had incontinentia pigmenti, a condition associated with the NEMO gene, abnormalities of which not only give rise to incontinentia pigmenti but can also be associated with immunodeficiency states9.

Of the eight maternal deaths associated with streptococcal infection, one was due to ascending genital tract infection by strep pneumoniae but the other cases were all from Group A. Though not directly comparable with this review, the HPA’s figure for streptococcal puerperal sepsis (3%) suggests that maternal death is a relatively rare complication of the infection. Coliform infections were usually associated with premature rupture of membranes or the insertion of cervical sutures.

The reports of 20 women who allegedly died from infection have been reviewed although not all of them are counted in Chapter 7 - Sepsis. In two cases no autopsy was conducted, one of whom died from necrotising fasciitis. Of the eighteen autopsy reports, nine were good or excellent and three adequate but six were poor or appalling.

All of the good or excellent reports had undertaken careful clinicopathological correlation and investigations relevant to the clinical circumstances. In most this involved appropriate microbiology and more specifically a search for the portal of entry. This search was inconclusive in several instances but in two cases there was clear histological evidence of inflammation centred on the genital tract and with a chorio-amniitis and funisitis in the placenta and cord. One other had not only demonstrated chorioamnitis due to enterococci and coliforms but had excluded streptococcal infection by polymerase chain reaction.

A frequent feature in the poor postmortem reports was inconsistency within the report itself. For instance, in one report the heart weighed 161 gm but was described as showing left ventricular hypertrophy. The most common deficiency however was a failure to address the clinical problems. In one death from streptococcal puerperal sepsis the woman had a preceding history of a sore throat and her lungs were described as consolidated but there was no histology taken to exclude either respiratory or genital tract infection. Furthermore, a laparotomy had revealed a dusky bowel and infarcted ovary but these features were not identified at the autopsy nor was there comment on the discrepancy. In another of the six poor postmortems there is serious doubt whether death was truly due to sepsis:

A woman with congenital heart block had routine antenatal care by her midwife and General Practitioner (GP) until late in pregnancy when pernicious anaemia was diagnosed. Only then was she referred to a cardiologist. Labour was induced but a caesarean section was performed for fetal distress. She called her GP a day or so after discharge because of vomiting and was readmitted and received intravenous antibiotics. A few days after discharge, she suddenly collapsed with dyspnoea and quickly died.

At the autopsy, performed some days after her death, there was a duodenal ulcer and the incision into the uterus was gaping although the sutures were intact. There was no description of local inflammation or peritonitis. A suggestion of retained placental tissue within the uterus and of lung basal consolidation was not confirmed by histology and no definite focus of infection was identified either macroscopically or histologically. No microbiological samples were taken. Her heart was mildly enlarged but there was no detailed examination and there was no detailed histology. Although there was a history of puerperal sepsis this had clinically responded well to antibiotics and there was no evidence of continuing infection after her final discharge from hospital. There was no definite evidence of sepsis at the autopsy and the mode of death was much more indicative of a sudden cardiac arrhythmia rather than sepsis.

Box 15.3

Pathology learning point: sepsis

Where possible the nature and portal of entry of the infection should be identified.

Other causes of Direct death

The cases reviewed in this category include nine deaths from ectopic pregnancy, two following miscarriage, two terminations of pregnancy, four anaesthetic deaths, three deaths from choriocarcinoma and a small number of individually rare causes. Autopsy confirmed the deaths from ectopic pregnancy and miscarriage, and one probable illegal abortion was identified. In this case a perforation was found in the lateral fornix in a woman who died of septicaemia from an unusual organism.

Autopsies for the anaesthetic deaths confirmed the absence of other causes of death and in one case demonstrated haemothorax from subclavian vein perforation during central venous line insertion. One death from choriocarcinoma deserves particular mention:

Following an intrauterine death at term, an externally normal infant was delivered. A detailed perinatal autopsy demonstrated a massive haemoperitoneum with a tumour on the liver which was shown to be choriocarcinoma. No primary site was identified in the placenta but eight weeks after delivery the mother was admitted to hospital with a brain haemorrhage. A brain biopsy demonstrated metastatic choriocarcinoma and, despite treatment, she died. 

The results of the perinatal autopsy report had not been sent to the family or any of her health professional carers. Indeed they were not known to them until the mother had been admitted with her fatal brain haemorrhage. Given that choriocarcinoma is curable if treated in time, this long delay represents substandard care. Sadly this may be a critical reflection on the paucity of paediatric pathologists that has now developed in the United Kingdom. Failure to demonstrate a primary site for choriocarcinoma is a rare, but well documented, event10, 11.

Other miscellaneous Direct deaths include one death in a woman diagnosed as having fatty liver, but who probably had Ehlers-Danlos syndrome. There was no autopsy in another Indirect case of Ehlers-Danlos but the diagnosis was made on the bowel following resection for mesenteric vein thrombosis.12

Indirect deaths

The leading overall, and Indirect cause of maternal death, is cardiac disease. The second most common death from Indirect causes is suicide. This is a reversal from the last Report.

Cardiac disease 
Obesity

Forty-eight women died from heart disease during pregnancy or within six weeks of delivery in this triennium and a further thirty four Late deaths occurring some months after childbirth were also considered by the Enquiry. Sixty-one of these deaths have been reviewed by the pathology assessor together with six cases of non-aortic dissections. Fifty four cases had available autopsy reports of which 29 were good or excellent, 11 adequate, 11 poor and three were appalling. As discussed later, they have been broadly subclassified into ischaemic heart disease and unascertained, arterial dissections, cardiomyopathies and miscellaneous. 

Obesity was present in 55% of these cases and morbid obesity was present a quarter of women who died of ischaemic heart disease/myocardial infarction. A similar picture was seen for deaths from sudden adult death syndrome (SADS). Abdominal obesity is the form of obesity most strongly associated with the metabolic syndrome with its associated increased risk of cardiovascular disease. Clinically there is increased waist circumference with raised triglycerides, total cholesterol and low concentrations of high-density lipoprotein cholesterol in the peripheral blood. Non-esterified fatty acids are also frequently raised. Raised levels of non-esterified fatty acids have also been associated with sudden death from cardiovascular disease13. Measurement of non-esterified fatty acids at postmortem is unreliable but a good proxy is measurement of abdominal circumference particularly in ratio to the hip circumference14. For women an abdominal girth greater than 88 centimetres would be a criterion helping to define the metabolic syndrome15.

Ischaemic and unascertained

Three of the 12 women who died from ischaemic heart disease and whose cases were assessed had no postmortem. Three autopsies were poor but the other six were good or better. All these women had clinical risk factors such as smoking, obesity or diabetes. The clinical history was often suggestive in the cases of women who died from sudden adult death syndrome (SADS). One woman, who had been investigated some years earlier for episodes of collapse that had been attributed to vasovagal faints, collapsed and died whilst talking to someone. Two other cases may have been due to SADS but cannot be adequately categorised; the coroner failed to release the autopsy report in one case whilst in the other the autopsy was so bad that an attributable cause may have been missed. Of the other cases of SADS where autopsy reports were available, one was adequate and another poor but the majority were very good and had clearly excluded other potential causes. For example:

A young girl was found dead a few months after delivery and a thorough, detailed, autopsy failed to demonstrate a cause of death. Cardiomyopathy was suspected but excluded after referral to a cardiac pathologist and so a diagnosis of SADS was made. Subsequently cardiac tissue that had been submitted for genetic studies showed a mutation which was not in the European Society of Cardiology database.

The pathogenicity in this particular case is not known and can obviously only be determined when more cases are identified. This would require more detailed investigation of cardiomyopathies and SADS. It is, however, vital that there is further screening of the family members when a young woman under the age of 40 years has an unascertainable cause of death. In one study there were cardiac abnormalities in 40% of the families associated with such deaths. The causes included electrical conduction problems, cardiomyopathies of all types and hypercholesterolaemia16.

Arterial dissections

There were 17 deaths from arterial dissection subject to pathological review. Of these eight were in the aorta, four in the splenic artery, three in the coronary arteries and one each in the external iliac and the basilar arteries.

All of the postmortems on the aortic dissections were performed to a high standard. Two women were morbidly obese, two had Marfanoid features, the father of another had died of aortic dissection and one had three separate aortic dissections with only the third rupturing. One further woman died of pulmonary embolus: the dissection had re-entered to create a double barreled aorta. In all but one case histology revealed features such as organisation of the blood within the aortic wall indicating that the dissection had started some days prior to final rupture though the duration of the clinical symptoms had rarely been greater than 48 hours. In one report on a woman with features of Marfan’s, tissues were submitted for genetic analysis but no causative mutation was found in the Fibrillin-1 gene. However this does not exclude Marfan’s as these mutations are only found in 77% of patients fulfilling the Gent diagnostic criteria17.

There was no autopsy in three of the four cases of splenic artery rupture. All three had had surgery for haemoperitoneum but the histology reports on the resected spleens were only available in two. The site of rupture was not identified nor was the histology of the splenic artery described in one report. In the other the woman had some years previously had a splenectomy and rupture was attributed to a false aneurysm of the splenic artery. Myxoid degeneration of the media localised to the splenic artery was found in the one case referred for autopsy.

In two of the three cases of coronary artery dissection referred for autopsy the postmortem was poor with no attempt to perform histology or exclude systemic or inherited disease. 

The case of the woman with a basilar artery rupture deserves mention if only because deaths from cerebral haemorrhage are now diagnosed on CT imaging and rarely come to autopsy even if the underlying cause is not known. The artery was described as abnormally tortuous close to its origin from the vertebral arteries and there was a ‘tear’ in the wall. There was no history of hypertension or pre-eclampsia (PET).

There are lessons from the case of a woman who died from a ruptured iliac artery associated with very friable tissues at surgery. The pathologist also commented on the difficulty with reconstruction as the sutures cut through the skin very easily. The histology of various organs was normal. Whilst Ehlers-Danlos Syndrome might have been the underlying cause, no tissue was submitted for genetic analysis and consent to retain the histology was refused. Consequently there is no possibility of any retrospective diagnosis should any other family members have problems in future.

Box 15.4

Pathology learning points: cardiac disease

Obesity should be identified and supportive evidence provided.

In unascertained and sudden unexpected cardiac deaths, the clinicopathological summary should strongly advise screening of close family members.

The possibility of inheritable causes of arterial dissection and rupture should be identified in the report and where appropriate investigated.

Cardiomyopathy

One of the anomalies of the international definition of maternal deaths is that it is restricted to death in pregnancy or within 42 days postpartum yet a peripartum cardiomyopathy, by definition, can arise within five months of delivery. Of the 17 deaths reported to this Enquiry, 16 were Late occurring more than 42 days postpartum. Four of these Late deaths were due to arrythmogenic right ventricular cardiomyopathy and the remainder were dilated/peripartum cardiomyopathy. There were no examples of hypertrophic cardiomyopathy. In all 13 of these have been reviewed for this Chapter but only nine had a postmortem. There was one poor and one appalling report and of the remainder two were adequate and five good or better. In one case citalopram probably contributed to death:

A woman collapsed and died a few months postpartum after multiple episodes of shortness of breath that had been attributed to her known agarophobia treated with citalopram and diazepam. At autopsy she had an enlarged dilated heart with numerous heart failure cells in the lungs. Toxicology showed non-lethal citalopram levels in the blood that can be associated with death when present with other pathology18.
Miscellaneous cardiac conditions

A total of 13 other deaths were reviewed, from nine different causes. There was no autopsy in two cases. In the others four were excellent, four adequate and three poor or worse. The most significant cases are the two deaths in Black or Minority Ethnic women from mitral stenosis due to rheumatic fever: a diagnosis that has not featured in previous triennial reports for some time.

Psychiatric causes of death

There were 98 assessed deaths caused, or contributed to, by a psychiatric disorder; 18 were due to suicide or substance abuse during pregnancy or in the first 42 days following delivery. A further 21 women died from suicide and 16 from an overdose of street drugs later in the first postnatal year. Ten other women died from violent causes and the rest from physical disease compounded by a psychiatric history.

The deaths of 29 women who committed suicide were considered by the pathology assessor. Only seven of these women died from an overdose of a prescribed or over-the-counter medication. As in previous Reports, the method of suicide was predominantly violent. Hanging and jumping from a height were the commonest methods of suicide, but there were also self-immolations, drownings and a death on the railway track.

These violent deaths usually occur in the community and the deceased will often no longer be in contact with the NHS maternity services. Consequently the pathologist may be the prime professional providing valuable information on the clinical circumstances surrounding the death relevant to the Enquiry. This of course has to be with consent of Her Majesty's Coroner which is usually, but not always given:

A prisoner had a miscarriage and, at her request, returned to prison a day later. The prison midwife was unable to see her straight away because of the prison rules and procedures relating to out of hours visits. She was found hanging in her cell some days later. The coroner refused access to the postmortem report apparently considering that the Confidential Enquiry was not an appropriately interested party.

Box 15.5

Pathology learning point: psychiatric deaths

In any violent death in a woman of child-bearing age, details of any pregnancy within the preceding year should be sought and the information included in the report.

Other Indirect deaths

These cover a wide spectrum of diseases but the main conditions considered in this section are cerebral haemorrhage/infarction, epilepsy, infections and endocrine/immunity disorders. The exceptional esoteric case is also identified.

Cerebral haemorrhage

Of the 17 deaths from cerebral haemorrhage reviewed only six had a postmortem.  The ‘view and grant’ Scottish example, in which alternative clinical diagnoses were not investigated, has already been identified but is not unique:

An obese multigravid woman with a long history of substance abuse including heroin, cocaine and amphetamines had pre-eclampsia controlled by labetalol. A few weeks after delivery her blood pressure was recorded as 130/90 mm/Hg but the following day she had a fit. Cerebral and subarachnoid haemorrhage was diagnosed on CT scan and she died shortly afterwards. Despite the known connection between cocaine and cerebral haemorrhage, no autopsy was authorised by the coroner.

All six autopsies were conducted to a very high standard. In half of the cases there were antecedent causes such as hypertension, primary erythrocytosis and haemorrhage into a vascular malformation. One death occurred some days after delivery in a woman with a two day history of worsening headaches. The pathologist carefully searched for and excluded features of pre-eclampsia and reviewed the literature around pregnancy-associated cerebral haemorrhage and stroke in the absence of pre-eclampsia19, s20. Another unique case had a family history of death from cerebral haemorrhage in pregnancy:

A first time mother booked saying that two close female family relatives had died from a brain haemorrhage in the latter half of their pregnancies. A thrombophilia screen of unknown detail had been negative. This history seems to have been ignored and she was not referred for assessment until she was admitted in the third trimester with vomiting, weakness and loss of balance. It was thought she had a neurological problem and she was transferred to a medical ward in another hospital where a CT scan showed brain swelling. She was then transferred to the local neurological centre where the cause of the swelling remained obscure even after a brain biopsy. A cerebral haemorrhage was thought to be the most likely diagnosis and despite intensive treatment she rapidly deteriorated and died.

At autopsy a cerebral vein thrombosis (CVT) deep in the thalamostriate vein was the cause of death. The superficial venous sinuses were patent confirming the imaging findings when she was alive. The pathologist was able to review material from one of the relatives’ autopsy and concluded that a deep cerebral CVT, not cerebral haemorrhage was the cause of her death. It was not possible to examine material from the other relative though her death was confirmed as ‘cerebral haemorrhage’ in pregnancy.

Even if this family has an undetected inherited thrombophilia, the occurrence of a deep CVT arising in late in pregnancy in two, and probably three family members and without any other clinical manifestations of a thrombotic tendency is striking. It also demonstrates that the modern imaging techniques used to diagnose cerebral haemorrhage do not provide all the answers and a good autopsy can still add diagnostic value.

Epilepsy

Ten of the women who died from epilepsy in this triennium had a postmortem; only one was poor in its macroscopic description but this and two others lacked postmortem toxicology and therefore are considered deficient. Seven deaths were classified as due to sudden unexpected death in epilepsy (SUDEP). In this condition death is usually not witnessed and may or may not occur during a seizure but is not due to a complication of the seizure such as aspiration. It is not clear whether pregnancy is a direct risk factor for SUDEP. One reason for this is that there is only limited information on drug compliance during pregnancy. Clinically antiepileptic therapy was discontinued or only intermittently taken in at least two cases but only confirmed by postmortem toxicology in one of these.

In three cases there was no postmortem toxicology and the results were not detailed for a fourth. As both the SUDEP study and the College Guidelines1 state, toxicology is indicated in these deaths. There was no underlying morphological abnormality in the brain in ten cases but detailed neuropathological examination was only performed in two.

Box 15.6

Pathology learning points: neurology
Cerebral haemorrhage

Cerebral haemorrhage is a final common pathway with several different causes: a well conducted postmortem can add value to even the most detailed clinical investigation.

Epilepsy

Deaths from epilepsy should have toxicological analysis of a postmortem blood sample.

Infection

Three deaths from tuberculosis, all in women from the Asian subcontinent, three other deaths from meningitis and four from pneumonia were reviewed for this Chapter. None of the women who died from tuberculosis had a postmortem. Two of the three deaths from meningitis were due to the pneumococcus and the organism was not ascertained in the third death. None had an autopsy.

Two of the deaths from pneumonia were due to the pneumococcus. One woman had positive blood cultures and classical lobar pneumonia after a two week history of a ‘flu-like’ illness. The other autopsy was poor in attributing the lobar pneumonia and her acute mitral and aortic endocarditis to meningitis. Histology was taken but no report was included and attempts by the regional assessor to obtain the histology report and slides failed. One death was due to a staphylococcal pneumonia complicated by invasive aspergillosis following an influenza B infection and the other an atypical pneumonia.

Endocrine and immunological causes

Apart from the diabetic women who died from ischaemic heart disease there were three other diabetic deaths. In one, a woman whose chaotic life style was associated with multiple episodes of ketoacidosis was found dead in bed a few days after discharge from hospital when her blood glucose had been 16mmol/L. Her baby had been stillborn. The poor autopsy excluded unnatural causes and the cause of death was given as unascertained but glucose and ketone sampling was not undertaken.

SLE/Antiphospholipid syndrome.

There were four deaths in this category. There was no autopsy for one and good autopsies addressed the clinical issues in two of the other deaths. The final case was a woman with a mid trimester stillbirth who was admitted with difficulty in breathing and chest pain. Clinical investigations suggested a diagnosis of SLE but the postmortem simply gave the cause of death as ARDS. As the local assessors comment: ‘It is obvious that the question needing an answer is whether the death was directly due to systemic lupus……the pathology report, albeit on a severely limited autopsy is substandard and does not give any evidence of attempting to elucidate the true nature of the underlying disease leading to death.

Miscellaneous Indirect deaths

Included in this category are deaths from a number of other causes as shown in Chapter 10, Indirect deaths. Of particular note were two deaths from liver rupture and bowel infarction related to Ehlers Danlos Syndrome (EDS). The chief features of EDS IV are thin lax skin, joint hypermobility, bleeding, blood vessel rupture particularly aortic dissection, intestinal rupture and preterm labour. If suspected then tissue from skin or aorta can be analysed for decreased type III collagen or skin for the mutation in the COL3A1 gene that is found in Ehlers Danlos type IV.

In another case a woman admitted acutely ill with epigastric pain and vomiting in mid pregnancy was initially diagnosed as having acalculous cholecystitis. It was later realised that she had serious alcohol dependency and she died soon after. Clinically this was the classic picture of acute alcoholic liver disease described over 40 years ago21 but her inadequate autopsy gave the cause of death as acute liver failure secondary to disseminated intravascular coagulation and acalculous cholecystitis. No histopathology was performed and so the probable diagnosis of acute severe steatohepatitis was missed.

Coincidental deaths

The need to assess these deaths, which now mainly comprise those from malignancy and road traffic accidents, has been questioned. However suicides were previously in this category but because of their unusual characteristics in the first year postpartum, these are now considered to be Indirect. The possibility that other causes may be more closely related to pregnancy than is currently thought cannot be completely discounted and therefore they are kept under review. Although the postmortems have a variety of features and causes of death, the generic theme about the lack of correlation with the clinical circumstances remains pertinent. They also provide additional and valuable information on generic aspects of maternal health such as domestic violence, ethnic discrimination and language difficulties. These features are discussed in many of the Chapters in this Report.

Acknowledgements

Earlier drafts of this Chapter were seen and commented on by Dr Robert Nairn, Dr Laurence Brown, Dr Leslie Murray and Dr Grainne McCusker.