Search

There have been many attempts to find predictors of the therapeutic response to the clomipramine treatment of obsessive-compulsive disorder. The majority of studies have failed to identify such predictors. Possible reasons for this failure include the small sample size of most studies, samples homogeneous with respect to the study factors of interest, and the use of statistical procedures that are insensitive to individual differences or that inadequately control for confounding. We have reanalyzed data from Ciba-Geigy's large, multicenter clinical trial of clomipramine for obsessive-compulsive disorder, using stratification and regression techniques to identify multiple prognostic factors and control for confounders. We assessed the relationship between therapeutic response and baseline measures such as severity of symptoms, type of symptoms (obsessions, compulsions, depression), length of illness, age of onset, and other demographic factors (age, race, and sex). We found age of onset to be a strong predictor of response to clomipramine: people who develop obsessive-compulsive disorder later in life have a better chance of responding than do those who become ill earlier, independent of length of illness. We also found that baseline depression is associated with response, but the association appears to be nonlinear.

Objective. —To ascertain whether depressive symptoms as determined by the Center for Epidemiologic Studies—Depression scale (CES-D) predict accelerated mortality and worse medical outcomes in patients infected with human immunodeficiency virus (HIV). Design. —Eight-year cohort study with semiannual follow-up. Setting. —Community volunteers. Participants. —A total of 1809 HIV-seropositive homosexual men without the acquired immunodeficiency syndrome (AIDS) who entered the Multicenter AIDS Cohort Study in 1984 or 1985. Eight-year follow-up data were available on 75% of eligible participants. Outcome Measures. —Times to AIDS, death, and prophylactic treatment, and slopes describing the decline in CD4 count for each individual participant. Results. —Using a conventional definition of depression (CES-D ≥16 at the first study visit), 21.3% of participants were classified as depressed. Depressed participants had lower CD4 counts and reported more AIDS-related symptoms. There were no significant differences between depressed and nondepressed participants on any of the outcome measures ( P >.05 in all cases). In contrast, men reporting AIDS-related symptoms had shorter times to AIDS and to death even after adjusting for CD4 counts ( P Conclusions. —We find no evidence that depressive symptoms independently prognosticate worse outcomes in HIV infection. Because of associations of depression with symptom reports, CD4 counts, and indicators of socioeconomic status, future studies of the relationship between depression and HIV outcome should consider these variables as confounders. ( JAMA . 1993;270:2563-2567)

The present study reports new and unexpected results of cognitive abnormalities among human immunodeficiency virus type 1 (HIV-1) asymptomatic subjects in the Multicenter AIDS Cohort Study. The major purpose of our analyses is to estimate the separate and combined effects of serostatus and education level on the prevalence of cognitive abnormality. Cognitive "abnormality" was defined as performance that deviated > or = 2 SDs below the mean of the total seronegative group on at least one of the five neuropsychological screening tests (Grooved Pegboard, Verbal Fluency, Digit Span, Symbol Digit Modalities, Rey Auditory Verbal Learning). The predicted prevalence of cognitive abnormality was 38% in seropositive individuals with no more than 12 years of education, compared with < 17% in the other education-serostatus groups. This interaction between education level and serostatus remained after controlling for the possible confounding effects of age, ethnicity, CD4 level, depression, prior drug history, and learning disability using logistic regression. To account for these findings, we suggest that low education might reflect an indirect index of lower reserve capacity (i.e., a risk factor) that lowers the threshold for neuropsychological abnormalities in cases of early HIV-1 infection.

Objective. —To examine the determinants of health insurance coverage for Latinos in the United States and how different targeted strategies for health care reform differentially affect the country's major ethnic groups, focusing on the implications for the Latino population. Design. —Data from the 1980 and 1990 Current Population Surveys were used to compare the insurance status of nonelderly ( Main Outcome Measures. —Percentage uninsured, percentage insured by Medicaid, and attributable fraction for covariates. Results. —Latinos have the worst health insurance coverage of any ethnic group in the country. Approximately 39% of Latinos are uninsured compared with 13.8% for the Anglo and 24% for the black population. Providing coverage to all the poor could reduce the uninsured rate for Anglos by about 23%, whereas the reduction among Latinos could be about 37% and among blacks about 42%. Similar reductions could be achieved by covering all workers and their minor dependents. Regardless of the approach to reform, however, Latinos would remain with high absolute rates of uninsured. Conclusions. —Differences in Medicaid eligibility, labor force characteristics, and family composition between Latinos and other ethnic groups suggest that policy initiatives may affect Latinos differently. Targeted strategies, such as employer mandates, "pay-or-play" programs, or Medicaid expansions, can improve coverage, but many Latinos could still remain uninsured. (JAMA. 1993;269:889-894)

Despite the widespread use of ecologic analysis in epidemiologic research and health planning, little attention has been given by health scientists and practitioners to the methodological aspects of this approach. This paper reviews the major types of ecologic study designs, the analytic methods appropriate for each, the limitations of ecologic data for making causal inferences and what can be done to minimize these problems, and the relative advantages of ecologic analysis. Numerous examples are provided to illustrate the important principles and methods. A careful distinction is made between ecologic studies that generate or test etiologic hypotheses and those that evaluate the impact of intervention programs or policies (given adequate knowledge of disease etiology). Failure to recognize this difference in the conduct of ecologic studies can lead to results that are not very informative or that are misinterpreted by others.

Despite considerable research into the statistical properties of matching in case-control studies, much disagreement remains regarding practical guidelines for the employment of the procedure. We critically examine some recent statistical recommendations, and review a number of theoretical and practical considerations that should be employed in designing matched studies. We conclude that determination of matching criteria and stratum sizes should always involve practical as well as theoretical considerations, and that global recommendations regarding matching cannot be justified on the basis of statistical considerations alone.

The authors describe a scale designed to measure five dimensions of delusional experience: conviction, extension, bizarreness, disorganization, and pressure. Reliability was adequate to excellent on four of the dimensions, but only fair on the dimension of bizarreness. In 52 delusional patients, no two dimensions correlated highly with each other, indicating that the dimensions were not redundant. Factor analysis identified two factors from the five dimensions--delusional involvement and delusional construct. On the basis of these results the authors suggest that delusions are a multidimensional phenomenon; the results have implications for the measurement of delusions in clinical research and for the understanding of the structure of psychotic experience.

The problems of who uses medical services and why, are important ones for investigators interested in studying chronic diseases, particularly if they wish to avoid systematic error when assembling a study population. These issues are important when studying pediatric diseases due to the uneven use of medical services by children and the tendency of lower socioeconomic groups to use hospital facilities rather than private practitioners. In order to address these problems, we must understand why families seek medical care for young children. Utilization research shows that a number of descriptive factors such as child's age, birth order, parental education, financial resources and perceived symptoms are related to service use. Additionally, psychosocial variables, such as distress, also predict utilization. Neither the descriptive nor psychosocial variables explain much of the variance in utilization. While some of this lack of explanatory power can be attributed to problems in measurement or study design, there are also conceptual and methodological issues that are not addressed in pediatric utilization research. This paper discusses four of these issues. It is our conclusion that two of these problems, the need for a new utilization taxonomy and the ambiguity of cause and effect, can be remedied. While more difficult to address, the inadequate conceptualization of social stress, psychological distress, and social support can be improved. However, measuring health status independently of utilization represents a major methodologic problem for which we currently have no ideal solution.

Consideration of factors involved in the selection of subjects is essential for evaluating the validity of a putative etiologic association. The purpose of this paper is to provide a quantitative conceptual framework for understanding selection bias; this framework integrates both epidemiologic and statistical considerations. Emphasis is given to specifying the conditions under which such bias is likely to occur, identifying the direction and magnitude of the bias, and illustrating how these features differ by type of study design.

A hospital-based case-control study was conducted between 1977 and 1983 of 189 men and 78 women with renal cell carcinoma at 18 hospital centers in six US cities and an equal number of controls matched on the basis of hospital, sex, race, age (+/- 5 years), and time of admission (+/- 12 months). Quetelet index (weight/height2) was found to be associated with renal cell cancer in both men and women. Crude odds ratios (ORs) for individuals in the upper tertile of the Quetelet index distribution (28 or more) were calculated to be 2.7 (95% confidence interval (CI) = 1.5-5.9) among men and 2.4 (95% CI = 1.2-6.8) among women compared to the lowest tertile. This relationship was found to be independent of the effects of cigarette smoking, chewing tobacco use, and decaffeinated coffee consumption. The use of chewing tobacco among men was positively associated with disease (OR = 4.0; 95% CI = 1.2-12.9). This was due primarily to a significant interaction between chewing tobacco use and cumulative cigarette smoking. The fitted odds ratio for persons who chewed tobacco and had 30 pack-years of cigarette smoke exposure, compared to never users of either tobacco product, was 26.0 (95% CI = 4.4-153.0) after adjusting for Quetelet index and decaffeinated coffee use. The odds ratio for decaffeinated coffee consumption among men and women combined was 1.9 (95% CI = 1.0-3.8) after controlling for other risk factors. A consistent dose-response was not found, however, by number of cups of decaffeinated coffee consumed per day. Those drinking 1-2 cups had an odds ratio of 2.0 while those drinking 3 cups or more per day had an odds ratio of 1.3, thus casting doubt on a causal interpretation of the finding. Alcohol drinkers had a lower, although not significantly different, rate of renal cell cancer than did never drinkers when the data were examined separately by sex. However, combining the data for both sexes yielded a crude odds ratio of 0.6 (95% CI = 0.4-1.0) for this association. There was no independent effect of tobacco smoking on the odds ratio for renal cell cancer. No significant differences between cases and controls were found for either the amount or duration of artificial sweetener use or the lifetime consumption of saccharin. Beverages such as caffeinated coffee, soft drinks, and tea, in addition to physical activity and occupation, were unrelated to the occurrence of disease.

In their reply, Kramer and Boivin state “The terms cohort, case-control, cross-sectional, prospective, and retrospective are used regularly by epidemiologic and clinical investigators to describe the design of their studies. It is clear that it is the concept of directionality that is essential to the distinctions indicated by this panoply of terms.” [5, p. 7181. Stating that “It is clear” is neither explanatory nor clarifying. In particular, Kramer and Boivin’s claim is not clear to us or, apparently, Dr Miettinen [4]. We and many others [e.g. 6101 regularly use all the terms they list without ever having recourse to the concept of “directionality.” Case-control studies are distinguished from cohort studies by their use of outcome-selective sampling. Crosssectional studies are distinguished from cohort studies by the data studied: the former collect and examine cross-sectional data (which allows only direct measurement of disease prevalence), whereas the latter collect and examine data taken over a span of cohort experience (which allows direct measurement of incidence as well). Prospective studies are based on ascertainment of events (exposures and outcomes) at the time they actually occur; retrospective studies are based on purely historical determinations; thus this distinction is one of timing. Since we and other [6-lo] can distinguish all the terms given by Kramer and Boivin [2, 51 without using the directionality concept, we ask them to please clarify just how directionality is “essential” to the distinctions among those terms. Kramer and Boivin also claim that “directionality remains a key concept for understanding sources of bias and the inferential problems arising therefrom” [ 1, p. 7 181. Unfortunately, they give no example in which directionality is in any way essential for understanding a source of bias or an inferential problem. We thus ask them to give an example in which, after considering sample selection, timing, and order of measurement, one needs directionality to understand a source of bias. Such an example may help clarify not only how directionality is essential (if it is) but also what it means.

Substantial differences in the use of pediatric medical resources reinforce the need for identifying and understanding factors that influence the use of medical services for children. This research assesses the simultaneous impact of sociodemographic characteristics, health attitudes and beliefs, psychologic distress, social stressors, and social networks on the use of pediatric acute care services during a 12-month period. Using a prospective longitudinal study design, data were obtained on 513 children and their families enrolled in a prepaid group practice. Linear modeling results showed that health attitudes and social networks were important predictors of acute care utilization in addition to child's age, birth order, baseline health status, and ethnic group. The authors were able to show significant effects for network size, dispersion, and tendency to use one's network members. Individuals with large nondispersed networks are more likely to use pediatric health services, apparently due to the transmission of the networks' pro-medical care health beliefs. Also the tendency to call on network members modifies an individual's propensity to seek care for minor pediatric medical problems and can make a difference by as much as 1.6 visits per year per child for acute care episodes.

The extension of case-control methods to the study of common outcomes has led to the development of several design and analysis techniques which do not employ the rare-disease assumption. Unfortunately, the principles underlying valid application of these techniques are more subtle than those first considered by Cornfield in the rare-disease setting, and appear to be easily misunderstood. We especially wish to caution that: The unrestricted inclusion of prevalent cases in the control group (as described by Hogue et al. for estimation of the risk ratio) will not make the odds ratio an unbiased estimate of the risk ratio (or anything else). In their response to our article, following, Hogue et al. describe restrictions on prevalence and duration necessary for the odds ratio from a case-exposure design to unbiasedly estimate the risk ratio in a stable population; these conditions were not mentioned in their original article, and in their new paper Hogue et al. do not provide mathematical proof that the conditions are sufficient to guarantee unbiasedness. Exclusion ("decontamination") of incident cases from the control group (as recommended by Hogue et al. for testing and test-based interval estimation) will result in improperly narrow risk-ratio confidence intervals whether or not the population is stable, and, in unstable populations, will generally lead to an invalid test. Methods that replace the rare-disease assumption with the stable-population assumption (such as case-exposure designs applied to open populations) will not yield unbiased results when the source population is a fixed cohort. (Of course, this will not be an issue for methods that are not based on either assumption, such as the case-base design applied to fixed cohorts, and the matched density design.) As each case-control design has certain practical implications for selection and interviewing, in choosing a design one should carefully consider practical issues (such as vulnerability to recall bias and ease of control selection) in addition to the statistical issues discussed here. In general, however, one should be wary of methods of studying incidence that involve the use of prevalent cases (such as the approach of Hogue et al.): prevalence is influenced by factors related to treatment, recovery, and fatality, and thus any etiologic study employing prevalent cases may be biased by such factors.(ABSTRACT TRUNCATED AT 400 WORDS)

Ecological bias is sometimes attributed to confounding by the group variable (ie the variable used to define the ecological groups), or to risk factors associated with the group variable. We show that the group variable need not be a confounder (in the strict epidemiological sense) for ecological bias to occur: effect modification can lead to profound ecological bias, whether or not the group variable or the effect modifier are independent risk factors. Furthermore, an extraneous risk factor need not be associated with the study variable at the individual level in order to produce ecological bias. Thus the conditions for the production of ecological bias by a covariate are much broader than the conditions for the production of individual-level confounding by a covariate. We also show that standardization or ecological control of variables responsible for ecological bias are generally insufficient to remove such bias.

This paper distinguishes between 2 concepts for measuring the incidence of disease: risk and rate. Alternative procedures for estimating these measures from epidemiologic data are reviewed and illustrated. An attempt is made to integrate statistical principles with epidemiologic methods while minimizing the use of higher mathematics. Several theoretical and practical criteria are discussed for choosing the appropriate incidence measure in the planning of a study and for selecting the best method of estimation in the analysis.

A general method is proposed for estimating the potential impact of a prevention program involving risk factor modification on the incidence of specific diseases in a target population. An evaluative framework for comparing alternative intervention strategies is also presented. On the basis of results from epidemiologic studies, the user must specify certain parameters regarding the distribution of the risk factor that is to be modified in the population, the magnitude of the association between the risk factor and disease, and the total risk of disease in the population. A quantitative measure, called the potential impact fraction, is derived to estimate the proportion of expected new cases that may be prevented under intervention programs of varying success. Estimates of this measure are then used to assess the potential efficacy, effectiveness, adequacy, and efficiency of planned intervention strategies. The method is illustrated with published data relating relative weight and coronary heart disease among middle-aged U.S. men, comparing different strategies of weight reduction. Key assumptions of the method and interpretation of results are discussed.

A cross-sectional study was conducted to identify predictors of tardive dyskinesia (TD) in a group of 180 psychiatric outpatients maintained on neuroleptic medications. The estimated prevalence of this involuntary movement disorder was 33% in the total study population. Using multiple logistic regression, we found that TD was independently related to five factors: being 55 yr of age and older; being male; using depot (injectable) neuroleptics; having 6 or more years of neuroleptic exposure; and having less than 6 months of psychiatric hospitalization. In addition, the effect of depot medication was much greater in white males than it was in other race-sex groups. We observed no other interaction effects between pairs of predictor variables, nor did we find significant independent effects of race, denture use, DSM III diagnosis, current neuroleptic dose and potency, percent time on neuroleptics, and recent use of antiparkinsonian drugs or lithium. This study is serving as a pilot investigation for a large prospective incidence study that has already begun among patients at risk of developing TD in the same source population.

To determine whether tardive dyskinesia (TD) is a single abnormal movement syndrome or multiple syndromes involving different anatomical areas, we examined 228 out-patients diagnosed with TD at the Connecticut Mental Health Center in New Haven. Application of factor analysis to the seven anatomical severity scores of the Abnormal Involuntary Movement Scale yielded three statistically independent factors involving abnormal movements primarily of the jaw-tongue, face-lips, and extremities-trunk. Using logistic regression to predict the severity of these factors, we found that the severity of the orofacial scores was positively associated with age, schizoaffective or affective disorder, and living alone, while severity of non-orofacial movement was positively associated with current neuroleptic dose, non-use of psychiatric medication, and living alone. Our findings suggest that orofacial and non-orofacial dyskinetic movements may involve distinct clinical syndromes of TD, each having a different set of prognostic and, possibly, aetiological determinants.

A statistically coherent view of confounding motivated by the over controversy over the proper control of confounding in the presence of prior knowledge is presented. Confounding by a covariate C in the presence of data on C is distinguished from confounding in the absence of data on C. A covariate C is defined to be a nonconfounder in the absence of data on C if the population parameter of interest can be unbiasedly estimated (asymptotically) absent data on C. Under this definition, C may be a confounder for some parameters of interest and a nonconfounder for others. If C is a confounder for a parameter of interest that has a causal interpretation, we call C a causal confounder. When data on C are available, C is defined to be a nonconfounder for a particular parameter of interest if and only if inference on the parameter of interest does not depend on the data through C. Bayesians, frequentists and pure likelihoodists will in general agree on the prior knowledge necessary to render C a non-confounder. In particular C will in general be a nonconfounder precisely when the crude data ignoring C are S-sufficient for the parameter of interest. The intuitive view held by many practicing epidemiologists that confounding by C represents a bias of the unadjusted crude estimator is in a sense correct provided inference is performed conditional on approximate ancillary statistics that measure the degree to which associations in the data differ due to sampling variability from those population associations known a priori.

Background: The International Lung Cancer Consortium (ILCCO) was established in 2004, based on the collaboration of research groups leading large molecular epidemiology studies of lung cancer that are ongoing or have been recently completed. This framework offered the opportunity to investigate the role of tobacco smoking in the development of bronchioloalveolar carcinoma (BAC), a rare form of lung cancer. Methods: Our pooled data comprised seven case-control studies from the United States, with detailed information on tobacco smoking and histology, which contributed 799 cases of BAC and 15,859 controls. We estimated the odds ratio of BAC for tobacco smoking, using never smokers as a referent category, after adjustment for age, sex, race, and study center. Results: The odds ratio of BAC for ever smoking was 2.47 (95% confidence interval [CI] 2.08, 2.93); the risk increased linearly with duration, amount, and cumulative cigarette smoking and persisted long after smoking cessation. The proportion of BAC cases attributable to smoking was 0.47 (95% CI 0.39, 0.54). Conclusions: This analysis provides a precise estimate of the risk of BAC for tobacco smoking. © 2010 The Author(s).

An epidemic of proteinuria in Pima Indians with Type 2 diabetes mellitus . Background The risk of proteinuria in Type 1 diabetes declined ≥30% over the past 50years, and improvements in metabolic control are believed to be largely responsible. Little is known about secular changes in the risk of proteinuria in Type 2 diabetes. Methods We examined trends in the incidence rate of proteinuria in Pima Indians ≥20years of age with diabetes diagnosed between January 1, 1955 and December 31, 1994. Results Among 1305 initially non-proteinuric diabetic subjects, 433 developed proteinuria during a median follow-up of 8.0years (range 0.8 to 30.2years). With subjects with diabetes diagnosed between 1955 and 1964 serving as the reference group, the rate of proteinuria was similar (rate ratio 1.0; 95% confidence interval, 0.79 to 1.3) in the cohort diagnosed between 1965 and 1974, 1.5 times as high (95% confidence interval, 1.1 to 2.0) in the cohort diagnosed between 1975 and 1984, and 1.9 times as high (95% confidence interval, 1.1 to 3.0) in the cohort diagnosed between 1985 and 1994, after adjusting for potential confounders in a generalized additive proportional hazards model. Between the first and last cohorts, plasma glucose concentration declined, on average, by 17% ( P = 0.0001) and the mean arterial pressure declined by 11% ( P = 0.0001). Conclusions The incidence rate of proteinuria in Pima Indians with Type 2 diabetes increased nearly twofold in the last 40years, despite improvements in plasma glucose and blood pressure. Rapidly changing environmental or behavioral factors must play an important role in the pathogenesis of diabetic renal disease in this population.

We collected data on all residents of San Diego County, California who were hospitalized for or died from a brain injury in 1981. The objectives were to assess the frequency of blood alcohol concentration (BAC) testing and the associations of BAC prevalence with the external cause of the brain injury and case outcome. We found that high BAC levels were most frequent among brain-injured subjects between the ages of 25 and 44 and among those subjects involved in motor vehicle crashes and assaults. Contrary to expectations, injury severity and hospital mortality were inversely related to BAC level, controlling for other predictors. We believe that these inverse associations might be due to differential rates of BAC testing by severity. Among brain-injured survivors with more severe injuries, however, we found that BAC level was positively associated with the prevalence of physician-diagnosed neurological impairment at discharge and with the length of hospitalization.

To quantify the impact of chronic exposure to neuroleptic medication on the occurrence of tardive dyskinesia (TD), we conducted a meta-analysis of data collected from 21 studies published between 1966 and 1985. The observed prevalence of dyskinesia was greater among exposed subjects in all 21 studies; we estimate that, on the average, the occurrence rate was 2.9 times greater in exposed persons than would be expected if they had been unexposed. We estimate that 65 per cent of exposed cases and 51 per cent of all cases in these investigations were caused by long-term neuroleptic exposure. Among adult United States residents in 1980, we estimate that there were approximately 193,000 neuroleptic-induced TD cases of which about 60 per cent occurred in outpatients. We also observed substantial heterogeneity of effect (rate ratio) across studies, however, partially explained, by changes and differences in the rate of dyskinesia, by differences in the frequency of certain effect modifiers, and by differences in diagnostic methods. Methodologic limitations of the studies and their possible effects on our results are discussed.

A few studies of chronic psychiatric inpatients have reported an inverse association between platelet monoamine oxidase (MAO) activity level and tardive dyskinesia (TD). In contrast to these earlier findings, we found no significant or consistent association between platelet MAO activity level and TD occurrence when we controlled for other TD predictors in a case-control study of 80 outpatients maintained on neuroleptic medications. We did find, however, that black subjects had significantly lower levels of MAO activity than did whites in an analysis controlling for age, sex, and neuroleptic dose.

A double-blind randomized clinical trial was conducted among 10 post-menopausal women with tardive dyskinesia (TD) to test the effect of estrogen replacement of the severity of abnormal movements and other outcome variables. After 3 weeks of treatment, the mean Abnormal Involuntary Movement Scale (AIMS) score decreased by 38% in the estrogen group and by 9% in the placebo group; the difference between groups was marginally significant (p less than 0.10). However, the small sample size and the imbalance between groups in baseline AIMS scores do not allow us to rule out the confounding effects of other prognostic factors. There were no significant differences between treatment groups for parkinsonian and psychological symptoms at any visit or for changes in these variables between visits. The findings of this preliminary trial are consistent with the results of other human and animal investigations, and they support the need for future research to understand the role of estrogens in the neuropathology and treatment of TD.

Until recently, inconsistent findings regarding the association between social status and coronary heart disease (CHD) have been explained primarily in terms of methodologic differences among studies. However, a closer examination of selected findings reveals that the association may have changed from positive (increasing risk with higher social status) to negative (increasing risk with lower social status) in certain urban populations as the absolute rate of CHD was also changing, following a period of rapid socio-ecologic change or “modernization”. The purpose of this study is to investigate a possible change in social status effect and to examine its biological basis for the white adult population of a rural county in Georgia between 1960 and 1974. While an apparent change in the association between social status and CHD was found for men in the predicted direction, no trend was found for women. The change was most dramatic among men under the age of 55 for whom the age-standardized ratio measure of association shifted from significantly greater than one (the null value) to significantly less than one. This result was concluded to be valid and is consistent with the finding of a substantial decline in CHD mortality during the same period for men in the study population but not for women. There is convincing evidence that the changing effect of social status was partially due to differential changes in certain biological risk factors of CHD, as intervening factors-particularly blood pressure increases among younger men during the 1960's. Alternatively, there was no support for the hypothesis that social status modified the effect of biological factors on CHD. Major sex differences in the findings and the implications of these results are discussed in the general context of social epidemiologic research.

We have prospectively followed over a 5-year period 434 volunteers who were at intake ambulatory, functional, presumably nondemented, and between 75 and 85 years of age. Fifty-six (an incidence of 3.53 per 100 person-years at risk) developed a progressive dementia: 32 met diagnostic criteria for Alzheimer's disease (AD) (an incidence of 2.0 per 100 person-years at risk), 15 had vascular or mixed dementia, and 9 had other disorders or remain undiagnosed. New cases of dementia were as common as myocardial infarction and twice as common as stroke. Risk factors for both dementia and AD were age (over 80) and gender (female); other reported risk factors such as family history, prior head injury, thyroid disease, maternal age, and smoking were not risk factors for AD in this elderly cohort. Prior stroke was the major risk factor for vascular or mixed dementia; diabetes and left ventricular hypertrophy but not a history of hypertension per se were also risk factors for vascular dementia. The major predictor of the development of AD was the mental status score on entry. The 58.5% of the cohort who made zero to two errors on a 33-item mental status test had a less than 0.6% per year chance of developing AD, whereas the 16% of the cohort with five to eight errors on this test developed AD at a rate of over 12% per year. Thus, it is possible to identify a large cohort of 80-year-olds who are at low risk for AD and a smaller cohort at very high risk.

A retrospective follow-up study was conducted to assess the impact of a psychosocial support program on survival with breast cancer. One-hundred and two nonparticipants were individually matched to 34 participants on several prognostic factors, and both groups were followed from date of cancer diagnosis (1971–1980) until December, 1981. Preliminary findings suggest a strong beneficial effect of the program on survival, which is statistically significant. However, this observed effect is due largely to a selection bias caused by the failure to match on the duration of the lag period between cancer diagnosis and program entry. Correcting for this bias in the analysis results in a small, nonsignificant program effect. We are not able to rule out a possible effect, however, because of the relative lack of statistical power and because of a modest, though nonsignificant benefit observed for women who entered the program shortly after diagnosis. Furthermore, the program might have other beneficial effects on the quality of life.

Background\ud \ud The high relapse and mortality rate of small-cell lung cancer (SCLC) fuels the need for epidemiologic study to aid in its prevention.\ud \ud Methods\ud \ud We included 24 studies from the ILCCO collaboration. Random-effects panel logistic regression and cubic spline regression were used to estimate the effects of smoking behaviors on SCLC risk and explore their non-linearity. Further, we explored whether the risk of smoking on SCLC was mediated through COPD.\ud \ud Findings\ud \ud Significant dose–response relationships of SCLC risk were observed for all quantitative smoking variables. Smoking pack-years were associated with a sharper increase of SCLC risk for pack-years ranged 0 to approximately 50. The former smokers with longer cessation showed a 43%quit_for_5–9 years to 89%quit_for_≥ 20 years declined SCLC risk vs. subjects who had quit smoking < 5 years. Compared with non-COPD subjects, smoking behaviors showed a significantly higher effect on SCLC risk among COPD subjects, and further, COPD patients showed a 1.86-fold higher risk of SCLC. Furthermore, smoking behaviors on SCLC risk were significantly mediated through COPD which accounted for 0.70% to 7.55% of total effects.\ud \ud Interpretation\ud \ud This is the largest pooling study that provides improved understanding of smoking on SCLC, and further demonstrates a causal pathway through COPD that warrants further experimental study.\ud \ud Abbreviations\ud COPD, chronic obstructive pulmonary disease; CPG, cigarettes per day; ILCCO, International Lung Cancer Consortium; MeSH, medical subject headings; NSCLC, non-small cell lung cancer; OR, odds ratio; SCLC, small cell lung cancer.

Item does not contain fulltext BACKGROUND AND METHODS: Familial aggregation of lung cancer exists after accounting for cigarette smoking. However, the extent to which family history affects risk by smoking status, histology, relative type and ethnicity is not well described. This pooled analysis included 24 case-control studies in the International Lung Cancer Consortium. Each study collected age of onset/interview, gender, race/ethnicity, cigarette smoking, histology and first-degree family history of lung cancer. Data from 24,380 lung cancer cases and 23,305 healthy controls were analysed. Unconditional logistic regression models and generalised estimating equations were used to estimate odds ratios and 95% confidence intervals. RESULTS: Individuals with a first-degree relative with lung cancer had a 1.51-fold increase in the risk of lung cancer, after adjustment for smoking and other potential confounders (95% CI: 1.39, 1.63). The association was strongest for those with a family history in a sibling, after adjustment (odds ratios (OR)=1.82, 95% CI: 1.62, 2.05). No modifying effect by histologic type was found. Never smokers showed a lower association with positive familial history of lung cancer (OR=1.25, 95% CI: 1.03, 1.52), slightly stronger for those with an affected sibling (OR=1.44, 95% CI: 1.07, 1.93), after adjustment. CONCLUSIONS: The occurrence of lung cancer among never smokers and similar magnitudes of the effect of family history on lung cancer risk across histological types suggests familial aggregation of lung cancer is independent of those risks associated with cigarette smoking. While the role of genetic variation in the aetiology of lung cancer remains to be fully characterised, family history assessment is immediately available and those with a positive history represent a higher risk group. 01 september 2012

This paper summarizes the results of three different studies of tardive dyskinesia conducted within a single-source, outpatient population at the Connecticut Mental Health Center in New Haven. Emerging themes from this work suggest that prevalence, severity and course of tardive dyskinesia are influenced--probably in different ways--by treatment, psychiatric illness, and other personal characteristics.

SUMMARY Validity and efficiency issues are considered with regard to the use of matching and random sampling as alternative methods of subject selection in follow-up and case-control studies. We discuss the simple situation involving dichotomous disease and exposure variables and a single dichotomous matching factor, and we consider the influence on efiiciency of a possible loss of subjects due to matching constraints. The decision to match or not should be motivated by efficiency considerations. An efficiency criterion based on a comparison of confidence intervals under matching and random sampling for the effect measure of interest (the risk ratio and risk difference in follow-up studies, and the odds ratio in case-control studies) leads to the following conclusions when the sampling method does not influence the size of the comparison group. In follow-up studies, matching on a confounder is expected to lead to a gain in efficiency over random sampling, while matching on a nonconfounder is not expected to result in a loss of efficiency. In case-control studies, the same conclusions hold, except that matching is not as advantageous as in follow-up studies and can lead to a loss of efliciency in some situations (usually of little practical importance). When matching reduces the size of the comparison group, there is likely to be a meaningful gain in efficiency due to random sampling only when the matched comparison group is at most 40-50% the size of the randomlysampled comparison group in a follow-up study, and at most 50-65% the size in a case-control study.

Injuries are an important health issue for children. Previous research, however, has presented confusing and conflicting results on the determinants of childhood injuries, particularly psychosocial predictors, largely due to methodologic problems. The purpose of this analysis, based on a prospective follow-up study of 532 children, was to identify factors related to injuries encountered in a prepaid group practice during a 12-month period. Using logistic regression, we found four factors independently associated with the risk of at least one treated injury: high activity level, high rate of pediatric utilization for non-injury-related visits during the follow-up period, occurrence of a treated injury during the year preceding the follow-up period, and negative attitude toward medical care providers by the child's mother. In addition, four factors were found to be independent predictors of injuries judged severe enough to always warrant medical care: occurrence of a treated injury in the preceding year, high rate of pediatric utilization for non-injury-related visits during the follow-up period, working more than 15 hours a week outside the home by the child's mother, and more life events reported by the mother for the year preceding the follow-up period. Since family stressors are related specifically to the risk of more severe injuries, which are unlikely to escape medical attention, we conclude that these factors probably are related to the occurrence of common injuries of early childhood and not exclusively to utilization behavior. We therefore suggest that children from families with these characteristics be targeted for injury prevention strategies.

• Twenty-one nonschizophrenic and 12 schizophrenic outpatients with tardive dyskinesia (TD) were followed up for a mean of 12.0 and 8.6 months, respectively, following discontinuation of neuroleptic therapy. Of the 33 patients, only one demonstrated complete reversal of TD. Cumulative survival curves of the length of time to first improvement (reduction in movement ratings by 50% of baseline) did not differ between the two groups. The median time to first improvement was seven months. If a patient can be kept off of a neuroleptic regimen for 18 months, the estimated probability of showing a 50% reduction in movement is 87.2%. In the nonschizophrenic group, depressed mood was negatively correlated with severity of abnormal movements.