2005, Number 3
Reliability and validity of the SCL-90 for the evaluation of psychopathology in women.
Lara MC, Espinosa SI, Cárdenas ML, Fócil M, Cavazos J
Full text
Language: Spanish
References: 20
Page: 42-50
PDF size: 90.97 Kb.
ABSTRACT
Women are especially vulnerable to anxiety and depressive disorders, as shown in the international literature and, in our country, in the National Survey of Psychiatric Epidemiology. However, it is important not only to identify the disorders as such, but also to identify the symptoms, that though not becoming a disorder generate illnesses in women and can respond to psychopharmacological or psychotherapeutic interventions.The Symptom Check List (SCL 90) is a widely used instrument. In this study, three groups of women were assessed: one group of the community (as reference group), a second one, of women diagnosed as depressed in a third level psychiatric institution, and another group of women with a temporomandibular disorder (TMD).
The main objective of this study was to identify the psychometric characteristics of the SCL-90.
The alpha of Cronbach was determined (internal consistency) by each one of the 9 subscales. The validity was determined through the following criteria: Validity of the SCL-90 as a method to detect depressed women. A ROC curve was made, comparing the depressed women and the women of the community. Validity of the SCL-90 as a measurement of depression. We hypothesized a high correlation with Beck’s depression scale and Hamilton’s depression scale.
Sensitivity to change: effect size. We hypothesized that the greater effect would be in the depression subscale.
For construct validity we compare the psychopathological profiles of the three groups of women. We hypothesized the existence of a gradient: the highest ratings in depressed women, the lowest in women of the community, and intermediate ratings in women with TMD. A factorial analysis was also carried out.
Women older than 18 years of age were included, who were literate and accepted to participate in this study. They had no cognitive deficit (mental retardation, dementia, psychosis, states of confusion) or any other problem that would impede that they answer the surveys. The three groups were made up as follows:
Women with TMD: Women with clinical diagnosis of Temporomandibular Disorder who went to an ambulatory service of Maxilofacial Surgery of the ISSSTEP (Social Security Hospital for State Government Employees), who had not had treatment for TMD during the past six months and who did not have degenerative arthritic illness. No previous psychiatric diagnosis.
Depressed women: Women attending to the National Psychiatric Institute who were diagnosed with Major Depressive Disorder and were prescribed anti-depressive treatment.
Women of the community: Mothers of the students at the secondary school Pablo Cassals, a field of work of the Community Psychiatric Service of the Psychiatric Hospital Fray Bernardino Alvarez.
Results: The interview was carried out with 289 women, 131 were depressed women, 96, women of the community, and 62, women with TMD. The mean (standard deviation) age was similar in the three groups: 36 (10), 37.7 (5.5), 37.6 (11.9) respectively.
Of the women with depression, 64% had a stable partner, 100% of the women in the community, and 61% of the women with TMD. 69% of the women with depression were dedicated to keeping house, 34% of the women with TMD, and 74% of the women of the community.
Internal consistency was determined through Cronbach’s alfa. The results of this analysis show us a high consistency of the instrument’s nine subscales. Except for the subscale of paranoid symptoms in the women of the community, the alpha coefficients of all subscales in the 3 groups were higher than 0.7.
Another form of establishing the validity of the instrument is through the determination of the cut-off point for patient identification. Through the ROC curve, 1.5 points in the subscale of depression was taken, as the best score to minimize the number of false positive and false negative results.
As a measurement of depression, it was observed that the correlations of the subscales with Hamilton’s depression scale and Beck’s depression inventory had the highest correlations with the subscale of depression. The correlations with Beck’s Inventory are higher than with Hamilton’s scale of depression. This information supports the convergent validity of the subscale of depression with conventional instruments for its assessment.
Another way of determining the validity of an instrument is to establish sensitivity to change. The size of the effect was calculated dividing the average of the difference (before-after) over the basal standard deviation of depressed women who participated in the longitudinal study. The size of the effect was higher in Hamilton’s scale of depression and second in the subscale of depression, which indicates that there is a specific answer to a specific anti-depressive intervention.
According to what has been hypothesized (construct validity), the highest ratings corresponded to the depressed women; the lowest to the women of the community, and the intermediate to the women with TMD. The differences between groups were statistically significant in each of the nine subscales and in the total indexes. The group of the community was significantly different from those of the depressed women and the women with TMD. The difference between these last two groups was also significant.
In an exploratory factorial analysis of the 90 reactives, 87 had factorial loads higher than .40 in a first factor that explained 46.77% of the total variance of the SCL 90.
Although 13 more factors with Eigen values higher than 1 were identified, the variance explained by each of them was lower than 5%; according to this technique, only one general factor is the best explanation for the checklist. The two reactives with small factorial loads were: “Do you believe that others should be blamed for your problems?”, “Do you hear voices that others don’t hear?” and “Excesive eating”.
Although the checklist was developed for the assessment of a psychopathological profile, some authors have proposed that it should be used as a measurement of general psychopathology. In this study, the checklist was evaluated as a method of assessing a psychopathological profile as originally proposed by Derogatis, as a method of assessing general psychopathology (with the three total indicators), and as a specific method of assessing depression.
The information that most solidly supports the validity of the List’s subscales is the differences in the profile that were observed among the three groups.
The points in all the subscales were higher for women with TMD than those for the women of the community, but lower than in depressed patients; this shows that the psychological state of women with TMD tends to be similar to that of women with psychiatric diagnosis of depression. In this regard, the advantage of using the List to measure a profile is evident, and not only to assess general psychopathology.
The Symptom Checklist proved to have the psychometric characteristics appropriate to be used as an instrument to assess a psychopathological profile, to identify patients with depression, and to measure the intensity of depression. As an instrument to measure change, it is sensitive to psychopharmacological interventions. Although the size of the effect observed in the depression subscale is lower than that observed with the Hamilton scale, it is notably higher than the one observed with Beck’s Depression Inventory.
With these psychometric characteristics, it is possible to use the List in any type of research, including those that assess anti-depressive interventions.
REFERENCES