FOR ANXIETY DISORDERS. MEASURED USING THE HOSPITALQUESTIONNAIRE (4...

15] for anxiety disorders. Measured using the HospitalQuestionnaire (4DSQ) developed to differentiate distressAnxiety and depression (HAD) Scale in the Netherlands,from two psychiatric illnesses (depression and anxiety)the one-year prevalence of depression and anxiety in theand from somatization. Together, these four symptomworking population is 7.1% and 8.2% for males and 6.2%and 10% for females [16].clusters account for the majority of the mental healthproblems in primary health care. According to Terluin,distress is the psychological squeal of strain caused byFor the recognition, prevention and treatment of mentalhealth problems, the underestimation of distress can beunsuccessfully coping with a stressor. Stressors can be thecommon cause for distress and depression or anxiety.regarded as unfavorable for several reasons.Under less favorable conditions, distress might be a pre-cursor for more serious psychiatric disorders. On the otherThe first reason is the imminent concomitance of distressand sickness absence. Distress as a main cause of sicknesshand, psychiatric illness can act as a stressor that aggra-vates strain and distress. That may explain why individu-absence can be labeled under 'adjustment disorders' fol-lowing the DSM IV classification [17]. In the Netherlands,als with depression and anxiety in many cases also exhibitdistress.approximately 30% of the employees who visit the occu-pational physician for sickness absence report mentalThe 4DSQ, a 50-item self-report questionnaire, has beenhealth problems [18] including common mental healthdeveloped for clinical and non-clinical populations withproblems like adjustment disorders, but also psychiatricdisorders such as anxiety and depressive disorders. Thepsychological complaints and has been validated in pri-majority of the employees absent for mental health rea-mary health care [31,32] and in occupational health care[7,29]. The four scales of the DSQ are internally consist-sons can be classified as having an adjustment disorder[19]. Nieuwenhuijsen et al. [20] demonstrated a percent-ent, with Cronbach's alphas ranging from .79 to .90. Thesubscale distress, the focus of this study, is associated withage of 59% in employees absent for mental health prob-job stressors and indicators of strain, which supports thelems. Prevention of – at least a part – of sickness absencethrough a reduction in high levels of distress is a challengeutility of the questionnaire for screening purposes. Sinceworking employees with a high rate of distress as a conse-for the occupational health professional and can be a ben-efit for employees and companies.quence of job stressors and strain, run a high risk of sick-ness absence, a cutoff point for distress can be helpful forthe identification – and maybe even monitoring – ofA second reason for a focus on distress is the high concur-employees at risk for sickness absence and for the selec-rence with anxiety and mood disorders [21-23], which intion of cases for support like stress management programsturn show a high degree of intercorrelation [24-26]. Dis-or treatment in order to prevent absenteeism.tress symptoms such as concentration problems, irritabil-ity and fatigue are common to both anxiety andThe use of a cutoff point [4,33] for inclusion in preventivedepression in the DSM IV diagnostic criteria [27].stress management programs has remarkably not beenA third reason for discerning distress is the implication forreported until now. Because of the size of the problem,reducing sickness absenteeism by applying interventionstreatment and guidance. The reduction of distress presum-ably has its own typical approach. In the past, 20% ofto reduce work-related stress is of great importance. Indi-patients reporting themselves sick with an adjustment dis-vidually focused programs aim to increase the employee'sorder due to distress did not return to work within onemental resilience [34], usually referred to as a stress man-agement training [35,36]. And although the term stressyear [28]. Van der Klink et al [17] demonstrated that anmanagement training may suggest a rather uniform set ofactivating intervention based on the principles of timeintervention strategies, it usually refers to a mixture ofcontingency and cognitive behavioral treatment was suc-cessful in reducing sick leave duration by 25–30% com-treatment techniques. To a certain extent these (work-related) stress interventions claim to reduce psychologicalMethodcomplaints [37-40], to increase individual quality of lifeSampleTwo samples of employees with presupposed differences[41-43], to reduce stress-related health care costs [44,45].and to reduce absenteeism [46-48]. Although such effectsin distress were used. Both employee samples worked in alarge telecom company in the Netherlands and wereof stress management interventions have been shown, theeffects on absenteeism are still subject to debate. Differ-approached by the company's Department of Occupa-tional Health.ences between the intervention programs as well as meth-odological differences between these studies – such as thelack of a control group, inadequate collection of data orThe first sample, representing the 'healthy workingemployees', were participants in an occupational healthdifferent study designs with different measures – aresurvey with a focus on occupational stress. Questionnairesbrought forward to explain these inconsistent results.were mailed to all employees of the company (N = 7,522).However, another important cause may be the lack of aThe questionnaires were completed by 3,852 employeescutoff point in most studies for selecting participants [34].It is a lamentable omission for current stress management(response rate 51%). The sample consisted mainly of men(91%), medium- or highly-educated employees (74%),programs and guidelines that we miss clear criteria for theand had a mean age of 43.9 years. At the moment at whichreferral of employees with a certain level of distress tothe employees filled in the questionnaire, 247 (6.4%)occupational health physicians or psychosocial careteams.were on sick leave; these were excluded from the sampleresulting in 3605 employees.In addition, the distress dimension of the 4DSQ and a cut-The second sample consisted of 280 employees who hadoff point can be used as a valid estimator for the preva-been on sick leave for at least two weeks and, in accord-lence of distress across demographic and occupationalsubgroups [29]. A well-founded cutoff point can be usedance with the procedure, were referred to their occupa-tional physician. To be included in the sample, employeesas a criterion to classify cases for research purposes. "Cut-off scores are used in a wide variety of settings to divide ahad to be on their first sickness leave because of stress atscore scale or other set of data into two or more categories,work or a stress-related disorder due to a recent identifia-with inferences made or actions taken on the basis of thisble psychosocial stressor at work. The employees had toclassification" as has been stated by Dwyer [49]. Thedemonstrate at least eight out of 16 distress symptoms ofchoice of such a categorization represented by one orthe 4DSQ scale (at level one or higher) that represent themore cutoff points, however, is a result of judgments. Onemain symptom categories of the DSM IV adjustment dis-order [17]. Exclusion criteria were a psychiatric diagnosisof the unwanted side-effects of this process of decisionsuch as an anxiety disorder or a depressive disorder andmaking may be the emergence of different cutoff points inphysical co-morbidity.different studies [50]. This makes comparisons acrossstudies extremely difficult or even impossible.MeasureConsequently, clarification of the process of decisionThe 4DSQ is a 50-item self-report questionnaire [7] thatidentifies four symptom dimensions: distress (16 items,making is indispensable. In this article we thereforedescribe explicitly the process by which we selected ane.g. "Did you feel easily irritated?"), depression (6 items,optimal cutoff score of a risk factor that gives the best sep-e.g. "Did you feel that you can't enjoy anymore?"), anxiety(12 items, e.g. "Were you afraid of anything when therearation between employees with high distress levelswas really no need for you to be afraid?") and somatiza-related to the risk for subsequent sickness absence due totion (16 items, e.g. "Did you suffer from excessive perspi-psychological complaints on the one hand, and employ-ees who are not at risk on the other. By doing this, theration?"). Participants are instructed to indicate how theyresults of this study can be compared with the results offelt during the previous week, and the items are scored ona 5-point Likert scale (from 0 = 'No' to 4 = 'Very often'). Inother studies.the application of the 4DSQ, to reduce the influence ofaggravating response tendencies on sum scores, all itemIn conclusion, the objective of the present study is toestablish an optimal cutoff point for distress measuredscores of '3' and '4' are recoded into a score of '2' beforeusing the corresponding scale of the 4DSQ, with the pre-calculating sum-scores per dimension. Thus, symptomsdiction of sickness absence as a criterion. The cutoff pointare rated as absent ('no': 0 points), doubtfully presentshould result in a measure that can be used as a credible('sometimes': 1 point) or present at a clinically significantselection instrument for stress management programs orlevel ('regularly/often/very often': 2 points). The factorother interventions to prevent sickness absence due toscore for distress ranges from 0–32, where a high scorespsychological complaints in occupational health practiceindicates substantial distress. The value for Cronbach'sand in future studies on distress and mental disorders.alpha for distress is .90.Analysesincrease of the positive predictive value, while the consid-Distress scores (means, standard deviations and percentileerable reduction in sensitivity decreases the negative pre-dictive value only marginally. Beforehand, we had madescores) were calculated for both samples. Considering thethe choice to set specificity at above 90%. Applied to aaim of identifying employees in a working environment atworking population, a screening test with this specificityrisk for sickness absence due to psychological complaints,can exclude a large majority of persons not at risk.and applying the recommendations of Dwyer [48], wefirst explore the test threshold which can discriminate wellbetween distressed employees without sickness absenceResultsThe demographics are specified in Table 1. For the firstdue to psychological problems and employees on sicksample, questionnaires were mailed to all employees ofleave because of stress or a stress-related disorder. Athe company (N = 7,522). The questionnaires were com-Receiver Operating Characteristic (ROC) analysis wasused to define a cutoff point, displaying the predictedpleted by 3,852 employees (response rate 51%). The sam-ple consisted mainly of men (91%), medium-or highly-probability of the target event – sickness absence. TheROC shows a range of cutoff points with correspondingeducated employees (74%), and had a mean age of 43.9years. At the moment at which the employees filled in thesensitivity and specificity.questionnaire 247 (6.4%) employees were on sick leave;these were excluded from the sample. The second sampleTo find with the ROC analysis the most optimal cutoff(n = 280) who had been on sick leave for at least twopoint that discriminates best between both groups, weweeks because of stress or stress-related disorder, con-formed in the first place a purposefully created artificialstudy population with an equal number of employeessisted of 66% men, 66% medium- or highly-educatedemployees, and the mean age was 41.9 years.from both samples: 280 'healthy working employees' ran-domly selected from the first sample, and in addition theTable 2 shows the means, standard deviations, and thetotal second sample of 280 employees on sick leave.Together, both populations form what we called thepercentile scores of distress for both samples (N = 3605and N = 280). As expected, employees on sickness absence'equal sample study population', in total 560 employees.due to psychological complaints scored significantlySecondly and in order to check, using a ROC analysis, thehigher on distress (Mean = 22.3, SD = 6.7) than the sam-ple of healthy working employees (Mean = 4.0, SD = 5.0)described ROC curve and its cutoff point in a representa-tive population, we formed a second artificial study pop-(T-test; p <.000).ulation similar to a working population with a normalprevalence of sickness absence due to psychological com-As can be seen from Table 3, the optimal cutoff score forplaints (2%). Therefore we added 72 employees (2% ofdistress, given a specificity that exceeds 90%, equals 10 inthe equal sample study population (N = 560). As