Complete data sets were available for 806 subjects. Mean age was 40 ± 12 years (range 18–81 years), and 460 subjects (57.1%) were female. Six hundred twenty-four were diagnosed with bipolar I disorder (77.4%) and 182 with bipolar II (22.6%). 17,937 visits of these subjects were available with corresponding YMRS and IDS-C, both rated at the same occasion. Female subjects contributed 9583 visits (53.4%) for analysis of the total score of the YMRS and 9593 visits (53.5%) for the analysis of the YMRS core items.

There were no significant demographic differences between the ODD and the EVEN sample. The ODD sample included 404 subjects with a mean age of 40 ± 12 years (range 18–81), and 235 subjects (58.2%) were female. In this sample, 320 subjects (79.2%) were diagnosed with bipolar I disorder. The EVEN sample included 402 subjects with a mean age of 40 ± 11 years (range 18–76), and 225 subjects (56.0%) were female. In this sample, 304 subjects (75.6%) were diagnosed with bipolar I disorder. 9099 (50.7%) visits were available from subjects with an odd registration number, 8838 (49.3%) visits were available from subjects with an even registration number.

Inspection of the plots “Average IDS-C score by YMRS score “and “Average IDS-C score by core-YMRS score “ suggested increasing severity of depressive symptoms with increasing YMRS rating scores in the ODDS sample (see Additional file 1: Fig. S5 and Fig. S6). Subsequently, we tested our primary and secondary hypothesis in the EVEN sample.

For verification of our hypotheses “(1) Subjects in a manic state have higher IDS-C-values than patients in a non-manic state” and “(2) Subjects in a hypomanic state have higher IDS-C values than patients in anon-manic state”, data were stratified into three groups defined by YMRS-scores (see Table 1). Average IDS-C Scores of these groups were compared with the t-Test.

Using YMRS total score

When grouping by YMRS total score, 8,102 YMRS ratings (92.3%) felt into the non-manic range between 0 and 11.5 points, 583 YMRS ratings (6.6%) in the hypomanic range between 12 and 20.5 points and 93 YMRS ratings (1.1%) in the manic range between 21 and 29.5 points. 18 YMRS-scores (0.2%) with ≥ 30 points (total) were excluded from final analysis as we were in reasonable doubt how reliable depressive symptoms have been reported while in a highly excited manic state. Forty-two visits (0.5%) had to be excluded because of missing data.

In the non-manic group of the EVEN-Sample the mean IDS-C score was 15.75 ± 12.6 points (range 0–66), in the hypomanic group 19.06 ± 12.4 points (range 0–63) and in the manic group 18.29 ± 11.9 points (range 0–59) (Table 2).

Table 2 Mean IDS-C scores for the YMRS-groups (total and core items) EVEN SAMPLE

For the YMRS total score, the difference between (hypo)manic groups and the non-manic reached statistical significance (p < 0.001 for hypomanic vs non-manic, and p < 0.05 for manic vs non-manic) (Table 3).

Table 3 Differences in mean IDS-C scores for the YMRS-groups (total and core items) EVEN SAMPLEUsing YMRS core-score (4-item and 3-item)

Using the YMRS total score may confound results as there is some overlap with depressive symptom rating scales, including the IDS-C, (e.g., decreased sleep could be due to either depression or (hypo)mania). Using the sum score of the four YMRS core items might minimize this risk. However, as irritability is rated both in the 4-item YMRS core and the IDS-C, we also checked results for a 3-item YMRS core (see “Methods” section).

We identified 7914 4-item YMRS core scores (90.1%) in the non-manic range between 0 and 5.5 points, 748 4-item YMRS core scores (8.5%) in the hypomanic range between 6 and 10 points and 119 4-item YMRS core scores (1.4%) in the manic range between 10.5 and 14.5 points. 39 4-item YMRS core scores (0.4%) with ≥ 15 points (total) were excluded as we were in reasonable doubt how reliable depressive symptoms have been reported while in a highly excited manic state, 18 visits (0.2%) had to be excluded because of missing data.

In the 4-item YMRS core score defined non-manic group of the EVEN-Sample the mean IDS-C score was 15.46 ± 12.42 points (range 0–64), in the hypomanic group 20.76 ± 13.12 points (range 0–66) and in the manic group 21,26 ± 13.23 points (range 0–59) (Table 2). All differences between all groups reached significance (p < 0.001) (Table 3).

Using only the sum score of the other 3 YMRS core items, we identified 8129 3-item YMRS core scores (92.17%) in the non-manic range between 0 and 4 points, 420 3-item YMRS core scores (4.76%) in the hypomanic range between 4.5 and 7.5 points and 230 3-item YMRS core scores (2.6%) in the manic range between 8 and 11.5 points. 39 3-item YMRS core scores (0.46%) with ≥ 12 points (total) were excluded as we were in reasonable doubt how reliable depressive symptoms have been reported while in a highly excited manic state, 18 visits (0.2%) had to be excluded because of missing data.

In the 3-item YMRS core score defined non-manic group of the EVEN-Sample the mean IDS-C score was 15.73 ± 12.52 points (range 0–62), in the hypomanic group 17.94 ± 12. points (range 0–64) and in the manic group 21.73 ± 13.93 points (range 0–64) (Table 2). All differences between all groups reached significance (p < 0.001) (Table 3).

Summarizing these results, mean IDS-C scores related to the severity of the manic state as measured with the core items of the YMRS (Table 2). When using the total YMRS score instead of the YMRS core score, we observed significantly higher depressive symptom severity in hypomanic and manic subjects compared to non-manic subjects, too, but no significant difference between hypomanic and manic probands.

Additional analyses in the full sample

After verification of the primary and secondary hypotheses, additional analyses were conducted using the full data set (ODD and EVEN sample).

IDS-C score in relation to mood state (euthymic, hypomanic, manic)

Plotting the average IDS-C sum score for the full data set as a function of the categories non-manic, hypomanic and manic as defined by the 4-item YMRS core score demonstrates significant differences between groups (p < 0.001) with the manic group scoring highest, and the non-manic group lowest on the IDS-C (Fig. 2).

Fig. 2

Average IDS-C scores by state (based on 4-item YMRS core-score), EVEN sample

Number of depressive symptoms

Plotting only the IDS-C sum score as a function of the YMRS score supplies an estimate of severity of depression, but does not reflect the full symptomatology, e.g., how many mental, social and physical domains are affected. Results for the mean number of depressive symptoms (IDS-C items scoring ≥ 1) in the non-manic, hypomanic and manic group as defined by 4-item YMRS core-scores are shown in Fig. 3. Table 4 illustrates that these results are stable and significant for all three definitions of mood state, by total YMRS score, 4-item and 3-item YMRS core score.

Fig. 3

Average IDS-C-Symptom count by state (based on 4-item YMRS core-score), EVEN sample. ***p < 0.001

Table 4 Mean IDS-C SYMPTOMCOUNT for the YMRS-groups (total and core items) in the full sampleGender effect on number of depressive symptoms

Previous analyses showed a female preponderance for mixed hypomania (Suppes et al. 2005) and numerically also for mixed mania (Vieta et al. 2014). To test whether this is also true in our study we compared numbers of IDS-C items scoring ≥ 1 as a function of manic severity separately for female and male subjects. Female patients were affected by more depressive symptoms than males, especially when manic symptoms were more severe. We observed a linear increase of depressive symptomatology with severity of manic core symptoms (average IDS-C-symptom count vs 4-item YMRS core score, R2 = 0.86), but not in males (R2 = 0.29) (Fig. 4). Thus, the increase of the number of depressive symptoms associated with an increase of YMRS-core scores was mainly carried by women. Factorial analysis of variance testing showed highly significant (p < 0.001) effects for all 3 factors, that is (1) YMRS 4 item core score (2) gender as well as (3) the combination of YMRS 4 item core score and gender.

Fig. 4

Average IDS-C-Score by 4 item YMRS CORE SCORE (male vs female)

Number of occasions and subjects fulfilling DSM-5 hypomanic or manic with mixed features specifier criteria

Finally, we examined at how many occasions (visits) DSM-5 hypomanic or manic with mixed features specifier criteria were likely to be satisfied, and the corresponding number of subjects. As a proxy we assigned the six depressive A-criteria listed in DSM-5 to corresponding specific IDS-C items (Table 5). Symptom criterion was counted as satisfied with a score of at least 1, except criterion 1 where the corresponding IDS-C items 5 or 10 should score at least 2 to satisfy also the duration criterion of DSM-5 (“majority of days”). To criterion 1, we also assigned two IDS-C items as sadness alone (item 5) might not cover all flavors of depressed mood; however, for fulfilling DSM-5 criterion 1 it was sufficient to score at least 2 in either item 5 or 10.

Table 5 Assignment of the six DSM-5 depressive symptoms for (hypo)mania with mixed features specifier to selected IDS-C items

Using the YMRS total scores for assignment to the non-manic, hypomanic and manic group, we found that 432 visits in the hypomanic group (40,6%) and 58 visits in the manic group (32,2%) fulfilled DSM-5 criteria for a (hypo)manic episode with mixed features specifier (≥ 3 depressive symptoms).

Applying the 4-item-YMRS core score to group subjects, we found that 652 visits in the hypomanic group (47.7%) and 121 visits in the manic group (48%) fulfilled DSM-5 criteria for a (hypo)manic episode with mixed features specifier (≥ 3 depressive symptoms). There was no significant difference at p ≤ 0.01 between the manic and hypomanic group. Examining gender preponderance, we found that in the hypomanic group 55.2% of visits of females and 39.1% of visits of males were likely to satisfy mixed specifier criteria, and in the manic group 53.5% of visits of females and 40.7% of visits of males. Differences between visits of females and males were significant at p ≤ 0.001 in the hypomanic group, but with the relatively low number of visits only a trend (p = 0.45) was observed in the manic group.

Next, we examined how many (hypo)manic subjects fulfilled the mixed feature specifier at least at one visit. As a result, we found that 272 subjects in the hypomanic group (64%) and 77 subjects in the manic group (50%) as defined by 4-item YMRS core score fulfilled DSM-5 criteria for a (hypo)manic episode with mixed features specifier (≥ 3 depressive symptoms) at some point in time. There was a significant difference at p ≤ 0.01 between the hypomanic and manic group. Examining gender preponderance, we found that in the hypomanic group 66.7% of females and 60.8% of males were likely to satisfy mixed specifier criteria at some point in time, and in the manic group 51.1% of females and 48.5% of males. There was no significant difference between female and male subjects.

Non-mood symptoms predict more severe depression in (hypo)manic patients

Previous work suggests that the non-mood-symptoms anxiety, agitation and irritability are more prevalent in manic patients suffering from a higher depressive burden and fulfilling DSM-5 mixed feature specifier (Suppes et al. 2017; Vieta et al. 2014). In our sample, we examined the correlation between severity of agitation and irritability vs. average IDS-C score. As a proxy for the degree of agitation, we used the YMRS item 2 and defined a score 0–2 as “no or low degree of agitation”, and a score > 2 as “agitated”. As a proxy for the degree of irritability, we used the YMRS item 5 and defined a score 0–4 as “no or low degree of irritability”, and a score > 4 as “irritable”. We could not test for anxiety as there is no appropriate YMRS item as an equivalent. Finally, we compared the average IDS-C value in the group of “High scorer for agitation and irritability”, scoring > 2 points for item 2 and > 4 points for YMRS item 5, against the rest of the sample. Confirming previous studies, we found that scoring high on YMRS item 2 (Agitation) or/and 5 (Irritability) significantly predicted more severe depression in (hypo)manic subjects (Table 6).

Table 6 Scoring high on YMRS item 2 (Agitation) or/and 5 (Irritability) predicted more severe depression in (hypo)manic subjectsPanic and agitation increase with hypomania

For clinicians it is relevant to know which depressive symptoms have the strongest association with the DSM-5 mixed feature specifier in (hypo-)mania. Therefore, we compared the average values for all IDS-C items between the 3 groups non-manic, hypomanic and manic (see Table 1) for subjects fulfilling criteria for the depressive mixed features specifier.

Compared to the non-manic group, the hypomanic group showed more than 50% increase of the average item score for the IDS-C items interpersonal sensitivity (+ 54%), early morning insomnia (+ 56%), irritable mood (+ 78%), panic/phobic Symptoms (+ 100%), psychomotor agitation (+ 120%).

Compared to the non-manic group, the manic group showed a greater than 50% increase of the average item score for the IDS-C items for anxious mood (+ 50%), sleep onset insomnia (+ 51%), sympathetic arousal (+ 57%), interpersonal sensitivity (+ 66%), suicidal ideation (+ 67%), mid-nocturnal insomnia (+ 67%), weight decrease (+ 67%), panic/phobic symptoms (+ 107%), irritable Mood (+ 114%), early morning insomnia (+ 128%), and psychomotor agitation (+ 206%).

Compared to the hypomanic group, the manic group showed the biggest increase of the average item score for the IDS-C items for mid-nocturnal Insomnia (+ 27%), suicidal ideation (+ 30%), weight decrease (+ 34%), psychomotor agitation (+ 39%) and early morning insomnia (+ 46%),