Advances in Psychiatric Treatment (1997), vol. 3, pp. 211-218 Life expectation in organic brain disease David Jolley & David Baxter The purpose of this review is to outline current with accuracy or certainty; for example, onset of knowledge on the life expectation of people illness or onset of symptoms. Similarly, death or suffering from organic brain disease, the tech discharge from hospital are hard, reliable outcome niques available for describing and comparing life events, whereas 'recovery from symptoms' is less expectation in populations, factors which are easy to determine. associated with longer and shorter lifeexpectation, Follow-up studies present a number of dif and the causes of death among patients with this ficulties in analysis. Some patients will have condition. reached the event of interest (death in studies of survival/mortality) while others have not. The period over which individuals have been followed usually varies and some patients will have been Survival analysis lost from the study, perhaps because they have moved away, withdrawn from the study, or become non-compliant with the treatment of Death is an event or outcome which occurs to an interest. Data relating to such individuals can be individual, and as such is similar to other outcomes included in analyses of survival/mortality but which include marriage, admission to hospital, needs to be 'censored' so that they contribute only discharge from hospital, or recovery from the time/experience during which they satisfy symptoms. These can be related in time to other inclusion criteria of the analysis. (previous) events - birth, onset of illness, referral Simple analyses of survival/mortality can be to a service, or onset of treatment. Analysis of the presented as: the proportion of patients reaching time between an onset and an outcome tells us the end-point (death) during a predetermined something of the experience of an individual, or period of follow-up; the median survival time for of groups of individuals who may be considered the cohort using lifetables or Kaplan-Meier curves; together because they share particular character or death rates per unit of time in follow-up (e.g. istics (they might be suffering from the same illness person-years at risk) (Altman, 1992). or receiving a common treatment regime). Having determined the survival experience of a The statistical techniques used to analyse these group of individuals, one is usually interested in experiences are described as survival analyses. The how this compares with a reference population usefulness of such analyses are determined by the (e.g. a non-exposed group, a disease-free group, validity and reliability of the basic data used. Some or a standard treatment group). Survival curves onset data are easily determined, and should be can be compared using the log rank test, which both accurate and reliable; for example, date of tests the null hypothesis that the groups being birth, or date of admission to hospital. Other onset compared are from the same population. Altern data, which are appealing because of their potential atively, a measure of difference between the link to biological events, are less easy to determine survival experiences of two populations can be David Jolley is Professor of Old Age Psychiatry, University of Wolverhampton and Wolverhampton Health Care NHS Trust (Penn Hospital, Penn Road, Wolverhampton, West Midlands WV4 5HN). Recently, he moved to Wolverhampton to re-establish mental health and elderly care services to a community model, in which to continue his teaching and research ideas developed during 20 years in Manchester. David Baxter is Senior Lecturer in Public Health Medicine at Manchester University, with research interests in the epidemiology of various disorders, including mental health problems. APT (1997), vol. 3, p. 212 follet/ & Baxter obtained by calculating the 'hazard ratio' (denoted 4.5 :1 to 3.5 :1 (Sangstad & Odegard, 1979). Similar 'R'), which is computi J from the observed number reductions in death rates and lowered excess of events (deaths) compared with the number of mortality ratios were achieved in mental hospitals events expected if the null hypothesis (that the two in the rest of Europe and North America, probably groups are from the same population) were true. because of improved conditions of hygiene, heating O, and food. R = o, Death rates among psychiatric in-patients continue to exceed those of the general population Confidence intervals can be calculated for by factors of between two and three. Many individual survival probabilities and for the unexpected deaths occur to young male patients difference between survival probabilities for two and these are most often due to suicide or other groups. non-accidental injuries. Advances in treatment methods and alterations in service styles in psychiatric practice have meant that many patients Mortality in mentally ill are now treated within their natural community. Thus, studies of survival/death rates among the populations mentally ill must utilise techniques which allow patients to be followed over periods of time through different modes of care. This can be Although mental illnesses are not usually looked achieved by a service-based register (Rorsman, on as causative of early death in the same way as 1974), record-linkage (Black et al, I985a,b,c), or by conditions such as cancer, heart disease or acute creating a short-term register to follow a limited infections, it has been known for many years that cohort of patients over a short time (Martin et al, survival among psychiatric patients is shorter than 1985). for the general population of similar age and in Studies of out-patients demonstrate that death the same calendar period. rates are higher than in the general population, Using the Norwegian Psychiatric Case Register, relative risk of death being more markedly raised Odegard demonstrated that mortality rates among among younger patients but remaining elevated the asylum populations at the turn of the century into later life (Odegard, 1953; Rorsman, 1974; Black were five times those of the general population et al, I985a,b,c; Zilber et al, 1989; Wood et al, 1991). (Odegard, 1953). Malzeberg reported similarly Reduced survival is found within all the major increased death rates among patients admitted to psychiatric diagnostic groupings: schizophrenia New York mental hospitals during the period (Zilber et al, 1989; Baxter, 1993), mood disorder 1929-1931, and undertook a longitudinal study of (Black et al, I985a,b,c; Zilber et al, 1989; Baxter, patients admitted to these hospitals between 1943 1993), personality disorder (Blackeíal, 1985; Zilber and 1944, which demonstrated an increased risk et al, 1989; Baxter, 1993), neurosis (Black et al, of death for all diagnostic groups over a four-year 1985í7,b,c;Baxter, 1993) and organic brain disease. follow-up period. The relative risk was highest for patients suffering from organic brain syndromes (Malzeberg, 1952). Most of the excess deaths within Mortality in organic brain mental hospitals were due to infectious diseases, including tuberculosis, and may have been related disease to impaired nutrition and the poor, overcrowded conditions under which 'inmates' were kept (Digby, 1983). The diagnosis of organic brain disease is made on Death rates among the general population of clinical grounds at a 'syndromal' level. That is, it England and other European countries have fallen brings together a number of conditions which progressively through the 20th century; by 1970 produce similar alterations in mental state: death rates among women had fallen to half the disorders of cognition, impaired memory, difficulty level in 1900 and death rates among men had fallen with registration and recall of information, poor by one-third in the same period (Cartwright et al, concentration and limited tolerance. There may be 1973). Improvements in life expectation were disorientation in time, place or person and even greater among male patients admitted to perseveration of thoughts, acts or feelings. Norwegian mental hospitals, so that their excess Emotional control may be impaired and there may mortality ratio fell from 5.1 :1 to 2.4 :1. For female be other abnormalities of mood, psychotic patients, however, the improvement in mortality phenomena or evidence of focal brain disorder was less, with the excess mortality falling from (Lishman, 1978). It is an acquired condition rather Organic brain disease APT (1997), vol. 3, p. 213 than one present from birth and is associated with alcohol. The age spectrum of these patients was damage to brain substrate from trauma, infection, mainly middle age: mean age for men 47.4 years; neoplasia, pressure or degenerative disorder. for women 51.0 years. The average period of Similar changes may be produced by alterations follow-up was 4.5 years. The estimated standard in the bodily milieu due to failure of the internal ised mortality rates (SMRs) for both male (1.75) homeostatic mechanisms. and female (2.1) patients were significantly raised In population studies the most common under at the P<0.001 level. SMRs were greatest among lying causes of organic brain disease are the patients aged under 40 years (male 8.23, female degenerative dementias such as Alzheimer's 15.63) and fell to below statistical significance in disease, multi-infarct (vascular) dementia and those aged over 69 years (male 1.05, female 1.12). Lewy body disease. Yet many studies are neither Death rates were particularly high during the three designed nor equipped to penetrate beneath the months after admission, but remained significantly syndromal level of classification, and findings are elevated throughout the 24 months after first expressed at this level. As more refined clinical and service contact. Expected deaths were computed investigatory tools have become available, recent using the mortality rates from the Iowa population. studies have restricted their interest to specific Magnusson (1989) addressed a very different conditions (often Alzheimer's disease).