Descriptive information

Of the 1,763 publications on bipolar disorder, which were included in the analysis, 1,415 were regular (21 early access) and 348 were review (9 early access) articles. Figure 1A illustrates the number of annual publications during the last decade of bipolar disorder research in Germany. The annual growth rate was 1.7%. Accordingly, German productivity in the field remained relatively constant for the whole decade, with no substantial increase or decrease concerning productivity. Figure 1B shows the number of articles partitioned according to the country of its corresponding author at the time of publication. 913 (51.8%) of the articles under German collaboration were indeed published by a corresponding author seated in Germany, of which 497 (54.4%) were single country publications (i.e., only German authors were involved in its production). Accordingly, a substantial part of publications with a German corresponding author had an international orientation (N = 416, 45.6%). The remaining 850 articles were collaboration projects with German authors headed by corresponding authors with affiliations in the USA (N = 177), United Kingdom (N = 177), Australia (N = 65), Italy (N = 55), Canada (N = 53), Spain (N = 40), Netherlands (N = 39), Switzerland (N = 36), and Denmark (N = 35).

Fig. 1

Scientific production. Annual scientific production under German collaboration remained relatively stable during the last decade (A). More than 50% of articles were published with a corresponding author seated in Germany. The remaining articles originated in international collaboration projects with the corresponding author most frequently being seated in the USA, UK, Australia, Italy, and Canada (B). Note: SCP = Single Country Publications, MCP = Multiple Country Publications, N.= Number

Analysis of the scientific networkAnalysis of the most productive institutions

In total, 3,355 different institutions were involved in production of the present publications. Five of the ten most frequently indicated (i.e., most productive) institutions were seated in Germany (N refers to their total number of mentions): University of Bonn (N = 355), Heidelberg University (including the Central Institute for Mental Health, ZI Mannheim; N = 321), Technical University Dresden (N = 288), University of Münster (N = 164) and Ludwig Maximilian University of Munich (N = 154). The other five were seated abroad, i.e., Kings College London (N = 315), University of Melbourne (N = 201), University of Toronto (N = 192), Karolinska Institute (N = 181), and University of Barcelona (N = 146). When analyzing the network structure of institutions that produced more than 25 publications within the last decade, five major clusters emerged (see Fig. 2; Table 1 for a full description of Cluster members).

Fig. 2

Collaboration network of most productive institutions. While two of the clusters – cluster 1 (depicted in red) and cluster 3 (depicted in blue) – were dominated by institutions seated in foreign countries, cluster 2 (depicted in green) connects a large number of German institutions. Note: Institutions that have published more than 25 publications in the database were included in this visualization

Table 1 Institutional collaboration networkAnalysis of the most productive authors and sources

In total, 11,253 authors contributed to the included articles. While only 25 articles were written by a single author, the remaining 1,738 articles were written under collaboration of several authors (mean = 15.5, SD = 32.6, max = 571). Of these, 58.5% (N = 1,031) were original scientific papers authored by 2–9 authors, 36.6% (N = 646) were team works authored by 10–49 authors and 4.9% (N = 86) were multi-center studies with more than 50 authors (Fig. 3A). The 10 most productive authors, or research teams led by the respective author, were Michael Bauer (TU Dresden, N = 154, Total number of citations (TC) = 7,981, h-index = 38), Marcella Rietschel (UMM Mannheim, N = 144, TC = 8,238, h-index = 41), Andreas Reif (University Hospital Frankfurt, N = 119, TC = 5,910, h-index = 29), Markus M. Nöthen (University of Bonn, N = 106, TC = 6,040, h-index = 34), Thomas G. Schulze (LMU Munich, N = 96, TC = 6,989, h-index = 33), Sven Cichon (Research Center Jülich, N = 89, TC = 6,963, h-index = 38), Eduard Vieta (University of Barcelona, N = 83, TC = 4,491, h-index = 31), Christoph U. Correll (Charité Berlin, N = 75, TC = 1,573, h-index = 22), Andrea Pfennig (TU Dresden, N = 75, TC = 2,909, h-index = 27), and Martin Alda (Dalhousie University, N = 68, TC = 4028, h-index = 29).Footnote 1 See Fig. 3B. The collaboration network of authors that published more than 25 publications in the last decade (Fig. 3C) consists of four clusters, which are all highly interconnected. See Table 2 for a full description of cluster members in the collaboration network.

Fig. 3

Author contributions. Most of the published articles were original scientific papers (A). Eight of the most productive corresponding authors were affiliated in Germany at least for a substantial part of the time accounted for in this analysis (B). The collaboration network of authors that published more than 25 publications in the last decade contained four major clusters, which were all highly interconnected (C)

Table 2 Author collaboration network

Articles were published in 420 journals with the ten most important sources being highly regarded (IF = WOS impact factor 2022 as retrieved from https://www.scijournal.org): Journal of Affective Disorders (N = 83, IF = 4.8), Translational Psychiatry (N = 63, IF = 6.2), European Archives of Psychiatry and Clinical Neuroscience (N = 58, IF = 5.3), International Journal of Bipolar Disorders (N = 58, IF = 4.3), Bipolar Disorders (N = 57, IF = 6.7), Frontiers in Psychiatry (N = 53, IF = 4.1), Molecular Psychiatry (N = 53, IF = 16.0), Psychiatry Research (N = 44, IF = 3.2), Schizophrenia Research (N = 40, IF = 4.9), and Journal of Psychiatric Research (N = 37, IF = 4.8). (Note: An additional short description of the twenty most cited articles which were referred to in the publications can be found in the supplementary material).

Conceptual structure – analysis of research topicsCo-word analysis

As has to be expected by the method of the literature research, the most frequently indicated keyword in the data base was bipolar disorder OR bipolar disorders (N = 605). This was followed by schizophrenia (N = 303), depression OR major depressive disorder OR major depression (N = 297), lithium (N = 86), psychosis (N = 79), mania (N = 66), fMRI OR functional magnetic resonance imaging (N = 53), cognition (N = 42), genetics (N = 42), and meta-analysis (N = 42). Clustering analysis of the co-occurence network of keywords that were mentioned at least ten times in the data base revealed seven major clusters (Fig. 4). An interpretation of the co-occurrence network suggests, that one large cluster (cluster 1, depicted in red) was concerned with early recognition and therapy of bipolar disorders, two of the clusters (cluster 2 and 3, depicted in blue and green) with neuroimaging, two of the clusters (cluster 4 and 7 depicted in yellow and orange) with (functional) genetics/genomics, one cluster (cluster 5 depicted in purple) with pharmacological interventions, and one cluster with (early) diagnosis (cluster 6 depicted in turqouise). The multitude of connections between the three clusters indicates, however, that a strict thematic separation of topics would be rather artificial. See Table 3 for a full list of keywords within each cluster.

Fig. 4

Co-word analysis network. While the Cluster 1 (depicted in red) is concerned with early recognition and therapeutical interventions, Clusters 2 (depicted in green) and 3 (depicted in green) are rather concerned with neuroimaging methods, Clusters 4 (depicted in yellow) and 7 (depicted in orange) are dominated dominated by genetic studies, Cluster 5 (depicted in purple) with pharmacological interventions and Cluster 6 (depicted in turqouise) with diagnosis and differentiation from psychosis. Note: The size of the nodes reflect the occurrence frequency of the respective keyword, thickness of the connecting line represents the strength of association

Table 3 Cluster members of the co-word analysisThematic map of author keywords

Several themes were identified by analyzing the provided author keywords. No clear motor theme could be identified, i.e., a theme that is important to and well established within the research field. However, some basic themes emerged, that were important for the research field, even though they were not highly developed, yet. Four clusters of basic themes could be identified, of which ‘schizophrenia, psychosis, fMRI, cognition, and hippocampus’ was of lowest density and thus a rather developing theme. The other three clusters were in the ‘middle field’ of thematic development – ‘bipolar disorder, major depression, mania, genetics, meta-analysis’, ‘comorbidity, ADHD, prevention, suicide, early recognition’, and ‘lithium, antipsychotics, antidepressants, bipolar depression, treatment’. Besides these basic themes, several niche themes were identified: ‘impulsivity, aggression, personality, toxoplasma gondii, kynurenine’, ‘social cognition, theory of mind, childhood trauma, social cognition’, ‘mental illness, diet, assessment’, and ‘sunlight, age of onset, solar insolation’. Those themes are highly developed but rather isolated within the research field (Fig. 5).

Fig. 5

Thematic map of author keywords. The thematic map of the field was characterized by basic themes, that are important for the whole community without showing a high cohesiveness (lower-right quadrant) and niche themes, that are highly developed but isolated to specific research sub-communities, i.e. without a high number of relations with other themes. Note: The thematic map uses sub-group analysis to identify specific themes that are prominent in the field and plots them in a thematic diagram based on Callon density (as a measure of the topic’s cohesiveness and thus development) and Callon centrality (as a measure of the degree of correlation with other topic’s and thus importance for the whole field). Four topologies of themes are plotted in four different quadrants: Motor themes (upper-right quadrant) are characterized by high centrality and high density. These themes are highly developed and important for the whole research field. Basic themes (lower-right quadrant) are high in centrality and low in density. These themes are thus important for the whole research field, even though the topics themselves are not highly developed (yet). Emerging / declining themes (lower-left quadrant) are low in centrality and low in density. Accordingly, these themes are both, weakly developed and marginal. Lastly, niche themes (upper-left quadrant) are highly developed, but isolated, i.e. they are characterized by well developed internal links (high density) but unimportant external links (low centrality)

Grant support of projects on bipolar disorder

Seven out of the 20 institutions responding to our survey explicitly stated that none of their research projects focused on bipolar disorders. There were a few projects that accounted for the majority of research activities: the BMBF-funded BipoLife consortium (bipolife.org), which however ran out in 2022 and thus cannot be counted in the present analysis; the same holds true for the EU-funded project Moodinflame (moodinflame.eu; Universities of Münster and Ulm).

A major current research structure on mood disorders is the Research Group FOR 2107, funded by the DFG (for2107.de), which revolves around the long-term course of affective disorders, not specific to but including bipolar disorder. The FOR2107 (ending in 2023) connects the Universities of Marburg (lead), Münster, Kiel, Bonn and the Central Institute for Mental Health Mannheim. It gave rise to three following individual grants by the DFG, all to PIs at the University of Münster. Four other collaborative research efforts by the DFG, RTG2862, iRTG2150, FOR2858 and iRTG2773, which had a transdiagnostic approach, had one subproject each that also included – amongst others – research questions on bipolar disorder. The Heisenberg-professorship, awarded to PI Hahn at the University of Münster, also supports research on bipolar disorder. Finally, there is one individual grant to PI Noethen at University of Bonn that includes bipolar disorder research. To conclude, apart from FOR 2107 (and its follow-ups), which has one research focus on bipolar disorder, there are only very few (six) research projects on bipolar disorder funded by the DFG.

Since the collaborative project “Bipo-Life” ran out, there is no national funding on bipolar disorder from the BMBF. In 2023, the German Center for Mental Disorders (DZPG) was founded, which is co-financed by the BMBF and which also includes research on bipolar disorders. However, the extent to which bipolar disorder is considered in the DZPG is as yet unclear. Only three of the centers that formed BipoLife are also members of the DZPG (Berlin, Munich, Tübingen). The collaborative BMBF project COMMITMENT includes, amongst others, research questions on bipolar disorder in one bioinformatic subproject. There are four collaborative projects on the European level, which are administered by the BMBF and focus on bipolar disorder. These are Plot-BD (PerMed framework), DynAMoND (EraNet Neuron framework), BIPCOM (PerMed framework), and UNMET (EraNet Neuron framework).

Two large-scale European Horizon 2020 projects on bipolar disorder have contributions from Germany, R-LiNK (ending in 2023) and PsychSTRATA. Intramural research projects were reported by six institutions. All members but one of the National Center of Affective Disorders (NCAD; www.ncad.health) reported research on bipolar disorders funded by either DFG and/or BMBF.