Limits...
Positive and negative relationship between anxiety and depression of patients in pain: a bifactor model analysis.

Xie J, Bi Q, Li W, Shang W, Yan M, Yang Y, Miao D, Zhang H - PLoS ONE (2012)

Bottom Line: In addition, we tested this hierarchical model with model fit comparisons with unidimensional, bidimensional, and tridimensional models.Compared with the three first-order models, the bifactor hierarchical model had the best model fit.This finding has not been convincingly demonstrated in previous research.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology, Fourth Military Medical University, Xi'an, People's Republic of China.

ABSTRACT

Background: The relationship between anxiety and depression in pain patients has not been clarified comprehensively. Previous research has identified a common factor in anxiety and depression, which may explain why depression and anxiety are strongly correlated. However, the specific clinical features of anxiety and depression seem to pull in opposite directions.

Objective: The purpose of this study is to develop a statistical model of depression and anxiety, based on data from pain patients using Hospital Anxiety and Depression Scale (HADS). This model should account for the positive correlation between depression and anxiety in terms of a general factor and also demonstrate a latent negative correlation between the specific factors underlying depression and anxiety.

Methods: The anxiety and depression symptoms of pain patients were evaluated using the HADS and the severity of their pain was assessed with the visual analogue scale (VAS). We developed a hierarchical model of the data using an IRT method called bifactor analysis. In addition, we tested this hierarchical model with model fit comparisons with unidimensional, bidimensional, and tridimensional models. The correlations among anxiety, depression, and pain severity were compared, based on both the bidimensional model and our hierarchical model.

Results: The bidimensional model analysis found that there was a large positive correlation between anxiety and depression (r = 0.638), and both scores were significantly positively correlated with pain severity. After extracting general factor of distress using bifactor analysis, the specific factors underlying anxiety and depression were weakly but significantly negatively correlated (r = -0.245) and only the general factor was significantly correlated with pain severity. Compared with the three first-order models, the bifactor hierarchical model had the best model fit.

Conclusion: Our results support the hypothesis that apart from distress, anxiety and depression are inversely correlated. This finding has not been convincingly demonstrated in previous research.

Show MeSH

Related in: MedlinePlus

The test information curve of the HADS based on the bifactor analysis for the general distress factor.X-axis represents severity of the general factor (theta), which had been standardized (0 being average, 1 being a standard deviation). The Y-axis represents the test information value. Test information is a kind of reliability criterion in IRT models, the bigger the test information value, the less measurement error, and better reliability. In contrast to models built using CTT, in IRT models, there is a test information value corresponding to every severity point, representing the reliability at that level of severity. We get the test information curve by connecting all these values.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3475698&req=5

pone-0047577-g001: The test information curve of the HADS based on the bifactor analysis for the general distress factor.X-axis represents severity of the general factor (theta), which had been standardized (0 being average, 1 being a standard deviation). The Y-axis represents the test information value. Test information is a kind of reliability criterion in IRT models, the bigger the test information value, the less measurement error, and better reliability. In contrast to models built using CTT, in IRT models, there is a test information value corresponding to every severity point, representing the reliability at that level of severity. We get the test information curve by connecting all these values.

Mentions: Bifactor analysis is a kind of IRT modeling that allows the assessment of direct fit for hierarchical models in which a general factor is separated from several specific factors [17]. Bifactor analysis can also provide us with standardized subject scores (0 being average and 1 being standard deviation, Figure 1) and main parameters for each item as well as the whole scale (Table 1). Factor loading has the same meaning as loadings in other kinds of factor analyses. Slope is a kind of discrimination parameter. Items with higher slopes are better at discriminating between patients with symptoms of different severity. The severity parameter is a kind of the location parameter. A larger severity parameter represents more severe symptoms. Test information is a kind of reliability criterion. Larger test information represents more accurate results. Unlike in classical testing theory (CTT), in which reliability for a scale is just one value, test information is a kind of function, with severity being X-axis, and information value being Y-axis. So we could know on what severity, the scale can get most accurate results.


Positive and negative relationship between anxiety and depression of patients in pain: a bifactor model analysis.

Xie J, Bi Q, Li W, Shang W, Yan M, Yang Y, Miao D, Zhang H - PLoS ONE (2012)

The test information curve of the HADS based on the bifactor analysis for the general distress factor.X-axis represents severity of the general factor (theta), which had been standardized (0 being average, 1 being a standard deviation). The Y-axis represents the test information value. Test information is a kind of reliability criterion in IRT models, the bigger the test information value, the less measurement error, and better reliability. In contrast to models built using CTT, in IRT models, there is a test information value corresponding to every severity point, representing the reliability at that level of severity. We get the test information curve by connecting all these values.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3475698&req=5

pone-0047577-g001: The test information curve of the HADS based on the bifactor analysis for the general distress factor.X-axis represents severity of the general factor (theta), which had been standardized (0 being average, 1 being a standard deviation). The Y-axis represents the test information value. Test information is a kind of reliability criterion in IRT models, the bigger the test information value, the less measurement error, and better reliability. In contrast to models built using CTT, in IRT models, there is a test information value corresponding to every severity point, representing the reliability at that level of severity. We get the test information curve by connecting all these values.
Mentions: Bifactor analysis is a kind of IRT modeling that allows the assessment of direct fit for hierarchical models in which a general factor is separated from several specific factors [17]. Bifactor analysis can also provide us with standardized subject scores (0 being average and 1 being standard deviation, Figure 1) and main parameters for each item as well as the whole scale (Table 1). Factor loading has the same meaning as loadings in other kinds of factor analyses. Slope is a kind of discrimination parameter. Items with higher slopes are better at discriminating between patients with symptoms of different severity. The severity parameter is a kind of the location parameter. A larger severity parameter represents more severe symptoms. Test information is a kind of reliability criterion. Larger test information represents more accurate results. Unlike in classical testing theory (CTT), in which reliability for a scale is just one value, test information is a kind of function, with severity being X-axis, and information value being Y-axis. So we could know on what severity, the scale can get most accurate results.

Bottom Line: In addition, we tested this hierarchical model with model fit comparisons with unidimensional, bidimensional, and tridimensional models.Compared with the three first-order models, the bifactor hierarchical model had the best model fit.This finding has not been convincingly demonstrated in previous research.

View Article: PubMed Central - PubMed

Affiliation: Department of Psychology, Fourth Military Medical University, Xi'an, People's Republic of China.

ABSTRACT

Background: The relationship between anxiety and depression in pain patients has not been clarified comprehensively. Previous research has identified a common factor in anxiety and depression, which may explain why depression and anxiety are strongly correlated. However, the specific clinical features of anxiety and depression seem to pull in opposite directions.

Objective: The purpose of this study is to develop a statistical model of depression and anxiety, based on data from pain patients using Hospital Anxiety and Depression Scale (HADS). This model should account for the positive correlation between depression and anxiety in terms of a general factor and also demonstrate a latent negative correlation between the specific factors underlying depression and anxiety.

Methods: The anxiety and depression symptoms of pain patients were evaluated using the HADS and the severity of their pain was assessed with the visual analogue scale (VAS). We developed a hierarchical model of the data using an IRT method called bifactor analysis. In addition, we tested this hierarchical model with model fit comparisons with unidimensional, bidimensional, and tridimensional models. The correlations among anxiety, depression, and pain severity were compared, based on both the bidimensional model and our hierarchical model.

Results: The bidimensional model analysis found that there was a large positive correlation between anxiety and depression (r = 0.638), and both scores were significantly positively correlated with pain severity. After extracting general factor of distress using bifactor analysis, the specific factors underlying anxiety and depression were weakly but significantly negatively correlated (r = -0.245) and only the general factor was significantly correlated with pain severity. Compared with the three first-order models, the bifactor hierarchical model had the best model fit.

Conclusion: Our results support the hypothesis that apart from distress, anxiety and depression are inversely correlated. This finding has not been convincingly demonstrated in previous research.

Show MeSH
Related in: MedlinePlus