Mortality associated with cysticercosis in a historical cohort from Britain

ABSTRACT Background: The burden of premature mortality associated with human cysticercosis is largely ignored mainly due to poor record-keeping in Taenia solium endemic regions. Objective: To document mortality and survival characteristics of an historical cohort with cysticercosis. Methods: The years of onset of symptoms and death untill 1957 were extracted from published reports of a British military cohort (n=450) examined in London in the early twentieth century. Data were entered into a Kaplan Meier survival analysis with the presence (or absence) of clinical manifestations as independent variables, which were then fitted into a Cox proportional hazards model to determine their significance. Results: Cysticercosis was responsible for 24 (52.2%) of 46 deaths in the first 15 years of follow-up in comparison to 7 (19.4%) of 36 deaths in the 20-40 years of follow-up period. In the univariate and Cox analyses, intracranial hypertension (hazard ratio [HR]: 8.26; CI: 4.71, 14.49), ocular cysticercosis (HR: 6.60; CI: 3.04, 14.33), and mental disorder (HR: 3.98; CI: 2.22, 7.13) but not epilepsy (HR: 0.66; CI: 0.20, 2.18) were associated with mortality. Over half of all deaths in the first 15 years of follow-up were attributed to cysticercosis. Conclusions: Several deaths occurred early after acquiring cysticercotic infection. Intracranial hypertension, ocular cysticercosis, and mental disorder but not epilepsy were predictors of mortality in this cohort.

of cysticercosis deaths in the United States (US) between 1990 and 2002 found that the mean age at death in people with cysticercosis was 41 years 3 . There is, however, little information on mortality associated with cysticercosis outside the US, mainly due to ineffective public health systems and lack of surveillance systems for life events [3][4][5][6] .
A large cohort of people with cysticercosis was systematically followed-up in Britain in the early twentieth century [7][8][9][10][11][12] . India was a British colony before it became independent in 1947, and British military personnel were frequently deployed in India before 1947. They and their families remained in India for varying periods of time, usually up to three years. Cysticercosis was present in Britain until it was eradicated, with only few reports in the eighteenth century. Hence, it is safe to assume that British military personnel acquired cysticercosis while serving in India, where it remained mostly undiagnosed. Upon their return, they were examined at the Queen Alexandra Military Hospital in London for neurological symptoms including seizures. The hospital specialized in the treatment of non-combat related disorders. After diagnosis, the returnees were retired from military service but were maintained on follow-up until 1957. The cohort was thus comprised of people diagnosed between 1921 and 1957.
We performed a secondary analysis of data contained in various published reports on this cohort from the 1920s to 1961.
Here we report survival rates and their relationship with various clinical presentations in this historical cohort.

cohort composition
This cohort consisted mainly of members of the British army and their families who had previously served in India and had contracted cysticercosis there. They were identified between 1921 and 1957 and were followed-up by army medics at Queen Alexandra Military Hospital, London 13,14 . The earliest report of five cases dates back to 1930 7 but the number increased to 71 cases in 1934 9 , 284 in 1944, 11 and to 450 in 1957 12 . Deployment of military personnel to India ceased in 1947 and the last case was diagnosed in 1949. In response to the initial reports, the Army supported continued scrutiny of the cases and longitudinal data were systematically collected. The final report was published in 1961 and contained data up to August 1957 12 .

clinical presentations
The final report contained the findings of the autopsy of 55 of the subjects who died before August 1957 12 . The report included the forms of presentation coded as epilepsy, focal neurological deficit, intracranial hypertension, mental disorder, ocular cysticercosis, subcutaneous involvement, lingual cysticercosis, muscular disorder, and acute febrile illness, as well as the year of symptom-onset and death. The report also categorized the causes of death into those due to cysticercosis and those due to other causes. Death was attributed to cysticercosis if the immediate or underlying cause was reported as related to epilepsy (status epilepticus, drowning, phenobarbital overdose, and suicide), intracranial hypertension, or cysticercotic meningitis. We used the reported year of symptom onset and year of death in those who died 12 . In those who remained asymptomatic, the year of exposure was assumed from the mid-point of their deployment in India. The original reports also provided data on asymptomatic cases, which were also included in examination of their association with survival 7-12 .

Statistical analysis
Data were entered into a Kaplan Meier survival analysis, and subjects were considered at risk of dying from the year of symptom onset. They exited at either death or censoring, whichever occurred earlier. Censoring was at the time of the last follow-up (August 1957). The association between the presence or absence of each category of clinical manifestations was examined in univariate analyses. The presence or absence of manifestations was then fitted as a covariate in a Cox proportional hazard model to estimate exponential individual coefficients and hazard ratios. The coefficients represent the hazard ratio for a unit change in the covariate. Stata version 15.1 (StataCorp, TX, USA) was used for analysis.

reSUltS
The final report provided details of 450 cases up to August 1957 12 . The diagnosis of cysticercosis in the cohort was established by biopsy of subcutaneous nodules in 108 (24%), by radiology of the limbs and skull in 313 (70%), by autopsy or surgery in 22 (5%), and by ocular cysticercosis in 2 12 .
Seven (47.6%) of 18 deaths in the first 5 years of follow-up, 9 (56.3%) of 16 deaths in the next five years, and 8 (66.7%) of 12 deaths in the 11-15 years follow-up period were attributed to cysticercosis. In comparison, the proportional mortality attributed to cysticercosis was comparatively lower from 20 years onwards

Survivor function and clinical presentations
The clinical features of the cohort are shown in Table 1. Epilepsy was the commonest presentation, with 413 (91.8%) affected people. Subcutaneous nodules were detected in 243 (54%) subjects. In all, only 15 (3.3%) subjects remained asymptomatic throughout the observation period.

DiScUSSiON
Evidence derived from this cohort has had a profound influence on the current understanding of cysticercosis 13,14 . Our analysis of mortality and survival provides further insights, as cysticercosis-related mortality is a major public health issue largely neglected partly due to the constraints of poor recordkeeping in endemic areas [3][4][5][6] . Despite the limitation of the absence of data on the age characteristics of the cohort, our findings suggest that cysticercosis substantially contributed to premature mortality. This is because cysticercosis usually occurs in young people and most deaths attributed to cysticercosis occurred in the first 15 years of exposure in this historical cohort [15][16][17][18][19][20] . Only very recently has the premature mortality associated with infestation been confirmed by an analysis of cysticercosis-related deaths in the USA, which estimated a mean age at death of 40.5 years 3 .
How does the mortality estimate of 4.6/1000 person-years in our cohort compare with current data? A Brazilian report estimated a crude mortality rate of 0.88-2.51/million people in 1995-2004 6 . In comparison, analysis of the nationwide inpatient sample from the USA produced an estimate of 0.1 deaths/million population 15 . When death certificates from the National Centre for Health Statistics were the source, mortality estimates for the period 1990-2002 amounted to 0.06/million population 3 .
Careful documentation of presentations and year of symptom-onset and death in the whole cohort provided an opportunity to assess the relationship between various clinical presentations and mortality. In the Cox model, intracranial hypertension, mental disorder, and ocular cysticercosis, but not epilepsy, were associated with reduced survival (Table 3). Intracranial hypertension is a severe manifestation of cysticercosis and might occur as a result of non-obstructive hydrocephalus associated with basal sub-arachnoid cysticercosis, obstructive hydrocephalus due to intraventricular cysticercosis, giant solitary cysticerci, or multiple parenchymal cysticerci with massive cerebral edema [16][17][18][19][20] . Two conditions, cysticercotic encephalitis and disseminated cysticercosis, characterized by massive brain infection and intracranial hypertension 18,19 are still associated with high case fatality rates. Obstructive hydrocephalus could have contributed to death in seven subjects who underwent neurosurgical treatment, as five of them died soon after probably due to the procedure 12 . A Polish report from the 1960s suggested that nearly two thirds of those with hydrocephalus resulting from basal arachnoidal cysticercosis died from the disease 20 . The only available surgical treatment for obstructive hydrocephalus at the time was ventriculostomy. Shunt procedures were implemented several decades later and led to a decline in mortality associated with hydrocephalus 21 . Likewise, the application of medical treatment in the form of repeated courses of albendazole signaled a major breakthrough in the management of subarachnoid cysticercosis with intracranial hypertension 22 . Besides intracranial hypertension, both mental disorders and ocular cysticercosis were associated with reduced survival. Mental disorders, described in many early reports, are also associated with heavy cerebral infection 12,18,23,24 . Presumably, many subjects in the cohort had heavy infection loads, which is supported by the fact that more than half of the subjects had subcutaneous nodules, a surrogate for a heavy cyst burden 12 .
The lack of an association of mortality with epilepsy is difficult to explain, especially in view of the well-established association between epilepsy and premature mortality (Tables 2 and 3; Figure 1) 25 . An original report analyzed the causes of death in a representative sample and found status epilepticus as the immediate cause of death in 18 cases (43%) 12 . A possible explanation arises from the small proportion of patients without epilepsy (n=37 of 450; 8%; Appendix, Table 1), which could inadvertently inflate the number of deaths in the subgroup without epilepsy in a cohort with severe cysticercosis. Alternatively, the occurrence of seizures might have led to earlier diagnosis (and thus better prognosis) by physicians familiar with epilepsy as a manifestation of cysticercosis [7][8][9] . A parallel can be drawn from studies of cerebral glioma, in which the occurrence of seizures is an independent predictor of improved survival 26 .
Our findings must be interpreted in light of the prevailing standards of care at the time. One discernible difference between the historical and modern series relates to the available diagnostic tools for neurocysticercosis. In this cohort, limb and skull x-rays and biopsies of subcutaneous nodules were the main diagnostic methods and were perhaps only able to diagnose the more severe infestations 7-12 . Currently, neuroimaging is almost always used to establish the diagnosis of neurocysticercosis and this has led to the recognition of milder forms of infestation with solitary (or 1-2 cysts) parenchymal cysticercosis, a condition with an excellent prognosis 27,28 . Some of the medications that are the standard of care today, such as the anthelminthic agents albendazole and praziquantel, and intravenous agents used to treat status epilepticus were not available in the early twentieth century [29][30][31][32][33] . Our study was also limited by the absence of data on the age of individual subjects, which precluded the estimation of age-adjusted mortality rates. The findings of the association between intracranial hypertension, mental disorders, and ocular cysticercosis and reduced survival are noteworthy and currently relevant. We believe that our analysis is not only of historical interest, as it also provides a useful estimate of mortality and survival in people with cysticercosis.

AcKNOWleDGeMeNtS
Data for the survival analysis performed in this study was obtained from: Dixon