Journal of Affective Disorders 162 (2014) 55–60
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Research report
Screening for bipolar disorder: Does gender distort scores and case-finding estimates? Gordon Parker a,b,n, Kathryn Fletcher a,b, Stacey McCraw a,b, Howe Synnott b, Paul Friend b, Philip B. Mitchell a,b, Dusan Hadzi-Pavlovic a,b a b
School of Psychiatry, University of New South Wales, Sydney, Australia Black Dog Institute, Prince of Wales Hospital, Randwick 2031, Sydney, Australia
art ic l e i nf o
a b s t r a c t
Article history: Received 18 February 2014 Received in revised form 20 March 2014 Accepted 21 March 2014 Available online 28 March 2014
Background: Gender differences in rates of bipolar disorder have been described, with most studies reporting males as over-represented in those diagnosed with a bipolar I disorder and females overrepresented in those diagnosed with a bipolar II disorder. This could reflect true differences in prevalence or measurement error emerging from screening or case-finding measures. We examine the possible contribution of the latter by examining one screening measure—the Mood Swings Questionnaire (MSQ). Methods: We analyse MSQ data from a large sample of age- and gender-matched bipolar I and bipolar II patients (and their composite group). Gender differences were examined in terms of prevalence and severity of MSQ symptoms, MSQ sub-scales scores and total MSQ scores, employing univariate and differential item functioning (DIF) analyses. Results: Both male and female bipolar I patients reported higher total MSQ and higher mysticism MSQ sub-scale scores than their male and female bipolar II counterparts. There were no gender differences when bipolar I, bipolar II and composite bipolar groups were separately examined on both total and subscale MSQ scores, suggesting that gender does not impact on MSQ scoring. When item analyses of bipolar I and II groups were undertaken separately, a number of differences emerged, but as few were consistent across bipolar sub-types such differences could reflect chance and failure to control for multiple comparisons. The over-representation of some items in females and some in males may have contributed to the comparable total and sub-scale scores. Limitations: Large sample size and only one measure (i.e. MSQ) examined. Conclusion: As total and sub-scale MSQ scores were uninfluenced by gender we can conclude that this screening test is not confounded by gender and, if representative of other such screening measures, would indicate that any differential prevalence of the bipolar disorders identified in community studies possibly reflects gender differences in their occurrence rather than artefactual consequences of screening measures having a gender bias. & 2014 Elsevier B.V. All rights reserved.
Keywords: Bipolar Gender Screening Measure
1. Introduction Gender differences in the prevalence of bipolar I (BP I) and bipolar II (BP II) disorders have been long described, albeit with some inconsistencies in findings across studies. In relation to BP I, Kawa et al. (2005) overviewed three epidemiological studies reporting equivalent rates of BP I in men and women. Diflorio and Jones (2010) reviewed studies reporting gender data since 1980, limiting their review to clinical studies with more than 20
n Correspondence to: Black Dog Institute, Prince of Wales Hospital, Randwick 2031, Sydney, Australia. E-mail address: [email protected] (G. Parker).
http://dx.doi.org/10.1016/j.jad.2014.03.032 0165-0327/& 2014 Elsevier B.V. All rights reserved.
subjects and epidemiological studies with more than 2000 subjects. They reported comparable 12-month and lifetime gender rates in eight studies but with three studies reporting higher rates of mania or of BP I disorder in men. In a large data set of multinational studies, Merikangas et al. (2011) also reported higher lifetime rates of BP I in males. Turning to BP II, Diflorio and Jones (2010) observed that most studies reported higher rates of hypomania or of BP II in women, a finding also quantified in the multinational report by Merikangas et al. (2011). By contrast, the large National Epidemiologic Survey on Alcohol and Related Conditions (NESARC; Grant et al., 2009) failed to identify any such gender difference. Kawa et al. (2005) also noted some inconsistencies across reports and suggested that any female preponderance (if present in the clinical studies) might
56
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
reflect differential treatment seeking—although this effect would not hold in the community-based studies. Thus, any gender difference in prevalence could reflect artefactual issues (e.g. women being more likely to seek help or admit to symptoms; measurement error) or be a valid finding. If valid, it might be expected that gender differences would either reflect gender differences in severity or in over-representation of certain symptoms. In relation to the last issues, Diflorio and Jones (2010) found no support for severity differences, but did find variable evidence of differential symptom reporting with some studies reporting men scoring higher on sexual interest, problem behaviours, excitement and grandiosity, and women scoring higher on lability of mood and hallucinations. If gender contributes to differential rates of some symptoms, then any measure comprising gender-influenced symptoms might then artefactually contribute to differential prevalence rates for bipolar disorder. We are not aware of any ‘case’ measure of bipolar disorder having being examined in terms of any gender impact on item affirmation. We now report a study examining this issue in relation to one screening measure for bipolar disorder—the Mood Swings Questionnaire (MSQ; Parker et al., 2006). The MSQ seeks to distinguish bipolar (I or II collectively) from unipolar depressed subjects by the presence and experiential quality of any ‘highs’, with the measure containing 46 items derived from the literature and clinical experience (Parker et al., 2006). The MSQ has an initial screener question with two probes: “Do you ever have mood swings, and as part of such swings, have times when (i) your mood is higher than your usual sense of happiness (a “high”) and (ii) you feel quite ‘wired’, ‘energised’, ‘elevated’, ‘expansive’, and possibly ‘irritable’? For those who affirm the screening question, 46 MSQ items are then presented and with the responder required to rate the experience of each (during such periods) on a standard threepoint scale (0 ¼no more than usual, 1 ¼somewhat more than usual; 2 ¼much more than usual), with the total score calculated by summing item scores. The accuracy of the MSQ-46 measure in differentiating bipolar from unipolar patients (in tertiary referral samples) is high, with QROC analyses (weighting false positives and negatives equally) from the original study (Parker et al., 2006) generating a cut-off score of 36 or more as having 80.4% sensitivity and 92.6% specificity. In a later study (Parker et al., 2012), QROC analyses identified a cut-off score of 35 or more and with that cutoff having 84.6% sensitivity and 78.3% specificity. In this report we examine for gender differences across each of the original 46 MSQ items, sub-scale scores (based on 27 of the total 46 MSQ items) and total MSQ scores.
2. Methods 2.1. Recruitment and assessment Over the 2005–2013 period, patients attending a Sydney-based tertiary referral clinic for diagnostic and management advice for mood disorders were invited to complete a series of questionnaires for research purposes. Written informed consent was obtained as per University of New South Wales Ethics Committee requirements. As part of routine clinical assessment prior to consultation with a psychiatrist, patients completed the Mood Assessment Program (MAP; Parker et al., 2008)—a computerised tool assessing a range of features including socio-demographic, mood disorder and treatment history details. The MAP contains the MSQ screening measure for bipolar disorder with the screening question noted above and its 46 items. For those who answer the screener question negatively on the MAP, the MSQ item set is not presented and the patient is assigned a ‘probable unipolar’
diagnosis. The current report is therefore limited to those who affirmed the screener question. Following consultation with a psychiatrist, diagnostic information was recorded. Inclusion criteria for the study were: aged 16–85, no current psychosis or underlying organic issues, good comprehension of English, and ability to provide written informed consent. On the basis of their clinical interview, MAP data and seeking any corroborative data, a clinical diagnosis of BP I was assigned by the assessing psychiatrist to those who had experienced psychotic manic episodes while those who had never been psychotic during their hypomanic periods were assigned a BP II diagnosis. 2.2. Analyses We initially undertook a number of univariate analyses involving two-tailed t and chi square tests using the SPSS statistical package. Subsequently we examined invariance of MSQ scores by gender by differential item functioning (DIF) analysis, and now detail our approaches. First we factor analysed the 46 items, treating the 0–2 ratings as categorical and retaining a solution with four obliquely-rotated factors (consistent with Parker et al., 2006). The solutions in the full sample and age-matched sample were in very close agreement and provided very good fits to the data (RMSEA ¼0.060 and 0.053, respectively). The items from the reduced 27-item set loaded essentially as would be expected. For purposes of examining DIF each item was assigned to a factor on the basis of the largest loading, and for 36 items this was quite clear. Some 10 items, however, either loaded weakly on all factors or loaded moderately on more than one factor and so were assigned to more than one factor. These items were therefore only weakly associated with the dimensions used to model item responses for DIF. We examined DIF in two different ways. First, we used a version of the MIMIC (Muthén, 1989) approach. A factor analytic model, with items allowed a non-zero loading only on the factors identified above, was fitted. The factors were regressed on gender which was included as a covariate. Item regressions on gender were fixed initially at zero and modification indices from an initial run used to identify items with possible DIF and which would subsequently be allowed a non-zero weight. All models were fitted using Mplus (Muthén and Muthén, 1998–2012). Second, following an assessor's suggestion, we fitted Rasch partial credit models (PCM) to the items and examined whether there were gender effects on the residuals—that is, on the differences between subjects’ responses on items and their predicted responses based on their level of the underlying dimension (latent trait) as estimated by the PCM. We ran a PCM for each set of items and for each sample. The obtained residuals were then regressed on gender and the latent trait and their interaction, with effects for gender suggesting DIF. While previous studies examining residuals (e.g. Tennant et al., 2004), have used ANOVA rather than multiple regression, we preferred the latter to avoid having to divide the latent trait values into groups. All estimates were obtained from the R package eRm (Mair et al., 2014).
3. Results Our provisional sample comprised 1231 patients who affirmed the MSQ screener question and proceeded to complete MSQ items. Females were over-represented (59.4%) and were significantly younger than male participants (respective mean ages¼33.4 vs. 37.4, t¼5.6, po0.001). Given the significant age difference, the sample was age and gender matched, yielding 994 participants (497 male, 497 female), with 82% exceeding the cut-off score of 36 or more for a putative bipolar diagnosis. All subsequent analyses
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
57
Table 1 MSQ item analyses by gender in combined bipolar sample. MSQ items
1. Feel more confident and capable 2. Have lots of ideas, plans and goals 3. See things in a new and exciting light 4. Feel very creative with lots of ideas and plans 5. Feel that ordinary things have developed special meaning 6. Feel constantly on the go, and that others can't keep up 7. Have a wish to increase consumption of stimulants (e.g. alcohol, coffee, cigarettes) 8. Become more creative 9. Become over-involved in new plans and projects 10. Become totally confident that everything you do will succeed 11. Feel things are very vivid and crystal clear 12. Talk more 13. Spend, or wish to spend, significant amounts of money 14. Find that your thoughts race 15. Notice lots of coincidences occurring 16. Feel special connections to people 17. Become too optimistic, thinking that nothing will go wrong in the future 18. Judge that you can make decisions in a flash 19. Note that your senses are heightened and your emotions are intensified 20. Work harder, being much more motivated 21. Race from one plan to another 22. Feel one with the world and nature 23. Feel incredibly happy 24. Set lots of goals 25. Find that natural things (e.g. beaches, parks) feel especially beautiful 26. Believe that things have a ‘special meaning’ 27. Say quite outrageous things 28. Feel ‘high as a kite’, elated, ecstatic and ‘the best ever’ 29. Feel irritated 30. Feel quite carefree, not worried about anything 31. Have a much increased interest in sex (whether thoughts and/or actions) 32. Feel very impatient with people 33. Take risks, fearing no consequences 34. Need less sleep 35. See things with absolute clarity 36. Feel ‘wired’, ‘hyper’ and speeded up 37. Laugh more and find lots more things humorous 38. Read special significance into things 39. Talk over people 40. Feel extremely energetic 41. Have quite mystical experiences 42. Dress more colourfully or flamboyantly 43. Do fairly outrageous things 44. Sleep less and not feel tired 45. Sing 46. Feel angry MSQ total score (Raw scores) MSQ-mood elevation (Raw scores) MSQ-Irritability (Raw scores) MSQ-Disinhibition (Raw scores) MSQ-Mysticism (Raw scores) n
Symptom present—% within gender indicating ‘yes’
Symptom severity—mean (SD)
Male
Female
χ2
Male
94.4 94.8 89.5 92.3 63.2 77.1 70.0 85.7 79.3 76.6 77.1 92.3 77.3 92.3 63.8 65.0 64.0 73.0 85.9 88.7 81.3 63.6 81.7 76.9 70.2 50.9 71.4 82.1 63.8 75.6 78.3 78.7 75.4 75.8 73.6 85.9 77.7 57.5 72.2 86.3 34.4 40.4 61.0 70.6 56.3 51.5 – – – – –
96.6 95.0 93.1 93.4 63.4 80.9 67.2 86.1 81.3 78.9 78.1 95.0 82.1 95.8 65.0 67.2 64.2 74.8 88.7 91.1 85.5 69.4 87.1 80.3 76.9 56.9 70.4 84.9 67.0 76.6 66.0 74.2 68.4 76.2 72.4 91.3 84.3 59.5 79.1 90.3 38.2 55.3 56.7 72.3 65.0 53.5 – – – – –
2.8 0.0 4.1* 0.4 0.0 2.2 0.9 0.0 0.6 0.7 0.1 2.9 3.6 5.2* 0.1 0.5 0.0 0.4 1.8 1.6 3.2 3.8 5.6* 1.7 5.6* 3.6 0.1 1.4 1.1 0.1 18.6** 2.7 6.1* 0.0 0.2 7.3* 7.1* 0.4 6.3* 3.9 1.6 22.1** 1.8 0.8 7.8* 0.1 – – – – –
1.6 1.7 1.5 1.6 0.9 1.2 1.1 1.4 1.2 1.2 1.2 1.5 1.3 1.6 0.9 0.9 0.9 1.1 1.3 1.4 1.3 0.9 1.2 1.1 1.0 0.7 1.1 1.3 0.9 1.1 1.2 1.2 1.1 1.1 1.0 1.4 1.2 0.8 1.0 1.3 0.5 0.5 0.9 1.1 0.8 0.7 52.2 14.5 5.4 6.3 3.8
Female
(0.6) 1.6 (0.5) (0.6) 1.7 (0.6) (0.7) 1.6 (0.6) (0.6) 1.6 (0.6) (0.8) 0.9 (0.8) (0.8) 1.3 (0.8) (0.8) 1.0 (0.8) (0.7) 1.3 (0.7) (0.8) 1.2 (0.7) (0.8) 1.2 (0.8) (0.8) 1.2 (0.8) (0.6) 1.7 (0.6) (0.8) 1.3 (0.8) (0.6) 1.7 (0.5) (0.8) 1.0 (0.8) (0.8) 1.0 (0.8) (0.8) 0.9 (0.8) (0.8) 1.1 (0.8) (0.7) 1.3 (0.7) (0.7) 1.5 (0.6) (0.8) 1.4 (0.7) (0.8) 1.0 (0.8) (0.7) 1.4 (0.7) (0.7) 1.2 (0.7) (0.8) 1.2 (0.8) (0.8) 0.8 (0.8) (0.8) 1.0 (0.8) (0.7) 1.3 (0.7) (0.8) 1.0 (0.8) (0.7) 1.1 (0.7) (0.8) 1.0 (0.8) (0.7) 1.1 (0.8) (0.8) 1.0 (0.8) (0.8) 1.2 (0.8) (0.7) 1.0 (0.7) (0.7) 1.5 (0.6) (0.8) 1.3 (0.7) (0.8) 0.8 (0.8) (0.8) 1.2 (0.8) (0.7) 1.5 (0.6) (0.7) 0.5 (0.7) (0.7) 0.8 (0.8) (0.8) 0.8 (0.8) (0.8) 1.1 (0.8) (0.8) 0.9 (0.8) (0.8) 0.8 (0.8) (19.9) 54.3 (18.6) (5.3) 14.8 (4.9) (2.7) 5.7 (2.6) (3.2) 6.5 (3.1) (3.0) 4.2 (3.1)
T 1.5 0.8 2.0* 0.8 0.1 1.8 1.9 0.0 0.3 0.5 0.0 3.1 1.2 2.7* 1.0 1.0 0.1 0.0 1.5 2.8 2.1 2.8 3.6* 2.8 3.2* 1.6 0.9 1.1 1.5 0.1 4.0** 0.7 2.8* 0.2 0.5 2.2* 3.2* 0.6 3.0* 2.8 0.7 4.6** 1.4 1.2 2.5* 0.5 1.7 1.0 1.9 1.1 1.7
po 0.05. p o0.001.
nn
were based on the age/gender-matched sample and with members having a mean age of 37.4 (SD¼12.0; age range 16–77) years. MSQ item scores for males and females were compared by two univariate strategies. First, scores for each item were re-coded to reflect the presence vs. absence of each symptom (0 vs. 1 or 2) and, second, scores for each item were averaged to allow symptom severity to be compared. In addition, gender comparisons were undertaken for total MSQ scores (summed score of all items), and for the four MSQ sub-scales (Mood Elevation, Irritability, Disinhibition, Mysticism)—as empirically identified and described in the development paper (Parker et al., 2006). Table 1 reports MSQ item and sub-scale comparative analyses for male and female participants.
Irrespective of the scoring method (presence/absence vs. symptom severity), female participants were more likely than male participants to endorse the following nine MSQ items: (i) see things in a new and exciting light, (ii) find that your thoughts race, (iii) feel incredibly happy, (iv) find that natural things (e.g. beaches, parks) feel especially beautiful, (v) feel ‘wired’, ‘hyper’ and speeded up, (vi) laugh more and find lots more things humorous, (vii) talk over people, (viii) dress more colourfully or flamboyantly, and (ix) sing more. Conversely, male participants were more likely than female participants to endorse the following two MSQ items: (i) have a much increased interest in sex (whether thoughts and/or actions), and (ii) take risks, fearing no consequences. Central to study objectives,
58
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
Table 2 MSQ item analyses by gender and bipolar disorder subtype (I or II). MSQ item
Bipolar I patients (n¼ 138) Symptom present—% within gender indicating ‘yes’
1. Feel more confident and capable 2. Have lots of ideas, plans and goals 3. See things in a new and exciting light 4. Feel very creative with lots of ideas and plans 5. Feel that ordinary things have developed special meaning 6. Feel constantly on the go, and that others can't keep up 7. Have a wish to increase consumption of stimulants (e.g. alcohol, coffee, cigarettes) 8. Become more creative 9. Become over-involved in new plans and projects 10. Become totally confident that everything you do will succeed 11. Feel things are very vivid and crystal clear 12. Talk more 13. Spend, or wish to spend, significant amounts of money 14. Find that your thoughts race 15. Notice lots of coincidences occurring 16. Feel special connections to people 17. Become too optimistic, thinking that nothing will go wrong in the future 18. Judge that you can make decisions in a flash 19. Note that your senses are heightened and your emotions are intensified 20. Work harder, being much more motivated 21. Race from one plan to another 22. Feel one with the world and nature 23. Feel incredibly happy 24. Set lots of goals 25. Find that natural things (e.g. beaches, parks) feel especially beautiful 26. Believe that things have a ‘special meaning’ 27. Say quite outrageous things 28. Feel ‘high as a kite’, elated, ecstatic and ‘the best ever’ 29. Feel irritated 30. Feel quite carefree, not worried about anything 31. Have a much increased interest in sex (whether thoughts and/or actions) 32. Feel very impatient with people 33. Take risks, fearing no consequences 34. Need less sleep 35. See things with absolute clarity 36. Feel ‘wired’, ‘hyper’ and speeded up 37. Laugh more and find lots more things humorous 38. Read special significance into things 39. Talk over people 40. Feel extremely energetic 41. Have quite mystical experiences 42. Dress more colourfully or flamboyantly 43. Do fairly outrageous things 44. Sleep less and not feel tired 45. Sing 46. Feel angry MSQ total score (Raw scores) MSQ-Mood Elevation (Raw scores) MSQ-Irritability (Raw scores) MSQ-Disinhibition (Raw scores) MSQ-Mysticism (Raw scores) n
po 0.05.
Bipolar II patients (n¼ 138) Symptom severity—mean (SD)
Symptom present—% within gender indicating ‘yes’
Symptom severity—mean (SD)
Male
Female χ2
Male
Female
t
Male
Female χ2
Male
93.0 93.0 85.9 90.1
95.5 98.5 92.5 95.5
0.4 2.6 1.6 1.5
1.6 1.7 1.6 1.6
1.6 1.7 1.6 1.7
0.5 0.2 0.0 1.3
100.0 97.2 94.4 94.4
97.0 94.0 94.0 94.0
2.2 0.8 0.0 0.0
1.6 1.7 1.5 1.7
1.6 1.6 1.6 1.6
0.1 1.0 0.2 0.3
77.5
70.1
1.0
1.3
1.1
1.3
59.2
71.6
2.4
0.9
1.0
1.0
78.9
82.1
0.2
1.3
1.4
0.2
81.7
80.6
0.0
1.3
1.3
0.6
69.0
67.2
0.1
1.2
1.0
1.0
63.4
68.7
0.4
1.0
1.1
0.5
87.3 84.5
91.0 79.1
0.5 0.7
1.5 1.4
1.5 1.3
0.2 0.8
91.5 83.1
88.1 83.6
0.5 0.0
1.4 1.3
1.4 1.4
0.2 -0.6
83.1
77.6
0.7
1.4
1.2
1.5
81.7
76.1
0.6
1.2
1.2
0.5
77.5 90.1 85.9
77.6 94.0 88.1
0.0 0.7 0.1
1.3 1.6 1.5
1.2 1.6 1.4
0.7 0.7 0.7
81.7 91.5 80.3
77.6 95.5 76.1
0.4 0.9 0.4
1.2 1.5 1.3
1.2 1.7 1.3
0.1 1.5 0.4
94.4 80.3 73.2 77.5
97.0 80.6 74.6 56.7
0.6 0.0 0.0 6.8*
1.6 1.3 1.2 1.3
1.7 1.3 1.0 0.9
1.0 0.5 0.9 2.9*
97.2 67.6 67.6 66.2
97.0 56.7 68.7 70.1
0.0 1.7 0.0 0.2
1.6 0.9 0.9 0.9
1.7 0.8 1.0 1.0
0.7 1.0 0.8 0.3
80.3
79.1
0.0
1.3
1.2
0.4
74.6
71.6
0.2
1.1
1.0
0.7
97.2
91.0
2.4
1.5
1.5
0.0
87.3
88.1
0.0
1.3
1.3
0.0
88.7 90.1 81.7 78.9 83.1 76.1
86.6 88.1 73.1 79.1 74.6 77.6
0.2 0.2 1.4 0.0 1.5 0.0
1.4 1.4 1.3 1.3 1.2 1.2
1.4 1.4 1.1 1.3 1.2 1.3
0.1 0.3 1.0 0.2 0.3 0.9
94.4 83.1 60.6 78.9 78.9 69.0
92.5 92.5 68.7 86.6 86.6 83.6
0.2 2.8 1.0 1.4 1.4 4.0*
1.6 1.4 0.8 1.3 1.2 0.9
1.6 1.5 1.0 1.4 1.4 1.3
0.6 0.9 1.3 1.2 1.4 2.5*
67.6
58.2
1.3
1.0
1.0
0.4
45.1
58.2
2.4
0.6
0.8
1.6
83.1 85.9
76.1 86.6
1.0 0.0
1.3 1.5
1.2 1.4
1.2 1.0
73.2 83.1
64.2 85.1
1.3 0.1
1.0 1.3
0.9 1.3
0.9 0.1
71.8 83.1
71.6 76.1
0.0 1.0
1.0 1.3
1.1 1.2
0.2 1.3
67.6 74.6
61.2 76.1
0.6 0.0
0.9 1.1
0.9 1.0
0.3 0.4
83.1
74.6
1.5
1.3
1.2
1.1
76.1
71.6
0.3
1.0
1.0
1.3
82.7 81.7 91.5 81.7 88.7 74.6
82.1 77.6 76.1 73.1 94.0 80.6
0.0 0.4 6.1* 1.4 1.2 0.7
1.2 1.3 1.5 1.2 1.5 1.2
1.2 1.2 1.3 1.1 1.6 1.3
0.1 0.7 2.0* 0.9 0.7 0.7
80.3 88.7 73.2 85.9 88.7 77.5
71.6 73.1 80.6 71.6 88.1 85.1
1.4 5.5* 1.0 4.2* 0.0 1.3
1.3 1.3 1.0 1.2 1.4 1.2
1.0 1.0 1.3 1.0 1.5 1.2
1.7 2.3* 2.0 1.6 0.2 0.4
74.6 78.9 84.5 62.0 56.3 70.4 85.9 54.9 57.7 – – – – –
80.6 86.6 86.6 64.2 59.7 67.2 76.1 74.6 59.7 – – – – –
0.7 1.4 0.1 0.1 0.2 0.2 2.2 5.8* 0.1 – – – – –
1.1 1.1 1.4 1.0 0.9 1.1 1.5 0.9 0.8 59.9 (22.7) 16.1 (5.9) 5.8 (2.5) 7.3 (3.9) 5.7 (3.3)
1.2 1.3 1.5 1.0 0.9 1.0 1.3 1.2 0.9 58.5 (22.6) 15.4 (5.6) 6.3 (2.7) 7.2 (3.5) 5.6 (3.4)
0.9 1.8 0.4 0.1 0.0 0.3 1.3 2.0* 1.0 0.4 0.7 1.1 0.2 0.2
49.3 76.1 87.3 32.4 33.8 54.9 67.6 54.9 59.2 – – – – –
53.7 85.1 91.0 44.8 56.7 58.2 82.1 64.2 47.8 – – – – –
0.3 1.8 0.5 2.2 7.3* 0.2 3.8 1.2 1.8 – – – – –
0.7 0.8 1.1 1.2 1.4 1.4 0.4 0.5 0.4 0.8 0.8 0.8 1.0 1.3 0.9 0.9 0.8 0.7 52.4 (16.7) 54.2 (17.6) 14.9 (4.4) 15.0 (4.8) 5.7 (2.4) 5.5 (2.3) 6.2 (3.1) 6.1 (2.8) 3.4 (2.5) 3.9 (3.0)
Female
t
0.9 1.1 0.3 1.0 3.1* 0.3 2.5* 0.0 0.8 0.6 0.2 0.4 0.1 1.0
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
59
no differences were observed between male and female participants in terms of total MSQ scores, nor were differences identified across any of the four MSQ sub-scales. As noted in the Introduction a number of studies have suggested an over-representation of BP I in males and of BP II in females. We therefore repeated our analyses in a sub-set of those clinically rated as having a BP I and, second, in those having a BP II disorder. For each group we had a sample set of 138 patients, with a mean age of 40.3 years and with 67 (49%) being female. Total MSQ scores were significantly higher for BP I than for BP II patients (t¼2.5, df¼ 255, po0.05). While MSQ sub-scale scores were consistently higher for male and female BP I patients than for BP II patients, the only significant result was for the mysticism sub-scale where male BP I patients returned higher scores than male BP II patients (t¼ 4.6, po0.001) as did female BP I vs. BP II patients (t¼3.1, po0.005). Within the bipolar sub-sets, total MSQ scores were not significantly different for male and female BP I patients (59.9 vs. 58.5, t¼0.4) or male and female BP II patients (52.4 vs. 54.2, t¼0.6). The MSQ subscale scores also derived no gender differences within the BP I and BP II subgroups. Turning to individual items (Table 2), only a few showed a gender impact (i.e. po0.05). For the BP I sub-set, males rated higher on levels of optimism and judging that nothing would go wrong and on needing less sleep, while females rated higher on singing—both in terms of prevalence of experiencing such a feature and rating it as more severe. For the BP II sub-set, males rated higher (in prevalence rate and severity) on taking risks, in the prevalence of reporting seeing things with absolute clarity and on severity of needing to sleep less and not feeling tired, while females rated higher (on prevalence and severity) on judging natural things feeling especially beautiful and item 42 (dressing more colourfully). Agreement between gender-related over-represented items in the BP I and BP II sub-sets with over-represented items in the overall bipolar sample was minimal. Turning to for DIF (MIMIC) analyses, three items were identified in both samples while the full sample had four additional items. For the two multivariate models, where all identified items (for the sample) were allowed to regress on gender simultaneously, we report the probability (p) value and the odds ratio (OR) for gender (male as reference level) corresponding to responding in a higher versus immediately lower category. The three common items were increased interest in sex (OR¼ 0.68, po0.001; and OR¼0.64, po0.001 in the full and age-matched samples respectively); take risks (OR¼0.71, po0.001; and OR¼0.67, po0.001); and dress more colourfully (OR¼1.36, po0.001; and OR¼ 1.37, po0.001), but the last was not well-defined by the factors. In the full sample, DIF was found for increased consumption of stimulants (OR¼0.83, po0.001); (ii) talk more (OR¼ 1.30, po0.001); set lots of goals (OR¼1.25, po0.001); and sing (OR¼1.20, po0.001) but the first two were not well defined by the factors. All models were fitted using Mplus (Muthén and Muthén, 1998–2012). In examining for gender effects on residuals, this method identified a large number of items (including all those above) as showing DIF—approximately two-thirds in the full sample, and with half of those not identified in the age-matched sample. Two caveats need to be made in considering these findings. First, with the very large sample sizes, the power to detect even very small effects is considerable. Second, while effects were significant, the multiple R2 for the models typically accounted for less than 2% of the variance and only rarely for 5–7%, suggesting that any DIF effects were quite small and usually trivial.
were distinctly influenced by gender and, if so, which might support the possibility that previously identified gender differences in lifetime rates of the bipolar disorders might not reflect differential disorder prevalence but be (entirely or partially) an artefact of measurement. As noted, we are not aware of any previous study that has examined whether screening or diagnostic measures of bipolar disorder are artefactually influenced by gender (i.e. by having gender-dimorphic behaviours or having items subject to biases such as amplification or denial being more likely by women or by men respectively). Our univariate analyses – undertaken on a very large sample – quantified no impact of gender on total MSQ scores or across the four MSQ sub-scales – whether analyses examined the combined bipolar sample or examined BP I and BP II subjects separately. While a number of items did show gender differences in prevalence and/or severity – and varied across the BP I and BP II subsets – such items were rarely differentiating in the whole bipolar sample. The DIF analyses identified a few more gender-weighted items. Such a lack of consistency is likely to reflect some true gender-dimorphic differences (perhaps BP II women being more likely to dress colourfully, talk more, sing more and set more goals, and BP II men more likely to have an increased interest in sex and be more likely to take risks) but is also likely to reflect both chance (i.e. there being multiple comparisons made without statistical correction) as well as the large sample size generating significant (but quite small) differences. Such nuances do not discount the main finding that the MSQ total score is not susceptible to any substantive gender impact and that bipolar hypomanic and manic episodes are not marked by features that differ distinctly between men and women.
4. Discussion
References
The key objective of the study was to determine if scores reported by those with a bipolar condition on the MSQ measure
Diflorio, A., Jones, I., 2010. Is sex important? Gender differences in bipolar disorder. Int. Rev. Psychiatry 22, 437–452.
5. Conclusion Taking all such considerations into account allows two relatively clear conclusions. First, at the ‘micro’ level, we have established that total and sub-scale MSQ scores are not subject to gender biases and that analyses involving the MSQ do not therefore need to control for any gender effect. We did find that females scored higher on some items and males on others, but such gender differences seemingly evened out across the MSQ total measure and scale scores. Second, if the MSQ is representative of other screening measures then a more ‘macro’ finding is suggested. In essence, studies reporting differential rates of bipolar disorders by gender are possibly detecting true gender differences in prevalence rather than reflect methodological error whereby the screening measure may affect distortions by having overrepresented gender-dimorphic behaviours, or is open to differential response biases such as denial or amplification. Role of funding source This study was supported by an NHMRC program grant (1037196). The NHMRC did not participate in the study design, nor in the data collection and analysis process, or in any phase of the manuscript preparation.
Conflict of interest No conflict declared.
Acknowledgements This study was funded by an NHMRC Program Grant (1037196). We thank Georgie McClure for data entry.
60
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
Grant, J.E., Desai, R.A., Potenza, M.N., 2009. Relationship of nicotine dependence, subsyndromal and pathological gambling, and other psychiatric disorders: data from the National Epidemiologic Survey on Alcohol and Related Conditions. J. Clin. Psychiatry 70, 334–343. Kawa, I., Carter, J.D., Joyce, P.R., Doughty, C.J., Frampton, C.M., Elisabeth, Wells, et al., 2005. Gender differences in bipolar disorder: age of onset, course, comorbidity, and symptom presentkation. Bipolar Disord. 7, 119–125. Mair, P., Hatzinger, R., Maier M.J., 2014. eRm: Extended Rasch Modeling. R Package Version 0.15-4. 〈http://erm.r-forge.r-project.org/〉. Merikangas, K.R., Jin, R., He, J.-P., Kessler, R.C., Lee, S., Sampson, N.A, et al., 2011. Prevalence and correlates of bipolar spectrum disorder in the World Mental Health Survey Initiative. Arch. Gen. Psychiatry 68, 241–251. Muthén, B., 1989. Latent variable modeling in heterogeneous populations. Psychometrika 54, 557–585. Muthén, L.K., Muthén, B.O., 1998–2012. Mplus User's Guide, seventh ed. Muthén & Muthén, Los Angeles, CA.
Parker, G., Fletcher, K., Hyett, M., 2008. The mood assessment program: a computerised diagnostic tool for deriving management plans for mood disorders. Med. J. Aust. 188, S126–S128. Parker, G., Hadzi-Pavlovic, D., Tully, L., 2006. Distinguishing bipolar and unipolar disorders: an isomer model. J. Affect. Disord. 96, 67–73. Parker, G., Graham, R., Hadzi-Pavlovic, D., Fletcher, K., Hong, M., Futeran, S., 2012. Further examination of the utility and comparative properties of the MSQ and MDQ bipolar screening measures. J. Affect. Disord. 138, 104–109. Tennant, A., Penta, M., Tesio, L., Grimby, G., Thonnard, J.L., Slade, A., Lawton, G., Simone, A., Carter, J., Lundgren-Nilsson, A., Tripolski, M., Ring, H., Biering-Sørensen, F., Marincek, C., Burger, H., Phillips, S., 2004. Assessing and adjusting for cross-cultural validity of impairment and activity limitation scales through differential item functioning within the framework of the Rasch model: the PRO-ESOR project. Med. Care 42 (1 Suppl.), I37–48.
Contents lists available at ScienceDirect
Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad
Research report
Screening for bipolar disorder: Does gender distort scores and case-finding estimates? Gordon Parker a,b,n, Kathryn Fletcher a,b, Stacey McCraw a,b, Howe Synnott b, Paul Friend b, Philip B. Mitchell a,b, Dusan Hadzi-Pavlovic a,b a b
School of Psychiatry, University of New South Wales, Sydney, Australia Black Dog Institute, Prince of Wales Hospital, Randwick 2031, Sydney, Australia
art ic l e i nf o
a b s t r a c t
Article history: Received 18 February 2014 Received in revised form 20 March 2014 Accepted 21 March 2014 Available online 28 March 2014
Background: Gender differences in rates of bipolar disorder have been described, with most studies reporting males as over-represented in those diagnosed with a bipolar I disorder and females overrepresented in those diagnosed with a bipolar II disorder. This could reflect true differences in prevalence or measurement error emerging from screening or case-finding measures. We examine the possible contribution of the latter by examining one screening measure—the Mood Swings Questionnaire (MSQ). Methods: We analyse MSQ data from a large sample of age- and gender-matched bipolar I and bipolar II patients (and their composite group). Gender differences were examined in terms of prevalence and severity of MSQ symptoms, MSQ sub-scales scores and total MSQ scores, employing univariate and differential item functioning (DIF) analyses. Results: Both male and female bipolar I patients reported higher total MSQ and higher mysticism MSQ sub-scale scores than their male and female bipolar II counterparts. There were no gender differences when bipolar I, bipolar II and composite bipolar groups were separately examined on both total and subscale MSQ scores, suggesting that gender does not impact on MSQ scoring. When item analyses of bipolar I and II groups were undertaken separately, a number of differences emerged, but as few were consistent across bipolar sub-types such differences could reflect chance and failure to control for multiple comparisons. The over-representation of some items in females and some in males may have contributed to the comparable total and sub-scale scores. Limitations: Large sample size and only one measure (i.e. MSQ) examined. Conclusion: As total and sub-scale MSQ scores were uninfluenced by gender we can conclude that this screening test is not confounded by gender and, if representative of other such screening measures, would indicate that any differential prevalence of the bipolar disorders identified in community studies possibly reflects gender differences in their occurrence rather than artefactual consequences of screening measures having a gender bias. & 2014 Elsevier B.V. All rights reserved.
Keywords: Bipolar Gender Screening Measure
1. Introduction Gender differences in the prevalence of bipolar I (BP I) and bipolar II (BP II) disorders have been long described, albeit with some inconsistencies in findings across studies. In relation to BP I, Kawa et al. (2005) overviewed three epidemiological studies reporting equivalent rates of BP I in men and women. Diflorio and Jones (2010) reviewed studies reporting gender data since 1980, limiting their review to clinical studies with more than 20
n Correspondence to: Black Dog Institute, Prince of Wales Hospital, Randwick 2031, Sydney, Australia. E-mail address: [email protected] (G. Parker).
http://dx.doi.org/10.1016/j.jad.2014.03.032 0165-0327/& 2014 Elsevier B.V. All rights reserved.
subjects and epidemiological studies with more than 2000 subjects. They reported comparable 12-month and lifetime gender rates in eight studies but with three studies reporting higher rates of mania or of BP I disorder in men. In a large data set of multinational studies, Merikangas et al. (2011) also reported higher lifetime rates of BP I in males. Turning to BP II, Diflorio and Jones (2010) observed that most studies reported higher rates of hypomania or of BP II in women, a finding also quantified in the multinational report by Merikangas et al. (2011). By contrast, the large National Epidemiologic Survey on Alcohol and Related Conditions (NESARC; Grant et al., 2009) failed to identify any such gender difference. Kawa et al. (2005) also noted some inconsistencies across reports and suggested that any female preponderance (if present in the clinical studies) might
56
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
reflect differential treatment seeking—although this effect would not hold in the community-based studies. Thus, any gender difference in prevalence could reflect artefactual issues (e.g. women being more likely to seek help or admit to symptoms; measurement error) or be a valid finding. If valid, it might be expected that gender differences would either reflect gender differences in severity or in over-representation of certain symptoms. In relation to the last issues, Diflorio and Jones (2010) found no support for severity differences, but did find variable evidence of differential symptom reporting with some studies reporting men scoring higher on sexual interest, problem behaviours, excitement and grandiosity, and women scoring higher on lability of mood and hallucinations. If gender contributes to differential rates of some symptoms, then any measure comprising gender-influenced symptoms might then artefactually contribute to differential prevalence rates for bipolar disorder. We are not aware of any ‘case’ measure of bipolar disorder having being examined in terms of any gender impact on item affirmation. We now report a study examining this issue in relation to one screening measure for bipolar disorder—the Mood Swings Questionnaire (MSQ; Parker et al., 2006). The MSQ seeks to distinguish bipolar (I or II collectively) from unipolar depressed subjects by the presence and experiential quality of any ‘highs’, with the measure containing 46 items derived from the literature and clinical experience (Parker et al., 2006). The MSQ has an initial screener question with two probes: “Do you ever have mood swings, and as part of such swings, have times when (i) your mood is higher than your usual sense of happiness (a “high”) and (ii) you feel quite ‘wired’, ‘energised’, ‘elevated’, ‘expansive’, and possibly ‘irritable’? For those who affirm the screening question, 46 MSQ items are then presented and with the responder required to rate the experience of each (during such periods) on a standard threepoint scale (0 ¼no more than usual, 1 ¼somewhat more than usual; 2 ¼much more than usual), with the total score calculated by summing item scores. The accuracy of the MSQ-46 measure in differentiating bipolar from unipolar patients (in tertiary referral samples) is high, with QROC analyses (weighting false positives and negatives equally) from the original study (Parker et al., 2006) generating a cut-off score of 36 or more as having 80.4% sensitivity and 92.6% specificity. In a later study (Parker et al., 2012), QROC analyses identified a cut-off score of 35 or more and with that cutoff having 84.6% sensitivity and 78.3% specificity. In this report we examine for gender differences across each of the original 46 MSQ items, sub-scale scores (based on 27 of the total 46 MSQ items) and total MSQ scores.
2. Methods 2.1. Recruitment and assessment Over the 2005–2013 period, patients attending a Sydney-based tertiary referral clinic for diagnostic and management advice for mood disorders were invited to complete a series of questionnaires for research purposes. Written informed consent was obtained as per University of New South Wales Ethics Committee requirements. As part of routine clinical assessment prior to consultation with a psychiatrist, patients completed the Mood Assessment Program (MAP; Parker et al., 2008)—a computerised tool assessing a range of features including socio-demographic, mood disorder and treatment history details. The MAP contains the MSQ screening measure for bipolar disorder with the screening question noted above and its 46 items. For those who answer the screener question negatively on the MAP, the MSQ item set is not presented and the patient is assigned a ‘probable unipolar’
diagnosis. The current report is therefore limited to those who affirmed the screener question. Following consultation with a psychiatrist, diagnostic information was recorded. Inclusion criteria for the study were: aged 16–85, no current psychosis or underlying organic issues, good comprehension of English, and ability to provide written informed consent. On the basis of their clinical interview, MAP data and seeking any corroborative data, a clinical diagnosis of BP I was assigned by the assessing psychiatrist to those who had experienced psychotic manic episodes while those who had never been psychotic during their hypomanic periods were assigned a BP II diagnosis. 2.2. Analyses We initially undertook a number of univariate analyses involving two-tailed t and chi square tests using the SPSS statistical package. Subsequently we examined invariance of MSQ scores by gender by differential item functioning (DIF) analysis, and now detail our approaches. First we factor analysed the 46 items, treating the 0–2 ratings as categorical and retaining a solution with four obliquely-rotated factors (consistent with Parker et al., 2006). The solutions in the full sample and age-matched sample were in very close agreement and provided very good fits to the data (RMSEA ¼0.060 and 0.053, respectively). The items from the reduced 27-item set loaded essentially as would be expected. For purposes of examining DIF each item was assigned to a factor on the basis of the largest loading, and for 36 items this was quite clear. Some 10 items, however, either loaded weakly on all factors or loaded moderately on more than one factor and so were assigned to more than one factor. These items were therefore only weakly associated with the dimensions used to model item responses for DIF. We examined DIF in two different ways. First, we used a version of the MIMIC (Muthén, 1989) approach. A factor analytic model, with items allowed a non-zero loading only on the factors identified above, was fitted. The factors were regressed on gender which was included as a covariate. Item regressions on gender were fixed initially at zero and modification indices from an initial run used to identify items with possible DIF and which would subsequently be allowed a non-zero weight. All models were fitted using Mplus (Muthén and Muthén, 1998–2012). Second, following an assessor's suggestion, we fitted Rasch partial credit models (PCM) to the items and examined whether there were gender effects on the residuals—that is, on the differences between subjects’ responses on items and their predicted responses based on their level of the underlying dimension (latent trait) as estimated by the PCM. We ran a PCM for each set of items and for each sample. The obtained residuals were then regressed on gender and the latent trait and their interaction, with effects for gender suggesting DIF. While previous studies examining residuals (e.g. Tennant et al., 2004), have used ANOVA rather than multiple regression, we preferred the latter to avoid having to divide the latent trait values into groups. All estimates were obtained from the R package eRm (Mair et al., 2014).
3. Results Our provisional sample comprised 1231 patients who affirmed the MSQ screener question and proceeded to complete MSQ items. Females were over-represented (59.4%) and were significantly younger than male participants (respective mean ages¼33.4 vs. 37.4, t¼5.6, po0.001). Given the significant age difference, the sample was age and gender matched, yielding 994 participants (497 male, 497 female), with 82% exceeding the cut-off score of 36 or more for a putative bipolar diagnosis. All subsequent analyses
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
57
Table 1 MSQ item analyses by gender in combined bipolar sample. MSQ items
1. Feel more confident and capable 2. Have lots of ideas, plans and goals 3. See things in a new and exciting light 4. Feel very creative with lots of ideas and plans 5. Feel that ordinary things have developed special meaning 6. Feel constantly on the go, and that others can't keep up 7. Have a wish to increase consumption of stimulants (e.g. alcohol, coffee, cigarettes) 8. Become more creative 9. Become over-involved in new plans and projects 10. Become totally confident that everything you do will succeed 11. Feel things are very vivid and crystal clear 12. Talk more 13. Spend, or wish to spend, significant amounts of money 14. Find that your thoughts race 15. Notice lots of coincidences occurring 16. Feel special connections to people 17. Become too optimistic, thinking that nothing will go wrong in the future 18. Judge that you can make decisions in a flash 19. Note that your senses are heightened and your emotions are intensified 20. Work harder, being much more motivated 21. Race from one plan to another 22. Feel one with the world and nature 23. Feel incredibly happy 24. Set lots of goals 25. Find that natural things (e.g. beaches, parks) feel especially beautiful 26. Believe that things have a ‘special meaning’ 27. Say quite outrageous things 28. Feel ‘high as a kite’, elated, ecstatic and ‘the best ever’ 29. Feel irritated 30. Feel quite carefree, not worried about anything 31. Have a much increased interest in sex (whether thoughts and/or actions) 32. Feel very impatient with people 33. Take risks, fearing no consequences 34. Need less sleep 35. See things with absolute clarity 36. Feel ‘wired’, ‘hyper’ and speeded up 37. Laugh more and find lots more things humorous 38. Read special significance into things 39. Talk over people 40. Feel extremely energetic 41. Have quite mystical experiences 42. Dress more colourfully or flamboyantly 43. Do fairly outrageous things 44. Sleep less and not feel tired 45. Sing 46. Feel angry MSQ total score (Raw scores) MSQ-mood elevation (Raw scores) MSQ-Irritability (Raw scores) MSQ-Disinhibition (Raw scores) MSQ-Mysticism (Raw scores) n
Symptom present—% within gender indicating ‘yes’
Symptom severity—mean (SD)
Male
Female
χ2
Male
94.4 94.8 89.5 92.3 63.2 77.1 70.0 85.7 79.3 76.6 77.1 92.3 77.3 92.3 63.8 65.0 64.0 73.0 85.9 88.7 81.3 63.6 81.7 76.9 70.2 50.9 71.4 82.1 63.8 75.6 78.3 78.7 75.4 75.8 73.6 85.9 77.7 57.5 72.2 86.3 34.4 40.4 61.0 70.6 56.3 51.5 – – – – –
96.6 95.0 93.1 93.4 63.4 80.9 67.2 86.1 81.3 78.9 78.1 95.0 82.1 95.8 65.0 67.2 64.2 74.8 88.7 91.1 85.5 69.4 87.1 80.3 76.9 56.9 70.4 84.9 67.0 76.6 66.0 74.2 68.4 76.2 72.4 91.3 84.3 59.5 79.1 90.3 38.2 55.3 56.7 72.3 65.0 53.5 – – – – –
2.8 0.0 4.1* 0.4 0.0 2.2 0.9 0.0 0.6 0.7 0.1 2.9 3.6 5.2* 0.1 0.5 0.0 0.4 1.8 1.6 3.2 3.8 5.6* 1.7 5.6* 3.6 0.1 1.4 1.1 0.1 18.6** 2.7 6.1* 0.0 0.2 7.3* 7.1* 0.4 6.3* 3.9 1.6 22.1** 1.8 0.8 7.8* 0.1 – – – – –
1.6 1.7 1.5 1.6 0.9 1.2 1.1 1.4 1.2 1.2 1.2 1.5 1.3 1.6 0.9 0.9 0.9 1.1 1.3 1.4 1.3 0.9 1.2 1.1 1.0 0.7 1.1 1.3 0.9 1.1 1.2 1.2 1.1 1.1 1.0 1.4 1.2 0.8 1.0 1.3 0.5 0.5 0.9 1.1 0.8 0.7 52.2 14.5 5.4 6.3 3.8
Female
(0.6) 1.6 (0.5) (0.6) 1.7 (0.6) (0.7) 1.6 (0.6) (0.6) 1.6 (0.6) (0.8) 0.9 (0.8) (0.8) 1.3 (0.8) (0.8) 1.0 (0.8) (0.7) 1.3 (0.7) (0.8) 1.2 (0.7) (0.8) 1.2 (0.8) (0.8) 1.2 (0.8) (0.6) 1.7 (0.6) (0.8) 1.3 (0.8) (0.6) 1.7 (0.5) (0.8) 1.0 (0.8) (0.8) 1.0 (0.8) (0.8) 0.9 (0.8) (0.8) 1.1 (0.8) (0.7) 1.3 (0.7) (0.7) 1.5 (0.6) (0.8) 1.4 (0.7) (0.8) 1.0 (0.8) (0.7) 1.4 (0.7) (0.7) 1.2 (0.7) (0.8) 1.2 (0.8) (0.8) 0.8 (0.8) (0.8) 1.0 (0.8) (0.7) 1.3 (0.7) (0.8) 1.0 (0.8) (0.7) 1.1 (0.7) (0.8) 1.0 (0.8) (0.7) 1.1 (0.8) (0.8) 1.0 (0.8) (0.8) 1.2 (0.8) (0.7) 1.0 (0.7) (0.7) 1.5 (0.6) (0.8) 1.3 (0.7) (0.8) 0.8 (0.8) (0.8) 1.2 (0.8) (0.7) 1.5 (0.6) (0.7) 0.5 (0.7) (0.7) 0.8 (0.8) (0.8) 0.8 (0.8) (0.8) 1.1 (0.8) (0.8) 0.9 (0.8) (0.8) 0.8 (0.8) (19.9) 54.3 (18.6) (5.3) 14.8 (4.9) (2.7) 5.7 (2.6) (3.2) 6.5 (3.1) (3.0) 4.2 (3.1)
T 1.5 0.8 2.0* 0.8 0.1 1.8 1.9 0.0 0.3 0.5 0.0 3.1 1.2 2.7* 1.0 1.0 0.1 0.0 1.5 2.8 2.1 2.8 3.6* 2.8 3.2* 1.6 0.9 1.1 1.5 0.1 4.0** 0.7 2.8* 0.2 0.5 2.2* 3.2* 0.6 3.0* 2.8 0.7 4.6** 1.4 1.2 2.5* 0.5 1.7 1.0 1.9 1.1 1.7
po 0.05. p o0.001.
nn
were based on the age/gender-matched sample and with members having a mean age of 37.4 (SD¼12.0; age range 16–77) years. MSQ item scores for males and females were compared by two univariate strategies. First, scores for each item were re-coded to reflect the presence vs. absence of each symptom (0 vs. 1 or 2) and, second, scores for each item were averaged to allow symptom severity to be compared. In addition, gender comparisons were undertaken for total MSQ scores (summed score of all items), and for the four MSQ sub-scales (Mood Elevation, Irritability, Disinhibition, Mysticism)—as empirically identified and described in the development paper (Parker et al., 2006). Table 1 reports MSQ item and sub-scale comparative analyses for male and female participants.
Irrespective of the scoring method (presence/absence vs. symptom severity), female participants were more likely than male participants to endorse the following nine MSQ items: (i) see things in a new and exciting light, (ii) find that your thoughts race, (iii) feel incredibly happy, (iv) find that natural things (e.g. beaches, parks) feel especially beautiful, (v) feel ‘wired’, ‘hyper’ and speeded up, (vi) laugh more and find lots more things humorous, (vii) talk over people, (viii) dress more colourfully or flamboyantly, and (ix) sing more. Conversely, male participants were more likely than female participants to endorse the following two MSQ items: (i) have a much increased interest in sex (whether thoughts and/or actions), and (ii) take risks, fearing no consequences. Central to study objectives,
58
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
Table 2 MSQ item analyses by gender and bipolar disorder subtype (I or II). MSQ item
Bipolar I patients (n¼ 138) Symptom present—% within gender indicating ‘yes’
1. Feel more confident and capable 2. Have lots of ideas, plans and goals 3. See things in a new and exciting light 4. Feel very creative with lots of ideas and plans 5. Feel that ordinary things have developed special meaning 6. Feel constantly on the go, and that others can't keep up 7. Have a wish to increase consumption of stimulants (e.g. alcohol, coffee, cigarettes) 8. Become more creative 9. Become over-involved in new plans and projects 10. Become totally confident that everything you do will succeed 11. Feel things are very vivid and crystal clear 12. Talk more 13. Spend, or wish to spend, significant amounts of money 14. Find that your thoughts race 15. Notice lots of coincidences occurring 16. Feel special connections to people 17. Become too optimistic, thinking that nothing will go wrong in the future 18. Judge that you can make decisions in a flash 19. Note that your senses are heightened and your emotions are intensified 20. Work harder, being much more motivated 21. Race from one plan to another 22. Feel one with the world and nature 23. Feel incredibly happy 24. Set lots of goals 25. Find that natural things (e.g. beaches, parks) feel especially beautiful 26. Believe that things have a ‘special meaning’ 27. Say quite outrageous things 28. Feel ‘high as a kite’, elated, ecstatic and ‘the best ever’ 29. Feel irritated 30. Feel quite carefree, not worried about anything 31. Have a much increased interest in sex (whether thoughts and/or actions) 32. Feel very impatient with people 33. Take risks, fearing no consequences 34. Need less sleep 35. See things with absolute clarity 36. Feel ‘wired’, ‘hyper’ and speeded up 37. Laugh more and find lots more things humorous 38. Read special significance into things 39. Talk over people 40. Feel extremely energetic 41. Have quite mystical experiences 42. Dress more colourfully or flamboyantly 43. Do fairly outrageous things 44. Sleep less and not feel tired 45. Sing 46. Feel angry MSQ total score (Raw scores) MSQ-Mood Elevation (Raw scores) MSQ-Irritability (Raw scores) MSQ-Disinhibition (Raw scores) MSQ-Mysticism (Raw scores) n
po 0.05.
Bipolar II patients (n¼ 138) Symptom severity—mean (SD)
Symptom present—% within gender indicating ‘yes’
Symptom severity—mean (SD)
Male
Female χ2
Male
Female
t
Male
Female χ2
Male
93.0 93.0 85.9 90.1
95.5 98.5 92.5 95.5
0.4 2.6 1.6 1.5
1.6 1.7 1.6 1.6
1.6 1.7 1.6 1.7
0.5 0.2 0.0 1.3
100.0 97.2 94.4 94.4
97.0 94.0 94.0 94.0
2.2 0.8 0.0 0.0
1.6 1.7 1.5 1.7
1.6 1.6 1.6 1.6
0.1 1.0 0.2 0.3
77.5
70.1
1.0
1.3
1.1
1.3
59.2
71.6
2.4
0.9
1.0
1.0
78.9
82.1
0.2
1.3
1.4
0.2
81.7
80.6
0.0
1.3
1.3
0.6
69.0
67.2
0.1
1.2
1.0
1.0
63.4
68.7
0.4
1.0
1.1
0.5
87.3 84.5
91.0 79.1
0.5 0.7
1.5 1.4
1.5 1.3
0.2 0.8
91.5 83.1
88.1 83.6
0.5 0.0
1.4 1.3
1.4 1.4
0.2 -0.6
83.1
77.6
0.7
1.4
1.2
1.5
81.7
76.1
0.6
1.2
1.2
0.5
77.5 90.1 85.9
77.6 94.0 88.1
0.0 0.7 0.1
1.3 1.6 1.5
1.2 1.6 1.4
0.7 0.7 0.7
81.7 91.5 80.3
77.6 95.5 76.1
0.4 0.9 0.4
1.2 1.5 1.3
1.2 1.7 1.3
0.1 1.5 0.4
94.4 80.3 73.2 77.5
97.0 80.6 74.6 56.7
0.6 0.0 0.0 6.8*
1.6 1.3 1.2 1.3
1.7 1.3 1.0 0.9
1.0 0.5 0.9 2.9*
97.2 67.6 67.6 66.2
97.0 56.7 68.7 70.1
0.0 1.7 0.0 0.2
1.6 0.9 0.9 0.9
1.7 0.8 1.0 1.0
0.7 1.0 0.8 0.3
80.3
79.1
0.0
1.3
1.2
0.4
74.6
71.6
0.2
1.1
1.0
0.7
97.2
91.0
2.4
1.5
1.5
0.0
87.3
88.1
0.0
1.3
1.3
0.0
88.7 90.1 81.7 78.9 83.1 76.1
86.6 88.1 73.1 79.1 74.6 77.6
0.2 0.2 1.4 0.0 1.5 0.0
1.4 1.4 1.3 1.3 1.2 1.2
1.4 1.4 1.1 1.3 1.2 1.3
0.1 0.3 1.0 0.2 0.3 0.9
94.4 83.1 60.6 78.9 78.9 69.0
92.5 92.5 68.7 86.6 86.6 83.6
0.2 2.8 1.0 1.4 1.4 4.0*
1.6 1.4 0.8 1.3 1.2 0.9
1.6 1.5 1.0 1.4 1.4 1.3
0.6 0.9 1.3 1.2 1.4 2.5*
67.6
58.2
1.3
1.0
1.0
0.4
45.1
58.2
2.4
0.6
0.8
1.6
83.1 85.9
76.1 86.6
1.0 0.0
1.3 1.5
1.2 1.4
1.2 1.0
73.2 83.1
64.2 85.1
1.3 0.1
1.0 1.3
0.9 1.3
0.9 0.1
71.8 83.1
71.6 76.1
0.0 1.0
1.0 1.3
1.1 1.2
0.2 1.3
67.6 74.6
61.2 76.1
0.6 0.0
0.9 1.1
0.9 1.0
0.3 0.4
83.1
74.6
1.5
1.3
1.2
1.1
76.1
71.6
0.3
1.0
1.0
1.3
82.7 81.7 91.5 81.7 88.7 74.6
82.1 77.6 76.1 73.1 94.0 80.6
0.0 0.4 6.1* 1.4 1.2 0.7
1.2 1.3 1.5 1.2 1.5 1.2
1.2 1.2 1.3 1.1 1.6 1.3
0.1 0.7 2.0* 0.9 0.7 0.7
80.3 88.7 73.2 85.9 88.7 77.5
71.6 73.1 80.6 71.6 88.1 85.1
1.4 5.5* 1.0 4.2* 0.0 1.3
1.3 1.3 1.0 1.2 1.4 1.2
1.0 1.0 1.3 1.0 1.5 1.2
1.7 2.3* 2.0 1.6 0.2 0.4
74.6 78.9 84.5 62.0 56.3 70.4 85.9 54.9 57.7 – – – – –
80.6 86.6 86.6 64.2 59.7 67.2 76.1 74.6 59.7 – – – – –
0.7 1.4 0.1 0.1 0.2 0.2 2.2 5.8* 0.1 – – – – –
1.1 1.1 1.4 1.0 0.9 1.1 1.5 0.9 0.8 59.9 (22.7) 16.1 (5.9) 5.8 (2.5) 7.3 (3.9) 5.7 (3.3)
1.2 1.3 1.5 1.0 0.9 1.0 1.3 1.2 0.9 58.5 (22.6) 15.4 (5.6) 6.3 (2.7) 7.2 (3.5) 5.6 (3.4)
0.9 1.8 0.4 0.1 0.0 0.3 1.3 2.0* 1.0 0.4 0.7 1.1 0.2 0.2
49.3 76.1 87.3 32.4 33.8 54.9 67.6 54.9 59.2 – – – – –
53.7 85.1 91.0 44.8 56.7 58.2 82.1 64.2 47.8 – – – – –
0.3 1.8 0.5 2.2 7.3* 0.2 3.8 1.2 1.8 – – – – –
0.7 0.8 1.1 1.2 1.4 1.4 0.4 0.5 0.4 0.8 0.8 0.8 1.0 1.3 0.9 0.9 0.8 0.7 52.4 (16.7) 54.2 (17.6) 14.9 (4.4) 15.0 (4.8) 5.7 (2.4) 5.5 (2.3) 6.2 (3.1) 6.1 (2.8) 3.4 (2.5) 3.9 (3.0)
Female
t
0.9 1.1 0.3 1.0 3.1* 0.3 2.5* 0.0 0.8 0.6 0.2 0.4 0.1 1.0
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
59
no differences were observed between male and female participants in terms of total MSQ scores, nor were differences identified across any of the four MSQ sub-scales. As noted in the Introduction a number of studies have suggested an over-representation of BP I in males and of BP II in females. We therefore repeated our analyses in a sub-set of those clinically rated as having a BP I and, second, in those having a BP II disorder. For each group we had a sample set of 138 patients, with a mean age of 40.3 years and with 67 (49%) being female. Total MSQ scores were significantly higher for BP I than for BP II patients (t¼2.5, df¼ 255, po0.05). While MSQ sub-scale scores were consistently higher for male and female BP I patients than for BP II patients, the only significant result was for the mysticism sub-scale where male BP I patients returned higher scores than male BP II patients (t¼ 4.6, po0.001) as did female BP I vs. BP II patients (t¼3.1, po0.005). Within the bipolar sub-sets, total MSQ scores were not significantly different for male and female BP I patients (59.9 vs. 58.5, t¼0.4) or male and female BP II patients (52.4 vs. 54.2, t¼0.6). The MSQ subscale scores also derived no gender differences within the BP I and BP II subgroups. Turning to individual items (Table 2), only a few showed a gender impact (i.e. po0.05). For the BP I sub-set, males rated higher on levels of optimism and judging that nothing would go wrong and on needing less sleep, while females rated higher on singing—both in terms of prevalence of experiencing such a feature and rating it as more severe. For the BP II sub-set, males rated higher (in prevalence rate and severity) on taking risks, in the prevalence of reporting seeing things with absolute clarity and on severity of needing to sleep less and not feeling tired, while females rated higher (on prevalence and severity) on judging natural things feeling especially beautiful and item 42 (dressing more colourfully). Agreement between gender-related over-represented items in the BP I and BP II sub-sets with over-represented items in the overall bipolar sample was minimal. Turning to for DIF (MIMIC) analyses, three items were identified in both samples while the full sample had four additional items. For the two multivariate models, where all identified items (for the sample) were allowed to regress on gender simultaneously, we report the probability (p) value and the odds ratio (OR) for gender (male as reference level) corresponding to responding in a higher versus immediately lower category. The three common items were increased interest in sex (OR¼ 0.68, po0.001; and OR¼0.64, po0.001 in the full and age-matched samples respectively); take risks (OR¼0.71, po0.001; and OR¼0.67, po0.001); and dress more colourfully (OR¼1.36, po0.001; and OR¼ 1.37, po0.001), but the last was not well-defined by the factors. In the full sample, DIF was found for increased consumption of stimulants (OR¼0.83, po0.001); (ii) talk more (OR¼ 1.30, po0.001); set lots of goals (OR¼1.25, po0.001); and sing (OR¼1.20, po0.001) but the first two were not well defined by the factors. All models were fitted using Mplus (Muthén and Muthén, 1998–2012). In examining for gender effects on residuals, this method identified a large number of items (including all those above) as showing DIF—approximately two-thirds in the full sample, and with half of those not identified in the age-matched sample. Two caveats need to be made in considering these findings. First, with the very large sample sizes, the power to detect even very small effects is considerable. Second, while effects were significant, the multiple R2 for the models typically accounted for less than 2% of the variance and only rarely for 5–7%, suggesting that any DIF effects were quite small and usually trivial.
were distinctly influenced by gender and, if so, which might support the possibility that previously identified gender differences in lifetime rates of the bipolar disorders might not reflect differential disorder prevalence but be (entirely or partially) an artefact of measurement. As noted, we are not aware of any previous study that has examined whether screening or diagnostic measures of bipolar disorder are artefactually influenced by gender (i.e. by having gender-dimorphic behaviours or having items subject to biases such as amplification or denial being more likely by women or by men respectively). Our univariate analyses – undertaken on a very large sample – quantified no impact of gender on total MSQ scores or across the four MSQ sub-scales – whether analyses examined the combined bipolar sample or examined BP I and BP II subjects separately. While a number of items did show gender differences in prevalence and/or severity – and varied across the BP I and BP II subsets – such items were rarely differentiating in the whole bipolar sample. The DIF analyses identified a few more gender-weighted items. Such a lack of consistency is likely to reflect some true gender-dimorphic differences (perhaps BP II women being more likely to dress colourfully, talk more, sing more and set more goals, and BP II men more likely to have an increased interest in sex and be more likely to take risks) but is also likely to reflect both chance (i.e. there being multiple comparisons made without statistical correction) as well as the large sample size generating significant (but quite small) differences. Such nuances do not discount the main finding that the MSQ total score is not susceptible to any substantive gender impact and that bipolar hypomanic and manic episodes are not marked by features that differ distinctly between men and women.
4. Discussion
References
The key objective of the study was to determine if scores reported by those with a bipolar condition on the MSQ measure
Diflorio, A., Jones, I., 2010. Is sex important? Gender differences in bipolar disorder. Int. Rev. Psychiatry 22, 437–452.
5. Conclusion Taking all such considerations into account allows two relatively clear conclusions. First, at the ‘micro’ level, we have established that total and sub-scale MSQ scores are not subject to gender biases and that analyses involving the MSQ do not therefore need to control for any gender effect. We did find that females scored higher on some items and males on others, but such gender differences seemingly evened out across the MSQ total measure and scale scores. Second, if the MSQ is representative of other screening measures then a more ‘macro’ finding is suggested. In essence, studies reporting differential rates of bipolar disorders by gender are possibly detecting true gender differences in prevalence rather than reflect methodological error whereby the screening measure may affect distortions by having overrepresented gender-dimorphic behaviours, or is open to differential response biases such as denial or amplification. Role of funding source This study was supported by an NHMRC program grant (1037196). The NHMRC did not participate in the study design, nor in the data collection and analysis process, or in any phase of the manuscript preparation.
Conflict of interest No conflict declared.
Acknowledgements This study was funded by an NHMRC Program Grant (1037196). We thank Georgie McClure for data entry.
60
G. Parker et al. / Journal of Affective Disorders 162 (2014) 55–60
Grant, J.E., Desai, R.A., Potenza, M.N., 2009. Relationship of nicotine dependence, subsyndromal and pathological gambling, and other psychiatric disorders: data from the National Epidemiologic Survey on Alcohol and Related Conditions. J. Clin. Psychiatry 70, 334–343. Kawa, I., Carter, J.D., Joyce, P.R., Doughty, C.J., Frampton, C.M., Elisabeth, Wells, et al., 2005. Gender differences in bipolar disorder: age of onset, course, comorbidity, and symptom presentkation. Bipolar Disord. 7, 119–125. Mair, P., Hatzinger, R., Maier M.J., 2014. eRm: Extended Rasch Modeling. R Package Version 0.15-4. 〈http://erm.r-forge.r-project.org/〉. Merikangas, K.R., Jin, R., He, J.-P., Kessler, R.C., Lee, S., Sampson, N.A, et al., 2011. Prevalence and correlates of bipolar spectrum disorder in the World Mental Health Survey Initiative. Arch. Gen. Psychiatry 68, 241–251. Muthén, B., 1989. Latent variable modeling in heterogeneous populations. Psychometrika 54, 557–585. Muthén, L.K., Muthén, B.O., 1998–2012. Mplus User's Guide, seventh ed. Muthén & Muthén, Los Angeles, CA.
Parker, G., Fletcher, K., Hyett, M., 2008. The mood assessment program: a computerised diagnostic tool for deriving management plans for mood disorders. Med. J. Aust. 188, S126–S128. Parker, G., Hadzi-Pavlovic, D., Tully, L., 2006. Distinguishing bipolar and unipolar disorders: an isomer model. J. Affect. Disord. 96, 67–73. Parker, G., Graham, R., Hadzi-Pavlovic, D., Fletcher, K., Hong, M., Futeran, S., 2012. Further examination of the utility and comparative properties of the MSQ and MDQ bipolar screening measures. J. Affect. Disord. 138, 104–109. Tennant, A., Penta, M., Tesio, L., Grimby, G., Thonnard, J.L., Slade, A., Lawton, G., Simone, A., Carter, J., Lundgren-Nilsson, A., Tripolski, M., Ring, H., Biering-Sørensen, F., Marincek, C., Burger, H., Phillips, S., 2004. Assessing and adjusting for cross-cultural validity of impairment and activity limitation scales through differential item functioning within the framework of the Rasch model: the PRO-ESOR project. Med. Care 42 (1 Suppl.), I37–48.
