Changing patterns of infant death over the last 100 years: autopsy experience from a specialist children's hospital

Introduction

Life expectancy has dramatically increased over the past century in the United Kingdom and, in particular, rates of infant mortality have shown a 20-fold reduction; from 107.2/1000 in 1909 ¹ to 4.6/1000 in 2009. ² The explanation for such a fall is multifactorial and includes changes in medical care and public health, hygiene, environment and food quality. Public health improvements included changes in sanitation, mass immunizations and the provision of medical services. ³

Observers explained the high rate of infant mortality as due to poverty, including poor housing, poor diets and poor environment. There were 20,920 reported infant deaths in London in 1900. ⁴ The first children's hospital in the UK with dedicated inpatient facilities for children was the Great Ormond Street Hospital for Children in London. It opened in 1852, staffed by two physicians and had 10 beds. It remains the oldest children's hospital in the English-speaking world. ⁵ Today, it is a tertiary centre for the investigation of paediatric disease, including childhood deaths.

Postmortem examinations have been performed at this hospital from its inception and all of the original leather-bound, handwritten pathology records have been retained. In some cases, samples for histology were also taken, although further specific details were unavailable. The first available documented postmortem was in 1860. To investigate the differences in causes of infant deaths investigated by autopsy in the same geographical region over a 100-year period, we examined autopsy findings from consecutive infant postmortem reports from the years 1909 and 2009.

Methods

Postmortem reports were reviewed from all infant cases undergoing autopsy at Great Ormond Street Hospital in 1909 and in 2009. Autopsy reports were retrieved from the current computerized pathology reporting system for the 2009 cases. The original, handwritten postmortem reports from 1909 are stored in the pathology archive of the hospital in leather-bound, locked volumes, with reports bound together by year (Figure 1). Reports from 1909 were manually searched to identify autopsies performed on infants (under 1 year of age). These handwritten reports were subsequently read and summarized, with patient age, main diagnosis of cause of death and major findings at autopsy recorded. Similar data were extracted from the cases in 2009.

OPEN IN VIEWER

Figure 1

An example of a handwritten postmortem report from a case in 1909 (a) with demonstration of a hand-drawn picture illustrating location of growth of a nodule (b)

The main pathological findings and categories of deaths undergoing autopsy examination were then compared between the time periods. Only deaths of liveborn infants were included. Cases of stillbirth and deaths of children over 1 year of age were excluded. Data were extracted from the records and the main pathologies and causes of death were classified according to information provided by the pathologist performing the autopsy. Cases were not re-reviewed or reclassified retrospectively. Statistical analysis of groups was undertaken using comparison of proportions test using a statistical software package (Statsdirect Version 2.7.8, by Statsdirect Ltd, Cheshire, UK). The study was approved by the Local Research Ethical Committee.

Results

In 1909, 357 autopsies were performed at the centre including autopsies on 178 infants (50%), compared to a total of 347 autopsies in 2009, which included 128 infants (37%). No postmortem examinations carried out in 1909 presented as intrauterine death, stillbirth or termination of pregnancy. In comparison, there were 155 such cases in 2009.

The age range for infant cases in 1909 varied from 1 day to 11 months. Dates of birth were not routinely recorded and as such, ages were provided in a variety of formats, including days, weeks, and months and in one case, ‘infant’. In comparison, the age range for cases during 2009 varied from birth to 362 days of age (Figure 2). There were significantly less infants autopsied under one month of age in 1909 compared to 2009 (20 and 51, respectively, comparison of proportions test P < 0.0001; 95% CI for difference 18.99–38.27%). One hundred and eight were boys in 1909 compared to 79 in 2009 (60% and 61%, respectively). The main cause of death provided by the pathologist performing the autopsy and patterns of disease identified at autopsy are provided in Table 1 and Figures 3a and 3b. Major pathologies encountered were significantly different between the cohorts of 1909 and 2009. Specifically, infection was the main finding leading to death in 131 (74%) and 26 (20%) respectively (comparison of proportions test P < 0.0001; 95% CI for difference 43.07–62.08%).

OPEN IN VIEWER

Figure 2

Comparison of age and gender distribution between 1909 and 2009 (U accounts for the case in 1909 reported as ‘infant’)

OPEN IN VIEWER

Figure 3

The primary causes of death of infants undergoing autopsy at Great Ormond Street in 1909 (a) and 2009 (b)

There were a higher proportion of deaths due to systemic sepsis with no focus of infection (regardless of any variation in diagnostic criteria between the two time periods) in 2009; 3% in 1909 versus 9% in 2009 (comparison of proportions test P < 0.05; 95% CI for difference 0.01–11.71%). Conversely, the number of deaths due to pneumonia was significantly greater in the 1909 group, representing 37% of deaths in infancy compared to 23% in 2009 (comparison of proportions test P < 0.0001; 95% CI for difference 23.70–39.77%). Similarly, in 1909 there were 19 (11%) deaths from meningitis (10 of which were due to tuberculosis) compared to no such cases in 2009 (comparison of proportions test P < 0.0001; 95% CI for difference 6.93–16.07%). Furthermore, 27 (20% of infectious deaths, 15% of all deaths) had implicated infectious agents in 1909 which almost never cause infant death currently, namely tuberculosis, diphtheria and syphilis (Figure 4).

OPEN IN VIEWER

Figure 4

A postmortem report of a child with numerous siblings demonstrating deaths from different pathology including diphtheria and blood poisoning

Assessment of deaths related to cardiovascular causes was more difficult since terminology was used which is no longer in practice, such as ‘congenital morbus cordis’ and congenital heart disease currently. Congenital morbus cordis was the primary cause of death in four cases in 1909 and was found in nine cases overall (5%), regardless of the final cause of death. In 2009, nine cases (10%) were identified, of which four had been operated. Other cardiac causes of death in 2009 included conditions unknown a century before, such as mitochondrial cardiomyopathy.

Of the three cases (2%) with non-infectious pulmonary pathology in 1909, two were recorded as lung atelectasis (one congenital), and one as prematurity, with no gestational age provided. In 2009, there were 12 ‘pulmonary’ deaths (9%) which would not have survived the initial neonatal period in 1909, including chronic lung disease of prematurity, pulmonary hypoplasia (due to congenital diaphragmatic hernia or other congenital anomalies), severe tracheomalacia and pulmonary haemorrhage. Only one case was not associated with either prematurity or underlying congenital anomalies, this case presented with extensive pulmonary haemorrhage.

Gastroenteritis and malnutrition accounted for 9% and 7%, respectively, of infant deaths in 1909 compared to only one case of gastroenteritis (0.8%) in 2009 (comparison of proportions test P < 0.0001 95% CI for difference 10.21–21.76%). Five cases (3%) had a cause of death related to neurological pathology in 1909 including cerebral palsy, hydrocephalus and ‘cerebral clot’ (the exact location was unclear from the report, and it could have been a sinus thrombosis, subdural or subarachnoid haemorrhage). In comparison, in 2009, there were 10 neuropathological deaths (8%); nine of which were related to intraventricular haemorrhage and/or hypoxic ischaemic encephalopathy. One case had a Vein of Galen aneurysm with subsequent secondary cardiac failure. Furthermore, three cases had associated findings suspicious for non-accidental injury.

Congenital structural anomalies of all types were identified in similar numbers in both cohorts, 21% of cases in 1909 compared to 19% of cases in 2009 (comparison of proportions test P = 0.58; 95% CI for difference –6.78–11.49%). On review of the types of congenital abnormalities, more autopsies of congenital cardiac disease are performed currently but conditions which are now surgically correctable, such as congenital pyloric stenosis and congenital biliary atresia, represented 8% of cases in 1909 but no cases in 2009. Finally, it should be noted that no cases were classified as ‘unknown’ in 1909 compared to 49 cases (38%) in 2009, these representing cases referred as sudden unexpected deaths in infancy (SUDI) or ‘cot-deaths’, a condition either not recognized as such in 1909 or in which no autopsy was carried out.

Discussion

The findings of this study demonstrate that although there are some significant differences in postmortem examination derived findings over the 100-year period, there are surprisingly similar findings in other aspects, despite dramatic advances in medicine and public health. Such advances include preventative measures, diagnostics and treatment; including the discovery of antibiotics, usage of oral rehydration salts and vaccine development, the use of X-rays for diagnosis, the discovery of DNA and the subsequent human genome project.

The advantage of this study comparing postmortem findings from the same children's hospital over such a long time period is that the location of Great Ormond Street Hospital has remained the same, despite the geographical location of its inpatients changing dramatically. In 1909, patients would have been admitted from Central London and surrounding boroughs for primary care. Today, Great Ormond Street Hospital is a tertiary hospital and manages a wide variety of conditions, ranging from complex congenital heart disease to primary immunodeficiency. This includes referral from across the UK as well as internationally. The Department of Paediatric Pathology also performs postmortem examinations for several other hospitals and on behalf of numerous Coroners, covering a wide area as part of the paediatric pathology service provision for London and the South East. This change in patient demographics makes direct comparison to 1909 data inappropriate. In particular, complex referral cases examined within the department in 2009 would not be representative of cases performed in 1909 as a consequence of changes in health care provision. However, this is one of very few studies to assess postmortem findings at a single centre over such a time period.

From a historical perspective of the past 100 years, numerous changes are evident. This includes the number of prematurity-related conditions in infants who survived the early neonatal period in 2009 in comparison to 1909, when such cases would had died in the perinatal period and not undergone autopsy at this centre. We can reasonably expect that an infant born below 32 weeks would be unlikely to have survived in 1909, compared to a >95% survival chance presently. Furthermore, infants with severe congenital abnormalities, such as congenital diaphragmatic hernia, would also not have survived a century ago. Interestingly, there was no significant difference in the overall frequency of detected congenital abnormalities; however, the type of abnormalities differed greatly. Many lethal abnormalities identified at autopsy in 1909 would have been successfully treated in the 21st century. For example, the treatment of hypertrophic congenital pyloric stenosis by pyloromyotomy has become a standard procedure, whereas gastroenterostomy was the most commonly performed operation until 1907. ⁶ It is likely that in 1909, cases of congenital obliteration of the bile ducts refers to primary biliary atresia. As such, no treatment was available until 1950s, since when the Kasai porto-enterostomy and liver transplantation have become the mainstay of treatment. ⁷

Over the past century, there have been changes in legislation relating to autopsy. Hospital autopsy numbers have generally decreased across the United Kingdom, partly due to reluctance of treating physicians to request a postmortem examination. A study published in 2001 revealed that of 100 consecutive deaths where a postmortem examination was not requested, the physician decided not to ask for autopsy in 22% of cases. Of these, the rationale for not asking in 50% of the cases was due to a presumed known cause of death and 18% did not have an autopsy due to the physician deciding it was not needed. ⁸ Legislation states that in cases of sudden unexpected death, all cases are required to be investigated by HM Coroner, which usually means a full autopsy including ancillary investigations. ⁹ Of the 128 infant autopsies performed at our centre in 2009, 79 (62%) presented with sudden unexpected death (either sudden unexpected early neonatal death or sudden unexpected death in infancy presentation). It is also worth highlighting the absence of fetal autopsies in 1909 compared to current practice. The examination of the fetus in cases of stillbirth and/or termination has become an important investigation for such deaths over the past 50 years, although recently the fetal necropsy rate has also decreased, ^10,11 despite the recognized value of such a procedure to provide additional diagnostic information in 30–40% of cases. ¹² Interestingly, in the last quarter of the 19th century, fertility rates started to decline, a drop described by classic demographic transition theory, in which that fertility declines with reduction in mortality rates. ¹³ With fluctuation in fertility aligned with World War I, World War II and the post-war baby boom, ¹⁴ along with an improvement in the health of the nation, there has been an increased focus on pregnancy loss. Discovering underlying causes, in a preventative role for not only providing answers but also aiding future pregnancies; has become a significant component of most practising paediatric pathologists today.

As expected, with advances in all aspects of medicine, there has been a change in disease burden. Establishment of Louis Pasteur's Germ Theory in 1878 ¹⁵ with subsequent changes in public health, followed by the development of antibiotics and vaccines has led to reduced rates of communicable diseases. Studies of disease have shown that as a population becomes modernized, rates of fatal infection generally decline towards zero, while most non-communicable diseases increase. ¹⁶ This point is apparent in the analysis of our data from the past 100 years, in particular, the significant drop in fatal infant infections. Indeed, immunization for diphtheria and preventative measures of tuberculosis, as well as the improvement of social housing have vastly reduced these diseases to such a level that in 2009 there were no infant deaths attributable to either cause. ² Along with modernization, there is progression and implementation of new laws as demonstrated by the Children's Charter in 1889, preventing cruelty to children. ¹⁷ Subsequent evolvement of the charter into the Children's Act of today has put the child first and foremost. It is likely that much child abuse was not reported in the 1900s and many such cases would not have come into hospital. Of the cases in 1909, one case was diagnosed with a ‘cerebral clot’, but the aetiology remains uncertain. There was a lack of clinical history and other further specific findings at postmortem examination, which would be accepted standards today. Indeed, increasing awareness on the part of the police, health and social services prevents the majority of cases from being missed, but this has led to increasing infant autopsies for ‘suspicious’ unexpected infant deaths. Part of the ongoing difficulty in current infant autopsies is the determination of whether certain findings indicate inflicted non-accidental injury.

As Great Ormond Street is sited in the centre of London with several surrounding under privileged areas, including Clerkenwell and Saffron Hill, in the surrounding areas in 1909, the presentation of cases with malnutrition and marasmus in a society where infection was so rife coupled with dietary inadequacy is not surprising. This would be a startling find today. In a history of Great Ormond Street Hospital, dietary recommendations from early in the last century included ‘… the provision of a roast beef dinner for one day a week, green vegetables to be provided one day a week, except when very expensive and a slice of bread and butter be given for dinner when desired…’ ⁵ The development of appropriate nutrition for infants, as evidenced by the wide range of formula feeds available on the market today, coupled with the improvement in hygiene and development of oral rehydration salts have reduced the presence of nutritional and gastroenteritis associated death.

As previously mentioned, sudden unexpected deaths are all now investigated on behalf of HM Coroner. Unexplained sudden unexpected death or SIDS (Sudden Infant Death Syndrome) is a relatively recent ‘diagnosis’ (though one of exclusion) introduced in 1969 and now widely used. ¹⁸ Before this time, such sudden deaths were noted and thought to be due to overlaying. In 1890, a change came about in the body of thinking with a new hypothesis of status thymico-lymphaticus, ¹⁹ in which there was overdevelopment of the lymphatic system. Subsequent proposed treatment led to children being treated with either thymectomy or radiotherapy. It can also be imagined that a large proportion of such deaths never arrived in hospital. Today, classification of sudden unexpected death in infancy is used as a collective term for such deaths in which using current investigative measures, no identifiable pathological process is found to satisfactorily determine cause of death. Numerous hypotheses have been put forwards, including infection, ^20,21 cytokine polymorphisms ²² and abnormal brain stem mechanisms. ^23,24

In summary, we have demonstrated both significant differences and similarities in autopsy cases of infants examined at one centre over the past 100 years, where the postmortem examination itself, which besides the use of ancillary investigations, remains essentially the same. ²⁵ Recent advances however, include development of the less-invasive autopsy, using MRI, ^26,27 the minimally invasive autopsy, using laparoscopy and the molecular autopsy. ²⁸ In another 100 years, we hope that there will not only be changes in mortality patterns but also improvements in the identification of currently ‘unknown’ causes of death.

DECLARATIONS