Measuring change with the General Health Questionnaire (GHQ). The problem of retest effects

Temporal Stability of the Personality Assessment Inventory: Investigating Potential Predictors

In this study, we explored the temporal stability of the Personality Assessment Inventory (PAI), which has not been comprehensively reexamined since it was first published. Our three specific aims were to determine the utility of PAI indicators of basic protocol validity (inconsistent responses [ICN] and highly unusual/unlikely responses [INF]) in identifying suspect responding; calculate the stability coefficients for each PAI scale and subscale across 3-, 6-, and 9-week spans; and explore whether profile stability across four measurements could be prospectively predicted. We administered the PAI to a sample of undergraduates (N = 579) at four separate timepoints. ICN and INF effectively identified likely attriters and inconsistent responders. All PAI full scales and subscales evidenced good test-retest reliability, with some small exceptions. Finally, all PAI clinical scales were correlated with profile instability although many of these correlations were no longer significant when controlling for mean clinical elevation of the profile. We interpreted these results as evidence for the utility of PAI validity scales, the temporal reliability of the PAI, and the role of psychopathology in response variability over time. We also discussed some preliminary evidence that this variability can be prospectively predicted, suggesting that it in part reflects substantive changes rather than random error variance.

Employment status, psychological needs, and mental health: Meta-analytic findings concerning the latent deprivation model

Marie Jahoda's latent deprivation model proposes that unemployed people have a worse mental health compared to employed people. This is because they suffer not only from a lack of the manifest function of employment (earning money), but also from a lack of five so-called latent functions of employment: Time structure, social contact, collective purpose (i.e., the sense of being useful to other people), status, and activity. In order to test the basic assumptions of this theory, a study based on meta-analytic methods was conducted. Results showed that employed people reported higher levels on all five latent functions, as well as on the manifest function, compared to unemployed people. They also report more latent functions than people who are out of the labor force (OLF). Moreover, OLF-people reported more manifest and latent functions than unemployed people. Specific analyses for three OLF-subgroups found retired people to be almost as deprived of the latent functions (but not the manifest function) as unemployed people, while students were more similar to employed people but still experienced some manifest and latent deprivation. For homemakers, the effect sizes pointed in the expected direction, but they were not significant. Thus, the proposition that employment is the best provider of the latent functions was generally endorsed, although homemakers need further scrutiny in future studies. All latent functions, as well as the manifest function, emerged as significant independent predictors of mental health, when the influence of the other manifest and latent functions was controlled. Together, the dimensions in the model explained 19% of variation in mental health.

Multicomponent Intervention Associated with Improved Emotional and Cognitive Outcomes of Marginalized Unemployed Youth of Latin America

Mass migration and people seeking political refuge are critical social issues facing Latin America. Ecuador has the largest population of recognized refugees in the region. Youths from a migration background have an increased risk of becoming NEET (Young people not in employment, education, or training). Such youths struggle more with mental health problems than non-NEET peers. Being a refugee, NEET further increases the risk of having mental health problems and may be linked to lower cognitive functioning, which could maintain exclusion and unemployment. This intervention study was performed with a group of young people of different nationalities who were refugees or belonged to other vulnerable groups attending a six-week employability-support intervention in Ecuador. In order to assess the impact of the intervention, a range of measures of executive cognitive function, mental health, and the potential for positive change were used. We found that post-intervention, the group reported significantly less psychological distress and better self-esteem, self-efficacy, and cognitive response inhibition than before the intervention. We conclude that multicomponent interventions may effectively improve the psychological functioning of vulnerable NEET groups in the Latin American context.

Parallel short forms for the assessment of activities of daily living in cardiovascular rehabilitation patients (PADL-cardio): development and validation

PURPOSE

To develop and validate parallel short forms for the assessment of activities of daily living in cardiac rehabilitation patients (PADL-cardio I & II).

METHOD

PADL-cardio I & II were developed based on a sample of 106 patients [mean age  =  57.6; standard deviation (SD) = 11.1; 72.6% males] using Rasch analysis and validated with a sample of 81 patients (mean age  =  59.1; SD  =  11.1; 88.9% males). All patients answered PADL-cardio and the Short Form 12 Health Survey.

RESULTS

Both versions of PADL-cardio are composed of 10 items. The fit to the Rasch model was given documented by a non-significant Item-trait interaction score (PADL-cardio I: χ² = 31.08, df  =  30, p  =  0.41; PADL-cardio II: χ² = 45.6, df  =  40, p  =  0.25). The two versions were free of differential item functioning. Person-separation reliability was 0.72/0.78 and unidimensionality was given. The two versions correlated with r = 0.98 and the correlation between PADL-cardio and the underlying item bank was 0.99 for both versions. Concurrent validity is indicated through correlations with the Short Form 12 Health Survey (r  = -0.37 to -0.40).

CONCLUSION

PADL-cardio provides a short and psychometrically sound option for the assessment of activities of daily living in cardiovascular rehabilitation patients. The two versions of PADL-cardio are equivalent. Hence, they can be used to reduce practice and retest effects in repeated measurement, facilitating the longitudinal assessment of activities of daily living. Implications for Rehabilitation New parallel test forms for the assessment of activities of daily living in cardiac rehabilitation (PADL-cardio I & PADL-cardio II) are available. PADL-cardio I & II consist of 10 items and are therefore especially timesaving. Concurrent validity is given through correlations with the Short Form Health Survey 12. Therapeutic success could be determined more precisely by the parallel forms reducing practice and retest effects.

The Effects of Repeated Administrations on Self-Report and Parent-Report Scales

The use of repeated measures to track the out-come of psychosocial interventions is becoming common practice in most organizations. One threat with this methodology is the retest artifact. This artifact is defined by a systematic change in scores for individuals who repeatedly complete an outcome instrument yet do not receive an) formal treatment. The retest artifact can call into question the validity of change associated with school-based treatment programs when this change is based upon repeated testing using outcome instruments. This study directly assessed the retest effect for both a youth and an adult psychosocial outcome measure. The frequency of retest administrations was systematically varied (weekly, biweekly, monthly, pre/post testing) over a 9-week period using a youth parent-report and an adult self-report Outcome Questionnaire. Results indicate a statistically significant improvement in functioning at the time of the second administration regardless of the lapse in time for the adult measure and a steady improvement in functioning for the parent-report youth measure. Implications of this finding for evaluating psychosocial interventions are discussed.

Development and validation of parallel short forms PaSA-cardio for the assessment of general anxiety in cardiovascular rehabilitation patients using Rasch analysis

OBJECTIVE

To develop and validate parallel short forms for the assessment of general anxiety in cardiovascular rehabilitation patients, that facilitate repeated measurement over time without contamination from residual practice effect variance.

DESIGN

Development of the parallel short forms using Rasch analysis. Validation study.

SETTING

Cardiac rehabilitation centres in Germany.

SUBJECTS

Cardiovascular rehabilitation patients.

INTERVENTIONS

Not applicable.

MAIN MEASURES

Parallel short forms PaSA-cardio, Hospital Anxiety and Depression Scale, Short Form Health Survey SF-12 and Structured Clinical Interview for DSM-IV Axis I Disorders.

RESULTS

Each version of the parallel short forms (PaSA-cardio-I and PaSA-cardio-II) comprises ten items. The two forms fitted to the Rasch model with a non-significant item-trait interaction (PaSA-cardio-I: chi-square = 39.49, degrees of freedom = 30, probability = 0.12; PaSA-cardio-II: chi-square = 26.56, degrees of freedom = 30, probability = 0.65). Person-separation reliability was 0.75/0.76. Unidimensionality could be verified. Correlation between the two models was 0.94 and 0.95, and correlations with the underlying item bank were 0.95 and 0.93. Validity could be confirmed. The area under the curve was between 0.88 and 0.97 for PaSA-cardio-I and between 0.92 and 0.95 for PaSA-cardio-II.

CONCLUSIONS

Assessment of general anxiety in cardiovascular rehabilitation patients with the PaSA-cardio was valid, economical and accurate. The two forms of the PaSA-cardio are equivalent and allow retest without contamination from residual practice effect variance.

Measurement Invariance of the Experiences in Close Relationships Questionnaire across Different Populations

The measurement invariance of the Experiences in Close Relationships (ECR) questionnaire was investigated across four samples of non-clinical subjects (N = 222), dermatological patients (N = 458), psychiatric inpatients (N = 156), and patients with drug-resistant epilepsy (N = 101). The results provided evidence of configural, metric, and scalar invariance across groups. Overall, our findings provide further support to the reliability and validity of the ECR.

Prospective cohort study of mental health during imprisonment

BACKGROUND

Mental illness is common among prisoners, but little evidence exists regarding changes in symptoms in custody over time.

AIMS

To investigate the prevalence and predictors of psychiatric symptoms among prisoners during early custody.

METHOD

In a prospective cohort study, 3079 prisoners were screened for mental illness within 3 days of reception. To establish baseline diagnoses and symptoms, 980 prisoners were interviewed; all remaining in custody were followed up 1 month and 2 months later.

RESULTS

Symptom prevalence was highest during the first week of custody. Prevalence showed a linear decline among men and convicted prisoners, but not women or remand prisoners. It decreased among prisoners with depression, but not among prisoners with other mental illnesses.

CONCLUSIONS

Overall, imprisonment did not exacerbate psychiatric symptoms, although differences in group responses were observed. Continued discussion regarding non-custodial alternatives for vulnerable groups and increased support for all during early custody are recommended.

Prospective evaluation of insomnia in prison using the Pittsburgh Sleep Quality Index: Which are the factors predicting insomnia?

Objectives. To measure the subjective sleep quality of prisoners complaining of insomnia and to compare it to the subjective sleep quality of prisoners who report "good sleep", in order to determine factors that predict insomnia and severity of insomnia. Participants and methods. PSQI and GHQ scores and patient history were obtained for 86 randomly chosen remanded prisoners complaining of insomnia and 61 randomly chosen prisoners who did not complain of insomnia. Results. PSQI total and component scores were significantly different between insomniac and not insomniac prisoners, except for C7 (daytime dysfunction). A history of sleeping problems before prison (odds ratio: 13.3), the subjective experience of having had stressful events during the past week (odds ratio: 8.5), being separated or divorced (odds ratio: 8.8), GHQ >10 (odds ratio: 8.8), a history of psychiatric problems (odds ratio: 8.3) and the consumption of opiates (odds ratio: 7.9), and to a lesser degree "no sports in prison" and stress related to judicial, familial and prison problems, were predictors of insomnia. We did not find any evidence in this study that work or consumption of caffeine in prison were factors that distinguished good sleepers from insomnia patients. Conclusions. Our study helps prison physicians to identify prisoners at risk for insomnia and to obtain some orientation for treatment decisions. Psychological support to reduce context related stress should be routinely offered to insomniac prisoners.

No significant retest effects in oral health-related quality of life assessment using the Oral Health Impact Profile

OBJECTIVE

To investigate retest effects in assessment of the Oral Health Impact Profile (OHIP) in patients requiring prosthodontic treatment.

MATERIAL AND METHODS

The German 49-item OHIP was administered in 21 patients requiring prosthodontic treatment on three pretreatment occasions: twice (t(1) and t(2)) on day 1, 1-2 h apart, and on day 2 (t(3)), 3-78 days later. Differences in reliability coefficients, OHIP summary scores, and item responses, and the influence of time on score differences were analyzed.

RESULTS

Some evidence of retest effects in oral health-related quality of life (OHRQoL) was found (the difference in reliability coefficients between t(1)-t(2) and t(2)-t(3) was 0.08; p=0.02).

CONCLUSIONS

The findings support repeated OHRQoL assessment over short periods of time when perceived oral health is rapidly changing. This is important because many oral conditions have acute symptoms and treatments have immediate OHRQoL effects that need to be measured.

Measuring quality of life in patients with breast cancer: a systematic review of reliable and valid instruments available in Japan

BACKGROUND

Little is known about the availability and psychometric properties of instruments to measure quality of life (QOL) in clinical research on Japanese patients with breast cancer. The purpose of this systematic review is to find reliable and valid instruments available in Japan, and to summarize their characteristics.

METHODS

Instruments available in Japan were found through a systematic search of the literature. Each instrument identified was evaluated for item development, reliability, validity, interpretability and utility.

RESULTS

Six questionnaires to measure health-related QOL (the QOL-ACD, the EORTC QLQ-C30, the EORTC QLQ-BR23, the FACT-B, the SF-36, the WHO/QOL-26) and five scales to quantify the psychological burden (the STAI, the POMS, the SDS, the HADS, the GHQ), for which reliability and validity have been documented, are available in Japanese. All instruments were developed in foreign countries except for the QOL-ACD. Two of the QOL questionnaires were specific to breast cancer (the EORTC QLQ-BR23, the FACT-B). Though the measurements can be interpreted in some manner, the meaning of change scores over time has been documented for only three instruments (the EORTC QLQ-C30, the FACT-B, and the GHQ).

CONCLUSIONS

The review provides grounds for designing and implementing quantitative research on QOL of breast cancer patients in Japan. Methodological challenges, however, continue, particularly for validating instruments with regard to various study populations of Japanese people and demonstrating the clinical importance of change scores.

Veränderungsmessung in der Psychotherapie mit dem Ergebnisfragebogen (EB-45)

Zusammenfassung: Für ein kontinuierliches Monitoring psychotherapeutischer Veränderungen sind kurze, valide und veränderungssensitive Instrumente erforderlich. Die Eignung des Ergebnisfragebogens (EB-45) zu diesem Zweck wurde in dieser Studie untersucht. Anhand einer Bevölkerungsstichprobe (n = 232), einer Stichprobe von Bahnpendlern (n = 615), einer Stichprobe ambulanter Psychotherapiepatienten (n = 671) und einer Stichprobe stationärer Psychotherapiepatienten (n = 960) wurden Aspekte der konvergenten und diskriminativen Validität sowie die Veränderungssensitivität des Ergebnisfragebogens untersucht. Weiter wurden für die externen Kriterien “Suizidgedanken” und “Arbeitsfähigkeit” Sensitivitäts-Spezifitätsanalysen durchgeführt. Für die beiden klinischen Stichproben zeigte sich auf einzelnen Skalen des EB-45 eine Abhängigkeit vom Geschlecht, Alter und der klinischen Diagnose. Mäßige Konkordanzen (κ ~0.4) ergaben sich mit der Symptom-Checkliste und deren Kurzform, dem Brief Symptom Inventory, geringe Konkordanzen mit dem Fragebogen zur Lebenszufriedenheit und dem Gießener Beschwerdebogen. Die Diskrimination zwischen den klinischen und nicht-klinischen Stichproben war auf den einzelnen Subskalen und der Gesamtskala gut. Eine Verringerung der Skalenwerte bei wiederholter Messung im Verlauf der Therapie konnte bei beiden klinischen Stichproben gezeigt werden. Die Ergebnisse bestätigen die Eignung des Ergebnisfragebogens für ein kontinuierliches Monitoring psychotherapeutischer Veränderungen in der klinischen Praxis im Rahmen der Qualitätssicherung.

Changes in waiting-list patients over time: data on some commonly-used measures. Beware!

When samples of psychiatric patients are assessed on 2 occasions on state and trait measures, without any formal intervention having yet taken place, a mean change in scores towards less psychopathology is often observed. This re-test effect, which is a potential threat to the validity of longitudinal studies, has not been given serious attention by researchers using repeated administrations. The present study addressed this issue by examining re-test effects in 2 independent clinical samples with the Symptom Check List-90-Revised and other widely-used state and trait measures. Time intervals between intake and re-take were 3 months and 11-350 days (M = 93 days, median = 81 days). Results indicated that the re-test effect occurred for most of the measures. Where significant testing effects were observed, 54-72% of the patients had time 2 scores that were lower than the average time I scores. The importance and practical and scientific implications of the findings are discussed. Eleven hypotheses concerning the nature of the re-test effect are offered, including mood-congruent associative processing, natural coping mechanisms, self-monitoring hypothesis and response-shift.

Assessing psychological distress in patients with skin diseases: reliability, validity and factor structure of the GHQ-12

OBJECTIVE

We studied reliability, validity and factor structure of the 12-item General Health Questionnaire (GHQ-12) in dermatological patients.

METHOD

Subjects attending a dermatological outpatient clinic were administered the GHQ-12 and the Skindex-29. A random subsample was mailed another copy of the questionnaires to be completed within 1 week.

RESULTS

A total of 2,579 subjects completed the GHQ-12. The internal consistency of the GHQ-12 was high (Cronbach's alpha = 0.88). Test-retest reliability on 137 subjects was also fairly satisfactory (intraclass correlation coefficient = 0.72), notwithstanding a re-test artefact probably related to reassurance from the physician. Evidence of construct validity was provided by the pattern of correlation between scores on the GHQ-12 and on the scales of Skindex-29, which were as hypothesized. Also, GHQ-12 scores of patients in different clinical groups varied as hypothesized, with patients affected by inflammatory skin diseases scoring significantly higher than patients with isolated skin lesions. We obtained both a two-factor and a three-factor solution: the latter seemed theoretically more appropriate, although the first may have practical advantages. A 'social dysfunction' factor emerged in both analyses, while the other items loaded on a 'general dysphoria' factor or on two separate factors interpreted as 'anxiety' and 'loss of self-esteem'.

CONCLUSIONS

The GHQ-12 is a reliable and valid instrument with a factor structure that is quite stable across different samples as well as across diverse cultures. Its brevity and easiness of completion make it a useful tool to assess psychological distress in dermatological patients.

Outcomes of Depression International Network (ODIN). Background, methods and field trials. ODIN Group

BACKGROUND

ODIN aims (a) to provide data on the prevalence, risk factors and outcome of depressive disorders in rural and urban settings within the European Union (EU) based on an epidemiological sampling frame; and (b) to assess the impact of two psychological interventions on the outcome of depression and on service utilisation and costs.

METHOD

Five centres across the EU are participating in ODIN. The centres are linked electronically and members meet regularly for training and strategic reviews. Urban and rural areas have been identified in each centre. The sampling frame is of adults aged 18-64, identified via primary care databases or electoral registers. Potential cases of depressive disorders are identified using the Beck Depression Inventory SCAN II and other validated measures are used to assign caseness against DSM-IV and ICD-10 criteria; assess comorbidity, disability, genetic/familial susceptibility, psychosocial stressors, personality traits and cognitive factors; and utilisation of local health care services. A randomised controlled trial of individual problem-solving treatment and a group educational programme is undertaken for respondents identified as cases of depressive disorder individuals are followed-up at six and 12 months.

RESULTS AND CONCLUSIONS

ODIN has already stimulated the development of an effective international research partnership.

Counselling in primary care: a randomised controlled trial

A randomised controlled trial of the effectiveness of counselling (In-House Counselled (IHC) group) compared to routine advice from General Practitioners (Routine Treatment (RT) group) was conducted with three practices and a total of 188 patients in East Sussex. Changes in interpersonal problems using the 32 item Inventory of Interpersonal Problems, symptoms of psychological distress using the Symptom Index and self esteem using Repertory Grids were compared between groups at four months and after a further four month follow-up. A questionnaire monitoring patient service satisfaction was given to those who had received In-House counselling. The number of counselling sessions, early withdrawals and refusers was also monitored. In order to facilitate interprofessional communication, the three counsellors and a GP representative from each practice met monthly for an Action Learning group, led by an external facilitator to provide a forum to discuss working practices. The group met six times for two and a half hours. An audit of the participants' medical notes was carried out at the end of the study to establish any changes in subsequent use of medical services and prescribing patterns. The results show that patients within both groups improved considerably, in line with similar studies. The IHC group was significantly less likely to be referred out to mental health services. However, there was no statistical difference between the groups on any of the measures or in changes in subsequent service use or prescribing patterns. This may have been a result of Action Learning Group producing more psychologically minded GPs. Patients in the IHC group were overwhelmingly in favour of counselling and stated that it had helped them with a variety of psychological problems.

Short-term outcome following referral to a psychiatric emergency service

The study described here is part of an evaluation of a pilot project concerning the implementation of three psychiatric crisis units in general hospitals in Belgium. The purpose was to evaluate the short-term outcome of a multidisciplinary crisis intervention for psychiatric patients referred to the emergency department. Patients were assessed with the General Health Questionnaire (GHQ-28) at the time of referral to the emergency department and again 1 month later. Patients referred for a psychiatric crisis intervention were compared with patients receiving short-term psychiatric inpatient treatment in another hospital. Patients referred to the emergency department showed a considerable degree of psychiatric disturbance. The General Health Questionnaire appeared to be a good measure for assessment of the "state" aspect of a psychiatric disturbance. The state of distress was significantly reduced one month after referral in both treatment conditions. Nevertheless, an important proportion of patients remained in a state of considerable distress. The results indicate that a short hospital-based crisis intervention approach is comparable with more traditional acute inpatient treatment. However, in the case of more severely distressed patients it may be insufficient. Several limitations of this study are also discussed (risk of overestimation of improvement, influence of time or pre-existing differences).

Empirical psychological modeling of chest pain: a comparative study

In this study the psychological profiles of 67 patients with noncardiac chest pain (NCA) and 47 patients with coronary artery disease (CAD) were analyzed to construct an empirical-psychological model that would be able to discriminate these two groups. All patients were suffering from chest pain at the time of referral by their general practitioner. The noncardiac patients were significantly younger, more often female, single, and nonsmokers. The two groups differed significantly on anxiety, somatization, obsessive compulsive behavior, psychoneuroticism, and hyperventilation. Logistic regression analysis on the variables jointly, showed that age, gender, anxiety, and hyperventilation contributed significantly to the model for discriminating between the two groups. Crossvalidation showed that the constructed model was stable.

A critical review of dimension-specific measures of health-related quality of life in cross-cultural research

This article reviews six dimension-specific health-related quality of life (HRQL) measures which have been used cross-culturally. The instruments reviewed are: the Beck Depression Inventory (BDI); the McGill Pain Questionnaire (MPQ); the Center for Epidemiologic Studies--Depression (CES-D); the Zung Self-Rating Depression Scale (SDS); the General Health Questionnaire (GHQ); and the Psychological General Well-Being Index (PGWB). These instruments primarily represent the psychological or emotional dimension of HRQL, and are scales that were developed and validated in the USA, Canada or the UK. The review of specific studies for each of the six instruments was not meant to be exhaustive, but rather to give an indication of the ways in which the instruments have been assessed or used in various countries. The focus throughout this article is on the psychometric properties (reliability, validity and responsiveness) of these scales in different cultures, as well as the processes used to translate the instruments from English into another language. Implications of the results of this review for cross-cultural use of dimension-specific HRQL instruments are drawn.

Health consequences of employment and unemployment: longitudinal evidence for young men and women

This study examines the impact of employment and unemployment on psychological health and well-being, as measured by the General Health Questionnaire (GHQ). Using longitudinal data, it traces health changes over time for a variety of groups and through a variety of labour market experiences: during sustained employment and unemployment, in the transition from school to employment and unemployment, and as people move between employment and unemployment and between satisfying and unsatisfying jobs. The results show that employed people report significantly lower levels of health disorder than students and the unemployed. These differences are largely unaffected by demographic attributes, living arrangements, socioeconomic status or immediate labour market experiences, and can be attributed to employment status itself rather than predisposing health differences. However, the health consequences of employment and unemployment are directly contingent upon quality of work. As a result, the highest levels of health risk are found amongst dissatisfied workers and the lowest levels amongst satisfied workers. In between these two extremes lie employed people neither satisfied nor dissatisfied with their jobs, unemployed people irrespective of duration, and students. These results indicate that what happens in the workplace has even more impact on a person's health than success or failure in finding a job and keeping it.

DSM-III mental disorders in general medical sector: a follow-up and incidence study over a two-year period

In three general medical settings (general practice, hospital medical wards and emergency rooms) about 20% of the adult attenders had a DSM-III mental disorder, mainly in the area of affective and anxious disorders. Some of these disorders were quite severe. Of those cases reassessed 1 year and 2 years after the first interview, less than a quarter reached a "no-diagnosis status". The chronicity of most cases dependent on the interplay not only of either relapse or duration of the main disorder but also of comorbidity and incidence of new disorders. A high incidence of more transient disorders in subjects who were well at first assessment was also found.

Multidimensional properties of the General Health Questionnaire

This analysis re-examines the factor structure of the 12-item GHQ for a large Australian sample of young people. It shows that oblique factor rotation better approaches the criterion of simple structure, and allows separate components of the GHQ to be identified and measured using factor scores. When the performance of these separate factors is compared with composite Likert scores, it is found that they do not behave uniformly in response to outside variables, both in cross-sectional and longitudinal analyses. These results suggest that there are advantages to be gained by using the multidimensional properties of the GHQ as well as a single severity score. The additional information this yields can provide new insights into the nature of psychiatric impairment within and between samples.

Concurrent validity of GHQ-28 and PSE as measures of change

Substantial cross-sectional correlations have been reported between the GHQ and PSE (and CIS) total scores. Although necessary, this is not a sufficient condition for assuming good validity of the GHQ as a severity measure in longitudinal and health care evaluation studies. For this purpose the GHQ should also accurately reflect changes in severity over time. To examine their concurrent validity, GHQ and PSE scores were compared, in a three-wave longitudinal study, among 175 new psychiatric out-patients. Using a longitudinal structural equation model that takes measurement error into account, the strength of both the cross-sectional and longitudinal relationship between GHQ and PSE were estimated. The GHQ performed remarkably well; changes in severity as defined by PSE-ID and PSE total score were clearly reflected by changes in GHQ scores. The revised scoring method of the GHQ proposed by Goodchild and Duncan-Jones did not yield superior results.