The limits of comparison: institutional mortality rates,long-term confinement and causes of death during the early twentieth century

The institution

The Ranchi Indian Mental Hospital (since 1998 Ranchi Institute for Neuro-Psychiatry and Allied Sciences) was established in 1925 and was the largest psychiatric facility in British India, confining on average about 1250 patients during the period from 1920 to 1940 (Ernst, forthcoming). Ranchi’s catchment areas consisted of the provinces of Bengal, Bihar and Orissa in north-east India, an area of 93 million inhabitants. Its superintendent from 1925 to 1940 was J.E. Dhunjibhoy, fully trained in the western mode of medicine at Bombay Medical College, and one of the first few Indians who were permitted to occupy senior positions in the Indian Medical Service (see Ernst, 2012).

Mortality and long-term confinement

The rate and pattern of patients’ mortality in mental institutions are important parameters in the assessment of clinical outcomes. As Tidemalm et al. (2008) and others have highlighted, they provide a way of measuring the effects of mental health care provision. Mortality is closely related not only to the care and attention an institution bestows on its inmates but also its patient intake and health status on admission. Recent research in western countries has shown that mortality among psychiatric patients in both institutional and community-based care settings remains high. In Sweden the excess mortality for 12,103 patients in the late 1990s was three times that of the wider population (Tidemalm et al., 2008: 1); a Norwegian study of psychiatric inpatients from 1980 to 1992 concluded that ‘mortality of psychiatric patients is still unsatisfactorily high’ (Hansen, Arnesen and Jacobsen, 1997: 186); assessment of data from seven German hospitals in the mid-1990s revealed that over six times more patients died than expected (Hewer, Rössler, Fätkenheuer and Löffler, 1995: 174); and an Italian team of epidemiologists concluded in 1997 (Valenti et al., 1997: 1227): ‘Longer periods of hospitalization and non-discharge from hospital are the main risk factors for death in psychiatric patients, who globally experience higher death rates than the general population for a wide spectrum of causes of death, whatever their diagnosis or gender.’ Mortality statistics and data on the length of confinement clearly appear to be important aspects of institutional analysis. In relation to a particular institution such as the Ranchi Mental Hospital during the 1920s and 1930s, such details of course need to be assessed alongside those of other hospitals in British India. Caution is necessary as none of them provided institutional figures that were adjusted for age, gender and patient intake. However, their equivalents in Britain did not do so either. The mental hospital figures as well as the recorded (and estimated) census data for the 1920s and 1930s are crude death rates (CMR), rather than age-adjusted, standardized mortality rates (SMR). Age-specific rates are essential for proper comparisons, in particular over time. However, as has been noted in a recent debate in 1998 and 1999 between epidemiologists and clinicians in several issues of the British Medical Journal, the crude rates reflect ‘the real number of deaths’ that actually occurred at a particular time. While clinicians argue that ‘Crude rates, without standardisation for age, are always misleading’ (Tunstall-Pedoe, 1998: 475), epidemiologists concede that ‘Clinicians and epidemiologists view crude death rates differently’ (Nakayama, Yoshiike and Yokoyama 1999). The current consensus, reflected in the British Medical Journal editor’s statement on the debate, is that ‘crude death rates are misleading in comparisons. However, clinicians and epidemiologists have different attitudes towards these health indicators.’ (BMJ editor, 1999).

The diverse assessments of the relative merit of CMR, on which most statistics from the early part of the twentieth century are based, highlight the limitations of institutional data from this period for comparative purposes. In addition, it is vital to be aware of the impact of admission criteria and policies as well as the socio-economic status and cultural context of the population from which patients are drawn. These are important factors in relation to the institution at Ranchi. For example, in contrast to institutions in England and Wales, the criminal element in Indian mental hospitals appears to have been more prominent, in the absence of a specialist facility such as Broadmoor and with legal provision that only allowed for confinement of those ‘too dangerous and troublesome to be left in their relatives’ charge, those found helpless wandering about, or who are unprotected’ (Ewens, 1908: 1). The average percentage of ‘criminal lunatics’ reported for the 22 mental institutions in British India in 1924 amounted to 30 per cent (Anon., 1927b: 539).

Despite these caveats, if we do set various colonial institutions’ statistics alongside each other – as Ranchi’s superintendent did in 1932, rightly or wrongly – a startling trend can apparently be discerned. Mortality in the Ranchi Indian Mental Hospital was consistently lower, not only as compared with a number of other institutions for Indians in northern India, but even in relation to its European twin, the European Mental Hospital just down the road, which employed a racially selective and class-specific admission policy, restricting its patient intake to Europeans and better-off patients of Eurasian background. The European institution was far better funded, and there prevailed no such pressures as the lack of accommodation and medical facilities characteristic of the Indian hospital. Nevertheless, in terms of the average percentages of death for the three years 1930–32, for example (see Fig. 1), Indian patients fared far better than the select group of Europeans and higher-class Eurasians next door.

OPEN IN VIEWER

Figure 1.

Average percentage of death to daily average of patients in different mental hospitals in British India, 1930–32 (RIMH, 1933: 3)

An average figure of about two per cent mortality is a great achievement – not only compared with the higher rates prevalent at other institutions, but also if we note that morbidity and mortality levels were high among the wider population. (Reported and estimated rates for Bengal could amount to anything between 1.9 and 4.5 per cent mortality; Meyer, Burn, Cotton and Risley, 1909: 512.) A rate that is equal to or even below that of the population at large, and less than half the one achieved at the neighbouring European institution, clearly put the Ranchi Indian Mental Hospital into its own league. This was the case not only during the early 1930s, but right up to 1940. The trend for Ranchi was favourably out of step also with the pattern reported for western psychiatric institutions, where institutional death rates were consistently well above those prevalent among the wider population. During the period 1930–32, for example, the percentage of deaths reported for mental institutions in England and Wales was 6.76 – over three times the Ranchi figure (Livesay, 1936: 1247).

In western countries, more variable and unstable mortality rates have been observed within and between mental hospitals, in contrast to more stable rates for the general population (Tokuhata and Stehman, 1958: 750). This was clearly different in British India, where generally a high degree of fluctuation prevailed in both institutional data and census statistics. Census officials in India reported that provincial mortality statistics revealed ‘great variation’ over place and time (Meyer et al., 1909: 513). In 1909 for example, it was noted that rates ‘vary greatly in different areas during the same period, and in the same areas from year to year; and while this is characteristic of the figures at all times, we have to distinguish the exaggerating effects of special morbific [sic] influences, such as famine.’ In the period 1881–1900, recorded mortality per 1000 of the general population of British India fluctuated between 18.8 and 36.6. The recorded details were considered not reliable, with deaths being under-reported. A ‘probable true normal rate’ was therefore estimated: it amounted to an extraordinary rate of 44.8 per 1000 living. This was clearly, as noted at the time, ‘much above the European standards if Austria-Hungary and Italy be excluded’ (Meyer et al., 1909: 513). Mortality statistics continued to pose a challenge in subsequent censuses, not least because age-specific details were difficult to obtain. The census report for 1941 still lacked reliable data on age cohorts, which made the calculation of age-specific mortality and reproduction rates a matter of speculation (Yeatts, 1943: 51–2). The Ranchi pattern of an apparently lower institutional death rate (in contrast to mortality among the wider population) may induce historians working on western institutions to reflect further on epidemiological issues concerning the state of health of those confined in public institutions. For those working on colonial medicine, some critical reflection may be needed on the persistent trope of western-style institutions as manifestations of hegemonic power and control and culturally inappropriate media of health provision.

Death by gender

Following the publication of Showalter’s The Female Malady in 1987, it has become somewhat of a mantra for gender historians in the west as well as other countries in the world to postulate a disproportionately higher likelihood for women to be admitted to psychiatric institutions, to stay there for longer and have a higher chance to die there – although data to substantiate these propositions are lacking in particular for the period considered here. At Ranchi, there emerges no particularly strong gendered pattern if we disaggregate the mortality figures. While the trend for men remained relatively stable over the period if set against overall patient numbers, the female death rate in contrast follows a slight upward trend during the period 1927–39. The latter correlates with increased overcrowding on the female wards during this period. Again this contrasts favourably with trends in western institutions – as well as some Indian ones – where the death rates for females were frequently higher. At the old institutions at Dacca and Berhampore, for example, figures in 1924 and 1925 were 2.88 and 6.31 per cent for men and 9.57 and 7.97 per cent for women, respectively (RIMH, 1931: 14–15). However, rates tended to fluctuate considerably at institutions, so that any generalizations on gender-specific rates of death are problematic, and more research is required.

At Ranchi, too, the number of male and female deaths varied from year to year, with a conspicuous rise in the number of female deaths in 1935. Although tragic for the individual women who died that particular year in excess if compared to previous and earlier periods, statistically the data were not significant. ² Indian doctors and authorities were used to considerable fluctuations in people’s health status as an outbreak of smallpox, influenza and cholera, or drought and famine conditions, for example, could lead to sudden increases in morbidity and mortality rates. Whether the raised mortality rate in 1935 is to be accounted for by epidemiological factors, normal statistical variation or mismanagement on the part of the acting superintendent is difficult to ascertain on the basis of the information available.

However, over the 10-year period from 1930 to 1939, the ratio of male to female deaths is almost exactly equal, implying that over a longer period no gender-specific trend prevailed in men’s and women’s likelihood of dying during their stay at Ranchi. ³ Furthermore, if we disaggregate the Ranchi rates, it emerges that both sexes were almost as likely to die within the first year of admission (10 per cent of men and 14 per cent of women), while an equal proportion of men and women who died had been confined for over 20 years. ⁴ The only sex-specific feature of note in the mortality statistics was a higher proportion of women being confined for 1–5 years (26 per cent of females in contrast to only 14 per cent of males), while men were more likely to have been institutionalized for 10–20 years (27 per cent of males versus 19 per cent of females). Assessment of the impact of culturally and gender-specific life cycles and events needs to be undertaken to interpret these trends (e.g. see Busfield, 1996; Ernst, forthcoming).

It is clear from the data that Lt.-Col. J.E. Dhunjibhoy, MB, BS, FCPS, the superintendent at Ranchi, was particularly successful in keeping patients of both genders alive, more so than any of his colleagues at the other institutions for Indians and at the European Mental Hospital – and perhaps even in western institutions. The latter would deserve further probing, not least because it might well ‘de-centre’ current assumptions about institutions in western countries having been an ideal model for and superior to those in non-western ones. An assessment of the varied factors that had an impact on patients’ chances of survival in different institutions would be necessary, as even Dhunjibhoy’s record was not consistently one of success. Prior to his assignment to Ranchi in 1925, he had been made superintendent at Berhampore in 1923. Mortality the next year, in 1924 was high, namely 6.31 per cent for men and 7.97 per cent for women. Although these figures were still well in line with those achieved at other institutions in India, they were considerably above those he could rightly boast about at Ranchi once he had established a stable clinical and managerial momentum at the newly opened institution. A doctor’s experience is clearly an important factor in regard to institutional outcomes. It is also likely that brand-new premises like those provided at Ranchi would have helped the superintendent to keep his patients in good physical condition, even during a period of austerity measures such as during the 1920s and 1930s. In contrast, Berhampore was, like many of the other mental hospitals, in a severe state of dilapidation, with unsanitary and run-down facilities.

‘Chronic patients’ and long-term confinement

Long periods of hospitalization have been considered an important ‘risk factor’ for death in psych-iatric patients. Long-term confinement has also been a focus for authors interested in gender issues, to identify discrimination against women. According to Showalter (1987), it was a feature that hit women disproportionately in western institutions (although this suggestion appears to be incorrect). ⁵ For England and Wales, the overcrowding of mental hospitals with ‘chronic’ cases from the 1870s onwards has been identified as marking the real start of the ‘great confinement of the mad’ (Porter, 1992: 119–25). The ‘warehousing’ of hopeless and incurable cases in ‘mammoth asylums’ or ‘museums for the collection of insanity’, where the ‘maintenance of order’ rather than attempts at cure were the main drivers, ‘many patients were simply left to rot’ (Scull, 1995) and psychiatrists deprived ‘innocent persons of liberty’ (Szasz, 2012: 349), continues to be seen as a major feature of mental hospitals. The assessment of patients’ length of hospitalization is therefore an important issue for a variety of reasons.

Alas, for Ranchi no composite or sex-specific data on patients’ duration of stay are available in the hospital reports. The only available quantitative data relate to those inmates who died or were discharged (as cured, improved or not improved). These patients are of course not representative of those staying behind. However, the data for those being discharged are indicative of a highly favourable trend for women in terms of their likelihood of being discharged early. Most of the women (namely 96 per cent) who were discharged in 1930, for example, had not spent more than three years in the hospital. In contrast, only 65 per cent of the men discharged during that year had been confined for less than three years. In other words, 35 per cent of men but only four per cent of women discharged had spent more than three years at Ranchi. To what extent these figures reflect the high number of ‘criminal lunatics’ present in the institution and legal conditions on the terms of their discharge is difficult to ascertain unless a case-specific analysis is undertaken. This would provide a better basis for comparisons with other institutions.

From the mortality figures and the limited statistical data sets on the length of institutionalization at Ranchi, it can be concluded that of those discharged the majority of women were released earlier than men. Further, women were no more likely than men to die in the hospital, nor had women who died at Ranchi been hospitalized for longer than their male fellow inmates. One important factor that had a bearing on long-term confinement was the fact that chronic patients were then – as they are now – not particularly ‘attractive’ subjects for medical care: if there was no ‘hope for recovery’, the main regime that medical staff could pursue was ‘custodial and occupational’ (RIMH, 1936: 10). This kind of institutional treatment was not the favoured approach for doctors as it implied that their professional expertise could not be employed in a fruitful way, and they were reduced to mere custodians of ‘hopeless cases’ and overseers of inmates’ occupational and recreational management. During a period such as the 1920s and 1930s, when research, expertise and specialist skills became ever more highly valued, engagement in management rather than medicine was not desirable. It was clearly not in the interest of medical superintendents to retain patients for unduly long periods of time. As with medical professionals today, practitioners were outcome orientated, being interested in cure efficiency whenever feasible. It is therefore not surprising that attempts were routinely made to get rid of as many chronic inmates as possible. For example, in 1932 the superintendent at Ranchi reported (RIMH, 1933: 2):

Special efforts were made to discharge 43 chronic and harmless patients to the care of their relatives who could be traced out by the Magistrates of their districts. As a result, 18 patients were discharged and in 14 cases no friends or relatives could be found in spite of vigorous enquiries instituted by the Magistrates concerned. In 11 cases inability for financial reasons to take charge of the patients was accepted by the enquiring Magistrates and they were, therefore, not discharged.

Despite such efforts to transfer inmates back into the community, the number of chronic patients confined for long periods remained high, right up to the beginning of World War II. Furthermore, as the Ranchi superintendent noted in 1939, the ‘so-called “new admissions” in spite of our giving them all the known modern scientific treatment of mental disorders’ were ‘chronic cases beyond any hopes of recovery’ (RIMH, 1940: 2). This had been so ever since the hospital was established in 1925. Another doctor had reported that on the opening of Ranchi, the ‘majority’ of patients were ‘chronic cases and merely transfers who had been for years already in residence in other Mental Hospitals, namely Berhampore, Dacca and Patna’ (RIMH, 1931: 3). Consequently, a ‘large number of infirm, incurable and homeless cases’ had accumulated ‘from year to year’ (RIMH, 1931: 2). He referred to ‘several patients’ who ‘have been continuing as inmates for more than 30 years’ and mentioned one who had died at Ranchi ‘at the age of about 80 years’. The latter had been in mental hospitals, the doctor pointed out, for 46 years.

Long-term patients were also a thorn in the side of government officials on account of the cost of maintaining them in an institution and the fact that accommodation in Ranchi’s male wards was notoriously overcrowded. Administrative restrictions were therefore imposed to counteract the tendency for chronic patients to accumulate over the years. However, this was only partially effective. Doctors continued to lament the disproportionately high rate of chronic patients under their care, and government, in pursuit of financial prudence, aimed to curtail long-term confinement to the absolute minimum. Institutional care remained restricted to those who constituted a danger and threat to the wider community. It is therefore unlikely that many of the patients – including the women among them – would have been locked away unduly.

Some gendered tendencies clearly prevailed in regard to the pattern of long-term confinement. Yet, in contrast to the trends alleged for western institutions, the situation at Ranchi was reversed. The bias was slightly in favour of women – if we think of confinement in a mental hospital as a bad thing rather than a medical service provision for the care of the mentally troubled. It is of course vital to explore further the institutional and policy-induced constraints and the social, cultural and community-specific factors that impacted on these numerical trends (see Ernst, forthcoming).

Causes of death

As was the practice in other institutions in India and in western countries, the annual and triennial reports at Ranchi provided details on the causes of death by gender. There are, however, problems, which are apparent from Table 1 for the year 1931.

OPEN IN VIEWER

Table 1.

Causes of death assigned, by gender, 1931 (RIMH, 1932: 4)

Causes of deaths	Male	Female
Malaria	2	–
Pneumonia	2	–
Tuberculosis (lungs)	2	–
Diabetes	1	–
*All other general diseases	7	5
Diseases of nervous system of which pathological basis is not known	1	–
Diseases of the circulatory system	–	1
Diseases of respiratory system	1	–
All other diseases of digestive system	3	1
Acute or chronic nephritis	2	–
Injury, general or local	1	–
Total	22	7

*

Emphasis added

In relation to the women in particular, it is clear that the majority of the causes of death were merely noted as residual conditions, on the basis of the exclusion of all other specified disease, as ‘all other general diseases’. This lack of specificity constitutes a major drawback for any attempt at an analysis. Historians of medicine and medical epidemiologists have been well aware of this issue (e.g. see Risse, 1997).

Even in the face of this caveat, it is tempting to reflect on a couple of issues relevant to comparative analysis. First, it may strike us – particularly so in regard to an Asian institution – that much-feared killers such as smallpox, cholera and plague were rarely among the stipulated causes of death. After all, the representations in western countries of these three diseases have been strongly bound up with ‘the East’ and other places imagined as unsanitary and unwholesome. However, the institution’s superintendent was well aware of the danger posed by these diseases and he took sanitary measures to prevent them from taking hold among patients (see Ernst, forthcoming).

Alongside ‘other general diseases’, those specified as having been prevalent during most years were respiratory diseases (TB and pneumonia), gastrointestinal conditions (dysentery and diarrhoea), and ‘debility’ of various kinds. From the perspective of those familiar with institutional trends in western institutions during the 1920s and 1930s, tuberculosis and pneumonia are of most interest. These are conditions that affect in particular those already suffering from a fragile state of health on account of other illnesses. They are also highly infectious and easily transmitted in closed institutional settings, even to those of a healthy physical condition. Lacking details from individual patients’ medical case notes, it is not clear if those who succumbed to these diseases had developed them on top of other conditions or if they constituted the immediate causes of death. It is also difficult to ascertain if these particular diseases were contracted inside the hospital or if they were present already on admission. Medical practitioners suggested that the latter was the case in mental hospitals in Britain (e.g. see Livesay, 1936: 1248). The Ranchi superintendent himself referred to observations from England, in particular to those that focused on the importance of dealing with patients’ physical conditions on admission (RIMH, 1935: 4). He also emphasized that at Ranchi 40 per cent of ‘so-called new admissions were not only suffering from chronic mental diseases but were equally bad in their physical health’. He kept reiterating that, in nearly every year’s report, ‘many were admitted in a very low state of health, markedly anaemic with consequent debility’ and that ‘tubercle bacilli were detected in the sputum of two newly admitted cases [out of a total of 126 that year] and one of them subsequently died’. More precise data are, however, lacking.

In western institutions, high death rates from pneumonia and TB were reported, with gradual downward trends towards the late 1930s. The surgeon in charge of the female section at St Andrew’s Mental Hospital in Norwich from 1922 to 1935, A.W.B. Livesay, for example, held that the steadily decreasing general mortality rate during this period correlated with the fall in deaths from phthisis or pulmonary tuberculosis. ‘Two-fifths of the reduction in the death rate’, he argued in an article in the British Medical Journal, ‘was attributed to the lessening of this disease’ (Livesay, 1936: 1248). Dr Percy Stokes of the department of applied statistics at University College London had earlier noted the ‘much higher mortality from tuberculosis in the mental hospitals’ (Stokes, 1924). He alerted medical practitioners in a letter published in the British Medical Journal that this made comparisons between hospital statistics and those pertaining to the wider population problematic, as inpatients tended to die from TB before they had a chance to develop any of the other diseases that affected those living outside institutions at later ages.

The high incidence of death from respiratory diseases persisted until well into the post-World War II period. In the early 1950s, sex- and age-adjusted mortality rates (SMR) for the state mental hospitals in Michigan revealed that even then deaths from pneumonia and tuberculosis among patients were about 30 and 13 times, respectively, as high as in the general population (Tokuhata and Stehman, 1958: 760). Death from tuberculosis among mental patients has become rare in modern institutions. Pneumonia, however, still figures highly, alongside cardiovascular disorders. For example, Hewer et al. showed in 1995 that in seven psychiatric hospitals in Germany roughly one-half of deaths were due to these two conditions.

The Ranchi data do appear to chime in with mortality trends prevalent in western institutions in regard to TB and pneumonia, as these have been assigned as causes of death in a significant number of cases, namely 41 and 21, respectively (out of a total of 272 deaths during the period 1926–33). Together with other illnesses of the respiratory system, the number amounted to 69 or 25 per cent of all classified causes of death. This is in contrast to 26 deaths (or 10 per cent) said to have been caused by a variety of parasitic and waterborne illnesses.

However, the high number of residual causes of death makes comparisons between institutions more difficult. What is more, despite worldwide attempts to make mortality diagnostics more uniform, there is evidence that at least some of the disease categories were used inconsistently (e.g. in the case of General Paralysis of the Insane (GPI), the occurrence of which among Indians the Ranchi superintendent had first denied and later confirmed.) The contemporary system of listing specific conditions such as encephalitis lethargica, kala-azar and leprosy alongside wide-ranging ones (e.g. general debility and senility) and of basing some categories on the symptoms observed (diarrhoea) and others on the identification of the location of disease (e.g. diseases of the circulatory system) does not in itself constitute a barrier to inter-institutional or longitudinal comparisons. However, as we cannot ascertain their consistent use by practitioners, it is difficult to ascertain with confidence any differential mortality patterns and changes over time.

There is however evidence of development towards a greater degree of precision in the identification of the causes of death, as the residual category ‘all other general diseases’ slowly disappears during the later 1930s – at least as a category figuring prominently in the main causes of death. Further investigation is needed to determine whether developments in laboratory diagnostics and post-mortem techniques during the 1930s had an impact on this trend or whether it was due to a more diligent application of a wider range of disease categories suggested in the various versions of the International List of Causes of Death (ICD). In the case of the gradual disappearance of ‘anaemia’, for example, as a main cause of death, increasing diagnostic precision and better use of laboratory testing prior to a patient’s death may have been involved. Raised awareness of the varied causes underlying observed cases of anaemia rather than confusion of symptoms with causes may have been another factor. In particular, the superintendent’s attention to patients’ diet, once the institution’s vegetable gardens, orchards and dairy were running on a large scale, may have eliminated anaemia caused by deficient nutrition, which was particularly common among those whose staple food had prior to hospital admission consisted of not much else but polished rice. The focus on the treatment of a wide range of parasitic worms with which patients presented on admission and the increasing use of ankylostomiasis (hookworm) as a separate category would have contributed to the disappearance of anaemia as a term used to identify the cause of death.

In regard to the causes assigned to patients’ death, comparisons between institutions in India and those in western countries are clearly difficult. We need to keep in mind also that during the 1920s and 1930s considerable changes occurred in the categories used for the registration of deaths. Not all clinicians were up to date with these, and there persisted a great deal of variation in diagnostic practices between practitioners (e.g., see: Farewell, Johnson and Armitage, 2006; Hardy and Magnello, 2004; Matthews, 1995). The debate on taxonomies of causes of death can be traced back to Sauvage’s (Francois Bossier de Lacroix) Nosologia Methodica (1763), Linnaeus’s Genera Morborum (1763) and William Cullen’s Synopsis Nosologiae Methodicae (1769), if not John Graunt’s work on the ‘Bills of Mortality’ (1662). During the nineteenth century, William Farr, statistician in the general register office of England and Wales, developed a classification for the International Statistical Congress (1853 in Brussels, 1855 in Paris), drawing mainly on Cullen’s categories. His nomenclature of the causes of death was arranged – or some would say, jumbled together – in five groups, namely: (1) epidemic; (2) constitutional (general); (3) local (by anatomical site); (4) developmental; and (5) result of violence.

This system, however incongruous it may appear to us now, constituted nevertheless an important step towards uniform classification. It eventually informed – alongside French, German and Swiss nomenclatures – the development under Jacques Bertillon (1851–1922), of the statistical services of Paris, of the ‘International List of Causes of Death’. This list was to result in the modern International Classification of Diseases (ICD), a standardized and regularly revised classification system that now encompasses not only causes of death (mortality) but also diseases and injuries (morbidity). ⁶ The process of standardization, however conceptually fraught and contested its various metamorphoses may have been (and still are considered to be), was an improvement on the even more unsatisfactory situation prevalent prior to 1893, before Bertillon’s system was adopted by at least some countries. A commentator in the British Medical Journal remarked (Anon., 1927a): ‘As late as 1893 no two countries in the world used exactly the same forms and methods for the statistical classification of the causes of death.’

The shortcomings of the situation, which are clearly detectable in the Ranchi data, were widely acknowledged. It was recognized that it did ‘not pose as an academic or scientific classification or proper nomenclature of diseases, but is intended to be a practical working list whereby the compilers of statistics can correctly tabulate the causes of death’ (Anon., 1927a, original italics). Bertillon’s system was the best that was available at the time and it constituted an improvement on previous practices. His manual was adapted for use in 1911 for England and Wales and in 1926 for Scotland and Northern Ireland (Manual, 1926). Absolute uniformity was not guaranteed, as each country exercised a ‘certain amount of latitude in adapting to its own special requirements the model of an abridged list’ provided by Bertillon (Anon., 1927a). There was also the difficulty of ‘constructing a satisfactory class of diseases on the ever-shifting sands of medical knowledge’ (Anon., 1927a). For our intention of gleaning insight into mortality patterns prevalent at Ranchi – let alone comparing them with those in other institutions – problems arise: ‘there is a considerable difference between a nomenclature of diseases classified as far as may be possible on a scientific basis and a list of causes of death’ (Anon., 1927a).

Conclusion

Some intriguing trends have emerged in the brief investigation of select statistical trends at the Ranchi Mental Hospital. Some echoed those prevalent in Britain (for example, significance of TB and pneumonia assigned as causes of death), others were more favourable (mortality rates). Gender-specific confinement rates were opposite to the trend assumed by gender historians following Showalter’s flawed conjectures for western institutions. All these require further investigation, not least because of the statistical and conceptual problems that are inherent in the data produced for any early twentieth-century institution and the potential limits these pose for comparative assessment. What is more, quantitative analyses also require a far more in-depth discussion of the local socio-cultural, political and administrative contexts as well as transnational forces and processes that frame institutional phenomena than has been attempted here.

Gender issues have had a high prominence in the above appraisal, being woven into some aspects of the analysis. This is partly due to the fact that gender historians assumed a comparative perspective well before this became, relatively recently, a more commonly mooted methodology. We therefore have a ‘narrative’ in relation to some aspects of gender and madness with which studies based on individual institutions can readily engage. However, such preliminary engagement indicates that this existing narrative may need to be reconsidered and adjusted, as well as broadened to include currently less well studied aspects (such as locality and culturally-specific admission and discharge policies), following a more critical engagement with the statistical and conceptual problems pertaining to institutional data.

Despite the merely explorative account provided here and the many caveats raised in relation to the suggestions made, there emerge some observations on how subsequent institutional histories might be framed. A number of problems have been highlighted that make quantitative comparisons between institutions during the early part of the twentieth century a troubled endeavour. These are related to: the nature of the available statistics (CMR instead of SMR, for example); the only gradual implementation of uniform and reliable classifications (variations in the nosologies for causes of death in different countries despite the development of the ICD); culturally and context-specific parameters (local admission and discharge priorities); and individual doctors’ varied diagnostic practices. The problems are relevant not only in regard to issues of gender, which have been the focus in some of the evidence presented above. They also need to be tackled in relation to any intra-institutional and inter-institutional qualitative assessment that is properly anchored in quantitative trends.

The problem of scientific classification and statistical analysis of medical data has not been adequately resolved to the present day. Nosologies, diagnostics and medical statistics have improved but are still far from perfect. However, as White (1974: 297) has pointed out: ‘If some clinicians have come to think that ‘making a diagnosis’ is more important than helping the patient to ameliorate or resolve his problems, so some epidemiologists may have been unduly concerned with descriptive and observational precision. Both mistake ends for means. Major Greenwood’s similarly pragmatic observation of 1943, quoted at the start of this article, on the futility of waiting for medical statistics to become reliable and conceptually valid continues to be quoted by modern medical epidemiologists, medical statisticians and clinicians alike. Following Greenwood’s and White’s (1974) contention, historians of medicine should not shy away from assessing available statistics, but do need to abstain from undue generalizations especially when data sets from different institutions are compared. Greenwood’s statement serves as an epitaph for the above examination of institutional trends at Ranchi. However, in order to arrive at sensible comparisons between different institutions, historians may have to join the rustic for a while, watching conceptual, statistical and administrative ebbs and flows and eddies carefully before crossing, connecting and comparing the institutional data of different institutions.

Given that a lacuna has been identified by contributors to a recent Special Issue of History of Psychiatry (2011, 23(2)), historians are likely in future to focus more on trends in psychiatry during the twentieth century. Any ‘narrative’ similar to the ones available for the late eighteenth and the nineteenth centuries is likely to emerge only when single institution-based studies are written with the intention of linking up with and exploring connections, continuities and discontinuities between other institutions rather than aiming at adding yet another, ‘internalist’, case-study to the, albeit limited, existing field. A comparative and transnational methodology therefore is bound to be integral to any new work.

However, although transnational methodologies are increasingly being coveted by historians of medicine, at least one of the trends in the Ranchi data indicates that some perceptual and conceptual re-orientation may be necessary. The transnational perspective would be myopic if it kept its eyes mainly on exchanges between a variety of western countries or transnational transfers from the western to ‘other’ countries. Studies of institutions in Europe and North America must look beyond the confining rim of Eurocentric self-containment and acknowledge (and relate their work to) other institutions around the world. They also ought to abstain from prematurely assuming that any exchange or transfer would necessarily be one-way, with western countries providing the blueprint or impetus for developments, and other nations following suit (some by free choice, like Japan; others by force of colonial dictum, as in the case of India). Nor should it be assumed that markers of clinical excellence and good institutional practice are unequivocally the hallmark of psychiatry in ‘the west’. As the case of Ranchi showed, favourable mortality rates may also be achieved by Indian doctors in ‘the east’.

The hitherto lacking and envisaged narrative for the history of twentieth-century psychiatry therefore not only needs to talk of, say, Britain, Germany, France, and the different states of north America as well as, for example, Japan, India and south America, but connect the trajectories of the former with those of the latter. Historians working on the history of western institutions during the twentieth century need to be aware that the onus is on them to abstain from uncritically reproducing hegemonic histories of the modern world in which western cultures and nations are posited by default as the centre or metropolis and the rest as peripheries whose social and scientific developments may be seen to be of exotic interest, but merely derivative and peripheral.