Bipolar Disorder
Figure 1: Differential Item Functioning for Suicidal Ideation/Attempt Between Bipolar I and Unipolar Depression (A) and Between Bipolar I and Bipolar II Depression (B)
AB
Suicidal ideation/attempt
1.0 1.0 Suicidal ideation/attempt
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
-3
-2
-10123 Depression severity
Bipolar I Unipolar
methodologies is that it allows one to examine the likelihood that a particular symptom will be endorsed at a particular level of depression severity. Thus, differences in symptom endorsement between groups can be evaluated while simultaneously equating for individual levels of depression symptom severity. Given that the majority of research comparing bipolar with unipolar depression has not equated groups on underlying depressive severity, and that several of the differences identified in the literature (e.g. atypical features, suicide risk) may be directly linked to severity, this benefit of IRT overcomes a particularly important limitation of prior research in this area. IRT also relies on large samples that represent the full range of the latent trait.50 Although this stipulation could be considered a limitation, in that IRT may not be as applicable to smaller clinical samples recruited from specialty settings, it could also be considered a strength in that it allows investigators to extend research focused on differences between bipolar and unipolar depression to large, representative community samples.
Another important consideration when applying IRT methodologies is that in such large samples, results that emerge as statistically significant may not be clinically significant. As such, it is important for investigators to consult published recommendations for effect size interpretation63
0.0
-3
-2
-1 0123 Depression severity
Bipolar I Bipolar II
described above. To address this concern, we have evaluated compound items of appetite/weight disturbance, sleep disturbance, and psychomotor disturbance, in contrast to their component parts, in our research.51–53
Although this approach is in keeping with
assignment of a diagnosis of a major depressive episode, as per DSM- IV, one trade-off in adopting such an approach is that conclusions regarding depressive subtypes (e.g. atypical or melancholic features) cannot be formed.
What Have We Learned (Thus Far) About Differences Between Bipolar and Unipolar Depression Using Item Response Theory?
We have recently applied IRT methodologies52,53
to evaluate differences
in DSM-IV depression symptom endorsement across three groups of individuals who participated in the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC):66
those reporting depressive that are especially useful in determining the clinical
applicability of findings. In our analyses, we decided a priori what we believed to be a minimally clinically significant effect size. Finally, as mentioned above, there are particular statistical assumptions for IRT that must be met prior to analysis. For the study of depressive symptoms, there is sufficient evidence in the literature that depression can be conceptualized as a unidimensional construct.64 Application of traditional IRT methodologies to the study of manic symptom expression, however, may be more limited as there is less support for a unidimensional model of mania.65
In addition, there may
be some limits to the study of depression using IRT methods with regard to the assumption of local independence of symptoms, as
18
symptoms and a history of mania (i.e. bipolar I depression), those reporting depressive symptoms and a history of hypomania (i.e. bipolar II depression), and those reporting depressive symptoms without a history of hypomania or mania (i.e. unipolar depression). To date, the NESARC represents the largest epidemiological survey of psychiatric conditions in the US, and therefore provided a rich opportunity to evaluate DIF across these groups. Methods for obtaining and assessing the sample have been detailed elsewhere.67,68
Consistent with our review above, we made several decisions prior to our analyses. First, we accounted for the assumption of local independence by evaluating the compound items of appetite/weight, sleep, and psychomotor disturbance, rather than their component parts. Second, we decided a priori what we believed to be a minimally clinically significant effect size for interpretation, following guidelines provided by Steinberg and Thissen.63 Hochberg procedure69
Third, we employed the Benjamini- to adjust p-values in order to account for risk of type I error due to multiple comparisons.
US PSYCHIATRY
Probability
Probability
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68