Psychiatry Research 228 (2015) 289–294
Contents lists available at ScienceDirect
Psychiatry Research journal homepage: www.elsevier.com/locate/psychres
The dimensional structure of cycling mood disorders Lloyd Balbuena 1, Marilyn Baetz, Rudy C. Bowen n Department of Psychiatry, University of Saskatchewan, 103 Hospital Drive, Saskatoon, Saskatchewan, Canada S7N 0W8
art ic l e i nf o
a b s t r a c t
Article history: Received 25 July 2014 Received in revised form 27 April 2015 Accepted 10 June 2015 Available online 26 June 2015
This study examines whether mood disorders differ fundamentally in terms of phase duration. Most clinically significant mood disorders are recurrent and cyclical. The phase duration of these cycles is part of the diagnostic criteria. Specifically, we determined whether a dimensional or taxonic latent structure better captures cycling mood disorders. 319 patients recruited from 5 psychiatrists and a psychoeducational program completed three questionnaires assessing aspects of mood cycling. These were the Affective Lability Scale-Short Form (ALS-SF), Mood Disorders Questionnaire (MDQ), and the Eysenck Neuroticism scale. Patient scores on these instruments were submitted to three taxometric procedures (MAMBAC, MAXEIG, and L-Mode). Comparison curve fit indices (CCFIs) were calculated to distinguish taxonic versus dimensional latent structure. In addition, graphs were produced for each procedure and compared with those of categorical or dimensional prototypes. The CCFIs of the three procedures ranged from 0.25 to 0.27, consistent with dimensional structure. The graphs closely resembled dimensional prototypes. Mood instability and other types of cycling moods probably conform to a dimensional latent structure. Patients with disorders featuring mood cycling might benefit from common treatments. & 2015 Elsevier Ireland Ltd. All rights reserved.
Keywords: Latent structure Mood instability Mood swings
1. Introduction The phase duration of low or high mood disorders is used by psychiatry to differentiate between mood disorders. If mood is stuck at severe low levels for two weeks or two years and no switch to high mood is reported, the diagnosis is major depression or persistent depressive disorder (American Psychiatric Association, 2013). Switching into elevated mood for one week characterizes mania, four days hypomania and shorter durations are referred to as brief hypomania or the bipolar spectrum (Angst, 1998; Cassano et al., 1999; American Psychiatric Association, 2013). In general, the DSM system shifted emphasis away from fluctuations in mood to a checklist of symptoms (Goodwin et al., 2007; American Psychiatric Association, 2013). However, the symptom differences between different categories of elevated mood (and depressed mood) are small, and syndrome severity appears to be continuous, so the diagnosis depends largely on where the cut in duration is made (Fiedorowicz et al., 2011; Abbreviations: ALS, Affective Lability Scale; CCFI, comparison curve fit indices; ENS, Eysenck Neuroticism Subscale; L-Mode, latent mode factor analysis; MDQ, Mood Disorders Questionnaire; MI, mood instability; MAMBAC, mean above minus below a cut; MAXEIG, Maximum eigenvalue n Corresponding author. Fax: þ 1 306 844 1504. E-mail address: [email protected] (R.C. Bowen). 1 Present address: Administrative Data Research Centre—Wales, Swansea University, Swansea, Wales SA2 8PP, United Kingdom. http://dx.doi.org/10.1016/j.psychres.2015.06.031 0165-1781/& 2015 Elsevier Ireland Ltd. All rights reserved.
American Psychiatric Association, 2013). Consequently, patients are assigned to diagnostic categories, which are assumed to have clearly defined boundaries. The critical question however is whether treatment and research are improved by these apparently arbitrary cuts in phase (Cassano et al., 1999). Defining what is meant by indicators, latent variable, taxon and category may be helpful to the general reader in understanding the taxometric method. Indicators are conditions such as fever or high blood pressure that are perceptible by the senses or with the aid of medical instruments. Latent variables such as aggression or mood instability (MI), on the other hand, can only be indirectly or partly inferred from more concrete phenomena such as yelling or facial expressions. MI is defined as frequent fluctuations of mood over time, which may be as brief as a few hours (Trull et al., 2008). It is usually identified with the borderline personality disorder (Marroquin, 2011) but has also been studied in patients with mood and anxiety disorders (Bowen et al., 2006), personality disorders (Trull et al., 2008), psychotic disorders (Marwaha et al., 2014) and in university students (Eid and Diener, 1999). Of crucial importance to our argument is the distinction between taxonic and dimensional structure. The distinction is captured by the intuitive notions of “differing in kind” (taxonic) and “differing by degree” (dimensional). Our concern is with latent structure, since categories such as “hot” or “cold” are artificially created by an arbitrary cut in temperature, which is continuous. As another example, “reptiles” and “mammals” are taxonic because they do not simply
290
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
differ by degree. They differ by degree in body temperature but there are innumerable and more fundamental differences. A taxonic structure may characterize the domain of mood disorders across diagnostic categories, but the boundaries are debatable. First, the distinction between “normal” people and people with mental disorders is usually made by applying a “distress or impairment” criterion to distinguish the group with mental disorders (American Psychiatric Association, 2013). The second is the distinction between regular psychiatric disorders and personality disorders. Terms that are applied to personality disorders like “enduring” “relatively stable across time” and “inflexible and pervasive” imply a stable course so phase duration is not specified (American Psychiatric Association, 2013). The third is the distinction between unipolar and bipolar mood disorders because of the shift to elevated mood in bipolar disorders (Perlis et al., 2011; Phillips and Kupfer, 2013). The boundary between unipolar and bipolar disorders is disputed and the demonstration of MI in patients with unipolar disorders further muddles the distinction (Bowen et al., 2004). Fourthly, within the broad category of major depression two subtypes called “melancholic” and “atypical” are contrasted. The first is characterized by “the lack of reactivity to usually pleasurable stimuli” while the latter shows “mood reactivity in response to actual or potential positive events.” (American Psychiatric Association, 2013). Our main concern in this paper is whether the categories formed using these distinctions are warranted. Three recent papers addressing the latent structure of elevated moods reached different but reconcilable conclusions. Meyer and Keller (2003) conducted a taxometric analysis of hypomania in two large samples (young adults in one, and adolescents in the other) and concluded that hypomania has a dimensional latent structure. Their study used the Hypomanic Personality Scale as the sole indicator of hypomania (Eckblad and Chapman, 1986). Ahmed et al. (2011) examined the latent structure of mania using the Collaborative Psychiatric Epidemiological Surveys (n 420,000). Three indicators of mania were used, each of which covered a separate symptom domain of mania defined by DSM-IV. In addition to mania, the researchers also assessed unipolar depression using another set of indicators. The authors concluded that a taxonic solution better fits the data but that significant dimensional variation within groups reflecting illness severity remained. Prisciandaro and Roberts (2011) used three different modeling strategies to analyze data from the Epidemiologic Catchment Area study and all lines of evidence converged on a dimensional solution for mania. It seems reasonable to conclude, at the minimum, that mood problems are dimensional within a homogeneous group of individuals. Whether the population as a whole is homogeneous remains to be seen. Given these inconsistent results and the disputable categorization of mood disorders, our hypothesis was that elevated mood has a dimensional latent structure.
2. Methods
Table 1 Descriptive statistics of participant scores in 5 instruments.
Mood Disorders Questionnaire* Affective Lability Scale* Short Eysenck Neuroticism Scale* PHQ-9 Penn State Worry Scale a
n
Mean (SD)
Range
Skewness
321 318 316 320 326
6.42 40.33 33.51 12.89 39.36
0–19 3–72 12–48 0–27 12–60
0.23 0.08 0.46 0.01 0.35
(3.94) (13.94) (8.78) (7.52) (13.29)
Participant scores in these three instruments were used in taxometric ana-
lysis.
We started with 345 participants but because of missing data in some questionnaires, the sample size was reduced to 319. The average age was 34.80 years and 52 per cent of our sample was female. Sixty-four percent scored 10 or higher on the PHQ-9 questionnaire, indicating depression in the range typical of patients diagnosed with major depression (Manea et al., 2012; American Psychiatric Association, 2013). Forty-seven percent scored seven or higher on the MDQ indicating possible hypomania (Hirschfeld et al., 2000). This is consistent with recent estimates of the proportion of patients with mood syndromes who report hypomanic symptoms (Angst, 2006; Zimmerman et al., 2010). Sixtythree percent scored 45 or higher on the Penn-State Worry Questionnaire (Meyer et al., 1990). This indicates the proportion of patients who reported high worry, but is not necessarily the same as generalized anxiety disorder because worry is common in major depression (Salzer et al., 2009). Table 1 shows the distribution statistics for the relevant scales. 2.2. Indicator construction We used three psychometrically validated scales measuring various aspects of mood instability—the Affective Lability ScaleShort Form (ALS-SF), the Mood Disorders Questionnaire (MDQ), and the Eysenck Neuroticism Subscale (ENS). The 18-item ALS-SF (Oliver and Simons, 2004) is an abridged version of the original instrument (Harvey et al., 1989) and assesses switches in moods across three factors: anxiety/depression, depression/elation, and anger. The MDQ (Hirschfeld et al., 2000) is a 15-item self-report instrument to assess switches to elevated or activated mood. Only the 13 symptom questions were used since it was not our objective to make a diagnosis but to study the indicators. The MDQ was developed to aid the recognition of bipolar I and II disorders (Hirschfeld et al., 2003). There are mixed opinions regarding its usefulness (see Zimmerman and Galione, 2011 for a review), with some studies questioning its usefulness for patients with impaired insight (Miller et al., 2004), and in distinguishing between bipolar I and II (Vieta et al., 2007). The ENS is composed of 12 questions that assess fluctuating mood and anxiety and depression. We used the ENS because it contains the MI factor that has been shown to predict suicidal ideation (Bowen et al., 2011). The Cronbach alphas were 0.94, 0.91, and 0.88 for the ALS-SF, ENS, and MDQ respectively. The items in each scale were summed to better meet the requirement for continuous variables in taxometric analysis.
2.1. Sample 2.3. Suitability of taxometric analysis The sample was a mixed group of outpatients from two main sources. The first group came from the practices of 5 general psychiatrists. A second group was composed of outpatients referred to a psycho-educational mood program. Patients were asked to complete a questionnaire consisting of 13 short scales before their first appointment and to bring this to the first appointment. All patients gave written consent for the use of their data for research. The study was approved by the University Behavioural Ethics Board.
We assessed the suitability of our data for taxometric analysis in two ways: first, by estimating the taxon base rate and second, by evaluating the reliability and validity of our indicators. Data with base rates lower than 0.5 are suboptimal for detecting a taxon (Meehl, 1995). Since the MDQ instrument provides the threshold score of 7 for hypomania, we used it as a criterion for estimating our taxon base rate of 0.46, which is very close to the ideal. Our indicators had acceptable skewness. Then, we examined
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
291
Table 2 Indicator reliabilities and correlations. Indicators
Reliability per procedure (in Cohen’s d)
Indicator correlations In the overall sample
MDQ ALS ENS
MAMBAC
MAXEIG
L-Mode
MDQ
ALS
1.65 2.4 1.53
1.68 2.23 1.63
1.64 2.34 1.56
1 0.56 0.34
1 0.65
participant scores in the three indicators. Our indicators exceeded Meehl’s minimum threshold of Cohen’s d¼ 1.25 (Meehl, 1995). Cohen’s d in this context measures the capability of indicators to distinguish putative taxon and complement members. It is comparable to the concept of statistical power in the sense that both are about detecting a difference if one exists. We then compared the indicator correlations in the full sample and within the putative taxon and complement groups, based on the estimated base rate. Ruscio et al. (2006) recommends that correlations within groups should be appreciably weaker than those in the full sample. Finally, we generated dimensional and taxonic comparison data for each analytical technique (described in the next section). The extent that graphs of dimensional and taxonic comparison data are distinguishable reflects the confidence in a conclusion one way or the other. Overall, our dataset and indicators proved to be well suited for taxometric analysis. The characteristics of our indicators are summarized in Table 2.
In the taxon (complement) ENS
MDQ
ALS
ENS
1
1 0.18 (0.11) 0.15 ( 0.04)
1 0.31 (0.38)
1
study, we performed L-Mode analysis by first visually inspecting the graph of factor scores and based on this graph assigning locations for the left and right modes. Because the graph did not produce any discernible modes, we left the program at the default setting of searching on either side of the mean factor score. In many cases, the graphs produced by the taxometric procedures do not fit the prototype shapes and experienced human readers are called upon to make a judgment. Fortunately, taxometric procedures produce a numerical statistic called the comparison curve fit index (CCFI) that ranges from 0 to 1 (Ruscio et al., 2006). CCFI values below 0.4 indicate a dimensional structure while those above 0.6 indicate a categorical structure. CCFI’s between 0.4 and 0.6 are equivocal and careful interpretation is recommended. All taxometric analyses were implemented in Ruscio’s taxometric (Ruscio et al., 2006) as implemented in R (R Core Team, 2014).
2.4. Statistical analysis 3. Results We submitted our indicators to three taxometric procedures: (1) mean above minus below a cut (MAMBAC); (2) Maximum eigenvalue (MAXEIG); and (3) latent mode factor analysis (L-Mode). In the MAMBAC procedure, successive cuts are made along an input variable in which subjects are rank ordered, thereby dividing the sample into intervals (Meehl and Yonce, 1994). Within each interval, the mean score in an output variable above the cut is subtracted from the mean score below the cut. This procedure distinguishes taxonic from dimensional structure because taxonic and dimensional data produce peaked and bowl-shaped graphs, respectively. The graph’s inflection point represents the score that best separates the taxon from its complement. We performed MAMBAC by assigning one indicator as output and the sum of the other two variables as input, thereby producing three curves. Fifty evenly spaced cuts beginning 25 cases from either extreme were made and 10 replications were performed. MAXEIG (Meehl and Yonce, 1996) calculates the association of two or more indicators. An input indicator is designated as the axis along which the cases are ordered by magnitude. The association between the other remaining indicators (output) is calculated at given intervals of the input. When the latent variable is dimensional, the association between output indicators should be constant along the input. By contrast, the association should be stronger in the interval where there is a highest mixing of taxon and complement members, if the latent variable is taxonic (Ruscio et al., 2006). We performed MAXEIG analysis with 25 windows having a 90 percent overlap with n ¼94 for each window. L-Mode (Waller and Meehl, 1998) has its origins in the Bartlett’s factor analytic method (Bartlett, 1937). In contrast to the sliding cuts approach taken by MAMBAC and MAXEIG, L-Mode relies on the distribution of factor scores of the latent variable. Factor scores have a mean of zero and standard deviation of 1. A bimodal shape supports a taxonic interpretation while unimodality supports dimensionality (Ruscio et al., 2006). For this
The three MI measures were moderate to strongly correlated (Pearson’s r ranged from 0.34 to 0.65). For MAMBAC, the averaged curve of the three input–output combinations more closely resembled simulated dimensional data (see Fig. 1). The graph produced by the MAXEIG was generally flat and likewise supported dimensionalit. (see Fig. 2). L-Mode produced a graph with a unimodal distribution of factor score. (see Fig. 3) The CCFIs of these three procedures were very consistent in supporting dimensionality: 0.25 for MAXEIG, 0.26 for L-Mode, and 0.27 for MAMBAC.
4. Discussion Our main finding is that naturally occurring cycling of moods probably has a dimensional latent structure. We used three psychometrically validated instruments in a clinical sample with an ideal taxon base rate and all three taxometric procedures supported a dimensional interpretation. A point also worth noting is that MI, as currently defined, refers to very short cycle phases of a few hours, but our findings suggest that cycling of moods is not qualitatively different across cycling of different phase durations (Trull et al., 2008). The scales originated from three conceptually distinct sources. The MDQ was derived from DSM-IV criteria for hypomania and the clinical experience psychiatrists familiar with bipolar mood disorders (Hirschfeld et al., 2000). The authors of the ALS were familiar with disorders with labile affect such as the Borderline Personality Disorder and were also familiar with the General Behavior Inventory that was developed to differentiate bipolar and unipolar mood disorders (Depue et al., 1981). Both the General Behavior Inventory and the ALS were originally validated on university students (Harvey et al., 1989). The Neuroticism Scale of the Eysenck Personality Inventory borrowed items from several other personality scales, including
292
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
Fig. 1. Comparison of the averaged MAMBAC curve for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
the Maudsley Personality Inventory that Eysenck had developed (Deary and Bedford, 2011). He emphasized the empirical rather than conceptual origins of the final version of the scale (Eysenck et al., 1969). At a minimum our results indicate that conditions characterized by cycling mood and mood instability are dimensional within the population of patients with mood disorders. In other words, categories defined as unipolar major depression, bipolar II, bipolar I and borderline personality disorders are probably artificial categories with a latent dimensional structure. This inference is similar
to Kraepelin’s position on variations of mood representing a “single morbid process” (MacKinnon and Pies, 2006). More recently, other authors have reached similar conclusions based on family history studies, patients’ symptoms, follow-up studies, pharmacological treatment outcome studies, and the lack of a clear diagnostic boundary. (Angst et al., 2003; MacKinnon and Pies, 2006; Rihmer et al., 2010; Phillips and Kupfer, 2013). The clinical implication is that treatments that benefit mood cycling of one kind might also benefit conditions with a different phase duration.
Fig. 2. Comparison of the averaged MAXEIG curve for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
293
Fig. 3. Comparison of the frequency distribution of estimated factor scores for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
Our study is subject to three limitations. The sample size was relatively small in terms of the taxometric method. With Meehl’s recommendation of n 4300, we did meet the minimum required to complete the analysis (Meehl, 1995). This limitation is partly offset by our high taxon base rate, which was close to the ideal. The main consequence of our small sample size was our inability to complete latent class analysis. A second limitation is that we did not have access to clinical diagnoses. This group of participants did not include a specified group of “normal” individuals, but there was a range of severity as seen in the range of scores (Table 1) and one frequently used way to access the program is self-referral. The third limitation is that we relied on retrospective self-report measures of mood instability. It was reported that the MDQ is useful when manic symptoms are being recalled within the last two years, but not those over the entire lifetime (Boschloo et al., 2013). We did not have information regarding how recent the episodes were for the study participants. This limitation is balanced by the correlation of retrospective and prospective measures of MI (Anestis et al., 2010) and that our sample consisted solely of psychiatric outpatients instead of community samples for whom the instruments are less reliable. Future studies should consider taxometric analysis of prospectively collected data.
5. Conclusion The results of this study indicate that segregating patients into different categories based on length of phase of mood cycles is not justified. This suggests that established treatments for mood instability might benefit the full range of mood conditions manifesting cycles of different phase lengths. Contributors Rudy Bowen conceptualized the study, critically reviewed the
literature, collected the data, provided the indicators of affective instability, interpreted the data, re-wrote the initial draft and wrote the discussion. Lloyd Balbuena, developed the analytical strategy, implemented the analysis, wrote the initial draft, addressed reviewer comments, and produced the tables and figures. Marilyn Baetz collected the data, provided constructive comments, critically re-wrote the draft and revised the discussion. All authors approved the manuscript for submission.
Conflict of interest The authors have no conflict of interests to declare.
References Ahmed, A.O., Green, B.A., Clark, C.B., Stahl, K.C., McFarland, M.E., 2011. Latent structure of unipolar and bipolar mood symptoms. Bipolar Disord. 13 (5–6), 522–536. American Psychiatric Association, 2013. Diagnostic and statistical manual of mental disorders, 5th ed. American Psychiatric Publishing, Arlington, VA. Anestis, M.D., Selby, E.A., Crosby, R.D., Wonderlich, S.A., Engel, S.G., Joiner, T.E., 2010. A comparison of retrospective self-report versus ecological momentary assessment measures of affective lability in the examination of its relationship with bulimic symptomatology. Behav. Res. Ther. 48 (7), 607–613. Angst, J., 1998. The emerging epidemiology of hypomania and bipolar II disorder. J. Affect. Disord. 50 (2–3), 143–151. Angst, J., 2006. Do many patients with depression suffer from bipolar disorder? Can. J. Psychiatry 51 (1), 3–5. Angst, J., Gamma, A., Endrass, J., 2003. Risk factors for the bipolar and depression spectra. Acta Psychiatr. Scand. Suppl. 418, 15–19. Bartlett, M.S., 1937. The statistical conception of mental factors. Br. J. Psychol. Gen. Secti. 28, 97–104. Boschloo, L., Nolen, W.A., Spijker, A.T., Hoencamp, E., Kupka, R., Penninx, B.W.,
294
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
Schoevers, D., 2013. The Mood Disorder Questionnaire (MDQ) for detecting (hypo)manic episodes: its validity and impact of recall bias. J. Affect. Disord. 151 (1), 203–208. Bowen, R., Baetz, M., Hawkes, J., Bowen, A., 2006. Mood variability in anxiety disorders. J. Affect. Disord. 91 (2–3), 165–170. Bowen, R., Baetz, M., Leuschen, C., Kalynchuk, L.E., 2011. Predictors of suicidal thoughts: mood instability versus neuroticism. Personal. Individ. Differ. 51 (8), 1034–1038. Bowen, R., Clark, M., Baetz, M., 2004. Mood swings in patients with anxiety disorders compared with normal controls. J. Affect. Disord. 78 (3), 185–192. Cassano, G.B., Dell’Osso, L., Frank, E., Miniati, M., Fagiolini, A., Shear, K., Pini, S., Maser, J., 1999. The bipolar spectrum: a clinical reality in search of diagnostic criteria and an assessment methodology. J. Affect. Disorder. 54 (3), 319–328. Deary, I.J., Bedford, A., 2011. Some origins and evolution of the EPQ-R (short form) neuroticism and extraversion items. Person. Individ. Differ. 50 (8), 1213–1217. Depue, R.A., Slater, J.F., Wolfstetter-Kausch, H., Klein, D., Goplerud, E., Farr, D., 1981. A behavioral paradigm for identifying persons at risk for bipolar depressive disorder: a conceptual framework and five validation studies. J. Abnormal Psychology 90 (5), 381–437. Eckblad, M., Chapman, L.J., 1986. Development and validation of a scale for hypomanic personality. J. Abnormal Psychol. 95 (3), 214–222. Eid, M., Diener, E., 1999. Intraindividual variability in affect: reliability, validity, and personality correlates. J. Person. Soc. Psychol. 76 (4), 662–676. Eysenck, H.J., Eysenck, S.B.G., Hendrickson, A., 1969. Personality Structure and Measurement. Routledge & Kegan Paul, London. Fiedorowicz, J.G., Endicott, J., Leon, A.C., Solomon, D.A., Keller, M.B., Coryell, W.H., 2011. Subthreshold hypomanic symptoms in progression from unipolar major depression to bipolar disorder. Am. J. Psychiatry 168 (1), 40–48. Goodwin, F.K., Jamison, K.R., Ghaemi, S.N., 2007. Manic-depressive Illness: Bipolar Disorders and Recurrent Depression, 2nd ed. Oxford University Press, New York, N.Y. Harvey, P.D., Greenberg, B.R., Serper, M.R., 1989. The affective lability scales—development, reliability, and validity. J. Clin. Psychol. 45 (5), 786–793. Hirschfeld, R.M., Holzer, C., Calabrese, J.R., Weissman, M., Reed, M., Davies, M., Frye, M.A., Keck, P., McElroy, S., Lewis, L., Tierce, J., Wagner, K.D., Hazard, E., 2003. Validity of the mood disorder questionnaire: a general population study. Am. J. Psychiatry 160 (1), 178–180. Hirschfeld, R.M., Williams, J.B., Spitzer, R.L., Calabrese, J.R., Flynn, L., Keck Jr., P.E., Lewis, L., McElroy, S.L., Post, R.M., Rapport, D.J., Russell, J.M., Sachs, G.S., Zajecka, J., 2000. Development and validation of a screening instrument for bipolar spectrum disorder: the Mood Disorder Questionnaire. Am. J. Psychiatry 157 (11), 1873–1875. MacKinnon, D.F., Pies, R., 2006. Affective instability as rapid cycling: theoretical and clinical implications for borderline personality and bipolar spectrum disorders. Bipolar Disord. 8 (1), 1–14. Manea, L., Gilbody, S., McMillan, D., 2012. Optimal cut-off score for diagnosing depression with the Patient Health Questionnaire (PHQ-9): a meta-analysis. Can. Med. Assoc. J. 184 (3) 304-304. Marroquin, B., 2011. Interpersonal emotion regulation as a mechanism of social support in depression. Clin. Psychol. Rev. 31 (8), 1276–1290. Marwaha, S., Broome, M.R., Bebbington, P.E., Kuipers, E., Freeman, D., 2014. Mood instability and psychosis: analyses of British national survey data. Schizophr. Bull. 40 (2), 269–277. Meehl, P.E., 1995. Bootstraps taxometrics-solving the classification problem in
psychopathology. Am. Psychol. 50 (4), 266–275. Meehl, P.E., Yonce, L.J., 1994. Taxometric Analysis.1. Detecting taxonicity with 2 quantitative indicators using means above and below a sliding cut (Mambac Procedure). Psychol. Rep. 75 (1) 2-2. Meehl, P.E., Yonce, L.J., 1996. Taxometric analysis.2. Detecting taxonicity using covariance of two quantitative indicators in successive intervals of a third indicator (MAXCOV procedure). Psychol. Rep. 78 (3), 1091–1227. Meyer, T.D., Keller, F., 2003. Is there evidence for a latent class called ’‘hypomanic temperament’? J. Affect. Disord. 75 (3), 259–267. Meyer, T.J., Miller, M.L., Metzger, R.L., Borkovec, T.D., 1990. Development and validation of the Penn State Worry Questionnaire. Behav. Res. Ther. 28 (6), 487–495. Miller, C.J., Klugman, J., Berv, D.A., Rosenquist, K.J., Ghaemi, S.N., 2004. Sensitivity and specificity of the Mood Disorder Questionnaire for detecting bipolar disorder. J. Affect. Disord. 81 (2), 167–171. Oliver, M.N.I., Simons, J.S., 2004. The affective lability scales: development of a short-form measure. Person. Individ. Differ. 37 (6), 1279–1288. Perlis, R.H., Uher, R., Ostacher, M., Goldberg, J.F., Trivedi, M.H., Rush, A.J., Fava, M., 2011. Association between bipolar spectrum features and treatment outcomes in outpatients with major depressive disorder. Arch. Gen. Psychiatry 68 (4), 351–360. Phillips, M.L., Kupfer, D.J., 2013. Bipolar disorder diagnosis: challenges and future directions. Lancet 381 (9878), 1663–1671. Prisciandaro, J.J., Roberts, J.E., 2011. Evidence for the continuous latent structure of mania in the Epidemiologic Catchment Area from multiple latent structure and construct validation methodologies. Psychol. Med. 41 (3), 575–588. Core Team, R., 2014. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Rihmer, Z., Akiskal, K.K., Rihmer, A., Akiskal, H.S., 2010. Current research on affective temperaments. Curr. Opin. Psychiatry 23 (1), 12–18. Ruscio, J., Haslam, N., Ruscio, A.M., 2006. Introduction to the Taxometric Method: A Practical Guide. Lawrence Erlbaum Associates, Mahwah, N.J. Salzer, S., Stiller, C., Tacke-Pook, A., Jacobi, C., Leibing, E., 2009. Screening for Generalized Anxiety Disorder in inpatient psychosomatic rehabilitation: pathological worry and the impact of depressive symptoms. Psychosoc. Med. 6, Doc02. Trull, T.J., Solhan, M.B., Tragesser, S.L., Jahng, S., Wood, P.K., Piasecki, T.M., Watson, D., 2008. Affective instability: measuring a core feature of borderline personality disorder with ecological momentary assessment. J. Abnormal Psychol. 117 (3), 647–661. Vieta, E., Sánchez-Moreno, J., Bulbena, A., Chamorro, L., Ramos, JL., Artal, J., Pérez, F., Oliveras, M.A., Valle, J., Lahuerta, J., Angst, J., 2007. EDHIPO (Hypomania Detection Study) Group, Cross validation with the mood disorder questionnaire (MDQ) of an instrument for the detection of hypomania in Spanish: the 32 item hypomania symptom check list (HCL-32). J. Affect. Disord. 101 (1-3), 43–55. Waller, N.G., Meehl, P.E., 1998. Multivariate Taxometric Procedures: Distinguishing Types from Continua. Sage Publications, Thousand Oaks, CA. Zimmerman, M., Galione, J.N., 2011. Screening for bipolar disorder with the Mood Disorders Questionnaire: a review. Harv. Rev. Psychiatry 19 (5), 219–228. Zimmerman, M., Ruggero, C.J., Chelminski, I., Young, D., 2010. Clinical characteristics of depressed outpatients previously overdiagnosed with bipolar disorder. Compr. Psychiatry 51 (2), 99–105.
Contents lists available at ScienceDirect
Psychiatry Research journal homepage: www.elsevier.com/locate/psychres
The dimensional structure of cycling mood disorders Lloyd Balbuena 1, Marilyn Baetz, Rudy C. Bowen n Department of Psychiatry, University of Saskatchewan, 103 Hospital Drive, Saskatoon, Saskatchewan, Canada S7N 0W8
art ic l e i nf o
a b s t r a c t
Article history: Received 25 July 2014 Received in revised form 27 April 2015 Accepted 10 June 2015 Available online 26 June 2015
This study examines whether mood disorders differ fundamentally in terms of phase duration. Most clinically significant mood disorders are recurrent and cyclical. The phase duration of these cycles is part of the diagnostic criteria. Specifically, we determined whether a dimensional or taxonic latent structure better captures cycling mood disorders. 319 patients recruited from 5 psychiatrists and a psychoeducational program completed three questionnaires assessing aspects of mood cycling. These were the Affective Lability Scale-Short Form (ALS-SF), Mood Disorders Questionnaire (MDQ), and the Eysenck Neuroticism scale. Patient scores on these instruments were submitted to three taxometric procedures (MAMBAC, MAXEIG, and L-Mode). Comparison curve fit indices (CCFIs) were calculated to distinguish taxonic versus dimensional latent structure. In addition, graphs were produced for each procedure and compared with those of categorical or dimensional prototypes. The CCFIs of the three procedures ranged from 0.25 to 0.27, consistent with dimensional structure. The graphs closely resembled dimensional prototypes. Mood instability and other types of cycling moods probably conform to a dimensional latent structure. Patients with disorders featuring mood cycling might benefit from common treatments. & 2015 Elsevier Ireland Ltd. All rights reserved.
Keywords: Latent structure Mood instability Mood swings
1. Introduction The phase duration of low or high mood disorders is used by psychiatry to differentiate between mood disorders. If mood is stuck at severe low levels for two weeks or two years and no switch to high mood is reported, the diagnosis is major depression or persistent depressive disorder (American Psychiatric Association, 2013). Switching into elevated mood for one week characterizes mania, four days hypomania and shorter durations are referred to as brief hypomania or the bipolar spectrum (Angst, 1998; Cassano et al., 1999; American Psychiatric Association, 2013). In general, the DSM system shifted emphasis away from fluctuations in mood to a checklist of symptoms (Goodwin et al., 2007; American Psychiatric Association, 2013). However, the symptom differences between different categories of elevated mood (and depressed mood) are small, and syndrome severity appears to be continuous, so the diagnosis depends largely on where the cut in duration is made (Fiedorowicz et al., 2011; Abbreviations: ALS, Affective Lability Scale; CCFI, comparison curve fit indices; ENS, Eysenck Neuroticism Subscale; L-Mode, latent mode factor analysis; MDQ, Mood Disorders Questionnaire; MI, mood instability; MAMBAC, mean above minus below a cut; MAXEIG, Maximum eigenvalue n Corresponding author. Fax: þ 1 306 844 1504. E-mail address: [email protected] (R.C. Bowen). 1 Present address: Administrative Data Research Centre—Wales, Swansea University, Swansea, Wales SA2 8PP, United Kingdom. http://dx.doi.org/10.1016/j.psychres.2015.06.031 0165-1781/& 2015 Elsevier Ireland Ltd. All rights reserved.
American Psychiatric Association, 2013). Consequently, patients are assigned to diagnostic categories, which are assumed to have clearly defined boundaries. The critical question however is whether treatment and research are improved by these apparently arbitrary cuts in phase (Cassano et al., 1999). Defining what is meant by indicators, latent variable, taxon and category may be helpful to the general reader in understanding the taxometric method. Indicators are conditions such as fever or high blood pressure that are perceptible by the senses or with the aid of medical instruments. Latent variables such as aggression or mood instability (MI), on the other hand, can only be indirectly or partly inferred from more concrete phenomena such as yelling or facial expressions. MI is defined as frequent fluctuations of mood over time, which may be as brief as a few hours (Trull et al., 2008). It is usually identified with the borderline personality disorder (Marroquin, 2011) but has also been studied in patients with mood and anxiety disorders (Bowen et al., 2006), personality disorders (Trull et al., 2008), psychotic disorders (Marwaha et al., 2014) and in university students (Eid and Diener, 1999). Of crucial importance to our argument is the distinction between taxonic and dimensional structure. The distinction is captured by the intuitive notions of “differing in kind” (taxonic) and “differing by degree” (dimensional). Our concern is with latent structure, since categories such as “hot” or “cold” are artificially created by an arbitrary cut in temperature, which is continuous. As another example, “reptiles” and “mammals” are taxonic because they do not simply
290
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
differ by degree. They differ by degree in body temperature but there are innumerable and more fundamental differences. A taxonic structure may characterize the domain of mood disorders across diagnostic categories, but the boundaries are debatable. First, the distinction between “normal” people and people with mental disorders is usually made by applying a “distress or impairment” criterion to distinguish the group with mental disorders (American Psychiatric Association, 2013). The second is the distinction between regular psychiatric disorders and personality disorders. Terms that are applied to personality disorders like “enduring” “relatively stable across time” and “inflexible and pervasive” imply a stable course so phase duration is not specified (American Psychiatric Association, 2013). The third is the distinction between unipolar and bipolar mood disorders because of the shift to elevated mood in bipolar disorders (Perlis et al., 2011; Phillips and Kupfer, 2013). The boundary between unipolar and bipolar disorders is disputed and the demonstration of MI in patients with unipolar disorders further muddles the distinction (Bowen et al., 2004). Fourthly, within the broad category of major depression two subtypes called “melancholic” and “atypical” are contrasted. The first is characterized by “the lack of reactivity to usually pleasurable stimuli” while the latter shows “mood reactivity in response to actual or potential positive events.” (American Psychiatric Association, 2013). Our main concern in this paper is whether the categories formed using these distinctions are warranted. Three recent papers addressing the latent structure of elevated moods reached different but reconcilable conclusions. Meyer and Keller (2003) conducted a taxometric analysis of hypomania in two large samples (young adults in one, and adolescents in the other) and concluded that hypomania has a dimensional latent structure. Their study used the Hypomanic Personality Scale as the sole indicator of hypomania (Eckblad and Chapman, 1986). Ahmed et al. (2011) examined the latent structure of mania using the Collaborative Psychiatric Epidemiological Surveys (n 420,000). Three indicators of mania were used, each of which covered a separate symptom domain of mania defined by DSM-IV. In addition to mania, the researchers also assessed unipolar depression using another set of indicators. The authors concluded that a taxonic solution better fits the data but that significant dimensional variation within groups reflecting illness severity remained. Prisciandaro and Roberts (2011) used three different modeling strategies to analyze data from the Epidemiologic Catchment Area study and all lines of evidence converged on a dimensional solution for mania. It seems reasonable to conclude, at the minimum, that mood problems are dimensional within a homogeneous group of individuals. Whether the population as a whole is homogeneous remains to be seen. Given these inconsistent results and the disputable categorization of mood disorders, our hypothesis was that elevated mood has a dimensional latent structure.
2. Methods
Table 1 Descriptive statistics of participant scores in 5 instruments.
Mood Disorders Questionnaire* Affective Lability Scale* Short Eysenck Neuroticism Scale* PHQ-9 Penn State Worry Scale a
n
Mean (SD)
Range
Skewness
321 318 316 320 326
6.42 40.33 33.51 12.89 39.36
0–19 3–72 12–48 0–27 12–60
0.23 0.08 0.46 0.01 0.35
(3.94) (13.94) (8.78) (7.52) (13.29)
Participant scores in these three instruments were used in taxometric ana-
lysis.
We started with 345 participants but because of missing data in some questionnaires, the sample size was reduced to 319. The average age was 34.80 years and 52 per cent of our sample was female. Sixty-four percent scored 10 or higher on the PHQ-9 questionnaire, indicating depression in the range typical of patients diagnosed with major depression (Manea et al., 2012; American Psychiatric Association, 2013). Forty-seven percent scored seven or higher on the MDQ indicating possible hypomania (Hirschfeld et al., 2000). This is consistent with recent estimates of the proportion of patients with mood syndromes who report hypomanic symptoms (Angst, 2006; Zimmerman et al., 2010). Sixtythree percent scored 45 or higher on the Penn-State Worry Questionnaire (Meyer et al., 1990). This indicates the proportion of patients who reported high worry, but is not necessarily the same as generalized anxiety disorder because worry is common in major depression (Salzer et al., 2009). Table 1 shows the distribution statistics for the relevant scales. 2.2. Indicator construction We used three psychometrically validated scales measuring various aspects of mood instability—the Affective Lability ScaleShort Form (ALS-SF), the Mood Disorders Questionnaire (MDQ), and the Eysenck Neuroticism Subscale (ENS). The 18-item ALS-SF (Oliver and Simons, 2004) is an abridged version of the original instrument (Harvey et al., 1989) and assesses switches in moods across three factors: anxiety/depression, depression/elation, and anger. The MDQ (Hirschfeld et al., 2000) is a 15-item self-report instrument to assess switches to elevated or activated mood. Only the 13 symptom questions were used since it was not our objective to make a diagnosis but to study the indicators. The MDQ was developed to aid the recognition of bipolar I and II disorders (Hirschfeld et al., 2003). There are mixed opinions regarding its usefulness (see Zimmerman and Galione, 2011 for a review), with some studies questioning its usefulness for patients with impaired insight (Miller et al., 2004), and in distinguishing between bipolar I and II (Vieta et al., 2007). The ENS is composed of 12 questions that assess fluctuating mood and anxiety and depression. We used the ENS because it contains the MI factor that has been shown to predict suicidal ideation (Bowen et al., 2011). The Cronbach alphas were 0.94, 0.91, and 0.88 for the ALS-SF, ENS, and MDQ respectively. The items in each scale were summed to better meet the requirement for continuous variables in taxometric analysis.
2.1. Sample 2.3. Suitability of taxometric analysis The sample was a mixed group of outpatients from two main sources. The first group came from the practices of 5 general psychiatrists. A second group was composed of outpatients referred to a psycho-educational mood program. Patients were asked to complete a questionnaire consisting of 13 short scales before their first appointment and to bring this to the first appointment. All patients gave written consent for the use of their data for research. The study was approved by the University Behavioural Ethics Board.
We assessed the suitability of our data for taxometric analysis in two ways: first, by estimating the taxon base rate and second, by evaluating the reliability and validity of our indicators. Data with base rates lower than 0.5 are suboptimal for detecting a taxon (Meehl, 1995). Since the MDQ instrument provides the threshold score of 7 for hypomania, we used it as a criterion for estimating our taxon base rate of 0.46, which is very close to the ideal. Our indicators had acceptable skewness. Then, we examined
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
291
Table 2 Indicator reliabilities and correlations. Indicators
Reliability per procedure (in Cohen’s d)
Indicator correlations In the overall sample
MDQ ALS ENS
MAMBAC
MAXEIG
L-Mode
MDQ
ALS
1.65 2.4 1.53
1.68 2.23 1.63
1.64 2.34 1.56
1 0.56 0.34
1 0.65
participant scores in the three indicators. Our indicators exceeded Meehl’s minimum threshold of Cohen’s d¼ 1.25 (Meehl, 1995). Cohen’s d in this context measures the capability of indicators to distinguish putative taxon and complement members. It is comparable to the concept of statistical power in the sense that both are about detecting a difference if one exists. We then compared the indicator correlations in the full sample and within the putative taxon and complement groups, based on the estimated base rate. Ruscio et al. (2006) recommends that correlations within groups should be appreciably weaker than those in the full sample. Finally, we generated dimensional and taxonic comparison data for each analytical technique (described in the next section). The extent that graphs of dimensional and taxonic comparison data are distinguishable reflects the confidence in a conclusion one way or the other. Overall, our dataset and indicators proved to be well suited for taxometric analysis. The characteristics of our indicators are summarized in Table 2.
In the taxon (complement) ENS
MDQ
ALS
ENS
1
1 0.18 (0.11) 0.15 ( 0.04)
1 0.31 (0.38)
1
study, we performed L-Mode analysis by first visually inspecting the graph of factor scores and based on this graph assigning locations for the left and right modes. Because the graph did not produce any discernible modes, we left the program at the default setting of searching on either side of the mean factor score. In many cases, the graphs produced by the taxometric procedures do not fit the prototype shapes and experienced human readers are called upon to make a judgment. Fortunately, taxometric procedures produce a numerical statistic called the comparison curve fit index (CCFI) that ranges from 0 to 1 (Ruscio et al., 2006). CCFI values below 0.4 indicate a dimensional structure while those above 0.6 indicate a categorical structure. CCFI’s between 0.4 and 0.6 are equivocal and careful interpretation is recommended. All taxometric analyses were implemented in Ruscio’s taxometric (Ruscio et al., 2006) as implemented in R (R Core Team, 2014).
2.4. Statistical analysis 3. Results We submitted our indicators to three taxometric procedures: (1) mean above minus below a cut (MAMBAC); (2) Maximum eigenvalue (MAXEIG); and (3) latent mode factor analysis (L-Mode). In the MAMBAC procedure, successive cuts are made along an input variable in which subjects are rank ordered, thereby dividing the sample into intervals (Meehl and Yonce, 1994). Within each interval, the mean score in an output variable above the cut is subtracted from the mean score below the cut. This procedure distinguishes taxonic from dimensional structure because taxonic and dimensional data produce peaked and bowl-shaped graphs, respectively. The graph’s inflection point represents the score that best separates the taxon from its complement. We performed MAMBAC by assigning one indicator as output and the sum of the other two variables as input, thereby producing three curves. Fifty evenly spaced cuts beginning 25 cases from either extreme were made and 10 replications were performed. MAXEIG (Meehl and Yonce, 1996) calculates the association of two or more indicators. An input indicator is designated as the axis along which the cases are ordered by magnitude. The association between the other remaining indicators (output) is calculated at given intervals of the input. When the latent variable is dimensional, the association between output indicators should be constant along the input. By contrast, the association should be stronger in the interval where there is a highest mixing of taxon and complement members, if the latent variable is taxonic (Ruscio et al., 2006). We performed MAXEIG analysis with 25 windows having a 90 percent overlap with n ¼94 for each window. L-Mode (Waller and Meehl, 1998) has its origins in the Bartlett’s factor analytic method (Bartlett, 1937). In contrast to the sliding cuts approach taken by MAMBAC and MAXEIG, L-Mode relies on the distribution of factor scores of the latent variable. Factor scores have a mean of zero and standard deviation of 1. A bimodal shape supports a taxonic interpretation while unimodality supports dimensionality (Ruscio et al., 2006). For this
The three MI measures were moderate to strongly correlated (Pearson’s r ranged from 0.34 to 0.65). For MAMBAC, the averaged curve of the three input–output combinations more closely resembled simulated dimensional data (see Fig. 1). The graph produced by the MAXEIG was generally flat and likewise supported dimensionalit. (see Fig. 2). L-Mode produced a graph with a unimodal distribution of factor score. (see Fig. 3) The CCFIs of these three procedures were very consistent in supporting dimensionality: 0.25 for MAXEIG, 0.26 for L-Mode, and 0.27 for MAMBAC.
4. Discussion Our main finding is that naturally occurring cycling of moods probably has a dimensional latent structure. We used three psychometrically validated instruments in a clinical sample with an ideal taxon base rate and all three taxometric procedures supported a dimensional interpretation. A point also worth noting is that MI, as currently defined, refers to very short cycle phases of a few hours, but our findings suggest that cycling of moods is not qualitatively different across cycling of different phase durations (Trull et al., 2008). The scales originated from three conceptually distinct sources. The MDQ was derived from DSM-IV criteria for hypomania and the clinical experience psychiatrists familiar with bipolar mood disorders (Hirschfeld et al., 2000). The authors of the ALS were familiar with disorders with labile affect such as the Borderline Personality Disorder and were also familiar with the General Behavior Inventory that was developed to differentiate bipolar and unipolar mood disorders (Depue et al., 1981). Both the General Behavior Inventory and the ALS were originally validated on university students (Harvey et al., 1989). The Neuroticism Scale of the Eysenck Personality Inventory borrowed items from several other personality scales, including
292
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
Fig. 1. Comparison of the averaged MAMBAC curve for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
the Maudsley Personality Inventory that Eysenck had developed (Deary and Bedford, 2011). He emphasized the empirical rather than conceptual origins of the final version of the scale (Eysenck et al., 1969). At a minimum our results indicate that conditions characterized by cycling mood and mood instability are dimensional within the population of patients with mood disorders. In other words, categories defined as unipolar major depression, bipolar II, bipolar I and borderline personality disorders are probably artificial categories with a latent dimensional structure. This inference is similar
to Kraepelin’s position on variations of mood representing a “single morbid process” (MacKinnon and Pies, 2006). More recently, other authors have reached similar conclusions based on family history studies, patients’ symptoms, follow-up studies, pharmacological treatment outcome studies, and the lack of a clear diagnostic boundary. (Angst et al., 2003; MacKinnon and Pies, 2006; Rihmer et al., 2010; Phillips and Kupfer, 2013). The clinical implication is that treatments that benefit mood cycling of one kind might also benefit conditions with a different phase duration.
Fig. 2. Comparison of the averaged MAXEIG curve for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
293
Fig. 3. Comparison of the frequency distribution of estimated factor scores for our research data (black line) superimposed over simulated taxonic (left panel) and dimensional (right panel) data. Gray band in the middle represents the middle 50 percent of the simulated data. Light outer lines represent minimum and maximum values of the simulated data.
Our study is subject to three limitations. The sample size was relatively small in terms of the taxometric method. With Meehl’s recommendation of n 4300, we did meet the minimum required to complete the analysis (Meehl, 1995). This limitation is partly offset by our high taxon base rate, which was close to the ideal. The main consequence of our small sample size was our inability to complete latent class analysis. A second limitation is that we did not have access to clinical diagnoses. This group of participants did not include a specified group of “normal” individuals, but there was a range of severity as seen in the range of scores (Table 1) and one frequently used way to access the program is self-referral. The third limitation is that we relied on retrospective self-report measures of mood instability. It was reported that the MDQ is useful when manic symptoms are being recalled within the last two years, but not those over the entire lifetime (Boschloo et al., 2013). We did not have information regarding how recent the episodes were for the study participants. This limitation is balanced by the correlation of retrospective and prospective measures of MI (Anestis et al., 2010) and that our sample consisted solely of psychiatric outpatients instead of community samples for whom the instruments are less reliable. Future studies should consider taxometric analysis of prospectively collected data.
5. Conclusion The results of this study indicate that segregating patients into different categories based on length of phase of mood cycles is not justified. This suggests that established treatments for mood instability might benefit the full range of mood conditions manifesting cycles of different phase lengths. Contributors Rudy Bowen conceptualized the study, critically reviewed the
literature, collected the data, provided the indicators of affective instability, interpreted the data, re-wrote the initial draft and wrote the discussion. Lloyd Balbuena, developed the analytical strategy, implemented the analysis, wrote the initial draft, addressed reviewer comments, and produced the tables and figures. Marilyn Baetz collected the data, provided constructive comments, critically re-wrote the draft and revised the discussion. All authors approved the manuscript for submission.
Conflict of interest The authors have no conflict of interests to declare.
References Ahmed, A.O., Green, B.A., Clark, C.B., Stahl, K.C., McFarland, M.E., 2011. Latent structure of unipolar and bipolar mood symptoms. Bipolar Disord. 13 (5–6), 522–536. American Psychiatric Association, 2013. Diagnostic and statistical manual of mental disorders, 5th ed. American Psychiatric Publishing, Arlington, VA. Anestis, M.D., Selby, E.A., Crosby, R.D., Wonderlich, S.A., Engel, S.G., Joiner, T.E., 2010. A comparison of retrospective self-report versus ecological momentary assessment measures of affective lability in the examination of its relationship with bulimic symptomatology. Behav. Res. Ther. 48 (7), 607–613. Angst, J., 1998. The emerging epidemiology of hypomania and bipolar II disorder. J. Affect. Disord. 50 (2–3), 143–151. Angst, J., 2006. Do many patients with depression suffer from bipolar disorder? Can. J. Psychiatry 51 (1), 3–5. Angst, J., Gamma, A., Endrass, J., 2003. Risk factors for the bipolar and depression spectra. Acta Psychiatr. Scand. Suppl. 418, 15–19. Bartlett, M.S., 1937. The statistical conception of mental factors. Br. J. Psychol. Gen. Secti. 28, 97–104. Boschloo, L., Nolen, W.A., Spijker, A.T., Hoencamp, E., Kupka, R., Penninx, B.W.,
294
L. Balbuena et al. / Psychiatry Research 228 (2015) 289–294
Schoevers, D., 2013. The Mood Disorder Questionnaire (MDQ) for detecting (hypo)manic episodes: its validity and impact of recall bias. J. Affect. Disord. 151 (1), 203–208. Bowen, R., Baetz, M., Hawkes, J., Bowen, A., 2006. Mood variability in anxiety disorders. J. Affect. Disord. 91 (2–3), 165–170. Bowen, R., Baetz, M., Leuschen, C., Kalynchuk, L.E., 2011. Predictors of suicidal thoughts: mood instability versus neuroticism. Personal. Individ. Differ. 51 (8), 1034–1038. Bowen, R., Clark, M., Baetz, M., 2004. Mood swings in patients with anxiety disorders compared with normal controls. J. Affect. Disord. 78 (3), 185–192. Cassano, G.B., Dell’Osso, L., Frank, E., Miniati, M., Fagiolini, A., Shear, K., Pini, S., Maser, J., 1999. The bipolar spectrum: a clinical reality in search of diagnostic criteria and an assessment methodology. J. Affect. Disorder. 54 (3), 319–328. Deary, I.J., Bedford, A., 2011. Some origins and evolution of the EPQ-R (short form) neuroticism and extraversion items. Person. Individ. Differ. 50 (8), 1213–1217. Depue, R.A., Slater, J.F., Wolfstetter-Kausch, H., Klein, D., Goplerud, E., Farr, D., 1981. A behavioral paradigm for identifying persons at risk for bipolar depressive disorder: a conceptual framework and five validation studies. J. Abnormal Psychology 90 (5), 381–437. Eckblad, M., Chapman, L.J., 1986. Development and validation of a scale for hypomanic personality. J. Abnormal Psychol. 95 (3), 214–222. Eid, M., Diener, E., 1999. Intraindividual variability in affect: reliability, validity, and personality correlates. J. Person. Soc. Psychol. 76 (4), 662–676. Eysenck, H.J., Eysenck, S.B.G., Hendrickson, A., 1969. Personality Structure and Measurement. Routledge & Kegan Paul, London. Fiedorowicz, J.G., Endicott, J., Leon, A.C., Solomon, D.A., Keller, M.B., Coryell, W.H., 2011. Subthreshold hypomanic symptoms in progression from unipolar major depression to bipolar disorder. Am. J. Psychiatry 168 (1), 40–48. Goodwin, F.K., Jamison, K.R., Ghaemi, S.N., 2007. Manic-depressive Illness: Bipolar Disorders and Recurrent Depression, 2nd ed. Oxford University Press, New York, N.Y. Harvey, P.D., Greenberg, B.R., Serper, M.R., 1989. The affective lability scales—development, reliability, and validity. J. Clin. Psychol. 45 (5), 786–793. Hirschfeld, R.M., Holzer, C., Calabrese, J.R., Weissman, M., Reed, M., Davies, M., Frye, M.A., Keck, P., McElroy, S., Lewis, L., Tierce, J., Wagner, K.D., Hazard, E., 2003. Validity of the mood disorder questionnaire: a general population study. Am. J. Psychiatry 160 (1), 178–180. Hirschfeld, R.M., Williams, J.B., Spitzer, R.L., Calabrese, J.R., Flynn, L., Keck Jr., P.E., Lewis, L., McElroy, S.L., Post, R.M., Rapport, D.J., Russell, J.M., Sachs, G.S., Zajecka, J., 2000. Development and validation of a screening instrument for bipolar spectrum disorder: the Mood Disorder Questionnaire. Am. J. Psychiatry 157 (11), 1873–1875. MacKinnon, D.F., Pies, R., 2006. Affective instability as rapid cycling: theoretical and clinical implications for borderline personality and bipolar spectrum disorders. Bipolar Disord. 8 (1), 1–14. Manea, L., Gilbody, S., McMillan, D., 2012. Optimal cut-off score for diagnosing depression with the Patient Health Questionnaire (PHQ-9): a meta-analysis. Can. Med. Assoc. J. 184 (3) 304-304. Marroquin, B., 2011. Interpersonal emotion regulation as a mechanism of social support in depression. Clin. Psychol. Rev. 31 (8), 1276–1290. Marwaha, S., Broome, M.R., Bebbington, P.E., Kuipers, E., Freeman, D., 2014. Mood instability and psychosis: analyses of British national survey data. Schizophr. Bull. 40 (2), 269–277. Meehl, P.E., 1995. Bootstraps taxometrics-solving the classification problem in
psychopathology. Am. Psychol. 50 (4), 266–275. Meehl, P.E., Yonce, L.J., 1994. Taxometric Analysis.1. Detecting taxonicity with 2 quantitative indicators using means above and below a sliding cut (Mambac Procedure). Psychol. Rep. 75 (1) 2-2. Meehl, P.E., Yonce, L.J., 1996. Taxometric analysis.2. Detecting taxonicity using covariance of two quantitative indicators in successive intervals of a third indicator (MAXCOV procedure). Psychol. Rep. 78 (3), 1091–1227. Meyer, T.D., Keller, F., 2003. Is there evidence for a latent class called ’‘hypomanic temperament’? J. Affect. Disord. 75 (3), 259–267. Meyer, T.J., Miller, M.L., Metzger, R.L., Borkovec, T.D., 1990. Development and validation of the Penn State Worry Questionnaire. Behav. Res. Ther. 28 (6), 487–495. Miller, C.J., Klugman, J., Berv, D.A., Rosenquist, K.J., Ghaemi, S.N., 2004. Sensitivity and specificity of the Mood Disorder Questionnaire for detecting bipolar disorder. J. Affect. Disord. 81 (2), 167–171. Oliver, M.N.I., Simons, J.S., 2004. The affective lability scales: development of a short-form measure. Person. Individ. Differ. 37 (6), 1279–1288. Perlis, R.H., Uher, R., Ostacher, M., Goldberg, J.F., Trivedi, M.H., Rush, A.J., Fava, M., 2011. Association between bipolar spectrum features and treatment outcomes in outpatients with major depressive disorder. Arch. Gen. Psychiatry 68 (4), 351–360. Phillips, M.L., Kupfer, D.J., 2013. Bipolar disorder diagnosis: challenges and future directions. Lancet 381 (9878), 1663–1671. Prisciandaro, J.J., Roberts, J.E., 2011. Evidence for the continuous latent structure of mania in the Epidemiologic Catchment Area from multiple latent structure and construct validation methodologies. Psychol. Med. 41 (3), 575–588. Core Team, R., 2014. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Rihmer, Z., Akiskal, K.K., Rihmer, A., Akiskal, H.S., 2010. Current research on affective temperaments. Curr. Opin. Psychiatry 23 (1), 12–18. Ruscio, J., Haslam, N., Ruscio, A.M., 2006. Introduction to the Taxometric Method: A Practical Guide. Lawrence Erlbaum Associates, Mahwah, N.J. Salzer, S., Stiller, C., Tacke-Pook, A., Jacobi, C., Leibing, E., 2009. Screening for Generalized Anxiety Disorder in inpatient psychosomatic rehabilitation: pathological worry and the impact of depressive symptoms. Psychosoc. Med. 6, Doc02. Trull, T.J., Solhan, M.B., Tragesser, S.L., Jahng, S., Wood, P.K., Piasecki, T.M., Watson, D., 2008. Affective instability: measuring a core feature of borderline personality disorder with ecological momentary assessment. J. Abnormal Psychol. 117 (3), 647–661. Vieta, E., Sánchez-Moreno, J., Bulbena, A., Chamorro, L., Ramos, JL., Artal, J., Pérez, F., Oliveras, M.A., Valle, J., Lahuerta, J., Angst, J., 2007. EDHIPO (Hypomania Detection Study) Group, Cross validation with the mood disorder questionnaire (MDQ) of an instrument for the detection of hypomania in Spanish: the 32 item hypomania symptom check list (HCL-32). J. Affect. Disord. 101 (1-3), 43–55. Waller, N.G., Meehl, P.E., 1998. Multivariate Taxometric Procedures: Distinguishing Types from Continua. Sage Publications, Thousand Oaks, CA. Zimmerman, M., Galione, J.N., 2011. Screening for bipolar disorder with the Mood Disorders Questionnaire: a review. Harv. Rev. Psychiatry 19 (5), 219–228. Zimmerman, M., Ruggero, C.J., Chelminski, I., Young, D., 2010. Clinical characteristics of depressed outpatients previously overdiagnosed with bipolar disorder. Compr. Psychiatry 51 (2), 99–105.
