Spanish Journal of Psychology (2014), 17, e107, 1–8. © Universidad Complutense de Madrid and Colegio Oficial de Psicólogos de Madrid doi:10.1017/sjp.2014.110
Symptom CheckList-90-R: Proposed Scales in a Probability Sample of Adolescents from the General Population Elena Felipe-Castaño1, Benito León-del-Barco1 and José Antonio López-Pina2 1 2
Universidad de Extremadura (Spain) Universidad de Murcia (Spain)
Abstract. To provide questionnaires for clinical assessment with scales adapted for adolescents would benefit clinical practice as well as research. The aim of this paper is to report normative data for adolescents on the SCL-90-R using a probability sample from the community. The participants were 1,663 adolescents, 845 girls and 818 boys, with an average age of 14.26 (SD = 1.36). They were selected through stratified cluster sampling with groups randomly selected from schools. Sampling error was estimated at 4% with a 95.5% confidence level. Cohen´s d effect sizes are reported for age-group. We found significant differences across participants according to gender and age on SCL-90-R Global Scores and Symptom Dimensions. Thus, we provide normative data, divided according to age and gender. Received 4 July 2013; Revised 7 May 2014; Accepted 16 June 2014 Keywords: SCL-90-R, normative data, adolescence, psychopathology, reliability.
Symptom checklists are utilized in both clinical practice and research to collect information about a patient and/or participant’s mental state. Among the most widely used is the Symptom Check-List Revised (SCL90-R) (Derogatis, 1994, 2002). Its objective is to gauge the respondent’s level of psychological well-being using 90 symptoms of varying severity. Altogether they comprise nine Symptom Dimensions, an Additional Scale, and three Global Scores, which are considered distress scores. Numerous studies have produced evidence for the SCL-90-R’s usefulness in clinical practice, specifically in screening (Bonicatto, Dew, Soria, & Seghezzo, 1997; Derogatis & DellaPietra, 1994) and in determining the quality or efficacy of psychological or medical treatments (Schmitz, Hartkamp, & Franke, 2000). However there is some controversy about its factorial validity (Bonynge, 1993). Examining the reliability of the SCL90-R’s dimensions, Derogatis, Rickels, and Rock (1976) found values of internal consistency, measured using Cronbach’s alpha, between .77 and .90. More recently, Schmitz, Hartkamp, Kruse et al. (2000) found internal consistency coefficients between .78, on the Hostility dimension, and .90 on the Depression dimension. In Spain, Caparrós-Caparrós, Villar-Hoz, Juan-Ferrer, and Viñas-Poch (2007) reported Cronbach’s alpha coefficients between .74 on the Paranoid Ideation dimension, and .88 on the Depression dimension, Correspondence concerning this article should be addressed to Elena Felipe-Castaño. Universidad de Extremadura (Spain). E-mail:
[email protected]
similar to the findings of Robles, Andreu, and Peña (2002). With respect to normative data, Derogatis (1994, 2002) reported scales in a population of adolescents and adults from the community as well as clinical populations, both outpatient and inpatient. In Spain, too, we have normative data from inpatient and outpatient psychiatric patients in populations of adults and adolescents from the community (De las Cuevas et al., 1991; Martínez-Azumendi, Fernández-Gómez, & BetiaFernández, 2001; Robles et al., 2002) as well as university students (Caparrós-Caparrós et al., 2007). The studies reviewed above focused on using the SCL-90-R in clinical assessment of adults, but did not take into account the normative characteristics of the adolescence stage of the lifespan. This could lead to confusion between the challenges associated with this point in development, and psychopathology. Clinical assessment of adolescents should be able to differentiate between the challenges that are characteristic of this stage, and those originating in earlier stages that later become stable and persistent (Navarro-Pardo, Meléndez, Sales, & Sancerni, 2012). Furthermore, they should enable researchers and clinicians to analyze such challenges’ progression, and their impact on the development of psychopathological symptomatology in adulthood (Widiger, De Clercq, & De Fruyt, 2009). With that in mind, and so that this stage of the lifespan may be carefully studied, assessment instruments adapted for adolescents must be made available (Zubeidat, Fernández-Parra, Ortega, Vallejo, & Sierra, 2009). Furthermore, like the SCL-90-R, such instruments
2 E. Felipe-Castaño et al. should have previously demonstrated clinical utility in adults (Bonicatto et al., 1997). That way, the same assessment tool can be used in adolescent and adult populations, and similarities and differences can be detected between the two age groups. In this study, we aim to propose scales differentiated by gender and age, collected in a probability sample of adolescents from the community, and to improve upon and expand the body of research on the SCL-90-R’s psychometric properties in adolescents. Method Participants The sample was comprised of 1,663 adolescents, 845 girls (50.80%) and 818 boys (49.20%), with an average age of 14.26 (SD = 1.36). Of those, 1,185 attended public school (71.3%) and 478 (28.7%) attended a charter school. Table 1 shows participants’ distribution in terms of gender, year in school, and age. Participants were selected through multistage statified cluster sampling, and groups were randomly selected from schools offering varied educational trajectories during Years 1 through 4 of ESO (obligatory secondary education). The strata considered were Teacher Resource Centers (TRCs), which divided the Spanish region of Extremadura into 18 geographical areas, each with designated secondary schools (IES [Secondary Education Institutions] or IESO [Obligatory Secondary Education Institutions]). Cluster sampling was carried out by randomly selecting two schools from each TRC. The number of participants was based on how many students were enrolled in each secondary school in Extremadura during the 2009–2010 school year (49,005 students, 3.4%). Sampling error was estimated at 4% with a 95.5% confidence level. We did not verify whether
or not participants had a history of psychopathological disorder or medical illness. Instruments Sociodemographic questionnaire with questions about age, gender, and year in school. Symptom Checklist, SCL-90-R (Derogatis, 1994, 2002). This 90-item questionnaire has a Likert-type response format with five choices (from 0 – not at all, to 4 –extremely). It asks adolescents how much the symptom described bothers them. It evaluates nine Symptom Dimensions: Somatization, ObsessiveCompulsive, Interpersonal Sensitivity, Depression, Anxiety, Hostility, Phobic Anxiety, Paranoid Ideation, and Psychoticism. It also includes seven additional items that together make up the Additional Scale, exploring aspects of diet and sleep; and three Global Scores: Global Severity Index (GSI), a general, non-specific measure of the intensity of overall psychological and psychosomatic distress; the Positive Symptom Total (PST), or sum of affirmative responses, refers to the breadth and diversity of psychopathology; and last, the Positive Symptom Distress Index (PSDI) indicates average symptom severity. Procedure To collect data, questionnaires were administered by classroom. Prior to data collection, we sought permission from each school, written informed consent from parents and teachers, and participants’ verbal informed consent. Participants were assured that their data would be kept confidential, and would only be used for research purposes. Administering the questionnaires took an average of approximately 15 minutes.
Table 1. Distribution of Participants According to Gender, Age, and Year in School Gender Girls n (%)
Boys n (%)
Total n (%)
Age
12 13 14 15 16 17
90 (5) 173 (10) 201 (12) 208 (13) 130 (7.80) 43 (2.60)
84 (5) 194 (12) 179 (11) 218 (13) 108 (6.50) 35 (2.10)
174 (10) 367 (22) 380 (23) 426 (26) 238 (14.30) 78 (4.70)
Year in School
1° ESO 2° ESO 3° ESO 4° ESO χ2 (3) = 10.84, p = .013
212 (12.70) 205 (12.03) 222 (13.30) 207 (12.40)
241 (14.50) 228 (13.70) 174 (10.05) 174 (10.50)
453 (27.20) 433 (26) 396 (23.80) 381 (22.90)
ESO = Obligatory Secondary Education.
SCL-90-R: Proposed Scales 3 Data collection took place over the course of three months during weekly tutoring sessions. Statistical Analysis Differences between groups according to gender, age, and year in school were analyzed using the chi-square statistic. The gender-age interaction effect on SCL-90-R Symptom Dimension scores and Global Scores was analyzed through ANOVA, with two fixed factors. Wilks’s lambda was used to analyze differences in the dependent variables, and partial eta-squared (η2) and Cohen’s d were computed as measures of effect size. Tukey’s test and Student’s t test were applied to calculate differences post-hoc, for age and gender respectively. Principal components analysis was conducted to determine the factor structure. Cronbach’s alpha was calculated as an index of scores’ internal consistency; values greater than or equal to .80 were considered good (Crocker & Algina, 1986). Normative data were calculated using percentiles. All analyses were done in SPSS 17.0. Results Differences were observed in participants’ distribution according to gender and year in school (see Table 1). Girls were underrepresented in Year 1 (adjusted standardized residuals = –2.0), and boys were underrepresented in Year 4 (adjusted standardized residuals = –2.3). Table 2 presents our descriptive analyses of the SCL-90-R Symptom Dimensions and Global Scores, for the total sample. The highest average score corresponded to the Obsessive-Compulsive dimension (0.59), followed by Hostility (0.52), and the lowest average score pertained to Phobic Anxiety (0.26), followed by Psychoticism (0.31). Regarding standard error, the kurtosis index yielded a value of 0.12; and the index of standard error for skewness produced a
value of 0.06. Both those values were under 1.96, indicating near normal distribution of scores. The internal consistency coefficients of Symptom Dimension scores were: Somatization (α = .83), ObsessiveCompulsive (α = .83), Interpersonal Sensitivity (α = .86), Depression (α = .87), Anxiety (α = .83), Hostility (α = .80), Phobic Anxiety (α = .76), Paranoid Ideation (α = .78), and Psychoticism (α = .84). According to these results, the sample’s scores were least consistent on the Phobic Anxiety dimension, and most consistent on the Depres sion dimension. With respect to the factor structure of the SCL-90-R, the indices of sampling adequacy indicate exploratory factor analysis can appropriately be applied to the data (Kaiser-Meyer-Okin: .972; Bartlett’s test of sphericity: .000). We utilized principal components analysis with orthogonal rotation, and followed Derogatis’s (1977) recommendation by using only scores with variablefactor correlation coefficients over .30. A 15-factor solution was obtained, explaining 58.85% of total variance. Of those 15, the first explained 33.63% of variance. On a second test, we limited the number of factors to nine, which explained 51.27% of total variance, the first one explaining 38%. Table 3 presents means and standard deviations on the SCL-90-R’s Symptom Dimensions and Global Scores, according to gender and age. The results showed significant differences as a function of gender (Wilks λ = 0.959, p < .001, η2 = .041) as well as age (Wilks λ = 0.891, p < .001, η2 = .023). As for the gender-age interaction, the only significant result was found in the Positive Symptom Distress Index (PSDI) (F = 2.40, p = .035, η2 = .007); this entails that the difference in PSDI scores between boys and girls was not the same in all age groups. Post-hoc comparisons revealed the following differences: in the female group, significant differences occurred between 13 and 16 years of age (p < .001), 14 and 16 years of age (p < .001), 13 and 17
Table 2. Descriptive Statistics. Dimensions and Global Scores on the SCL-90-R in the Total Sample of Participants (N = 1663) SCL90R Somatization Obsessive-Compulsive Interpersonal Sensitivity Depression Anxiety Hostility Phobic Anxiety Paranoid Ideation Psychoticism Global Severity Index (GSI) Positive Symptom Total (PST) Positive Symptom Distress Index (PSDI)
Minimum .00 .00 .00 .00 .00 .00 .00 .00 .00 .00 0 .00
Maximum
M
SD
Skewness
Kurtosis
3.42 3.50 3.78 5.15 3.10 4 3 3.67 3.40 2.97 90 4
0.49 0.59 0.49 0.46 0.42 0.52 0.26 0.49 0.31 0.46 24.31 1.35
0.56 0.62 0.64 0.62 0.53 0.67 0.45 0.63 0.49 0.51 21.32 .73
1.57 1.40 2.10 2.17 2.03 1.74 2.72 1.84 2.29 1.85 0.83 .45
2.39 1.98 5.33 5.72 4.79 3.13 8.55 3.74 5.74 3.98 –0.13 1.96
4 E. Felipe-Castaño et al.
Table 3. Descriptive Statistics Pertaining to Dimensions and Global Scores on the SCL90R by Gender and Age Age 12
SCL90R SOM OBS INT DEP ANX HOST PHOB PAR PSY GSI PST PSDI
13
14
15
16
17
Girls
Boys
Girls
Boys
Girls
Boys
Girls
Boys
Girls
Boys
Girls
Boys
n = 90
n = 84
n = 173
n = 194
n = 201
n = 179
n = 208
n = 218
n = 130
n = 108
n = 43
n = 35
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
M (SD)
.45 (.49) .55 (.51) .47 (.63) .50 (.69) .35 (.42) .42 (.63) .29 (.42) .42 (.62) .27 (.40) .41 (.45) 23.17 (18.75) 1.37(.74)
.35 (.42) .49 (.57) .43 (.60) .39 (.59) .31 (.43) .48 (.68) .2 1(.36) .41 (.59) .31 (.52) .38 (.46) 20.32 (20.56) 1.53(.86)
.46 (.57) .57 (.63) .47 (.63) .40 (.51) .38 (.50) .43 (.60) .29 (.48) .44 (.61) .31 (.54) .42 (.51) 23.98 (22.16) 1.21(.69)
.35 (.50) .43 (.58) .30 (.59) .35 (.54) .29 (.47) .33 (.55) .22 (.43) .35 (.55) .24 (.47) .33 (.47) 17.56 (20.56) 1.22(.86)
.48 (.51) .56 (.58) .45 (.56) .43 (.55) .40 (.50) .51 (.66) .22 (.37) .45 (.54) .27 (.45) .43 (.46) 23.97 (20.20) 1.24(.69)
.36 (.45) .45 (.53) .33 (.49) .30 (.48) .34 (.48) .44 (.63) .18 (.37) .34 (.54) .27 (.45) .34 (.43) 19.35 (.19.23) 1.19(.77)
.61 (.64) .69 (.66) .64 (.68) .57 (.64) .52 (.58) .60 (.70) .27 (46) .61 (.66) .33 (.47) .55 (.54) 29.41 (21.09) 1.42(.56)
.45 (.53) .55 (.59) .43 (.61) .40 (.58) .39 (.53) .54 (.66) .19 (.40) .48 (.62) .27 (.51) .42 (.50) 23.05 (21.27) 1.25(.65)
.77 (.60) .85 (.67) .74 (.81) .73 (.77) .61 (.65) .73 (.74) .44 (.61) .71 (.71) .40 (.49) .68 (.61) 34.61 (21.70) 1.58(.54)
.49 (.59) .68 (.67) .58 (.72) .53 (.71) .45 (.61) .63 (.75) .32 (.52) .57 (.72) .39 (.55) .52 (.58) 26.35 (22.50) 1.38(.76)
.64 (.52) .85 (.65) .59 (.50) .64 (.60) .56 (.51) .73 (.56) .36 (.54) .77 (.69) .38 (.39) .63 (.49) 33.20 (21.06) 1.66(.52)
.46 (.52) .56 (.49) .40 (.47) .40 (.49) .38 (.51) .54 (.67) .17 (.25) .51 (.70) .33 (.52) .43 (.45) 22.94 (20.19) 1.33(.66)
SOM = Somatization; OBS = Obsessive-Compulsive; INT = Interpersonal Sensitivity; DEP = Depression; ANX = Anxiety; HOST = Hostility; PHOB = Phobic Anxiety; PAR = Paranoid Ideation; PSY = Psychoticism; GSI = Global Severity Index; PSDI = Positive Symptom Distress Index, and PST = Positive Symptom Total.
SCL-90-R: Proposed Scales 5 male and female groups, specifically 12 to 14 years old and 15 to 17 years old (see Tables 4 and 5).
years of age (p = .003), and 14 and 17 years of age (p = .007) such that girls 16 to 17 years old scored higher than boys of the same age. As for the boys, significant differences were found between the groups 12 to 13 years old (p =.012), 12 to 14 years old (p = .006), and 12 to 15 years old (p = .040) such that 12-year-old boys scored higher than girls of the same age. Regarding gender, statistically significant differences were found on all Global Scores and Symptom Dimensions such that on average, girls scored higher than boys across the board, except on Psychoticism, which showed no significant differences (see Table 4). As for age, we found statistically significant differences on all Global Scores and Symptom Dimensions, with the 16-year-old group scoring higher than all other age groups. In general, the older participants were, the higher they scored. There were two exceptions. First, on the Hostility dimension, the 13-year-old group scored higher than the 15 to 17-year-olds. Second, the 12-year-old group had higher PSDI scores than the 13 to 14-year-old group (see Tables 3 and 4). To estimate the effect size of differences between the 16- and 12-year-old groups, and the 13- and 14-year-old groups, Cohen’s d was calculated (Table 5), yielding values between 0.27 (Psychoticism) and 0.47 (PST). According to Cohen (1988), these results are moderate, such that the average subject in the highest-scoring group scored higher than 50% of their own group, and higher than 69% of the lowest-scoring group on SCL-90-R Symptom Dimensions as well as Global Scores. These results were taken into account when percentiles were calculated, so they were separately computed for boys and girls, and by age group within the
Discussion The present study’s objective was to propose scales on the SCL-90-R, collected in a probability sample of adolescents from the community. The results illustrate that the symptom areas most widespread in the sample were Obsessive-Compulsive and Hostility, while Phobic Anxiety and Psychoticism were least represented. Our results are consistent with the findings of Caparrós-Caparrós et al. (2007) and Gempp and Avendaño (2008) in university students (except in their case, Depression took the place of the Hostility dimension), and the findings of Robles et al. (2002) in an inpatient non-psychiatric clinical sample. The reliability scores obtained on Symptom Dimen sions indicated highly acceptable levels of internal consistency similar to those reported by CaparrósCaparrós et al. (2007), Derogatis et al. (1976), and Robles et al. (2002). Regarding gender differences, girls had higher Global Scores and Symptom Dimension scores than boys, except on the Psychoticism dimension. Our results are in line with those of Caparrós-Caparrós et al. (2007), and González de Rivera, De las Cuevas, Gracia-Marco, and Henry-Benítez (1991), who also reported non-significant scores on the Paranoid Ideation dimension. Some studies have questioned the notion that Psychoticism and Paranoid Ideation consitute distinct dimensions (Prunas, Sarno, Preti, Madeddu, & Perugini, 2012), so we believe this discrepancy is of little significance. Differences between male and female manifestations
Table 4. ANOVA by Age and Gender. Dimension Scores and Global Scores on the SCL-90-R Age
Gender
SCL90R
F (1,1663)
η2
F (5,1663)
η2
Significant Post-hoc Comparisons Using Tukey’s Test
SOM OBS INT DEP ANX HOST PHOB PAR PSY GSI PST PSDI
25.08*** 18.46*** 14.17*** 18.64*** 13.38*** 3.87* 13.59*** 11.16*** 0.905 15.45*** 28.20*** 5.418*
.015 .011 .009 .011 .008 .002 .008 .007 .001 .009 .017 .003
7.75*** 8.30*** 7.01*** 7.41*** 6.04*** 7.65*** 5.05*** 8.20*** 2.34* 8.09*** 9.20*** 7.73***
.023 .025 .021 .022 .018 .023 .015 .024 .007 .024 0.27 .023
16>12***, 16>13***; 16>14***; 15 >13*; 15>14* 16>12***; 16>13***;16>14*** 16>12*; 16>13***;16>14***; 15>14* 16>12*; 16>13***; 16>14***; 16>15* 16>12**; 16>13***; 16>14**; 1312**; 16>13***; 16>14**; 13>17*; 13>15** 16> 13*; 16>14***; 16>15** 16>12**; 16>13***; 16>14***; 17>13*; 17>14* 16>14* 16>12***; 16>13***; 16>14***; 15>13* 16>12***; 16>13***; 16>14***; 15>13**; 15>14* 16>13***; 16>14***; 13