Characteristics of the sample
The sample consisted of 422 Italian students, 303 women (71.8%) and 119 men (28.2%). The mean age of participants was 20.70 years old (±3.44), the minimum age was 18 years. The mean body mass index (BMI), based on self-reported weight and height, was 21.83 (±3.42). Participants who had a high school diploma were mainly represented in our sample (60,3%). Finally, only 46.3% of the sample performed physical activity, while 53,7% did not engage in any sport activity. Table 1 describes clinical variables scores evaluated in our sample (i.e. DOS, ORTO-15, DEQ, BDI-II, OCI-R) by descriptive statistics such as mean and standard deviation.
Table 1 Descriptive statistics and internal consistency of the measures used assessing orthorexia nervosa (DOS and ORTO-15), disordered eating (DEQ), depression (BDI-II) and obsessive and compulsive symptoms (OCI-R).
Variable
|
M
|
SD
|
Cronbach’s alpha
|
DOS
|
15.60
|
5.35
|
.888
|
ORTO-15
|
38.32
|
3.96
|
.808
|
DEQ
|
24.17
|
20.47
|
.933
|
BDI-II
|
10.51
|
9.14
|
.896
|
OCI-R
|
10.06
|
9.73
|
.892
|
Confirmatory factor analysis
CFA with 415 participants was conducted for the one-factor model. Seven cases with missing on all variables were not included in the analysis.
The CFA yielded ambiguous results: χ2(35) = 179.212, p<0.001; RMSEA = 0.100, 90% CI =0.085-0.114, p<0.001; CFI = 0.970; TLI = 0.962; SRMR =0.054). Specifically, RMSEA was widely above acceptable thresholds. Thus, we examined potential sources for this not acceptable model fit and found that two error covariance (item 6 and 10; item 4 and 7) had large and significant MI value. Both pairs of items have a similar meaning and measure similar aspect of the ON construct: in particular item 6 and 10 refer to the consequences of unhealthy eating, while item 4 and 7 refer to the social consequences of orthorexia nervosa. Accordingly, we re-ran a model with this two error covariances freely estimated. The revised model showed the following fit indices: χ2(33) = 112.565, p<0.001; RMSEA = 0.076, 90% CI =0.061-0.092, p=0.003; CFI = 0.984; TLI = 0.978; SRMR = 0.043.
Table 2 shows the standardized factor loadings that were all above .70. All the factor loadings resulted statistically significant (p<0.001).
Table 2. Confirmatory factor analysis and internal consistency results. All the factor loadings are standardized and statistically significant (p<0.001)
|
Standardized factor loadings
|
α if item is deleted
|
Item total correlations
|
1. Eating healthy food is more important to me than indulgence/enjoying the food
|
0.732
|
.877
|
.631
|
2. I have certain nutrition rules that I adhere to
|
0.790
|
.878
|
.644
|
3. I can only enjoy eating foods considered healthy
|
0.820
|
.875
|
.671
|
4. I try to avoid getting invited over to friends for dinner if I know that they do not pay attention to healthy nutrition
|
0.768
|
.881
|
.582
|
5. I like that I pay more attention to healthy nutrition than other people
|
0.834
|
.873
|
.695
|
6. If I eat something I consider unhealthy, I feel really bad
|
0.713
|
.877
|
.639
|
7. I have the feeling of being excluded by my friends and colleagues due to my strict nutrition rules
|
0.723
|
.889
|
.457
|
8. My thoughts constantly revolve around healthy nutrition and I organize my day around it
|
0.911
|
.868
|
.763
|
9. I find it difficult to go against my personal dietary rules
|
0.842
|
.874
|
.676
|
10. I feel upset after eating unhealthy foods
|
0.716
|
.879
|
.598
|
Tests of gender factorial invariance and of gender differences
According to the revised model, factorial invariance tests across gender were examined. Invariance results are shown in Table 3.
Table 3. Results of the measurement invariance tests.
Models
|
ΔCHI
|
CFI
|
ΔCFI
|
TLI
|
SRMR
|
RMSEA and 90% Confidence Interval, p(RMSEA<0.05)
|
Configural invariance
|
|
.990
|
|
.986
|
.053
|
.064 (.046-.081), p=.102
|
Partial metric invariance
|
6.484 (p>0.05)
|
.990
|
0
|
.987
|
.054
|
.061 (.044-.078), p=.138
|
Partial scalar invariance
|
15.769 (p>0.05)
|
.990
|
0
|
.990
|
.055
|
.054 (.036-.070). p=.341
|
The first level (i.e., configural invariance) was achieved with the following fit indices: χ2(66) = 121.474, p<0.001; RMSEA = 0.066, 90% CI =0.046-0.081, p=0.102; CFI = 0.990; TLI = 0.986; SRMR = 0.053.
When constraints on loadings were introduced, an inspection of the MI revealed that there were three constraints not tenable (factor loadings on item 7, item 1, and item 9). After they were relaxed, a partial metric invariance model was achieved (ΔCHI = 6.484, p>.05; ΔCFI = 0).
When scalar invariance model was tested, an examination of the MI revealed that introduced constraints were not tenable (thresholds of item 6, item 4, and item 10). These constraints were relaxed and a partial scalar invariance model was obtained (ΔCHI = 15.769, p>.05; ΔCFI = 0). Given all of the above, the latent means difference across gender was examined. To achieve this, mean value was constrained to zero for the male group (i.e., reference group), while in the female group was freely estimated. The result highlighted a non-significant difference (p>.05).
Validity and reliability
The reliability of the I-DOS, estimated by Cronbach’s α, was .888, showing a strong internal consistency. Moreover, all the items showed a moderate or high correlation with the total items ranged from .457 to .763 (Table 2).
The I-DOS total score had strong and statistically significant correlations with ORTO-15 total score (r=-.573; p<.001), where lower ORTO-15 score indicated higher levels of orthorexia tendencies and behaviors. Significant correlations were also found with disordered eating symptoms (DEQ total score, r=597; p<.001), with obsessive and compulsive symptoms (OCI-R total score, r=.229; p<.001), and with the sum score of depressive symptoms (BDI-II total score, r=.262; p<.001). Regarding the association between BMI and orthorexic eating behavior (I-DOS total score), we found a statistically non-significant correlation (r=.079, p>0.05).
Table 4 presents the correlations between the I-DOS and other constructs.
Table 4. Correlations between orthorexia nervosa (DOS and ORTO-15), disordered eating (DEQ), depression (BDI-II), obsessive and compulsive symptoms (OCI-R) and body mass index (BMI) total scores
|
ORTO-15
|
DEQ
|
BDI-II
|
OCI-R
|
BMI
|
I-DOS
|
-.573**
|
.597**
|
.262**
|
.229**
|
.079
|
Note; ** = p < 0.001.
Distribution of an estimate of orthorexia nervosa in our sample
Participants mean score of the I-DOS was 15.60 (±5.35), scores ranging from a minimum value of 10 and a maximum value of 37. Using the original version’s cut-off points9, 3.2% of the study participants would be considered having ON (total score greater than 30), 4,9% would be at risk of ON (total score between 25 and 29), while no risk of ON was observed in 91.9% of the sample (total score less than 25).