The utility of latent trait models in psychiatric epidemiology

Differential item functioning in the cognitive screener used in the Longitudinal Aging Study in India

INTRODUCTION

The Longitudinal Aging Study in India (LASI) was initiated to capture data to be comparable to the Health and Retirement Survey (HRS) and hence used study instruments from the HRS. However, a rigorous psychometric evaluation before adaptation of cognitive tests may have indicated bias due to diversities across Indian states such as education, ethnicity, and urbanicity. In the present analysis, we evaluated if items show differential item functioning (DIF) by literacy, urbanicity, and education status.

METHODS

We calculated proportions for each item and weighted descriptive statistics of demographic characteristics in LASI. Next, we evaluated item-level measurement differences by testing for DIF using the alignment approach implemented using Mplus software.

OBSERVATION

We found that cognitive items in the LASI interview demonstrate bias by education and literacy, but not urbanicity. Items relating to animal (word) fluency show DIF. The model rates correct identification of the prime minister as the most difficult binary response item whereas the day of the week and numeracy items are rated comparatively easier.

CONCLUSIONS

Our study would facilitate comparison across education, literacy and urbancity to support analyses of differences in cognitive status. This would help future instrument development efforts by recognizing potentially problematic items in certain subgroups.

The Adequacy of Measurement of Short and Long-Term Consequences of False-Positive Screening Mammography

Objectives:

The aim of this study is to review quantitative studies on women's experiences of consequences of false-positive screening mammography to assess the adequacy of the most frequently used instruments for measuring short-term and long-term psychological consequences.

Methods:

Relevant papers reporting quantitative studies on consequences of false-positive screening mammography were identified using MEDLlNE, CINAHL, EMBASE and PsycInfo databases. Articles citing development and psychometric properties of the most frequently used measures were also retrieved. Finally, the review focused on studies that had used at least one of the most frequently used measures.

Results:

Twenty-three relevant studies were identified. The most commonly used measures were the General Health Questionnaire (GHQ), the Hospital Anxiety and Depression Scale (HADS), the Psychological Consequences Questionnaire (PCQ) and the State-Trait Anxiety Inventory (STAI). One or more of these was used in 17 of the 23 studies.

Conclusions:

The GHQ, the HADS and the STAI have problems with language, content relevance, and content coverage in studies of false-positive screening mammography. These instruments should not be used to measure psychological consequences of any kind of cancer screening. The PCQ is an adequate questionnaire for measuring short-term consequences, and the PCQ is preferable to other measures because of its higher sensitivity. However, there is little evidence that the PCQ is able to adequately detect all long-term consequences of screening mammography. Given the inadequacy of the measurement instruments used, any current conclusions about the long-term consequences of false-positive results of screening mammography must remain tentative.

Positive versus negative mental health in emerging adulthood: a national cross-sectional survey

Background

The dual continuum model suggests that positive mental health not only implies the absence of mental illness, but also constitutes an entity of its own. Measures that encompass both positive and negative mental health in young adults are rare. Thus, we assessed whether dimensions corresponding to positive and negative mental health could be identified in a sample of young individuals. Additionally, we explored how such dimensions were associated with potential health-related factors.

Methods

We obtained data from the Swedish National Public Health Survey 2004–2009 (23,394 women, 18,274 men, aged 16–29 years). We used exploratory factor analysis (EFA) to identify relevant factors revealed by the 12-item General Health Questionnaire (GHQ-12) and confirmatory factor analysis (CFA) to verify the factor structure. We tested the significance of the difference between effects of potential health-related factors on positive mental health (PMH) and negative mental health (NMH).

Results

The EFA for the GHQ-12 revealed a two factor model with negative items that had high positive loadings on one factor and lower negative loadings on the other factor. The positive items had loading trends that were opposite those of the negative items. The fit of this model was supported by the CFA, which yielded a significantly better match than a unidimensional model. When we investigated the associations between GHQ-scores and potential predictors of health, we found that most potential predictors had significant and opposing effects on both PMH and NMH; with the strongest effects from suicidal ideation and perceived humiliation.

Conclusions

Our results could be seen to indicate that positive and negative mental health are distinct and complementary constructs. Still, the results of our factor analysis may specifically reflect the wording of the items. We conclude that the GHQ-12 is an appropriate tool for its original purpose, to detect “psychiatric morbidity”. More refined measures, including predictors of health, are needed to assess PMH and validate the bidimensionality hypothesis.

Psychological functioning and adherence to the recommended dose of physical activity in later life: results from a national health survey

BACKGROUND

Official health organizations have established the dose of physical activity needed for preserving both physical and psychological health in old age. The objective of this study was to explore whether adherence to the recommended criterion of physical activity accounted for better psychological functioning in older adults in Israel.

METHODS

A random sample of 1,663 (799 men) Israelis reported their physical activity routine, and based on official guidelines were divided into sufficiently active, insufficiently active, and inactive groups. The General Health Questionnaire (GHQ) was used for assessing mental health and the Mini-Mental State Examination (MMSE) for assessing cognitive functioning.

RESULTS

Factor analysis performed on the GHQ yielded two factors - positive and negative. Logistic regressions for the GHQ factors and for the MMSE were conducted for explaining their variance, with demographic variables entered first, followed by health and then physical activity. The explained variance in the three steps was Cox and Snell R2 = 0.022, 0.023, 0.039 for the positive factor, 0.066, 0.093, 0.101 for the negative factor, and 0.204, 0.206, 0.209 for the MMSE.

CONCLUSIONS

Adherence to the recommended dose of physical activity accounted for better psychological functioning beyond demographic and health variables; however, the additional explained variance was small. More specific guidelines of physical activity may elucidate a stronger relationship, but only randomized controlled trials can reveal cause-effect relationship between physical activity and psychological functioning. More studies are needed focusing on the positive factor of psychological functioning.

Personality disorder diagnostic thresholds correspond to different levels of latent pathology

To meet a diagnosis for a particular personality disorder (PD), among other requirements, a person must exhibit a specified minimum number of the features outlined for that PD, otherwise referred to as the diagnostic threshold. Despite many years of research on the DSM PDs, there is little empirical basis for the chosen thresholds. The present study used mechanisms of item response theory (IRT) to link the diagnostic thresholds of six PDs to their corresponding level of latent PD pathology. Consistent with our hypothesis, analyses of the data from 41,227 participants revealed that PD diagnostic thresholds corresponded to a wide range of latent pathology. For example, the diagnostic threshold for schizoid PD corresponded to 1.54 SDs of pathology whereas the threshold for dependent PD corresponded to 2.72 SDs of pathology. The current analyses have demonstrated for the first time that the latent pathology associated with each PD threshold varies widely, and thus has quantified what others accurately have speculated. Implications and considerations for the future directions of PD classification are discussed.

The value of item response theory in clinical assessment: a review

Item response theory (IRT) and related latent variable models represent modern psychometric theory, the successor to classical test theory in psychological assessment. Although IRT has become prevalent in the measurement of ability and achievement, its contributions to clinical domains have been less extensive. Applications of IRT to clinical assessment are reviewed to appraise its current and potential value. Benefits of IRT include comprehensive analyses and reduction of measurement error, creation of computer adaptive tests, meaningful scaling of latent variables, objective calibration and equating, evaluation of test and item bias, greater accuracy in the assessment of change due to therapeutic intervention, and evaluation of model and person fit. The theory may soon reinvent the manner in which tests are selected, developed, and scored. Although challenges remain to the widespread implementation of IRT, its application to clinical assessment holds great promise. Recommendations for research, test development, and clinical practice are provided.

Should substance use disorders be considered as categorical or dimensional?

AIMS

This paper discusses the representation of diagnostic criteria using categorical and dimensional statistical models. Conventional modeling using categorical or continuous latent variables in the form of latent class analysis and factor (IRT) analysis has limitations for the analysis of diagnostic criteria.

METHODS

New hybrid models are discussed which provide both categorical and dimensional representations in the same model using mixture models. Conventional and new models are applied and compared using recent data for Diagnostic and Statistical Manual of Mental Disorders version IV (DSM-IV) alcohol dependence and abuse criteria from the National Epidemiologic Survey on Alcohol and Related Conditions. Classification results from hybrid models are compared to the DSM-IV approach of using the number of diagnostic criteria fulfilled.

RESULTS

It is found that new hybrid mixture models are more suitable than latent class and factor (IRT) models.

CONCLUSIONS

Implications for DSM-V are discussed in terms of reporting results using both categories and dimensions.

Item response mixture modeling: application to tobacco dependence criteria

This paper illustrates new hybrid latent variable models that are promising for phenotypical analyses. The hybrid models combine features of dimensional and categorical analyses seen in the conventional techniques of factor analysis and latent class analysis. The paper focuses on the analysis of categorical items, which presents especially challenging analyses with hybrid models and has recently been made practical in the Mplus program. The hybrid models are typically seen to fit data better than conventional models of factor analysis (IRT) and latent class analysis. An illustration is given in the form of analysis of tobacco dependence in a general population survey.

The Short Mood and Feelings Questionnaire (SMFQ): a unidimensional item response theory and categorical data factor analysis of self-report ratings from a community sample of 7-through 11-year-old children

Item response theory (IRT) and categorical data factor analysis (CDFA) are complementary methods for the analysis of the psychometric properties of psychiatric measures that purport to measure latent constructs. These methods have been applied to relatively few child and adolescent measures. We provide the first combined IRT and CDFA analysis of a clinical measure (the Short Mood and Feelings Questionnaire--SMFQ) in a community sample of 7-through 11-year-old children. Both latent variable models supported the internal construct validity of a single underlying continuum of severity of depressive symptoms. SMFQ items discriminated well at the more severe end of the depressive latent trait. Item performance was not affected by age, although age correlated significantly with latent SMFQ scores suggesting that symptom severity increased within the age period of 7-11. These results extend existing psychometric studies of the SMFQ and confirm its scaling properties as a potential dimensional measure of symptom severity of childhood depression in community samples.

Chronicity of Psychological Strain in Occupational Settings and the Accuracy of the General Health Questionnaire

This article examines the proposition that the traditional scoring method of the General Health Questionnaire (GHQ) underestimates the number of respondents classified as "cases." A revised "chronic" scoring method (the CGHQ) is used and demonstrates superior construct validity and greater sensitivity. A comparison of the CGHQ and GHQ also shows the CGHQ to be a superior criterion measure. These claims are demonstrated by survey data gathered from 3 occupational samples (Ns=11,637, 2,253, and 2,124). Results show that the CGHQ is more appropriate as a screening instrument for psychological morbidity. Tests of construct validity also favor the CGHQ with only a slight advantage for predictive validity in terms of variance explained. The more desirable statistical properties of the CGHQ result in a reduction of significant interaction terms and are strongly recommended in future studies as a means of controlling Type I errors when tests of moderation are examined.

A brief diagnostic screening instrument for mental disturbances in general medical wards

OBJECTIVE

Mental illness is prevalent among general hospital ward patients but often goes unrecognised. The aim of this study was to validate the SCL-8d as a brief questionnaire for mental disturbances for use in general hospitals.

METHODS

The study included 2040 patients, 18 years or older, consecutively admitted to 11 general internal medicine wards in seven European countries. All patients were screened on admission by means of the SCL-8d questionnaire. The psychometric performance (i.e., the internal validity) of the SCL-8d scale was tested using modern item response theory (IRT) in the form of the Rasch model.

RESULTS

Differences between sample characteristics were considerable. Even so, the SCL-8d scale showed a remarkable, statistically significant fit in terms of internal homogeneity (P>.01) in all individual settings, except in Spain and Germany where the item "Everything is an effort" had to be excluded to obtain a fit. When pooling data from all centres, an excellent statistical significance of fit (P>.05) was obtained by exclusion of the "Effort" item. The scale was homogeneous as to gender (P>.05), but not age as it performed better among young patients than among patients older than 60 years (P<.01). In these two patient groups both internal and external homogeneity (gender, median age) was achieved. The SCL-8d sum score showed a marked correlation with current and previous treatment for mental illness.

CONCLUSION

Apart from the "Effort" item ranking differently on the latent severity dimension as to age, the SCL-8d seems very robust from a psychometric point of view. Besides being short, the SCL-8d scale contains only emotional symptoms. It would therefore seem to be an excellent diagnostic tool for use in medical settings.

Are higher rates of depression in women accounted for by differential symptom reporting?

BACKGROUND

The gender difference in prevalence and incidence rates of depression is one of the most consistent findings in psychiatric epidemiology. We sought to examine whether any gender differences in symptom profile might account for this difference in rates.

METHOD

This study was a population-based 13-year follow-up survey of community-dwelling adults living in East Baltimore in 1981. Subjects were the continuing participants of the Baltimore Epidemiologic Catchment Area Program. Participants interviewed between 1993 and 1996 with complete data on depressive symptoms and covariates were included (n = 1727). We applied structural equations with a measurement model for dichotomous data (the MIMIC-multiple indicators, multiple causes-model) to compare symptoms between women and men, in relation to the nine symptom groups comprising the diagnostic criteria for major depression, adjusting for several potentially influential characteristics (namely, age, self-reported ethnicity, educational attainment, marital status, and employment).

RESULTS

There were no significant gender differences in the self-report of depression symptoms even taking into account the higher level of depressive symptoms of women and the influence of other covariates. For example, women were no more likely to endorse sadness than were men, as evidenced by a direct effect coefficient that was not significantly different from the null [adjusted estimated direct effect of gender on report of sadness = 0.105, 95% confidence interval (-0.113, 0.323)].

CONCLUSIONS

Men and women in this community sample reported similar patterns of depressive symptoms. No evidence that the presentation of depressive symptoms differs by gender was found.

Cross-validation, predictive validity, and time course of the Benzodiazepine Dependence Self-Report Questionnaire in a benzodiazepine discontinuation trial

The Benzodiazepine Dependence Self-Report Questionnaire (Bendep-SRQ) measures the severity of benzodiazepine (BZ) dependence on four domains: awareness of problematic use, preoccupation with the availability of BZ, lack of compliance with the therapeutic regimen, and withdrawal. Although promising results of the Bendep-SRQ have been obtained in cross-sectional studies, no attention has been paid to its clinical relevance during BZ withdrawal, i.e., predictive validity and time course. We performed cross-validation and evaluated the predictive validity and time course on 180 long-term BZ users who were taking part in a general practice BZ discontinuation trial. Three of the four domains had good scalability. Some concerns arose about the preoccupation scale, which emphasizes the need for cross-validation in clinically relevant populations. All scales showed excellent reliability (subject discriminability, item discriminability), while construct and discriminant validity were adequate. All four scales contributed significantly to the prediction of whether complete abstinence would be achieved directly after taking part in the discontinuation program. This prediction was independent of the other prognostic variables, except for those in the domain problematic use. The scales problematic use and preoccupation showed good sensitivity to changes during follow-up. The insensitivity of the scale, lack of compliance can be explained by low baseline scores in our population, while the insensitivity of the withdrawal scale was probably the result of the study design. In conclusion, our study indicated the clinical relevance of the Bendep-SRQ before and during a BZ discontinuation trial. We recommend the use of the Bendep-SRQ in discontinuation therapy and research into the field of BZ addiction.

Toward a dimensional and psychometrically-informed approach to conceptualizing psychopathology

Most modern research on psychopathology is framed by the categorical model of mental disorders embodied in the Diagnostic and statistical manual of mental disorders, 4th edition (DSM-IV; American Psychiatric Association, 1994). Nevertheless, the categorical model of the DSM is incompatible with robust empirical observations about psychopathological variation, such as the tendency for multiple, putatively distinct mental disorders to occur in the same persons. The goal of this invited essay is to outline a potential alternative to the DSM's categorical approach: a dimensional and psychometrically-informed approach to conceptualizing and studying psychopathology. We discuss the advantages of the dimensional approach, some modern statistical technologies that can be recruited in the service of the dimensional approach, and ways of integrating the dimensional approach with a behavioral approach to psychopathology assessment and research.

Gender differences in the relationship between depression and suicidal ideation in young adolescents

OBJECTIVE

This study examined the risk relationship between depressive symptomatology and suicidal ideation for young adolescent males and females.

METHOD

A large cohort of students in their first year of high school completed the Center for Epidemiological Studies Depression Scale (CES-D) and the Adolescent Suicide Questionnaire. The risk relationship between depressive symptomatology and suicidal ideation was modelled using non-parametric kernel-smoothing techniques.

RESULTS

Suicidal ideation was more frequently reported by females compared with males which was partly explained by females having higher mean depression scores. At moderate levels of depression females also had a significantly higher risk of suicidal ideation compared with males and this increased risk contributed to the overall higher levels of female ideation.

CONCLUSIONS

The risk relationship between depressive symptomatology and suicidal ideation is different for young adolescent males and females. The results indicate that moderate levels of depressive symptomatology can be associated with suicidal ideation (especially among young females) and that for these young people a suicide risk assessment is required.

Modern psychometric methods for detection of differential item functioning: application to cognitive assessment measures

Cognitive screening tests and items have been found to perform differently across groups that differ in terms of education, ethnicity and race. Despite the profound implications that such bias holds for studies in the epidemiology of dementia, little research has been conducted in this area. Using the methods of modern psychometric theory (in addition to those of classical test theory), we examined the performance of the Attention subscale of the Mattis Dementia Rating Scale. Several item response theory models, including the two- and three-parameter dichotomous response logistic model, as well as a polytomous response model were compared. (Log-likelihood ratio tests showed that the three-parameter model was not an improvement over the two-parameter model.) Data were collected as part of the ten-study National Institute on Aging Collaborative investigation of special dementia care in institutional settings. The subscale KR-20 estimate for this sample was 0.92. IRT model-based reliability estimates, provided at several points along the latent attribute, ranged from 0.65 to 0.97; the measure was least precise at the less disabled tail of the distribution. Most items performed in similar fashion across education groups; the item characteristic curves were almost identical, indicating little or no differential item functioning (DIF). However, four items were problematic. One item (digit span backwards) demonstrated a large error term in the confirmatory factor analysis; item-fit chi-square statistics developed using BIMAIN confirm this result for the IRT models. Further, the discrimination parameter for that item was low for all education subgroups. Generally, persons with the highest education had a greater probability of passing the item for most levels of theta. Model-based tests of DIF using MULTILOG identified three other items with significant, albeit small, DIF. One item, for example, showed non-uniform DIF in that at the impaired tail of the latent distribution, persons with higher education had a higher probability of correctly responding to the item than did lower education groups, but at less impaired levels, they had a lower probability of a correct response than did lower education groups. Another method of detection identified this item as having DIF (unsigned area statistic=3.05, p<0.01, and 2.96, p<0.01). On average, across the entire score range, the lower education group's probability of answering the item correctly was 0.11 higher than the higher education group's probability. A cross-validation with larger subgroups confirmed the overall result of little DIF for this measure. The methods used for detecting differential item functioning (which may, in turn, be indicative of bias) were applied to a neuropsychological subtest. These methods have been used previously to examine bias in screening measures across education and ethnic and racial subgroups. In addition to the important epidemiological applications of ensuring that screening measures and neuropsychological tests used in diagnoses are free of bias so that more culture-fair classifications will result, these methods are also useful for the examination of site differences in large multi-site clinical trials. It is recommended that these methods receive wider attention in the medical statistical literature.

The Child Behavior Checklist (CBCL) and related material: standardization and validation in Danish population based and clinically based samples

The Child Behavior Checklist (CBCL) and related material, developed by Achenbach and Edelbrock in Vermont, was validated in a mailed survey. A population based sample of 779 children between the age of four and 17 years was compared to a sample of 146 children referred for child psychiatric service. Danish children scored very much like our Scandinavian and German neighbours, but low compared to most others. The CBCL mean 'total behavior problem score' in the population was 17.7. The checklists, especially the parent and teacher versions, provided good construct validity. Youths generally reported more emotional problem behavior than their parents and teachers did about them. In general, parents and youths agree more, reporting emotional problems, and parents and teachers agree more, when scattering externalizing behavior. Short screening constructs are introduced, and by the use of latent trait analysis, four clinically relevant sub-scales were generated. Predictive value, sensitivity, specificity and clinical validity must be undertaken in a future two-phase study.

Psychometric properties of the General Health Questionnaire (GHQ-12) in a German primary care sample

OBJECTIVE

The aim of this investigation was to examine the psychometric properties of the German 12-item General Health Questionnaire (GHQ-12) in a primary care sample.

METHOD

A sample (n = 421) of adult out-patients was screened using the GHQ-12. A standardized clinical interview (SCID) was conducted with all screened patients. Reliability, validity and factor analysis of the GHQ-12 were evaluated. Item characteristics were examined using item characteristics curves. Item bias analysis was performed using contingency tables.

RESULTS

The German version of the GHQ-12 is a reliable instrument and performed well in detecting cases of psychological disorders. Factor analysis replicated the findings of earlier studies. Item characteristics curves and item bias analysis indicated that the individual items should be assessed carefully. One item was biased in relation to age, while another item showed a low positive response rate.

CONCLUSION

The GHQ-12 as a whole is a reliable questionnaire and can be a useful screening tool in primary care.

Correlates of expected success at adherence to health regimen of people with IDDM

Patient non-adherence is a well-recognized although poorly understood phenomenon that affects patients in all areas of health care. Failure to comply with health regimens is extremely costly both in economic terms and the health status of individuals. This study assesses factors which correlate with the expected success of health regimen adherence in 146 insulin-dependent diabetes mellitus (IDDM) subjects. The results indicate that success in complying with a health regimen is associated with good family support and rapport with health professionals, an absence of chronic stress and the capacity to take up the challenges posed by the disease. Health professionals have a role in engendering optimism, in maintaining enthusiasm, and facilitating and encouraging maintenance in health behaviours.

Screening for somatization and hypochondriasis in primary care and neurological in-patients: a seven-item scale for hypochondriasis and somatization

The aim of this study was to investigate the internal and external validity of the Whiteley Index as a screening instrument for somatization illness. A 14-item version of the Whiteley Index for hypochondriacal traits was given to 99 of 191 consecutive primary care patients, aged 18-65 years, and to 100 consecutive patients, aged 18-60 years, admitted for the first time to a neurological ward. The primary care sample was, in addition, interviewed by means of the SCAN (Schedules for Clinical Assessment in Neuropsychiatry) psychiatric interview. The GPs and the neurologists were asked to rate various characteristics of the patients that might indicate somatization. The internal validity of the Whiteley Index was tested by means of latent structure analysis. On this basis, a reduced seven-item scale (Whiteley-7 scale) and two subscales (i.e., an Illness Conviction and Illness Worrying scale, each with three items) were constructed. All three had a high internal validity fitting into the very restricted Rasch statistical model (p>0.05) and an acceptable transferability between most of the subpopulations investigated. In the primary care population, the Whiteley-7 and the Illness Conviction scales at cut-point 0/1 showed 1.00 and 0.87 sensitivity and 0.65 and 0.87 specificity, respectively, using as "gold standard" the fulfillment of criteria for at least one ICD-10 somatoform disorder, and 0.71 and 0.63 sensitivity and 0.62 and 0.87 specificity, respectively, as gold standard for the fulfillment of criteria for at least one DSM-IV somatoform disorder, excluding the NOS diagnostic group. The Illness Worrying subscale showed less impressive performance in this respect. The agreement between the Whiteley-7 scale including the two subscales and neurologists' rating and the GPs' rating and the somatization subscale on the SCL-90 was modest or worse. It may be concluded that the Whiteley-7 scale and the Illness Conviction subscale had acceptable psychometric profiles, and both seem to be promising screening tools for not only hypochondriasis but also for somatoform disorders in general.

Psychiatric epidemiology

Statisticians working in psychiatric epidemiology regularly confront a variety of problems that are rare in other branches of epidemiology. These problems range from the frequent absence of 'objective' biological markers and consequent reliance on informant and self-reports of considerable fallability, to the perhaps corresponding frequent use of multivariate statistical methods such as latent variable modelling. Equally, such difficulties can be regarded as challenges that offer the statistician a critical role.

Validation of the Bech-Rafaelsen Mania Scale using latent structure analysis

The essential criteria of internal validity have not been sufficiently evaluated for any mania rating scale, although the fulfillment of such criteria is a prerequisite for summing the item scores to give a total score reflecting the severity of mania, and for comparing total scores across patient groups that differ with regard to variables such as age and sex. This study investigated the internal validity of the Bech-Rafaelsen Mania Scale (MAS), based on the ratings of 100 consecutively admitted drug-free DSM-III-R manic patients. Application of logistic latent structure models did not statistically confirm the additivity of the MAS. However, a modified MAS (the MAS-M) arising from the analyses fulfilled the measurement model. Transferability of the MAS-M across age and sex was also confirmed. The MAS-M showed an acceptable concurrent validity and an adequate sensitivity in discriminating between responders and non-responders among patients participating in a drug trial. The MAS-M presented here is the first mania rating scale that has been shown to fulfil statistical criteria for internal validity.

Evaluation of the MOS SF-36 Physical Functioning Scale (PF-10): II. Comparison of relative precision using Likert and Rasch scoring methods

This study examined the relative precision (RP) of two methods of scoring the 10-item Physical Functioning Scale (PF-10) from a large sample of patients (n = 3445) of the Medical Outcomes Study. Based on a Likert scaling model, the PF-10 summated scoring method was compared with a Rasch Item Response Theory (IRT) scaling model in which raw scores were transformed into a latent trait variable of physical functioning. Potential differences between scoring methods were hypothesized to be attributed to: (1) the logarithmic nature of the Rasch transformation; (2) the unevenness of the PF-10 item distributions; and (3) reduction of within-group variance. RP ratios favored the Rasch model in discriminating between patients who differed in disease severity. The Rasch and Likert scoring models performed similarly for tests involving sensitivity to change over a two-year follow-up period. In all comparisons, differences between methods were most apparent in clinical groups whose scores most approximated the extremes of the score distribution. Further research is necessary to test for differences between scoring models in discrimination and sensitivity to change among clinical groups whose scores are sufficiently spread across the continuum of physical functioning, in particular patients with either very high or low physical functioning. The Rasch model of scoring may have important implications for the clinical interpretation of individual scores at all ranges of the scale.

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.

Psychiatric morbidity in primary public health care: a Nordic multicentre investigation. Part I: method and prevalence of psychiatric morbidity

The prevalence of mental illness in five different Scandinavian primary care populations was investigated in this study. Patients consecutively consulting their general practitioner a particular week-day were included in the study. Initially the SCL-25 was applied and next the high scores and a sample of the low scores were interviewed by the PSE. In the analysis the screening procedure was first validated. The internal validity of the SCL was tested by means of Rasch latent structure analysis and the external validity tested by ROC/QROC analysis. Based on this, a short 8-item version of the SCL was developed. The prevalence of mental illness in all centres was 0.26 with a minimum of 0.14 in Nacka and a maximum of 0.34 in Turku.

The structure of common psychiatric symptoms: how many dimensions of neurosis?

In order to replicate and elaborate the two-dimensional model of depression and anxiety underlying the structure of common psychiatric symptoms proposed by Goldberg et al. (1987), we carried out latent trait analyses on PSE symptom data of the original Manchester study and two recent Dutch studies. We used the same analytical strategy as Goldberg et al. to facilitate comparison with the earlier work. It was found that a more comprehensive set of common psychiatric symptoms caused an extra, third dimension to emerge, so that the earlier anxiety dimension became split between a specific anxiety axis characterized by situational and phobic anxiety and avoidance, and a non-specific anxiety axis characterized by free-floating anxiety, various symptoms relating to tension, irritability and restlessness. It is argued that three dimensions are sufficient to account for the covariance between common psychiatric symptoms. A fairly consistent correlation between the non-specific anxiety and the depression dimension was found across sites, as well as independence of the specific anxiety dimension from the other two dimensions. Furthermore, the depression dimension was robust with similar symptom profiles across samples, but there appeared to be local differences in the structure of anxiety symptoms.

The Psychogeriatric Assessment Scales: a multi-dimensional alternative to categorical diagnoses of dementia and depression in the elderly

The Psychogeriatric Assessment Scales (PAS) provide an assessment of the clinical changes seen in dementia and depression. Principal components analysis and latent trait analysis were used to develop a set of scales to summarize these clinical changes. There are three scales derived from an interview with the subject (Cognitive Impairment, Depression, Stroke) and three from an interview with an informant (Cognitive Decline, Behaviour Change, Stroke). Results are reported on the reliability and validity of these scales using data from clinical samples in Sydney and Geneva and a population sample from Canberra. The scales were found to have excellent validity when judged against clinical diagnoses of dementia and depression and could distinguish Alzheimer's from vascular dementia. Cut-off points were developed to indicate correspondence between scale scores and clinical diagnoses. Percentile rank norms were developed from the Canberra population sample. The PAS is easy to administer and score and can be used by lay interviewers after training. It is intended for application both in research and in services for the elderly.

Item bias in cognitive screening measures: comparisons of elderly white, Afro-American, Hispanic and high and low education subgroups

A study of item bias in standard cognitive screening measures was conducted in a sample of Afro-American, Hispanic and non-Hispanic white elderly respondents who were part of a dementia case registry study. The methods of item-response theory were applied to identify biased items. Both cross-cultural and high and low education groups were examined to determine which items were biased. Out of 50 cognitive items examined from six widely used cognitive screening measures, 16 were identified as biased for either high and low education groups or ethnic/racial group membership.

The two scaled versions of the Chinese General Health Questionnaire: a comparative analysis

The four scales of the Chinese 28-item General Health Questionnaire (GHQ-28), and the five scales of the Chinese GHQ-20 were separately subjected to confirmatory factor analysis (in individual sample analyses and simultaneous multi-sample analyses) based on four Chinese samples of 150 general psychiatric patients, 549 school teachers, 653 university undergraduates, and 1082 secondary school students. While the two scaled versions were comparable in discriminating patients from nonpatients, the five scales of the GHQ-20 assessing five correlated dimensions were interpreted as mapping salient symptom areas in the Hong Kong Chinese population as they were empirically derived in the Chinese setting. The need for using samples from different cultural settings in multi-sample confirmatory analysis was discussed.

Is the General Health Questionnaire (12 item) a culturally biased measure of psychiatric disorder?

There have been suggestions that some self-administered questionnaires designed to assess psychiatric disorder tend to overestimate prevalence in samples from Latin America. This phenomenon may be obscured when the General Health Questionnaire (GHQ) is used, as it is recommended that researchers determine the threshold in each setting by comparing the GHQ with a standardised interview. Reports in the literature suggest that Latin American samples have a higher threshold for case definition using the GHQ than that found in British samples. The present study confirmed this finding when comparing the 12-item GHQ in a Chilean primary care sample with a sample of primary care attenders from the United Kingdom. The increase in GHQ scores in the Chilean sample persisted after adjustment for age, sex, marital status and the score on the Revised Clinical Interview Schedule (CIS-R). The increase in scores seen in the Chilean sample was only found in that half of the GHQ that asks about negative aspects of mental health.

A latent trait analysis of an inventory designed to detect symptoms of anxiety and depression using an elderly community sample

An 18-item inventory designed by Goldberg et al. (1987) to detect symptoms of anxiety and depression was administered to an elderly general population sample. Latent trait analysis was used to assess the dimensionality of the inventory and the location and discriminatory ability of the symptoms. The items showed different patterns of discrimination in this group compared to the sample of general practice attenders on which the inventory was developed. Overall, the items did define two correlated dimensions of anxiety and depression. In addition, a third dimension of sleep disturbance was detected. Both individual scales and the total symptom scores were sensitive and relatively specific detectors of depressive disorders assessed according to ICD-10 and DSM-III-R criteria. The retention of sleep items on their original scales did not affect the sensitivity of the scales to detect depressive disorders. A two-step administration procedure suggested for use in the administration of the scales was investigated but found to be sensitive to differences between the current sample and the sample on which the inventory was developed. This symptom inventory can be recommended for use in epidemiological investigations as a brief, valid and acceptable method of detecting elevated levels of anxiety and depression in elderly persons.

The generalizability of the dependence syndrome across substances: an examination of some properties of the proposed DSM-IV dependence criteria

DSM-III-R and proposed DSM-IV schemes for the diagnosis of psychoactive substance use disorders are based largely on the dependence syndrome concept. However, there is an absence of empirical support for the generalizability of the dependence syndrome across substances. This study examines how consistently proposed DSM-IV dependence criteria function to measure dependence across seven substances: alcohol, cannabis, cocaine, stimulants, hallucinogens, sedatives and opiates. Using structured research diagnostic interviews, dependence diagnoses were determined for 295 American subjects in treatment for alcohol/drug problems. Several factor analytic techniques were used to assess whether criteria formed single dimensions and how consistently individual criteria measured dependence across substances. The ability and consistency of criteria to measure a continuum of severity across substances were also assessed. Only subjects who used the substance at least six times were entered in the analyses. Overall, results provide strong support for the DSM approach for alcohol, cannabis, cocaine, stimulants, sedatives and opiates, but not for hallucinogens. Results indicate that a single strong factor adequately described the criteria for these six substances. All criteria loaded strongly and uniformly on single factors indicating that all were good measures of dependence. Criteria provided a dimensional measure of severity based on several indices for these substances. In addition, four criteria provided relatively stable indicators of high or low severity across these substances. Results did not support the use of dependence criteria for hallucinogens as these criteria did not form a single factor. Results suggest that very few hallucinogen users experience an inability to cut down or control use, a key indicator of loss of control.

Latent trait modelling of symptoms of schizophrenia

Latent trait models were fitted to data for 149 schizophrenic or schizophreniform inpatients rated on the Scale for the Assessment of Positive Symptoms (SAPS) and the Scale for the Assessment of Negative Symptoms (SANS) using the Rasch Extended Logistic Model. It was found that a reduction in the numbers of rating categories, from six to three or four led to consistencies in response while deletion of several items led to consistent scales of symptoms that accorded with an item response characterization. Behaviours included in the final scales varied in the numbers of categories, and in the range of symptom level covered by a category. Relationships between scores representing symptoms were found to be modelled better by a factor structure that included a third overlapping 'cognitive' factor in addition to the now traditional positive and negative factors, than by the original positive and negative factors alone.

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.

The Chinese General Health Questionnaire in a psychiatric setting: the development of the Chinese scaled version

The development of the Chinese scaled GHQ-20 was based on the factor analysis of the item responses of 150 general psychiatric patients with the Chinese GHQ-60. The five subscales assessing the symptom areas of dysphoric functioning, health concern, anxious coping, sleep problems, and suicidal ideas were found to discriminate significantly the sample of patients from another sample of 653 nonpatients. As a scale, the Chinese GHQ-20 was also found to be comparable to the full Chinese GHQ-60 in screening for individuals who might benefit from psychiatric consultation. The implications for its use as a refined instrument to monitor therapeutic change of patients in different symptom areas are discussed.

The factor structure of the Turkish version of the General Health Questionnaire

The 60-item General Health Questionnaire (GHQ) was translated into Turkish and administered to a community sample of 437 Turkish-speaking immigrants resident in Melbourne, Australia. The factor structures of the 60-item and 28-item versions of the GHQ were examined to determine the cross-cultural validity of the four subscales of the 28-item GHQ "anxiety/insomnia", "social dysfunction", "severe depression" and "somatic complaints". Four-factor principal components analyses yielded factors which corresponded to similar underlying traits, but the pattern of symptom loadings differed in several ways. Insomnia was less closely associated with anxiety, and general illness ratings such as "not feeling perfectly well" were not uniquely associated with somatic symptoms. The "anxiety/insomnia" and "severe depression" factors overlapped, with many symptoms partially correlated with both factors. These factors appeared to be due to variations in the frequency of occurrence of these symptoms rather than a qualitative distinction between anxiety and depression.

Dimensions of neurosis

Factor analyses of the General Health Questionnaire have attempted to interpret the factors as measuring anxiety, depression and social functioning. Data from two large community surveys were used to conduct an unrotated principal components analysis of the 30-item General Health Questionnaire. A general factor, indicating overall severity of psychiatric disorder, accounted for around 30% of the variance. The next most important factor, accounting for about 8% of the variance, was bipolar with the positive ('less than usual') items of the General Health Questionnaire having positive coefficients and the negative ('more than usual') items having negative coefficients. It is suggested that the concepts of positive and negative mental health derive empirical support from the results and may prove to be a useful classification of dimensions of mental health in the community.

Determinants of somatization in primary care

This study explores possible determinants of somatization in primary care. Hypotheses were tested on samples of 'somatizers', 'psychologizers' and controls recruited by epidemiological procedures. Although 'somatizers' were found to be similar to 'psychologizers' in many respects, they were (i) less depressed; (ii) reported lower levels of social dissatisfaction, social stress and less dependence on their relatives; (iii) more likely to have an unsympathetic attitude towards mental illness and less likely to consult a doctor about psychological symptoms, and (iv) more likely to have received medical in-patient care as an adult before they had consulted their doctor with their current illness. These findings are discussed in the context of previous research.

Depression sub-typing: unitary, binary or arbitrary?

The strongest statistical support for the binary view of depression has been provided by factor (principal components) analytic studies which delineate a bipolar factor with features interpreted as reflecting "endogenous depression" and "neurotic depression" at opposing poles. We review the seminal studies to suggest instead that the bipolar factor has generally polarised depression and anxiety, and that no such entity or symptom complex of "neurotic depression" has been isolated. Instead "neurotic depression" has been defined principally by features of anxiety and personality style. We argue that the suggested entity is, in fact, a pseudo-entity, being no more than a residual group of non-depressive features without any significant intrinsic depressive characteristics. We support our interpretation by showing comparable solutions in published studies of depressives alone, contrasted with separate analyses of anxious and depressed patients. We also report two studies in which the "neurotic depressive" pole is made to appear and disappear by the inclusion and exclusion of anxiety items. As factor analytic studies have defined the "residual" pole so variably, we argue that some features held to distinguish neurotic depression are of no utility and that such a diagnosis is meaningless. We suggest that the clinician should not proceed (after excluding endogenous depression) to conclude that the default option is necessarily an entity "neurotic depression" and that instead a heterogeneous group of options (e.g. anxiety, personality disorder) require review. If the "neurotic depressive" type of the multivariate analytic studies is a pseudo-entity, then a modified unitary view of depression may be valid.

Dimensional approach to obsessive-compulsive disorder in childhood and adolescence

Based on a review of the charts of child and adolescent psychiatric patients, a group of patients with obsessive-compulsive symptoms and a control group were identified. Obsessive-compulsive symptoms described in the charts were classified according to form using the checklist of Khanna et al. By use of latent trait models, a 6-item obsessive-compulsive rating scale was established revealing a dimensional rather than a categorical distribution of obsessive-compulsive symptoms. The correlation with other clinical features was investigated.

The NH & MRC Social Psychiatry Research Unit, at the Australian National University, Canberra, 1975-90

In 1974, the National Health and Medical Research Council (NH & MRC) in Australia reviewed what initiatives might be undertaken to promote medical research relevant to the needs of the population. It noted that Australia had contributed with distinction in some areas, such as the neurosciences and immunology, whereas fields such as epidemiology and psychiatry were much less developed scientifically. As the principal source of funding for medical research, the NH & MRC had hitherto supported projects, individuals and a small number of institutions (e.g. the Walter and Eliza Hall, the Florey and the Baker Institutes). The initiative adopted in 1974, as an additional commitment, was to establish some research units in areas of major relevance for public health. These were intended to become centres of excellence in fields where more expertise was needed at a national level.

The Interval General Health Questionnaire and its relationship with psychiatric disorder in community and general practice samples

This paper contrasts the usual scoring of the GHQ-30 with scoring based upon a criterion approach applied to only 12 of the GHQ items that takes account of the 'diagnostic significance' of the chosen items. Repeat assessments with this shorter form of the questionnaire (the I-GHQ), coupled with procedures to assess change, can be used to provide summary measures of symptom course in those research situations where available interview time precludes the use of more formal assessment methods.

An evaluation of the psychoactive substance dependence syndrome in its application to opiate users

The Psychoactive Substance Abuse and Dependence (PSDA) section of the revised, 3rd edition of the American Psychiatric Association's Diagnostic and Statistical Manual for Mental Disorders (DSM-III-R) bears a close correspondence to the conceptualization of a dependence syndrome as posited by Edwards, Arif & Hodgson (1981). Further, DSM-III-R represents a major shift in psychodiagnostics. The present study investigated hypothesized properties of the dependence syndrome and examined the characteristics of DSM-III-R criteria with a sample of opiate users. Findings indicated general support for the unidimensional postulate of the dependence syndrome but did not support the weighting of syndrome elements within DSM-III-R. The implications of such findings for the proposed DSM-IV are discussed.

The validity of diagnostic systems for common mental disorders: a comparison between the ID-CATEGO and the DSM-III systems

It is argued that latent trait analysis provides a way of examining the construct validity of diagnostic concepts which are used to categorize common mental illnesses. The present study adds two additional aspects of validity using multiple discriminant analysis applied to two widely used taxonomic systems. Scales of anxiety and depression derived from previous latent trait analyses are applied to individuals reaching criteria for 'caseness' on the ID-CATEGO system and the DSM-III system, both at initial diagnosis and six months later. The first multiple discriminant analysis is carried out on the initial scale scores, and the results are interpreted in terms of concurrent validity. The second analysis uses improvement scores on the two scales and relates to predictive validity. It is argued that the ID-CATEGO system provides a better classification for common mental illnesses than the DSM-III system, since it allows a better discrimination to be made between anxiety and depressive disorders.

The limitations of multivariate statistical methods in the mensuration of human misery

Multivariate statistical methods have been widely used in the analysis of the multiple symptom data which are routinely collected in psychiatric research on the classification of depressive illnesses. The most commonly used methods, those of factor analysis and discriminant function analysis, were introduced into research on the classification of depressive illness with unreasonably high expectations about what they could achieve. The failure to realize these expectations has produced scepticism in some quarters about the usefulness of multivariate methods in psychiatric research. When evaluated more circumspectly, multivariate statistical methods have made a contribution to our understanding of depressive illnesses, and they will continue to do so, if they are used with more reasonable expectations.

Dimensional approach to infantile autism in mentally retarded adults

By use of latent trait models, a 7-item autism rating scale was established, based on the Medical Research Council's Schedule of Handicaps, Behaviours and Skills. Data from a community-based sample of 302 mentally retarded adults were assigning good internal validity to the autism scale, and revealing a distribution of symptoms consistent with dimensional, rather than categorical, distribution of infantile autism. A total of 23 (7.6%) of the sample were earlier diagnosed as suffering from infantile autism. Using this clinical diagnosis as a reference, the autism rating scale, with a cut-off point at greater than or equal to 2, established diagnoses with a sensitivity of 0.83 and specificity of 0.89.

Measuring behaviour disorder in mental handicap

By use of latent-trait item analysis (the Rasch model), the internal validity of the behaviour section of the MRC-HBS schedule was analysed. The set of behavioural symptoms in this rating scale was found to measure a one-dimensional structure. Thus addition of item scores provides a meaningful index. Such a numerical measure for the degree of behaviour disorder facilitates comparison between different patients or groups, and furthermore, forms the basis for quantitative description of the outcome of therapeutic intervention. A diagnosis of behaviour disorder was established in 17.2% of 302 mentally-handicapped adults. However, some kind of deviant behaviour was found in 41%, and was correlated with origin and degree of retardation, epilepsy, and place of living. The distribution of the symptoms strongly indicates that organic brain damage is a major aetiological cause.