At the Frontiers of Modeling Intensive Longitudinal Data: Dynamic Structural Equation Models for the Affective Measurements from the COGITO Study.

With the growing popularity of intensive longitudinal research, the modeling techniques and software options for such data are also expanding rapidly. Here we use dynamic multilevel modeling, as it is incorporated in the new dynamic structural equation modeling (DSEM) toolbox in Mplus, to analyze the affective data from the COGITO study. These data consist of two samples of over 100 individuals each who were measured for about 100 days. We use composite scores of positive and negative affect and apply a multilevel vector autoregressive model to allow for individual differences in means, autoregressions, and cross-lagged effects. Then we extend the model to include random residual variances and covariance, and finally we investigate whether prior depression affects later depression scores through the random effects of the daily diary measures. We end with discussing several urgent-but mostly unresolved-issues in the area of dynamic multilevel modeling.

The different trajectories of antipsychotic response: antipsychotics versus placebo.

BACKGROUND: It is generally accepted that antipsychotics are more effective than placebo. However, it remains unclear whether antipsychotics induce a pattern or trajectory of response that is distinct from placebo. We used a data-driven technique, called growth mixture modelling (GMM), to identify the different patterns of response observed in antipsychotic trials and to determine whether drug-treated and placebo-treated subjects show similar or distinct patterns of response. METHOD: We examined data on 420 patients with schizophrenia treated for 6 weeks in two double-blind placebo-controlled trials using haloperidol and olanzapine. We used GMM to identify the optimal number of response trajectories; to compare the trajectories in drug-treated versus placebo-treated patients; and to determine whether the trajectories for the different dimensions (positive versus negative symptoms) were identical or different. RESULTS: Positive symptoms were found to respond along four distinct trajectories, with the two most common trajectories ('Partial responder' and 'Responder') accounting for 70% of the patients and seen proportionally in both drug- and placebo-treated. The most striking drug-placebo difference was in the 'Dramatic responders', seen only among the drug-treated. The response of negative symptoms was more modest and did not show such distinct trajectories. CONCLUSIONS: Trajectory models of response, rather than the simple responder/non-responder dichotomy, provide a better statistical account of how antipsychotics work. The 'Dramatic responders' (those showing >70% response) were seen only among the drug-treated and make a significant contribution to the overall drug-placebo difference. Identifying and studying this subset may provide specific insight into antipsychotic action.

Trajectories of change in depression severity during treatment with antidepressants.

BACKGROUND: Response and remission defined by cut-off values on the last observed depression severity score are commonly used as outcome criteria in clinical trials, but ignore the time course of symptomatic change and may lead to inefficient analyses. We explore alternative categorization of outcome by naturally occurring trajectories of symptom change. METHOD: Growth mixture models were applied to repeated measurements of depression severity in 807 participants with major depression treated for 12 weeks with escitalopram or nortriptyline in the part-randomized Genome-based Therapeutic Drugs for Depression study. Latent trajectory classes were validated as outcomes in drug efficacy comparison and pharmacogenetic analyses. RESULTS: The final two-piece growth mixture model categorized participants into a majority (75%) following a gradual improvement trajectory and the remainder following a trajectory with rapid initial improvement. The rapid improvement trajectory was over-represented among nortriptyline-treated participants and showed an antidepressant-specific pattern of pharmacogenetic associations. In contrast, conventional response and remission favoured escitalopram and produced chance results in pharmacogenetic analyses. Controlling for drop-out reduced drug differences on response and remission but did not affect latent trajectory results. CONCLUSIONS: Latent trajectory mixture models capture heterogeneity in the development of clinical response after the initiation of antidepressants and provide an outcome that is distinct from traditional endpoint measures. It differentiates between antidepressants with different modes of action and is robust against bias due to differential discontinuation.

Factor analysis of alcohol abuse and dependence symptom items in the 1988 National Health Interview Survey.

The findings presented in this report are of general interest for the development of survey instruments for alcohol use disorders. They show which items represent the same dimension of alcohol problems when presented to respondents in a general population survey setting. The note determines the major dimensions underlying the complete set of 41 symptom items in the 1988 National Health Interview Survey, relates the item sets of the DSM- and ICD-criteria to these dimensions, and studies the measurement characteristics of items not used for DSM- or ICD-criteria.

Factor analysis of ICD-10 symptom items in the 1988 National Health Interview Survey on Alcohol Dependence.

The present paper analyzes data from the 1988 National Health Interview Survey (NHIS88) of US adults. This general population survey is important in that it is the largest survey to date that collects detailed information on alcohol consumption and alcohol problems. The NHIS88 is a multi-stage complex sample household study with a sample size of about 44,000 adults of which about 20,000 reported that they were current drinkers. Symptom items were used for classification of individuals into categories of alcohol abuse and alcohol dependence based on the DSM-III-R, DSM-IV, and ICD-10 classification. The present analyses are based on 20 items corresponding to the ICD-10 classification, the elements of which are closely related to the Edwards-Gross alcohol dependence syndrome concepts. A single dimension corresponding to alcohol problems in general accounted for much of the correlations among these items, but additional dimensions of interest were also identified. Of the ICD-10 elements only Impaired Control, Tolerance, and Withdrawal could be identified as separate dimensions.

The dimensionality of alcohol abuse and dependence: factor analysis of DSM-III-R and proposed DSM-IV criteria in the 1988 National Health Interview Survey.

Decisions on the final version of the DSM-IV alcohol abuse and dependence criteria will be determined largely by the APA's substance abuse field trials, conducted primarily in treated, clinical samples. Among the major objectives of the field trials are to study the boundaries between abuse and dependence, and to identify specific criteria that define the abuse category. The decisions on revisions of the abuse and dependence criteria in DSM-IV should, however, be informed by data from non-treated or general population samples as well. The present study addresses the field trial objectives using recent data from a large general population survey, the 1988 National Health Interview Survey (NHIS88). The paper reports on factor analyses to assess the dimensions underlying the DSM-III-R and DSM-IV dependence and abuse criteria as operationalized in the NHIS88. The focus of the analyses is on whether models with more than one dimension are needed and if so, the correspondence of the dimensions to criteria sets defined in the DSM-III-R and DSM-IV. The analyses show that a two-dimensional model is required. The dimensions are interpreted as abuse and dependence, but the sets of criteria that define each of the dimensions show important deviations from the criteria sets used in the DSM definitions.

Psychometric evaluation of diagnostic criteria: application to a two-dimensional model of alcohol abuse and dependence.

Psychometric investigations of diagnostic criteria can be helpful in the refinement of psychiatric instruments. This paper illustrates a methodology for investigating the measurement properties of a set of diagnostic criteria. The analyses are based on a two-dimensional factor analysis model for alcohol abuse and dependence. Based on this model, the methodology shows how cutpoints for diagnoses can be evaluated and defined, with which precision the criteria measure abuse and dependence, how well abuse without dependence can be measured, if the criteria should be weighted or not, if additional criteria are needed to improve measurement, if a smaller number of criteria could be used with almost as good results, and if diagnoses can be made with reliable results. The application of the methodology to the study of alcohol abuse and dependence in general population surveys shows important implications for diagnosis and prevalence estimation based on DSM criteria.

The development of heavy drinking and alcohol-related problems from ages 18 to 37 in a U.S. national sample.

OBJECTIVE: The purpose of this study is to add to the understanding of the development of heavy alcohol use and alcohol-related problems by examining data from the National Longitudinal Survey of Youth (NLSY), a general population sample that contains information on alcohol use for the ages 18-37. A key question in this study is how background characteristics of the individual influence this development and whether the influence of these background characteristics changes over time. METHOD: The data used in this study are a general population sample (N = 7,859) from the National Longitudinal Survey of Youth (NLSY). This study uses a multivariate outcome approach that focuses on individual variation in trajectories over age. The statistical analysis uses random coefficients in a latent variable framework. Across-age changes in the importance of the influence of background variables on the outcomes are modeled using varying centering points. RESULTS: A key finding is that dropping out of high school has no effect on alcohol-related problems for individuals in their mid-twenties, but is associated with significantly increased levels of alcohol-related problems for individuals in their mid-thirties. In contrast, going on to college is associated with lower levels of heavy drinking when individuals reach their late twenties and their thirties. Strong gender and ethnicity effects seen in the twenties diminish when individuals reach their thirties. CONCLUSIONS: The trajectory analysis expands the knowledge of problematic alcohol development for individuals in their late twenties and thirties. The increasing detrimental effect of dropping out of high school up to the age 37 endpoint of the study raises questions about the effects in later life of dropping out of high school.

Adolescent and young adult antisocial behavior and adult alcohol use disorders: a fourteen-year prospective follow-up in a national survey.

OBJECTIVE: Data from the National Longitudinal Survey of Youth (NLSY) are used to examine the association between antisocial behaviors (ASB) reported in youth (15-22 years old) and alcohol use disorders (AUD) 14 years later in a large (N = 7,326) representative national sample. METHOD: Structural equation modeling generalized to dichotomous outcomes was used to assess the associations between latent variables of ASB with latent variables of AUD and background variables. RESULTS: Exploratory factor analysis of 17 ASB items yielded three factors having clear interpretations with the literature-property offenses, person offenses and illicit substance involvement. When examined in the context of the multivariate structural equation model, several independent associations between ASB and AUD symptoms and covariates were found. Although there were significant and independent effects for each ASB factor on each of the alcohol use disorder factors, the strength of the association was strongest for the effects of early illicit substance involvement on alcohol abuse and dependence. CONCLUSIONS: Both illicit substance involvement and delinquency other than illicit substance involvement reported in 1980 were associated with alcohol use disorders 14 years later.

Alcohol use among college students: the effects of prior problem behaviors and change of residence.

OBJECTIVE: This article examines the relationship between prior problem behaviors and change in residence on alcohol use patterns among college students. METHOD: Measures of alcohol consumption (e.g., drinking frequency, average consumption and frequency of heavy episodic drinking were related to residence patterns and prior problem behaviors (e.g., conduct problems. illicit substance involvement and early age at onset of alcohol consumption). Subjects (N = 2,465; 51% women) were a subsample drawn from the National Longitudinal Study of Labor Market Experience in Youth. The analysis was conducted using a linear growth model for continuous outcomes with time-invariant and time-varying covariates for each of the drinking measures. RESULTS: The results of the structural equation analysis yielded significant and direct effects related to residence patterns and prior problem behaviors. Problem behaviors were related to drinking measures; however, there was no evidence for a mediational hypothesis. Neither was there systematic evidence that the relationship between prior problem behavior and alcohol use was mediated by residence patterns. The analysis of change in residence was related to both time-specific and longer-term influences on alcohol use. CONCLUSIONS: The presence of direct and independent effects for both dispositional and high-risk environmental factors in collegiate drinking practices support targeted and diverse strategies for prevention activities.

The dimensionality of alcohol abuse and dependence: a multivariate analysis of DSM-IV symptom items in the National Longitudinal Survey of Youth.

OBJECTIVE: This article examines the factor structure of 22 symptom items used to configure the criteria of DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition) alcohol abuse and dependence and relates the factor structure to background characteristics. METHOD: Data for this study were drawn from the National Longitudinal Study of Labor Market Experience in Youth (NLSY). The symptom items were related to the covariates using the statistical technique of structural equation modeling generalized to dichotomous outcomes. The present model is a special case of structural equation modeling, a multiple causes and multiple indicators (MIMIC) model, in which one or more latent variables (i.e., alcohol abuse and dependence) intervene between a set of observed background variables predicting a set of observed response variables (i.e., DSM-IV symptom items). RESULTS: The results of the structural equation analysis provide further support for two dimensions underlying the DSM-IV symptom items. Although the two-factor dimension bore a strong resemblance to the DSM-IV conceptions of abuse and dependence, there were notable differences in the item content of the symptom items for each dimension. The dependence dimension drew upon items related to the abuse criteria for continued drinking despite social problems and recurrent drinking resulting in failure to fulfill role obligations. The abuse dimension drew upon items related to the abuse criterion for hazardous drinking and the dependence criterion for larger amounts over time. The two factors were shown to have different relationships to the background variables. Alcohol dependence was related to family history of alcoholism and educational status. Age was not related to dependence and inversely related to alcohol abuse. CONCLUSIONS: Findings from this study replicate the two-dimensional model for DSM-IV criteria found in other studies and provide further support for the validity of alcohol dependence in general population samples. A major implication of the factor structure in the present study relates to the different classification of cases that would otherwise be obtained with DSM-IV criteria. These departures were shown to affect abuse, which retained only 40% of DSM-IV diagnoses, more strongly than dependence, which retained 91% of DSM-IV diagnoses.

The influence of changes in marital status on developmental trajectories of alcohol use in young adults.

OBJECTIVE: Multiple group latent curve analysis was used to assess the impact of changes in marital status on alcohol use trajectories in young adults and to test if these effects varied across ethnicity and gender. METHOD: Four years of data were obtained from a sample of young adults (N = 4,052; 54% male) drawn from the National Longitudinal Survey of Youth. Alcohol use and marital status were assessed once per year and covariates included age, gender, education and ethnicity. RESULTS: Latent curve models indicated that there was an overall nonlinear negative alcohol use trajectory across the four time points and that becoming married was reliably associated with an added down-turn to this trajectory. Multiple group models indicated that there was an interaction between ethnicity and marital status in the prediction of alcohol growth trajectories, but there was no interaction with gender. CONCLUSIONS: Becoming married for the first time exerted a unique effect on the overall developmental trajectory of alcohol use over time. This effect held for both ethnic groups but was reliably stronger for white compared to black respondents. This interaction may be attributable to lower levels of alcohol use reported by black respondents, or may be related to individual differences in reactivity to social influences by blacks relative to whites.

The relation between heavy alcohol use and bar patronage: a latent growth model.

OBJECTIVE: Random effects latent growth (LG) models were used to study the relation between changes in heavy alcohol use and changes in the frequency of bar patronage over a 3-year period. Previous research has identified a close link between alcohol use and bar patronage, but these cross-sectional findings limit inferences about possible temporal ordering. LG models are highlighted and compared with the more traditional but restricted cross-lagged models. METHOD: Three annual measures of heavy alcohol use and frequency of bar patronage were administered to a sample of 3071 adults (62% male) who participated in the National Longitudinal Study of Youch from 1982 to 1984. Background variables included age, gender, ethnicity and marital status. RESULTS: Heavy alcohol use and bar patronage were characterized by negative longitudinal growth trajectories and there was significant individual variability in these changes over time. Changes in heavy alcohol use closely paralleled corresponding changes in bar patronage. Finally, the Time 1 measures of each construct were inversely predictive of subsequent changes over time in the second construct. CONCLUSIONS: These findings suggest that heavy alcohol users tended to continue to patronize bars, and bar patronage tended to encourage continued heavy alcohol use. This relation was particularly strong for singles and males. Compared to more traditional cross-lagged models, LG modeling appears to be a much stronger analytic technique for studying growth over time. It is recommended that LG models be used for future studies of change in alcohol use over time.

Finite mixture modeling with mixture outcomes using the EM algorithm.

This paper discusses the analysis of an extended finite mixture model where the latent classes corresponding to the mixture components for one set of observed variables influence a second set of observed variables. The research is motivated by a repeated measurement study using a random coefficient model to assess the influence of latent growth trajectory class membership on the probability of a binary disease outcome. More generally, this model can be seen as a combination of latent class modeling and conventional mixture modeling. The EM algorithm is used for estimation. As an illustration, a random-coefficient growth model for the prediction of alcohol dependence from three latent classes of heavy alcohol use trajectories among young adults is analyzed.

Integrating person-centered and variable-centered analyses: growth mixture modeling with latent trajectory classes.

BACKGROUND: Many alcohol research questions require methods that take a person-centered approach because the interest is in finding heterogeneous groups of individuals, such as those who are susceptible to alcohol dependence and those who are not. A person-centered focus also is useful with longitudinal data to represent heterogeneity in developmental trajectories. In alcohol, drug, and mental health research the recognition of heterogeneity has led to theories of multiple developmental pathways. METHODS: This paper gives a brief overview of new methods that integrate variable- and person-centered analyses. Methods discussed include latent class analysis, latent transition analysis, latent class growth analysis, growth mixture modeling, and general growth mixture modeling. These methods are presented in a general latent variable modeling framework that expands traditional latent variable modeling by including not only continuous latent variables but also categorical latent variables. RESULTS: Four examples that use the National Longitudinal Survey of Youth (NLSY) data are presented to illustrate latent class analysis, latent class growth analysis, growth mixture modeling, and general growth mixture modeling. Latent class analysis of antisocial behavior found four classes. Four heavy drinking trajectory classes were found. The relationship between the latent classes and background variables and consequences was studied. CONCLUSIONS: Person-centered and variable-centered analyses typically have been seen as different activities that use different types of models and software. This paper gives a brief overview of new methods that integrate variable- and person-centered analyses. The general framework makes it possible to combine these models and to study new models serving as a stimulus for asking research questions that have both person- and variable-centered aspects.

Genetic Tobit factor analysis: quantitative genetic modeling with censored data.

Parameters of quantitative genetic models have traditionally been estimated by either algebraic manipulation of familial correlations (or familial mean squares), biometric model fitting, or multiple-group covariance structure analysis. With few exceptions, researchers who have used these methods for the analysis of twin data have assumed that their data were multinormal and, consequently, have used normal-theory estimation methods. It is shown that normal-theory methods produce biased genetic and environmental parameter estimates when data are censored. Specifically, with censored data, (1) normal-theory estimates of narrow-sense heritability are either positively or negatively biased, whereas (2) estimates of shared-familial environmental variance are always biased downward. An alternative method for estimating genetic and environmental parameters from censored twin data is proposed. The method is called genetic Tobit factor analysis (GTFA) and is an extension of the Tobit factor analysis model developed by Muthén (Br. J. Math. Stat. Psychol. 42, 241-250, 1989). Using a Monte Carlo design, the performance of GTFA is compared to traditional quantitative genetic methods in both large and small data sets. The results of this study suggest that GTFA is the preferred method for the genetic modeling of censored data obtained from twins.

The application of latent curve analysis to testing developmental theories in intervention research.

The effectiveness of a prevention or intervention program has traditionally been assessed using time-specific comparisons of mean levels between the treatment and the control groups. However, many times the behavior targeted by the intervention is naturally developing over time, and the goal of the treatment is to alter this natural or normative developmental trajectory. Examining time-specific mean levels can be both limiting and potentially misleading when the behavior of interest is developing systematically over time. It is argued here that there are both theoretical and statistical advantages associated with recasting intervention treatment effects in terms of normative and altered developmental trajectories. The recently developed technique of latent curve (LC) analysis is reviewed and extended to a true experimental design setting in which subjects are randomly assigned to a treatment intervention or a control condition. LC models are applied to both artificially generated and real intervention data sets to evaluate the efficacy of an intervention program. Not only do the LC models provide a more comprehensive understanding of the treatment and control group developmental processes compared to more traditional fixed-effects models, but LC models have greater statistical power to detect a given treatment effect. Finally, the LC models are modified to allow for the computation of specific power estimates under a variety of conditions and assumptions that can provide much needed information for the planning and design of more powerful but cost-efficient intervention programs for the future.

Age differences in the symptoms of depression: a latent trait analysis.

Our hypothesis was that older adults are less likely than younger adults to acknowledge dysphoria or anhedonia even at the same level of depression. Study subjects were 3,141 participants in Baltimore, Maryland, and 3,469 participants in the Durham-Piedmont region of North Carolina who had complete data on symptoms of depression active in the one month prior to interview, as well as several covariates thought to be related to depression. The effect of age on the endorsement of the dysphoria/anhedonia stem question from the section on Major Depression in the Diagnostic Interview Schedule was estimated in the two independently gathered samples employing structural equations with a measurement model. The results indicate that, even accounting for differences due to overall level of depressive symptoms, as well as gender, minority status, educational attainment, marital status, employment status, and cognitive impairment, dysphoria was less likely to be endorsed by persons 65 years of age and older. This bias against older adults may account in part for the low rates of Major Depression reported for older persons from epidemiologic studies employing the standard diagnostic criteria.

Latent Variable Modeling of Longitudinal and Multilevel Substance Use Data.

This article demonstrates the use of a general model for latent variable growth analysis which takes into account cluster sampling. Multilevel Latent Growth Modeling (MLGR4) was used to analyze longitudinal and multilevel data for adolescent and parent substance use measured at four annual time points. An associative LGM model was tested for alcohol, marijuana, and cigarette use with a sample of 435 families. Hypotheses concerning the shape of the growth curve and the extent of individual differences in the common trajectory over time were tested. The effects of marital and family status and socio-economic status on family levels of substance use were also examined. Findings are discussed in terms of family-level substance use and similarities in developmental trajectories across substances, and the impact of contextual factors on family levels of substance use and development.