Prognostic studies include clinical studies of variables predictive of future events as well as epidemiological studies of aetiological risk factors. As multiple similar studies accumulate it becomes increasingly important to identify and evaluate all of the relevant studies to develop a more reliable overall assessment. For prognostic studies this is not straightforward.

Box 1 summarises the clinical importance of information on prognostic factors. Many of the issues discussed are also relevant to aetiological studies, especially cohort ones. Some features of prognostic studies lead to particular difficulties for the systematic reviewer. Firstly, in most clinical prognostic studies the outcome of primary interest is the time to an event, often death. Meta-analysis of such studies is rather more difficult than that for binary data or continuous measurements. Secondly, in many contexts the prognostic variable of interest is often one of several prognostic variables. When examining a variable of interest researchers should consider other prognostic variables with which it might be correlated. Thirdly, many prognostic factors are continuous variables, for which researchers use a wide variety of methods of analysis.

Since 1991 the BMJ has had a policy of not publishing trials that have not been properly randomised, except in rare cases where this can be justified.1 Why?

The simplest approach to evaluating a new treatment is to compare a single group of patients given the new treatment with a group previously treated with an alternative treatment. Usually such studies compare two consecutive series of patients in the same hospital(s). This approach is seriously flawed. Problems will arise from the mixture of retrospective and prospective studies, and we can never satisfactorily eliminate possible biases due to other factors (apart from treatment) that may have changed over time. Sacks et al compared trials of the same treatments in which randomised or historical controls were used and found a consistent tendency for historically controlled trials to yield more optimistic results than randomised trials.2 The use of historical controls can be justified only in tightly controlled situations of relatively rare conditions, such as in evaluating treatments for advanced cancer.

This is a section of the book The Challenge of Epidemiology: Issues and Selected Readings. Edited by four eminent epidemiologists, this book consolidates, for the first time, a core of landmark articles on the evolution, scope and limitations, uses, and prospects of epidemiology. An outstanding feature of the book is the inclusion of the editors' assessments of the realm of epidemiology, where it is and where it should be going. It represents a useful tool for both students and practicing professionals and provides a much-needed frame of reference for reorienting the practice of epidemiology. The book is a collection of 91 articles, grouped in five parts. This article deals with the health, in particular pediatric and dental, aspects of water supplies that contain fluoride ion. A brief discussion on the metabolism of fluorides is provided.

Background. Few studies have examined the extent to which populations of suicides and attempted suicides are similar, or different. This paper compares suicides and serious suicide attempts in terms of known risk factors for suicidal behaviour.

Methods. Using case-control methodology, risk factors for suicidal behaviour were examined in 202 individuals who died by suicide, 275 individuals who made medically serious suicide attempts and 984 randomly selected control subjects. Based on data from significant others, measures used spanned sociodemographic factors, childhood experiences, psychiatric morbidity and psychiatric history, exposure to recent stressful life events and social interaction.

Results. Multiple logistic regression identified the following risk factors that were common to suicide and serious suicide attempts: current mood disorder; previous suicide attempts; prior out-patient psychiatric treatment; admission to psychiatric hospital within the previous year; low income; a lack of formal educational qualifications; exposure to recent stressful interpersonal, legal and work-related life events. Suicides and suicide attempts were distinguished in the following ways: suicides were more likely to be male (OR = 1·9, 95 Conclusions. Suicides and medically serious suicide attempts are two overlapping populations that share common psychiatric diagnostic and history features, but are distinguished by gender and patterning of psychiatric disorder.

Background For many years it has been claimed that observational studies find stronger treatment effects than randomized, controlled trials. We compared the results of observational studies with those of randomized, controlled trials.

Methods We searched the Abridged Index Medicus and Cochrane data bases to identify observational studies reported between 1985 and 1998 that compared two or more treatments or interventions for the same condition. We then searched the Medline and Cochrane data bases to identify all the randomized, controlled trials and observational studies comparing the same treatments for these conditions. For each treatment, the magnitudes of the effects in the various observational studies were combined by the Mantel-Haenszel or weighted analysis-of-variance procedure and then compared with the combined magnitude of the effects in the randomized, controlled trials that evaluated the same treatment.

Results There were 136 reports about 19 diverse treatments, such as calcium-channel-blocker therapy for coronary artery disease, appendectomy, and interventions for subfertility. In most cases, the estimates of the treatment effects from observational studies and randomized, controlled trials were similar. In only 2 of the 19 analyses of treatment effects did the combined magnitude of the effect in observational studies lie outside the 95 percent confidence interval for the combined magnitude in the randomized, controlled trials.

Conclusions We found little evidence that estimates of treatment effects in observational studies reported after 1984 are either consistently larger than or qualitatively different from those obtained in randomized, controlled trials.

A method is described for the determination of a measure of relative risk from vital statistical data. If the frequency of disease in a population is linearly related to the level of exposure to a given factor, then a measure of the relative risk can be estimated from the slope and intercept of the regression line. For example, when the exposure is measured in terms of the proportion of the population exposed to the factor, then the relative risk is equal to [Formula: see text] This offers an indirect but simple and inexpensive method for estimating relative risk. It should be used with caution, particularly where confounding factors may be responsible for the apparent association between disease and factor. Applications of the method to estimate the relative risk of (a) circulatory diseases in women using oral contraceptives and (b) ovarian cancer in women with different average family sizes, both yielded relative risk estimates comparable with those obtained from case-control and prospective studies.

OBJECTIVE: Major depression is a frequent and disabling psychiatric disorder in the United States. This report examines the prevalence and risk factor profile of both pure and comorbid major depression according to data from the National Comorbidity Survey. METHOD: To estimate the prevalence of psychiatric comorbidity in the United States, a national sample of 8,098 persons 15-54 years of age from the 48 conterminous states was surveyed with a modified version of the Composite International Diagnostic Interview. Results: From the survey data the prevalence of current (30-day) major depression was estimated to be 4.9%, with a relatively higher prevalence in females, young adults, and persons with less than a college education. The prevalence estimate for lifetime major depression was 17.1%, with a similar demographic distribution. Both 30-day and lifetime prevalence estimates were higher than estimates from the earlier Epidemiologic Catchment Area study. When pure major depression was compared with major depression co-occurring with other psychiatric disorders, the risk factor profiles exhibited clear differences. CONCLUSIONS: These findings suggest a greater burden of major depression in community- dwelling persons than has been estimated from previous community samples. The risk factor profile showed significant differences between persons with pure and combined major depression.

This article traces the development of the design of experiments from origins in the mind and professional experience of R.A. Fisher between 1922 and 1926. The article indicates how the analysis of variance procedure stimulated design, being justified by the principle of randomization that Fisher introduced with the analysis, and exploited by his use of blocking and replication. The article indicates the radically new form and efficiency of factorial block designs, shows the further advantages accruing to factorial arrangements through confounding, and suggests how Fisher's close collaboration with experimenters stimulated these developments.

BACKGROUND: The importance of stressful life events and long-term difficulties in the onset of episodes of unipolar depression is well established for young and middle-aged persons, but less so for older people. METHOD: A prospective case-control study was nested in a large community survey of older people. We recruited 83 onset cases during a 2-year period starting 2 1/2 years after the survey, via screening (N = 59) and GP monitoring (N = 24), and 83 controls, a random sample from the same survey population. We assessed depression with the PSE-10 and life stress exposure with the LEDS. RESULTS: Risk of onset was increased 22-fold by severe events and three-fold by ongoing difficulties of at least moderate severity. Severe events accounted for 21% of all episodes but ongoing difficulties for 45%. The association of onset with life stress, often health-related such as death, major disability and hospitalization of subject or someone close, was most pronounced in the cases identified by screening. While a clear risk threshold for events was found between threat 2 and 3 (on a scale of 1-4), the risk associated with difficulties increased more gradually with severity of difficulty. Compared with controls, severe events involved a larger risk for cases without a prior history of depression (OR = 39.48) than for cases with (OR = 8.86). The opposite was found for mild events (OR = 2.94 in recurrent episodes; OR = 1.09 in first episodes). The impact of ongoing difficulties was independent of severity of episode and history of depression. CONCLUSION: Although the nature of life stress in later life, in particular health-related disability and loss of (close) social contacts, is rather different from that in younger persons, it is a potent risk factor for onset of a depressive episode in old age. Severe events show the largest relative risk, but ongoing difficulties account for most episodes. The association of severe events with onset tends to be stronger in first than in recurrent episodes. Mild events can trigger a recurrent episode but not a first one.

Psychiatric case-identification in general populations allows us to study both individuals with functional psychiatric disorders and the populations from which they come. The individual level of analysis permits disorders to be related to factors of potential aetiological significance and the study of attributes of the disorders that need to be assessed in non-referred populations (an initially scientific endeavour). At the population level valid case identification can be used to evaluate needs for treatment and the utilization of service resources (a public health project). Thus, prevalence is of interest both to scientists and to those responsible for commissioning and planning services (Brugha et al. 1997; Regier et al. 1998). The quality of case identification techniques and of estimates of prevalence is thus of general concern (Bartlett & Coles, 1998).

Structured diagnostic interviews were introduced into general population surveys in the 1970s as a method `to enable interviewers to obtain psychiatric diagnoses comparable to those a psychiatrist would obtain' (Robins et al. 1981). The need to develop reliable standardized measures was partly driven by an earlier generation of prevalence surveys showing rates ranging widely from 10·9% (Pasamanick et al. 1956) to 55% (Leighton et al. 1963) in urban and rural North American communities respectively. If the success of large scale psychiatric epidemiological enquiries using structured diagnostic interviews and standardized classifications is measured in terms of citation rates it would seem difficult to question. But the development of standardized interviews of functional psychiatric disorders has not solved this problem of variability: the current generation of large scale surveys, using structured diagnostic interviews and serving strictly defined classification rules, have generated, for example, 12-month prevalence rates of major depression in the US of 4·2% (Robins & Regier, 1991) and 10·1% (Kessler et al. 1994). This calls into question the validity of the assessments, such that we must reopen the question of what they should be measuring and how they should do it.

BACKGROUND: Comparisons of structured diagnostic interviews with clinical assessments in general population samples show marked discrepancies. In order to validate the CIS-R, a fully structured diagnostic interview used for the National Survey of Psychiatric Morbidity in Great Britain, it was compared with SCAN, a standard, semi-structured, clinical assessment. METHODS: A random sample of 1882 Leicestershire addresses from the Postcode Address File yielded 1157 eligible adults: of these 860 completed the CIS-R; 387 adults scores > or = 8 on the CIS-R and 205 of these completed a SCAN reference examination. Neurotic symptoms, in the previous week and month only, were enquired about. Concordance was estimated for ICD-10 neurotic and depressive disorders, F32 to F42 and for depression symptom score. RESULTS: Sociodemographic characteristics closely resembled National Survey and 1991 census profiles. Concordance was poor for any ICD-10 neurotic disorder (kappa = 0.25 (95% CI, 0.1-0.4)) and for depressive disorder (kappa = 0.23 (95% CI, 0-0.46)). Sensitivity to the SCAN reference classification was also poor. Specificity ranged from 0.8 to 0.9. Rank order correlation for total depression symptoms was 0.43 (Kendall's tau b; P < 0.001; N = 205). DISCUSSION: High specificity indicates that the CIS-R and SCAN agree that prevalence rates for specific disorders are low compared with estimates in some community surveys. We have revealed substantial discrepancies in case finding. Therefore, published data on service utilization designed to estimate unmet need in populations requires re-interpretation. The value of large-scale CIS-R survey data can be enhanced considerably by the incorporation of concurrent semi-structured clinical assessments.

Background. The success of large scale surveys depends on well designed questionnaires and the skills of lay interviewers. Discrepancies in prevalence rates between epidemiological surveys and poor agreement between survey interviewer and clinician diagnostic interviews are giving rise to increasing concern among researchers, public health planners and policy developers. New approaches to information collection are called for. The feasibility of training experienced survey interviewers in semi-structured, clinical, diagnostic interviewing has never been investigated systematically across the range of neurotic and psychotic disorders. Methods. Eight experienced survey interviewers from the Office for National Statistics (ONS) were selected and underwent extended training in a Survey Form of SCAN (SCAN-SF). Sixty-four adults, including a majority of psychiatric in-patients were assessed by ONS interviewers and reinterviewed within a week by SCAN-trained clinicians. Feedback was sought from interviewers and trainers. Results. Trainers found lay interviewers coped at least as well with psychotic as with neurotic symptoms. Concordance for any disorder was 0.74 (95% CI: 0.57 to 0.91); for any specific psychotic disorder 0.63 (0.40 to 0.86); for any specific neurotic disorder 0.63 (0.43 to 0.83). Sensitivity ranged from 0.6 to 0.9 and specificity from 0.8 to 0.9. There was no evidence of rater bias. Conclusions. These preliminary findings are very promising. However, before the SCAN-SF, administered by carefully trained lay interviewers, can be recommended in large scale surveys, further evaluations of its feasibility and reliability in the general population are needed.

BACKGROUND: We examined whether children who are diagnosed as having schizophrenia in adulthood could be distinguished from their peers on performance in elementary school. METHODS: We used a case-control study design nested within a population-based birth cohort of all individuals born in Helsinki, Finland, between January 1, 1951, and December 31, 1960. Case ascertainment was from 3 national health care registers. Elementary school records were obtained for 400 children who were diagnosed as having schizophrenia in adulthood and for 408 controls. Results were analyzed for the 4 years of schooling (ages 7-11 years) that were common to all pupils. School subjects were entered into a principal components analysis and produced 3 factors: academic, nonacademic, and behavioral. These factors were compared between cases and controls after adjusting for sex and social group. Eligibility for high school and progression to high school were investigated among cases and controls. RESULTS: Cases performed significantly worse than controls only on the nonacademic factor (which loaded sports and handicrafts). There were no differences between the groups on the academic or behavioral factors, and there were no significant clinical correlates of factor scores. Cases were significantly less likely than controls to progress to high school, despite similar eligibility. CONCLUSIONS: Poor performance in sports and handicrafts during elementary school, which may indicate a motor coordination deficit, appears to be a risk factor for later schizophrenia. Poor academic performance in elementary school was not a risk factor for schizophrenia in this study. Lack of expected progression to high school among cases, despite good academic grades, provides evidence for deteriorating premorbid functional adjustment in schizophrenia.

Background In the hierarchy of research designs, the results of randomized, controlled trials are considered to be evidence of the highest grade, whereas observational studies are viewed as having less validity because they reportedly overestimate treatment effects. We used published meta-analyses to identify randomized clinical trials and observational studies that examined the same clinical topics. We then compared the results of the original reports according to the type of research design.

Methods A search of the Medline data base for articles published in five major medical journals from 1991 to 1995 identified meta-analyses of randomized, controlled trials and meta-analyses of either cohort or case-control studies that assessed the same intervention. For each of five topics, summary estimates and 95 percent confidence intervals were calculated on the basis of data from the individual randomized, controlled trials and the individual observational studies.

Results For the five clinical topics and 99 reports evaluated, the average results of the observational studies were remarkably similar to those of the randomized, controlled trials. For example, analysis of 13 randomized, controlled trials of the effectiveness of bacille Calmette-Guérin vaccine in preventing active tuberculosis yielded a relative risk of 0.49 (95 percent confidence interval, 0.34 to 0.70) among vaccinated patients, as compared with an odds ratio of 0.50 (95 percent confidence interval, 0.39 to 0.65) from 10 case-control studies. In addition, the range of the point estimates for the effect of vaccination was wider for the randomized, controlled trials (0.20 to 1.56) than for the observational studies (0.17 to 0.84).

Conclusions The results of well-designed observational studies (with either a cohort or a case-control design) do not systematically overestimate the magnitude of the effects of treatment as compared with those in randomized, controlled trials on the same topic.

This study aimed to test the procedural validity of the psychosis module of the Composite International Diagnostic Interview (CIDI) by comparing it with diagnostic checklists completed by experienced clinicians. Seventy-five subjects were interviewed using the interviewer-administered version of the CIDI. Their clinician(s) then completed diagnostic checklists for DSMIV and ICD10 diagnoses of schizophrenia. Agreement was measured at the diagnostic, criterion and subcriterion levels. The validity standard (diagnostic checklist) was shown to be reliable with interrater agreement between the clinicians for the diagnosis of schizophrenia being excellent (K = 0.82 for DSMIV and K = 0.71 for ICD10). The agreement between the CIDI and the clinician checklists varied with sensitivities for DSMIV subcriteria ranging from 0.18 (negative symptoms) to 0.93 (bizarre delusions) and specificities ranging from 0.38 (catatonia) to 0.95 (disorganised speech). A similar pattern was found for ICD10 subcriteria: sensitivity varied from 0.19 (neologisms) to 0.90 (persistent delusions) and specificity varied from 0.39 (catatonia) to 0.95 (negative symptoms). The poorest levels of agreement were found for symptoms requiring interviewer judgement. The CIDI showed good agreement with clinician checklist diagnoses when the criteria were based on questions asked of the subjects. When the interviewer was required to make judgement of behaviours, the agreement between the CIDI and the clinician checklists was lower, resulting in overall poor agreement between the CIDI and the clinician checklists. Suggestions for improving the validity of the psychosis module of the CIDI are made.

Background Previous research has found an association between obstetric complications and schizophrenia, but in many studies the sample size was limited, and no assessment of specific exposures was possible.

Aims To assess the role of different complications, and in particular to distinguish between disordered foetal development and hypoxia at birth.

Method From the Stockholm County In-Patient Register and community registers, we identified 524 cases of schizophrenia and 1043 controls, matched for age, gender, hospital and parish of birth. Data on obstetric complications were obtained from birth records.

Results There was a strong association between signs of asphyxia at birth and schizophrenia (OR 4.4; 95 Conclusions Signs of asphyxia at birth are associated with an increased risk of schizophrenia in adults.

Objective : To assess the hazards associated with long term use of tobacco. Design : Prospective study of mortality in relation to smoking habits assessed in 1951 and again from20time to time thereafter, with causes sought of deaths20over 40 years (to 1991). Continuation of a study that was last reported after 20 years' follow up (1951-71). Subjects : 34 439 British male doctors who replied20to a postal questionnaire in 1951, of whom 10 000 had died during the first 20 years and another 10 000 have died during the second 20 years. Results - Excess mortality associated with20smoking was about twice as extreme during the second half of the study as it had been during the first half. The death rate ratios during 1971-91 (comparing continuing cigarette smokers with lifelong20non-smokers) were approximately threefold at ages 45-64 and twofold at ages 65-84. The excess mortality was chiefly from diseases that can be caused by smoking. Positive associations with smoking were confirmed for death from cancers of the mouth, oesophagus, pharynx, larynx, lung, pancreas, and bladder; from chronic obstructive pulmonary disease and other respiratory diseases; from vascular diseases; from peptic ulcer; and (perhaps because of confounding by personality and alcohol use) from cirrhosis, suicide, and poisoning. A negative association was confirmed with death20from Parkinson's disease. Those who stopped20smoking before middle age subsequently avoided almost all of the excess risk that they would otherwise have suffered, but even those who stopped smoking in middle age were subsequently at substantially less risk than those who continued to smoke. Conclusion : Results from the first 20 years of this study, and of other studies at that time, substantially20underestimated the hazards of long term use of tobacco. It now seems that about half of all regular cigarette smokers will eventually be killed by their habit.

Including all relevant material - good, bad, and indifferent - in meta -analysis admits the subjective judgments that meta-analysis was designed to avoid. Several problems arise in meta-analysis: regressions are often non -linear; effects are often multivariate rather than univariate; coverage can be restricted; bad studies may be included; the data summarised may not be homogeneous; grouping different causal factors may lead to meaningless estimates of effects; and the theory-directed approach may obscure discrepancies. Meta-analysis may not be the one best method for studying the diversity of fields for which it has been used.

Why do we undertake systematic reviews of a given field? The most important reason is perhaps that we are concerned about a particular theory and wish to know how the evidence for and against stacks up. There are also practical reasons; single studies often use small numbers of subjects, and basing our estimates of effect sizes on large numbers of studies drastically lowers the fiducial limits around our estimates. Systematic reviews can be of several different kinds: traditional reviews, often not very systematic, and frequently biased; meta-analyses, including (we hope) all relevant material, good, bad, and different, and leading to an estimate of effect size*RF 1-3*; best-evidence synthesis4; and the hypothetico-deductive approach,5 in which the effort is directed at evaluating the evidence for and against a given theory, in an attempt to solve the problem of why contradictory results appear, rather than simply averaging often incompatible data.

Gender differences in social support tend to suggest that women have larger social networks and both give and receive more support than men. Nevertheless, although social support has been identified as protective of mental health, women have higher rates of psychological distress than men. We examine the prospective association between social support and psychological distress by gender in a cohort study of middle aged British Civil Servants, the Whitehall II study. In this sample we found that women have a larger number of close persons than men although men have larger social networks. We also found that the effects of marital status, social support within and outside the workplace and social networks on subsequent occurrence of psychological distress were similar for men and women independently of baseline mental health status.

The last decades have shown a rapid increase in nonresponse rates. For this reason it is important to study nonresponse and think about it critically. In this article we investigated whether nonresponse affected estimates of the levels of adjustment problems in children and the correlations between these outcomes. The nonresponse was caused by parents who refused permission to interview their children at school, parents who did not return a questionnaire, teachers who did not complete the questionnaire, and parents who refused to participate in an in-depth study, with nonresponse rates of 9%, 69%, 25%, and 46% respectively. The sample consisted of 1282 children aged 4-5 years and the dependent measures were peer-rated sociometric status, self-rated wellbeing at school, and teacher-rated behaviour problems. Despite considerable nonresponse in some conditions our results showed hardly any evidence for bias. This suggested that bias cannot simply be inferred from the amount of nonresponse and that standard rules such as nonresponse rates higher than 50 % are not acceptable lack a scientific basis. Instead, we argue that to assess nonresponse bias the specific conditions and analyses of the study will need to be considered and special measures may be required.

BACKGROUND: Understanding variations in the incidence of schizophrenia is a crucial step in unravelling the aetiology of this group of disorders. The aims of this review are to systematically identify studies related to the incidence of schizophrenia, to describe the key features of these studies, and to explore the distribution of rates derived from these studies. METHODS: Studies with original data related to the incidence of schizophrenia (published 1965-2001) were identified via searching electronic databases, reviewing citations and writing to authors. These studies were divided into core studies, migrant studies, cohort studies and studies based on Other Special Groups. Between- and within-study filters were applied in order to identify discrete rates. Cumulative plots of these rates were made and these distributions were compared when the underlying rates were sorted according to sex, urbanicity, migrant status and various methodological features. RESULTS: We identified 100 core studies, 24 migrant studies, 23 cohort studies and 14 studies based on Other Special Groups. These studies, which were drawn from 33 countries, generated a total of 1,458 rates. Based on discrete core data for persons (55 studies and 170 rates), the distribution of rates was asymmetric and had a median value (10Mesh-terms: Cohort Studies; Female; Humans; Incidence; Male; Rural Health; Schizophrenia :: epidemiology; Sex Distribution; Transients and Migrants :: statistics & numerical data; Urban Health;

It has been suggested that the quality of clinical trials should be assessed by blinded raters to limit the risk of introducing bias into meta-analyses and systematic reviews, and into the peer-review process. There is very little evidence in the literature to substantiate this. This study describes the development of an instrument to assess the quality of reports of randomized clinical trials (RCTs) in pain research and its use to determine the effect of rater blinding on the assessments of quality. A multidisciplinary panel of six judges produced an initial version of the instrument. Fourteen raters from three different backgrounds assessed the quality of 36 research reports in pain research, selected from three different samples. Seven were allocated randomly to perform the assessments under blind conditions. The final version of the instrument included three items. These items were scored consistently by all the raters regardless of background and could discriminate between reports from the different samples. Blind assessments produced significantly lower and more consistent scores than open assessments. The implications of this finding for systematic reviews, meta-analytic research and the peer-review process are discussed.

Schizophrenia has been linked with childhood psychological abnormalities since it was first described, but studies of associations have not used population samples and so may be subject to bias. We have studied associations between adult-onset schizophrenia and childhood sociodemographic, neurodevelopmental, cognitive, and behavioural factors within a cohort of 5362 people born in the week March 3-9, 1946. Childhood data were gathered prospectively and case ascertainment was independent of routine follow-up of this cohort. 30 cases of schizophrenia arose between ages 16 and 43 years (cumulative risk 0.63% [95% CI 0.41-0.86%]). Milestones of motor development were reached later in cases than in controls, particularly walking (difference in means 1.2 months [0.1-2.3], p = 0.005), and up to age 15, cases had more speech problems than had controls (odds ratio 2.8 [0.9-7.8], p = 0.04). Low educational test scores at ages 8, 11, and 15 years were a risk factor, with significant linear trends across population distributions; risk was not confined to very low scores. Solitary play preference at ages 4 and 6 years predicted schizophrenia (odds ratios 2.1, 2.5, p = 0.05). At 13 years cases rated themselves as less socially confident (p for trend, 0.04). At 15 years, teachers rated cases as being more anxious in social situations (p for trend 0.003), independent of intelligence quotient. A health visitor's rating of the mother as having below average mothering skills and understanding of her child at age 4 years was a predictor of schizophrenia in that child (odds ratio 5.8 [0.8-31.8], p = 0.02). Differences between children destined to develop schizophrenia as adults and the general population were found across a range of developmental domains. As with some other adult illnesses, the origins of schizophrenia may be found in early life.

The quality of controlled trials is of obvious relevance to systematic reviews. If the "raw material" is flawed then the conclusions of systematic reviews cannot be trusted. Many reviewers formally assess the quality of primary trials by following the recommendations of the Cochrane Collaboration and other experts. 1 2 However, the methodology for both the assessment of quality and its incorporation into systematic reviews and meta-analysis are a matter of ongoing debate.3-5 In this article we discuss the concept of study quality and the methods used to assess quality.

BACKGROUND: We seek to estimate lifetime prevalence and demographic correlates of nonaffective psychosis in the US population assessed by a computer-analyzed structured interview and a senior clinician. METHODS: In the National Comorbidity Survey, a probability subsample of 5877 respondents were administered a screen for psychotic symptoms. Based on the response to this screening, detailed follow-up interviews were conducted by mental health professionals (n = 454). The initial screen and clinical reinterview were reviewed by a senior clinician. Results are presented for narrowly (schizophrenia or schizophreniform disorder) and broadly (all nonaffective psychoses) defined psychotic illness. RESULTS: One or more psychosis screening questions were endorsed by 28.4% of individuals. By computer algorithm, lifetime prevalences of narrowly and broadly defined psychotic illness were 1.3% and 2.2%, respectively. Of those assigned a narrow diagnosis by the computer, the senior clinician assigned narrow and broad diagnoses to 10% and 37%, respectively. By clinician diagnosis, lifetime prevalence rates of narrowly and broadly defined psychosis were 0.2% and 0.7%, respectively. A clinician diagnosis of nonaffective psychosis was significantly associated with low income; unemployment a marital status of single, divorced, or separated; and urban residence Clinician confirmation of a computer diagnosis was predicted by hospitalization, neuroleptic treatment, duration of illness, enduring impairment, and thought disorder. CONCLUSIONS: Lifetime prevalence estimates of psychosis in community samples are strongly influenced by methods of assessment and diagnosis. Although results using computer algorithms were similar in the National Comorbidity Survey and Epidemiologic Catchment Area studies, diagnoses so obtained agreed poorly with clinical diagnoses. Accurate assessment of psychotic illness in epidemiologic samples may require collection of extensive contextual information for clinician review.

OBJECTIVE: To provide an overview of the World Health Organization (WHO) International Consortium in Psychiatric Epidemiology (ICPE), to introduce the World Mental Health 2000 (WMH2000) Initiative and to discuss methodological issues that the ICPE is grappling with in planning the WMH2000 Initiative. METHOD: We review the history, mission and organization of the ICPE and the rationale behind the WMH2000 Initiative. We review methodological research underlying major design and implementation decisions regarding the WMH2000 surveys. RESULTS: The ICPE is an international consortium created to facilitate cross-national comparative epidemiological research using the WHO Composite International Diagnostic Interview (CIDI). The first-phase core ICPE surveys, which we are currently analysing, include over 33 000 interviews in seven countries, with an additional set of over 30 000 interviews in seven countries ready to be added to the master file within the next year. The WMH2000 Initiative will include a third series of CIDI surveys that include an anticipated 100000 additional interviews in 10 countries. A series of complex methodological challenges confront us in designing and implementing the WMH2000 surveys. These include issues in the conceptualization and measurement of impairment and disablement, the implementation of standardized quality control procedures across countries, and the blending of epidemiological and clinical interviewing methods to obtain a valid cross-national characterization of disorder prevalences. Our current plans regarding these issues are discussed. CONCLUSION: Valid and representative general population epidemiological data on patterns, predictors and adverse consequences of psychiatric disorders are needed as a foundation for public health initiatives. The efforts of the ICPE promise to provide data of this sort for many regions in the world. Formidable methodological and logistical challenges arise in implementing this agenda, but we are confident that these challenges can be met by building on the firm foundation already established in the ongoing ICPE collaboration.

OBJECTIVE: The authors used self-report data to study patterns and predictors of treatment contact after the first onset of DSM-III-R mood, anxiety, and addictive disorders. METHOD: Data from the National Comorbidity Survey, a general population survey of 8,098 respondents, were used. Disorders were assessed by using a modified version of the Composite International Diagnostic Interview. Age at onset and age at first treatment contact were assessed retrospectively. RESULTS: There was great variation across disorders in lifetime probability of treatment contact. Most treatment contact was delayed; the median delay time was between 6 and 14 years across the disorders considered here. Probability of treatment contact was inversely related to age at onset and increased in younger cohorts. The effects of sociodemographic variables were modest and inconsistent across disorders. CONCLUSIONS: The majority of people with the disorders considered here eventually make treatment contact. However, delay was pervasive. Further research is needed on the determinants of delay and on the low probability of lifetime treatment contact among people with early-onset psychiatric disorders.

To describe causally predictive relationships, model parameters and the data used to estimate them must correspond to the social context of causal actions. Causes may act directly upon the individual, during a contact between individuals, or upon a group dynamic. Assuming that outcomes in different individuals are independent puts the causal action directly upon individuals. Analyses making this assumption are thus inappropriate for infectious diseases, for which risk factors alter the outcome of contacts between individuals. Transmission during contact generates nonlinear infection dynamics. These dynamics can so attenuate exposure-infection relationships at the individual level that even risk factors causing the vast majority of infections can be missed by individual-level analyses. On the other hand, these dynamics amplify causal associations between exposure and infection at the ecological level. The amplification and attenuation derive from chains of transmission initiated by exposed individuals but involving unexposed individuals. A study of household exposure to the only vector of dengue in Mexico illustrates the phenomenon. An individual-level analysis demonstrated almost no association between exposure and infection. Ecological analysis, in contrast, demonstrated a strong association. Transmission models that are devoid of any sources of the ecological fallacy are used to illustrate how nonlinear dynamics generate such results.

OBJECTIVE: The authors tested the hypothesis that fluoxetine, when used in combination with relapse prevention psychotherapy, directly reduces relapse frequency and severity for alcoholics. METHOD: The authors conducted a randomized, placebo-controlled trial of fluoxetine (up to a maximum of 60 mg/day) for 12 weeks in combination with weekly psychotherapy for 101 alcohol-dependent subjects who were not selected on the basis of comorbid major depression. Outcomes were measured at the end of treatment and 6 months after treatment. RESULTS: Placebo- treated subjects were more complaint with the medication regimen and remained in the study longer than fluoxetine-treated subjects. There was significantly less alcohol consumption in both groups during treatment than before treatment. These effects persisted during the posttreatment period. Although fluoxetine treatment had no significant effects on alcohol consumption, it reduced Hamilton Depression Rating Scale scores more than placebo treatment among subjects with current major depression. CONCLUSIONS: Fluoxetine at a dose of 60 mg/day is probably not of use for relapse prevention in alcoholics with mild to moderate alcohol dependence and no comorbid depression. In alcoholics with major depression, the drug may reduce depressive symptoms. Subsequent studies with fluoxetine should probably focus on more severely alcohol-dependent subjects or those with comorbid depression.

OBJECTIVE: To summarise comparisons of randomised clinical trials and non-randomised clinical trials, trials with adequately concealed random allocation versus inadequately concealed random allocation, and high quality trials versus low quality trials where the effect of randomisation could not be separated from the effects of other methodological manoeuvres. DESIGN: Systematic review. SELECTION CRITERIA: Cohorts or meta-analyses of clinical trials that included an empirical assessment of the relation between randomisation and estimates of effect. DATA SOURCES: Cochrane Review Methodology Database, Medline, SciSearch, bibliographies, hand searching of journals, personal communication with methodologists, and the reference lists of relevant articles. MAIN OUTCOME MEASURES: Relation between randomisation and estimates of effect. RESULTS: Eleven studies that compared randomised controlled trials with non-randomised controlled trials (eight for evaluations of the same intervention and three across different interventions), two studies that compared trials with adequately concealed random allocation and inadequately concealed random allocation, and five studies that assessed the relation between quality scores and estimates of treatment effects, were identified. Failure to use random allocation and concealment of allocation were associated with relative increases in estimates of effects of 150% or more, relative decreases of up to 90%, inversion of the estimated effect and, in some cases, no difference. On average, failure to use randomisation or adequate concealment of allocation resulted in larger estimates of effect due to a poorer prognosis in non-randomly selected control groups compared with randomly selected control groups. CONCLUSIONS: Failure to use adequately concealed random allocation can distort the apparent effects of care in either direction, causing the effects to seem either larger or smaller than they really are. The size of these distortions can be as large as or larger than the size of the effects that are to be detected.

Case-control studies involve 'cases' being compared to 'controls' with respect to 'exposures', possible aetiological (or associated) factors. Associations between a disease and an exposure can be explained by chance, reverse causality, confounding and biases or, lastly, by causality. However, confounders as well as information and selection biases can be adjusted for at the design (or analysis) stage of the study. The strength of an association can be measured by means of relative risk, calculated indirectly using the odds ratio. Well conducted control studies should produce accurate estimates of relative risks in many psychiatric investigations.

Many of the standardized interviews currently used in psychiatry require the interviewer to use expert psychiatric judgements in deciding upon the presence or absence of psychopathology. However, when case definitions are standardized it is customary for clinical judgements to be replaced with rules. The Clinical Interview Schedule was therefore revised, in order to increase standardization, and to make it suitable for use by 'lay' interviewers in assessing minor psychiatric disorder in community, general hospital, occupational and primary care research. Two reliability studies of the revised Clinical Interview Schedule (CIS-R) were conducted in primary health care clinics in London and Santiago, Chile. Both studies compared psychiatrically trained interviewer(s) with lay interviewer(s). Estimates of the reliability of the CIS-R compared favourably with the results of studies of other standardized interviews. In addition, the lay interviewers were as reliable as the psychiatrists and did not show any bias in their use of the CIS-R. Confirmatory factor analysis models were also used to estimate the reliabilities of the CIS-R and self-administered questionnaires and indicated that traditional measures of reliability are probably overestimates.

In a survey of eighty-eight county districts within England and Wales, rates of Alzheimer's disease in people under the age of 70 years were estimated from the records of the computerised tomographic (CT) scanning units that served these districts. Rates were adjusted to compensate for differences in distance from the nearest CT scanning unit and for differences in the size of the population served by the units. Aluminium concentrations in water over the past 10 years were obtained from water authorities and water companies. The risk of Alzheimer's disease was 1.5 times higher in districts where the mean aluminium concentration exceeded 0.11 mg/l than in districts where concentrations were less than 0.01 mg/l. There was no evidence of a relation between other causes of dementia, or epilepsy, and aluminium concentrations in water.

In medical research data are often collected serially on subjects. The statistical analysis of such data is often inadequate in two ways: it may fail to settle clinically relevant questions and it may be statistically invalid. A commonly used method which compares groups at a series of time points, possibly with t tests, is flawed on both counts. There may, however, be a remedy, which takes the form of a two stage method that uses summary measures. In the first stage a suitable summary of the response in an individual, such as a rate of change or an area under a curve, is identified and calculated for each subject. In the second stage these summary measures are analysed by simple statistical techniques as though they were raw data. The method is statistically valid and likely to be more relevant to the study questions. If this method is borne in mind when the experiment is being planned it should promote studies with enough subjects and sufficient observations at critical times to enable useful conclusions to be drawn. Use of summary measures to analyse serial measurements, though not new, is potentially a useful and simple tool in medical research.

Background Methods of suicide and suicide rates in England and Wales have fluctuated considerably since the 1960s.

Aims To review the changes that have occurred in suicide rates in England and Wales between 1960 and 1997.

Method Suicide rates, derived from total annual suicides and the estimated annual resident population, were obtained from the Office for National Statistics.

Results Suicide rates decreased in both genders between the early 1960s and the mid-1970s. The rate for males then increased between 1975 and 1990, while the rate for females continued to fall. Between 1990 and 1997, the rate decreased for males and females in all age groups, particularly for those using motor vehicle exhaust gas; the latter finding is associated with increasing use of catalytic converters.

Conclusions Following the increase in suicide among males until 1990 there was a decrease for both genders between 1990 and 1997, consistent with the `Health of the Nation' target.

Researchers are increasingly interested in studying the effects of the social environment on health.1 The concept of social capital has been put forward as one explanation for why some communities work better than others, with benefits for the whole of the local population.2 Social capital is applied to those features of a community that promote cohesion and a sense of belonging and that enable its members to cooperate. Similarly, criminologists have argued that the level of social organisation in a neighbourhood, or the degree to which residents are able to realise common goals and exercise social control, links the social composition of a neighbourhood and rates of deviant behaviour.3 We investigated how individual's reports of social capital and social disorganisation are associated with health outcomes among men and women aged 16 to 54 from a representative cross section of British households.

Future health prospects depend increasingly on globalisation processes and on the impact of global environmental change. Economic globalisation-entailng deregulated trade and investment-is a mixed blessing for health. Economic growth and the dissemination of technologies have widely enhanced life expectancy. However, aspects of globalisation are jeopardising health by eroding social and environmental conditions, exacerbating the rich-poor gap, and disseminating consumerism. Global environmental changes reflect the growth of populations and the intensity of economic activity. These changes include altered composition of the atmosphere, land degradation, depletion of terrestrial aquifers and ocean fisheries, and loss of biodiversity. This weakening of life-supporting systems poses health risks. Contemporary public health must therefore encompass the interrelated tasks of reducing social and health inequalities and achieving health-sustaining environments.

There is evidence that the quality of controlled clinical trials affects the results that are obtained. A systematic approach to the assessment of quality is required for psychiatric research. This study set out to develop an instrument for the assessment of the quality of controlled trials of interventions for depressive and non-psychotic conditions. A pilot study led to the development of a scale containing 23 items covering a wide range of aspects of quality including objective formulation, design, presentation of results, analysis and quality of conclusions. Scoring criteria were devised and the scale was then subjected to reliability testing using a random sample of trials of treatment for depression and neurosis. The scale showed moderate inter-rater reliability and results were comparable to those obtained with shorter instruments. It was quick and easy to use. There were significant correlations between year of publication and overall quality score with later studies showing higher quality. The instrument developed here provides a systematic approach to the assessment of quality for use in critical appraisal of individual studies and meta-analysis. However, the scoring process should be used cautiously since inter-rater agreement was only moderate.

BACKGROUND: This study examined the nature of ecological associations between `religiousness' and suicide rates (1985-1994) in the 11 provinces in the Netherlands.

METHODS: indices of religiousness, obtained from a nationwide survey, were used as aggreg ate predictors of provincial suicide rates in weighted linear regressions, and as individual-level predictors of suicide acceptance in logistic regressions. Socio demographic confounding was controlled for.

RESULTS: Orthodox beliefs and religious affiliation were the best predictors of lower suicide acceptance in individuals and of lower suicide rates in provinces. The ecological association was most pronounced in the least religious parts of the country givingrise to a curvilinear ecological regression line.

CONCLUSIONS: Curvilinear ecological regression lines arise when mean levels of exposure affect individual risk above and beyond personal exposure i.e. when there is ecological effect modification. This study demonstrates that such contextual effects, respons ible for cross-level bias, apply to the association between suicide and religious ness. Variation, from context to context, of the effects of exposure to psychosocial risk or protective factors for outcomes such as suicide, has important implications for research and prevention.

Background. Negative associations between religion and suicide, in individuals and countries, may be mediated by the degree to which suicide is tolerated.

Methods. Linear regression was used to examine ecological associations between suicide tolerance, religion and suicide rates in 19 Western countries in 1989/90. Logistic regression was used to study associations between suicide tolerance and strength of religious belief in 28085 individuals in these countries. The concept of effect modifying function was used to examine whether the strength of the association between suicide tolerance and religious belief in individuals depended on the extent of religious belief in their country.

Results. Higher female suicide rates were associated with lower aggregate levels of religious belief and, less strongly, religious attendance. These associations were mostly attributable to the association between higher tolerance of suicide and higher suicide rates. In the 28085 subjects suicide tolerance and the strength of religious belief were negatively associated even after adjustment for other religious and sociodemographic variables and general tolerance levels (odds ratios: men 0·74 (95 Conclusions. Ecological associations between religious variables and suicide rates are stronger for women than men, stronger for measures of belief than observance and mediated by tolerance of suicide. In individuals, stronger religious beliefs are associated with lower tolerance of suicide. Personal religious beliefs and, for men, exposure to a religious environment, may protect against suicide by reducing its acceptability.

Les problèmes de brutalité entre enfants et jeunes à l'école sont particulièrement préoccupants dans les pays Scandinaves et, de plus en plus, dans d'autres pays également. Des larges enquêtes réalisées par l'auteur, il ressort que du premier au neuvième grade, près de 9% des élèves sont très régulièrement victimes de brutalités et que 6-7% des élèves en agressent d' autres régulièrement. En démocratie, c'est un droit fondamental des enfants que d'être protégés contre l'oppression et les humiliations répétées impliquées par ces pratiques. L'auteur a mis au point un programme d'intervention scolaire contre les brutalités, dont les effets ont été évalués dans 42 écoles pendant deux ans. Les résultats montrent que la fréquence des problèmes de violence a diminué de 50-70%. En outre, l'importance des comportements antisociaux en général, tels que le vandalisme, le vol, l'alcoolisme et les absences non autorisées, a notablement diminué. Les principaux contenus du programme et ses principes-clés sont présentés. L'objectif primordial du programme peut être décrit comme une restructuration de l'environnement social. Le programme met l'accent sur les comportements et les attitudes caractérisés par l'association d'engagements positifs des enseignants et des parents, des limites strictes aux comportements inacceptables (nous n'acceptons pas la brutalité dans notre classe/école), et l'utilisation cohérente de sanctions elles-mêmes non brutales en cas de violation des règles. Les résultats positifs obtenus sont mis sur le compte de changements structuraux relatifs aux occasions et aux bénéfices des comportements de brutalité.

A travers le modèle de la dépression, cette étude montre la médiocre validité des estimations de la fréquence des troubles mentaux dans la vie d'un individu, selon les rapports «Epidemiologic Catchment Area» du NIMH

OBJECTIVES. An association of smoking with depression and anxiety has been documented in adult smokers. This study examines this association in a representative group of teenage smokers. METHODS. A two-stage cluster sample of secondary school students in Victoria, Australia, were surveyed by using a computerized questionnaire, which included a 7-day retrospective diary for tobacco use and a structured psychiatric interview. RESULTS. Subjects reporting high levels of depression and anxiety were twice as likely to be smokers after the potential confounders of year level, sex, alcohol use, and parental smoking were controlled for. Regular smokers were almost twice as likely as occasional smokers to report high levels of depression and anxiety. In a stratified analysis, an association between regular smoking and psychiatric morbidity was found in girls of all ages but for boys only in the youngest group. CONCLUSIONS. The cross-sectional association is consistent with the use of smoking by teenage girls as self-medication for depression and anxiety. Therefore, future health promotional campaigns might consider strategies that attend to perceived psychological benefits of smoking.

Most randomised clinical trials require periodic monitoring of the accumulating data. While the efficiency of trial management is enhanced by data monitoring, ethical reasons should primarily dictate the need to terminate or change a trial in response to interim findings. This article focuses on the ethical dilemma of when to stop a clinical trial and places statistical stopping rules in the context of such ethical decision making. Other issues include the organisation of data monitoring committees and the problems of premature publication and exaggerated estimation in trials that stop early. Several topical examples are used to convey the relevance of these issues to current practice.

This study aims to describe factors associated with cognitive decline among 2584 subjects, aged 65-74, who were followed up for 54 months in the Medical Research Council Elderly Hypertension Trial (1982-1989). The subjects completed a cognitive test, the Paired Associate Learning Test (PALT), five times over this period. Decline on the PALT was associated with advanced age, male sex, rural residence, depression and low intelligence. These effects were modified by gender and level of pre-morbid intelligence. Advanced age, rural residence and number of cigarettes smoked daily were only associated with PALT decline among women of below median intelligence. The association between depression and PALT decline was only apparent in women of below median intelligence and men of above median intelligence. While these findings are consistent with other research into cognitive decline, they differ in some ways from reported risk factors for dementia, suggesting aetiological separateness. That women were more vulnerable than men to the effects of age and smoking raises the question of the impact on cognition of accelerated atherosclerosis after the menopause.

Background. Population-based studies suggest substantial co-morbidity between physical illness and depression in late-life. However, a causal relationship has not been established. If a relationship exists, it is important to establish which aspects of poor health determine risk for depression, and which factors confer vulnerability or resilience in the face of poor health. We investigate the role of disablement, measured as impairment, disability and handicap.

Methods. A prospective population-based cohort study, comprising an index assessment and 1 year follow-up, of all residents aged 65 years or over, of an electoral ward in London, UK (N=889).

Results. The prevalence of SHORT-CARE pervasive depression was 17·7 Conclusions. It seems likely that disablement, specifically handicap, is the chief cause of onsets of depression in late-life. Genetic predisposition, early adversity and serious life events may play a less prominent role than in earlier life. Effective prevention of late-life depression requires attention at the structural level to the sources of handicap within communities.

The National Institute of Mental Health multisite Epidemiologic Catchment Area (ECA) program is described in the context of four previous psychiatric epidemiologic surveys that included a combined total of 4,000 subjects from Stirling County, the Baltimore Morbidity Study, Midtown Manhattan, and the New Haven third-wave survey. The ECA program is distinguished by its sample size of at least 3,500 subjects per site (about 20,000 total); the focus on Diagnostic Interview Schedule-defined DSM-III mental disorders; the one-year reinterview-based longitudinal design to obtain incidence and service use data; the linkage of epidemiologic and health service use data; and the replication of design and method in multiple sites. Demographic characteristics of community and sample populations are provided for New Haven, Conn, Baltimore, and St Louis.

A new interview schedule allows lay interviewers or clinicians to make psychiatric diagnoses according to DSM-III criteria, Feighner criteria, and Research Diagnostic Criteria. It is being used in a set of epidemiological studies sponsored by the National Institute of Mental Health Center for Epidemiological Studies. Its accuracy has been evaluated in a test-retest design comparing independent administrations by psychiatrists and lay interviewers to 216 subjects (inpatients, outpatients, ex-patients, and nonpatients).

Aetiology confronts two distinct issues: the determinants of individual cases, and the determinants of incidence rate. If exposure to a necessary agent is homogeneous within a population, then case/control and cohort methods will fail to detect it: they will only identify markers of susceptibility. The corresponding strategies in control are the 'high-risk' approach, which seeks to protect susceptible individuals, and the population approach, which seeks to control the causes of incidence. The two approaches are not usually in competition, but the prior concern should always be to discover and control the causes of incidence.

To compare the use of randomized controls (RCTs) and historical controls (HCTs) for clinical trials, we searched the literature for therapies studied by both methods. We found six therapies for which 50 RCTs and 56 HCTs were reported. Forty-four of 56 HCTs (79 percent) found the therapy better than the control regimen, but only 10 of 50 RCTs (20 percent) agreed. For each therapy, the treated patients in RCTs and HCTs of the same therapy was largely due to differences in outcome for the control groups, with HCT control patients generally doing worse than the RCT control groups. Adjustment of the outcomes of the HCTs for prognostic factors, when possible, did not appreciably change the results. The data suggest that biases in patient selection may irretrievably weight the outcome of HCts in favor of new therapies. RCTs may miss clinically important benefits because of inadequate attention to sample size. The predictive value of each might be improved by reconsidering the use of p less than 0.05 as the significance level for all types of clinical trials, and by the use of confidence intervals around estimates of treatment effects.

OBJECTIVE: Psychiatric disability has been defined largely from measures that focus on serious mental illness. This practice may have led to substantial underestimation of the total impact of mental disorders on community health. In this study a generic measure of mental health-related disability was used to examine disabilities attributable to various common mental disorders. METHODS: Data were drawn from the Australian National Survey of Mental Health and Wellbeing, a household survey of 10,641 adults that assessed participants for 14 DSM-IV disorders with use of the Composite International Diagnostic Interview. Screening instruments were used to identify likely cases of ICD-10 personality disorder, neurasthenia (an undifferentiated somatoform disorder), and psychosis. Mental health disability was assessed with the Medical Outcomes Study 12-item Short Form (SF-12) mental health summary scale, which was administered to all participants. RESULTS: Disability was significantly greater among participants with a current psychiatric diagnosis, and disability varied by type of disorder. Diagnosis remained a strong predictor of disability after sociodemographic factors and physical illness were controlled for. Disorders found to be independently associated with disability were depression, panic disorder, agoraphobia, social phobia, generalized anxiety disorder, alcohol dependence, and drug dependence. CONCLUSIONS: Substantial proportions of persons with mental disorders that are not usually classified as major mental disorders reported moderate and severe disability. A generic measure of mental health-related disability was able to detect variations in disability among persons with different diagnoses. Although such a measure is not as sensitive as a disorder-specific measure developed for use in psychiatric populations, it can facilitate comparison of disability across common mental disorders.

Prevalence rates for psychiatric disorders in the elderly are presented from the initial cross-sectional stage of a longitudinal community study of the incidence of dementia in the city of Liverpool. Together with five other centres in the UK the MRC-ALPHA project forms part of the MRC multicentre incidence study of dementia and cognitive decline. An age- and sex-stratified random sample of 5222 subjects aged 65 was interviewed at home using the Geriatric Mental State-AGECAT package to provide computer diagnoses. The overall age-standardized prevalence rates for organic disorder (4.7%) depressive illness (10.0%) and the neuroses (2.5%) are consistent with levels found in previous smaller studies that have used MS-AGECAT. Each of these diagnoses is more common in females than males. A rise in organic disorder with age is confirmed as continuing into the oldest age groups for both sexes. An apparent decline with age observed for depression and neurosis diagnoses disappears when organic cases are excluded from the analysis.

Recent empirical evidence supports the importance of adequate randomization in controlled trials. Trials with inadequate allocation concealment have been associated with larger treatment effects compared with trials in which authors reported adequate allocation concealment. While that provides empirical evidence of bias being interjected into trials, trial investigators rarely document the sensitive details of subverting the intended purpose of randomization. This article relates anonymous accounts of deciphering assignment sequences before allocation based on experiences acquired from epidemiologic workshops for physicians. These accounts run the gamut from simple to intricate operations, from transillumination of envelopes to searching for code in the office files of the principal investigator. They indicate that deciphering is something more frequent than a rare occurrence. These accounts prompt some methodological recommendations to help prevent deciphering. Randomized controlled trials appear to annoy human nature-if properly conducted, indeed they should.

OBJECTIVE-To determine if inadequate approaches to randomized controlled trial design and execution are associated with evidence of bias in estimating treatment effects. DESIGN-An observational study in which we assessed the methodological quality of 250 controlled trials from 33 meta-analyses and then analyzed, using multiple logistic regression models, the associations between those assessments and estimated treatment effects. DATA SOURCES-Meta-analyses from the Cochrane Pregnancy and Childbirth Database. MAIN OUTCOME MEASURES-The associations between estimates of treatment effects and inadequate allocation concealment, exclusions after randomization, and lack of double-blinding. RESULTS-Compared with trials in which authors reported adequately concealed treatment allocation, trials in which concealment was either inadequate or unclear (did not report or incompletely reported a concealment approach) yielded larger estimates of treatment effects (P < .001). Odds ratios were exaggerated by 41% for inadequately concealed trials and by 30% for unclearly concealed trials (adjusted for other aspects of quality). Trials in which participants had been excluded after randomization did not yield larger estimates of effects, but that lack of association may be due to incomplete reporting. Trials that were not double-blind also yielded larger estimates of effects (P = .01), with odds ratios being exaggerated by 17%. CONCLUSIONS-This study provides empirical evidence that inadequate methodological approaches in controlled trials, particularly those representing poor allocation concealment, are associated with bias. Readers of trial reports should be wary of these pitfalls, and investigators must improve their design, execution, and reporting of trials.

Studies that show a significant effect of treatment are more likely to be published, be published in English, be cited by other authors, and produce multiple publications than other studies.1-8 Such studies are therefore also more likely to be identified and included in systematic reviews, which may introduce bias.9 Low methodological quality of studies included in a systematic review is another important source of bias.10

All these biases are more likely to affect small studies than large ones. The smaller a study the larger the treatment effect necessary for the results to be significant. The greater investment of time and money in larger studies means that they are more likely to be of high methodological quality and published even if their results are negative. Bias in a systematic review may therefore become evident through an association between the size of the treatment effect and study sizesuch associations may be examined both graphically and statistically.

Objective Because of the pressure for timely, informed decisions in public health and clinical practice and the explosion of information in the scientific literature, research results must be synthesized. Meta-analyses are increasingly used to address this problem, and they often evaluate observational studies. A workshop was held in Atlanta, Ga, in April 1997, to examine the reporting of meta-analyses of observational studies and to make recommendations to aid authors, reviewers, editors, and readers.

Participants Twenty-seven participants were selected by a steering committee, based on expertise in clinical practice, trials, statistics, epidemiology, social sciences, and biomedical editing. Deliberations of the workshop were open to other interested scientists. Funding for this activity was provided by the Centers for Disease Control and Prevention.

Evidence We conducted a systematic review of the published literature on the conduct and reporting of meta-analyses in observational studies using MEDLINE, Educational Research Information Center (ERIC), PsycLIT, and the Current Index to Statistics. We also examined reference lists of the 32 studies retrieved and contacted experts in the field. Participants were assigned to small-group discussions on the subjects of bias, searching and abstracting, heterogeneity, study categorization, and statistical methods.

Consensus Process From the material presented at the workshop, the authors developed a checklist summarizing recommendations for reporting meta-analyses of observational studies. The checklist and supporting evidence were circulated to all conference attendees and additional experts. All suggestions for revisions were addressed.

Conclusions The proposed checklist contains specifications for reporting of meta-analyses of observational studies in epidemiology, including background, search strategy, methods, results, discussion, and conclusion. Use of the checklist should improve the usefulness of meta-analyses for authors, reviewers, editors, readers, and decision makers. An evaluation plan is suggested and research areas are explored.

BACKGROUND: Suggestive findings of an earlier study that prenatal nutritional deficiency was a determinant of schizophrenia prompted us to undertake a second test of the hypothesis using more precise data on both exposure and outcome. METHODS: Among persons born in the cities of western Netherlands during 1944 through 1946, we compared the risk for schizophrenia in those exposed and unexposed during early gestation to the Dutch Hunger Winter of 1944/1945. The frequency of hospitalized patients with schizophrenia at age 24 to 48 years in the exposed and unexposed birth cohorts was ascertained from a national psychiatric registry. RESULTS: The most exposed birth cohort, conceived at the height of the famine, showed a twofold and statistically significant increase in the risk for schizophrenia (relative risk [RR] = 2.0; 95% confidence interval [CI] = 1.2 to 3.4; P < .01) in both men (RR = 1.9; 95% CI = 1.0 to 3.7; P = .05) and women (RR = 2.2; 95% CI = 1.0 to 4.7; P = .04). Among all birth cohorts of 1944 through 1946, the risk for schizophrenia clearly peaked in this exposed cohort. CONCLUSION: Prenatal nutritional deficiency may play a role in the origin of some cases of schizophrenia.

In many randomised trials researchers measure a continuous variable at baseline and again as an outcome assessed at follow up. Baseline measurements are common in trials of chronic conditions where researchers want to see whether a treatment can reduce pre-existing levels of pain, anxiety, hypertension, and the like.

Statistical comparisons in such trials can be made in several ways. Comparison of follow up (post-treatment) scores will give a result such as "at the end of the trial, mean pain scores were 15 mm (95% confidence interval 10 to 20 mm) lower in the treatment group." Alternatively a change score can be calculated by subtracting the follow up score from the baseline score, leading to a statement such as "pain reductions were 20 mm (16 to 24 mm) greater on treatment than control." If the average baseline scores are the same in each group the estimated treatment effect will be the same using these two simple approaches. If the treatment is effective the statistical significance of the treatment effect by the two methods will depend on the correlation between baseline and follow up scores. If the correlation is low using the change score will add variation and the follow up score is more likely to show a significant result. Conversely, if the correlation is high using only the follow up score will lose information and the change score is more likely to be significant. It is incorrect, however, to choose whichever analysis gives a more significant finding. The method of analysis should be specified in the trial protocol.

Objective: To determine whether poverty and unemployment increase the likelihood of or delay recovery from common mental disorders, and whether these associations could be explained by subjective financial strain. Design: Prospective cohort study. Setting: England, Wales, and Scotland. Subjects: 7726 adults aged 16-75 living in private households. Main outcome measures: Common mental disorders were assessed using the general health questionnaire, a self assessed measure of psychiatric morbidity. Results: Poverty and unemployment (odds ratio 1.86, 95Conclusions: Poverty and unemployment increased the duration of episodes of common mental disorders but not the likelihood of their onset. Financial strain was a better predictor of future psychiatric morbidity than either of these more objective risk factors though the nature of this risk factor and its relation with poverty and unemployment remain unclear.

The advent of more explicit diagnostic criteria and the growing interest in "lifetime" rates of mental disorders has made imperative an accurate determination of time-related diagnostic criteria. We used data from two independent test-retest studies of the Diagnostic Interview Schedule (DIS) and the Composite International Diagnostic Interview (CIDI) to study the reliability of different time-related questions in these fully standardized diagnostic interviews. With two exceptions (anxiety disorders and alcohol-related questions), the test-retest reliability of most time-related questions in both interviews was judged to be satisfactorily high. Furthermore, the validity of time-related questions in the DIS (age at symptom onset, duration and frequency of illness episodes) was examined by comparing them with detailed "consensus" ratings done independently by different clinicians for 207 former psychiatric inpatients. A surprisingly high concordance was found for former psychotic patients except for those still severely disturbed at the follow-up investigation. Some severe restrictions were also found for nonpsychotic disorders with regard to judgment of the age at onset of phobias, panic attacks, and depression. For a more valid assessment of time-related symptom information, the use of specific memory aids is suggested.

The impact of cigarette smoking on stroke incidence was assessed in the Framingham Heart Study cohort of 4255 men and women who were aged 36 to 68 years and free of stroke and transient ischemic attacks. During 26 years of follow-up, 459 strokes occurred. Regardless of smoking status and in each sex, hypertensive subjects had twice the incidence of stroke. Using the Cox proportional hazard regression method, smoking was significantly related to stroke after age and hypertension were taken into account. Even after pertinent cardiovascular disease risk factors were added to the Cox model, cigarette smoking continued to make a significant independent contribution to the risk of stroke generally and brain infarction specifically. The risk of stroke increased as the number of cigarettes smoked increased. The relative risk of stroke in heavy smokers (greater than 40 cigarettes per day) was twice that of light smokers (fewer than ten cigarettes per day). Lapsed smokers developed stroke at the same level as nonsmokers soon after stopping. Stroke risk decreased significantly by two years and was at the level of nonsmokers by five years after cessation of cigarette smoking.