The St. Göran Project: A Multipronged Strategy for Longitudinal Studies for Bipolar Disorders

Longitudinal observational studies have been instrumental in advancing our understanding of bipolar disorder, addressing a wide range of scientific questions [1‒4]. While cross-sectional studies are valuable, they lack the ability to track time-dependent effects. Unresolved research questions such as those related to disease progression, long-term treatment efficacy, individual variability in illness course, and clinical presentation require longitudinal data for answers. Further, research on brain morphology and neurochemistry, cognitive and neuropsychiatric deficits, as well as the genetic architecture of bipolar disorder can benefit significantly from follow-up data. Clinically, there is an imperative to increase our understanding of the importance of co-morbid conditions, the trajectory of illness progression, and the variable treatment responses and functional outcomes. The Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) exemplifies the potential of longitudinal clinical outpatient investigations, aiming to identify the most effective treatments and treatment combinations for managing bipolar disorder [5].

In 2005, inspirated by the STEP-BD study, the St. Göran Bipolar Project (SBP) was launched as a longitudinal outpatient study investigation. This project aimed to enhance clinical assessments with brain imaging, collection of blood and cerebrospinal fluid (CSF), thereby providing a comprehensive dataset for analysis. The primary objective was to identify predictive factors associated with long-term outcomes of individuals with bipolar disorder, with a particular focus on cognitive function, relapse rates, treatment responses, and overall functional outcomes. By integrating brain imaging, biochemical analyses of CSF and blood, and genetics, SBP adopted a multifaceted approach to examine bipolar disorder from multiple perspectives. The published findings from the study reflects both the breadth and depth of the scope of its research goals, offering insights into various facets of bipolar disorder. As ongoing research continues to unfold, we anticipate that the data collected will yield further significant scientific insights in the coming years. This paper provides an overview of the research methods utilized in the SBP and summarizes the research findings that have emerged since its commencement.

The SBP was established through a collaboration between the Karolinska Institutet and the St. Göran bipolar outpatient clinic at the Northern Stockholm Psychiatry in 2005. Subsequently, in 2009, a second study center was started through a partnership between the Sahlgrenska University hospital in Gothenburg and the Institute of Neuroscience and Physiology at the University of Gothenburg.

Participants in the SBP study were consecutively enrolled from new or ongoing individuals treated at the respective bipolar outpatient clinics in Stockholm (SBP-S) and Gothenburg (SBP-G), Sweden. These outpatient units provide long-term, often lifelong, follow-up care for individuals diagnosed with bipolar disorder. Diagnoses were confirmed after an initial evaluation using the Patient History Form (see below). Individuals meeting the inclusion criteria (see below) were eligible for inclusion. Prior to inclusion, potential participants were informed about the study’s objectives, procedures, risks, and benefits. The voluntary nature of research participation was emphasized, and alternatives for non-research treatment were provided. Patients retained the right to withdraw their consent or exit the study at any time during their participation.

For each included patient, age- and sex-matched controls were randomly selected by Statistics Sweden [6] and contacted by mail. Statistics Sweden recommended contacting seven potential controls per study participant as typically only one out of seven responds positively to recruitment requests. Indeed, out of those who received the invitation, 14% responded, which is thus comparable to response rates in similar studies according to Statistics Sweden. Trained nurses conducted a telephone screening to exclude people with severe mental health issues, neurological problems, and severe substance abuse. Nurses were instructed to exclude candidates with a clear history of psychiatric hospitalization or significant psychiatric contact that would preclude their suitability as controls. Those who passed the telephone screening were scheduled for a 1-day assessment as part of the study.

Study procedures (Fig. 1) were performed when patients were deemed to be in a mood-stable state. During visits for blood sampling and lumbar puncture, participants were assessed using the Montgomery-Asberg Depression Rating Scale (MADRS) [7] and Young Ziegler Mania Rating Scale (YMRS) [8]. Starting in 2012, these assessments were also administered during the neuropsychological test sessions.

The STEP-BD project employed the structured interview instrument Affective Disorder Evaluation (ADE) [5]. The ADE includes a social anamnesis, medical history, and the affective module of the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-fourth edition (DSM-IV). It documents the number and characteristics of lifetime affective episodes. With the permission of the originator Gary S. Sachs (Bipolar Clinic and Research Program, Harvard Medical School, Boston, MA, USA), we modified the ADE to suit Swedish conditions. Specifically, we added a more detailed assessment of participants’ occupational level including details about sick leave. We also included questions pertaining to the Swedish weapon law, as well as details about non-pharmacological treatment including electroconvulsive therapy, psychotherapy, and patient education. This modified structured interview instrument, referred to as the Patient History Form, underwent minor refinements during the study, with a more thorough revision in 2014.

Further, the structured Mini International Neuropsychiatry Interview (MINI) [9] was administered to screen for psychiatric diagnoses other than bipolar disorder. The information for the Patient History Form and the MINI was collected by board-certified psychiatrists, residents in psychiatry, or a trained specialist nurse. Assessments were supplemented with information from medical records and, when needed, interviews with next of kin. A final best-estimate diagnostic decision was made at a diagnostic case conference by a consensus panel (n = 2–5) of experienced board-certified psychiatrists specialized in bipolar disorder.

Participants also completed self-report versions of the Alcohol Use Disorders Identification Test (AUDIT) [10] and the Drug Use Disorders Identification Test (DUDIT) [11]. Furthermore, both the symptom and function domains of the Global Assessment of Functioning (GAF) scales [12] were scored to assess the overall psychosocial functioning and the SCID-II Screen Questionnaire [13] was used to screen for personality disorders. The median time for completion of baseline procedures for patients, i.e., from inclusion to last visit, was 144 days (IQR: 256).

Controls completed a structured interview to collect information on dwelling, educational and occupational level including details about sick leave, questions pertaining to the Swedish weapon law, family history of psychiatric disorders, as well as reproductive and medical history. In addition, both the symptom and function domains of the GAF scales were scored to assess the overall psychosocial functioning. The MINI was administered to exclude psychiatric morbidity, and the SCID-II Screen Questionnaire was used to screen for personality disorders. Substance abuse was screened for at the telephone interview by the nurse, in the psychiatric interview, by AUDIT and DUDIT, as well as by determining serum levels of carbohydrate-deficient transferrin. The median time for completion of baseline procedures for controls, i.e., from inclusion to last visit, was 0 days (all assessments completed during a single day, IQR: 3).

New patients referred to the specialized outpatient clinics were evaluated and diagnosed using the Patient History Form and the MINI before being considered for inclusion in the study. Patients who met the criteria for any bipolar spectrum disorder (see below) and whose diagnoses were confirmed during the diagnostic case conference were invited to participate in the study while continuing treatment at the clinic. Additionally, ongoing patients already receiving treatment at the outpatient clinics could be invited to participate. In these cases, informed consent was obtained prior to conducting a full diagnostic interview using the Patient History Form and the MINI.

The inclusion criteria for patient participation in the SBP were as follows: (1) age of 16 years or older; (2) a diagnosis meeting the DSM-IV Criteria for bipolar I, bipolar II, bipolar not otherwise specified, cyclothymia, or schizoaffective disorder of the manic/bipolar subtype, and (3) completed the basic diagnostic evaluation in the Patient History Form. Exclusion criteria were as follows: (1) unwillingness or inability to comply with study requirements or (2) not being competent to give informed consent in the investigator’s opinion. Participants were not obliged to participate in all the study’s investigations and could choose which components they wished to participate in.

Overconsumption of alcohol as revealed by carbohydrate-deficient transferrin (>2.0%) or responses indicating large consumption (>8 standard drinks per time more than 2 times per week) and/or amnesia and/or loss of control more than once per month were exclusion criteria for controls. Other exclusion criteria were neurological conditions (other than mild migraine), untreated endocrinological disorders, pregnancy, dementia, recurrent depressive disorder, and a family history of schizophrenia or bipolar disorder in first-degree relatives. We did not exclude controls with past minor depressive episodes, isolated episodes of panic disorder, eating disorders, or obsessive-compulsive disorder that had remitted spontaneously or with brief psychotherapy counseling.

The aim of the study was to conduct follow-up assessments with study participants at 7 and 14 years after their baseline visits. The time interval was chosen to balance study site workload, provide sufficient follow up length, and have an acceptable rate of study dropouts. Study participants who did not attend the 7-year follow-up were still invited to the 14-year follow-up, provided they had not requested to withdraw from the study. The Patient History Form was modified for these follow-up visits. Changes included the addition of data collection regarding medication, episode relapse, inpatient care, suicide attempts that occurred during the follow-up period. Controls completed a shortened version of the modified Patient History Form at the follow-up visits.

The St. Göran study incorporates several rating scales. While some of these scales have been consistently used from study start throughout follow-up visits, other rating scales have been introduced or existing ones removed during the study. Table 1 provides an overview of rating scales across visits for both study centers.

Delusional ideation and dissociative experiences were measured using the Peters et al. [14] Delusions Inventory (PDI) and the Dissociative Experiences Scale (DES) [15]. Anxiety was measured using the short version of the Spielberger State-Trait Anxiety Inventory (STAI-S) [16]. Patients in the SBP-S cohort were assessed using the Seasonal Pattern Assessment Questionnaire (SPAQ) [17] at baseline. The Edinburgh Handedness Inventory (EHI) [18] was used to ascertain handedness. Gender traits were assessed in the SBP-S control group at the 7-year follow-up using the Bem Sex-Role Inventory (BSRI) [19]. Sexual orientation was assessed using the Kinsey Scale [20]. Data on personality were collected using the SCID-II Screening Questionnaire, the Swedish Universities Scales of Personality (SSP) [21], and the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego Autoquestionnaire (TEMPS-A) [22]. Subjective well-being was rated using the Psychological Well-Being Index (PGWB) [23]. Study persons view of life was rated using the Sense of Coherence Scale (SOC) [24]. Functional level and quality of life were assessed using the Sheehan Disability Scale (SDS) [25], Interpersonal Support Evaluation List (ISEL) [26], and EQ-5D questionnaires [27]. The Karolinska Sleep Questionnaire (KSQ) [28] was used to collect information on sleep difficulties. Creativity was measured using the Barron Welsh Art Scale (BWAS) [29], the Creative Achievement Questionnaire (CAQ) [30], and the Berlin Intelligence Structure Model test (BIS-4) [31]. Everyday behavior associated with executive function was assessed by the Behavior Rating Inventory of Executive Function-Adult version (BRIEF-A) questionnaire [32]. Finally, lithium response was measured using the Retrospective Assessment of the Lithium Response Phenotype Scale (Alda scale) [33].

As shown in Table 2, the presence of autism spectrum traits was assessed using the Autism-Spectrum Quotient Test (AQ test) [34] and the Ritvo Autism and Asperger Diagnostic Scale-14 Screen (RAADS-14) [35]. ADHD symptoms were rated using the Adult ADHD Self-Report Scale (ASRS) [36], the Brown Attention-Deficit Disorder Scales (BADDS) [37], and the Wender Utah Rating Scale (WURS) [38]. Further, controls in the SBP-S cohort completed the Barkley self-assessment test for ADHD symptoms [39] at the 7-year follow-up visit.

Study participants were assessed with a comprehensive neuropsychological test battery covering several cognitive domains [40]. In the SBP-S cohort, the tests included the Claeson-Dahl Verbal Learning and Retention test (CDT), the Color-Word Interference Test, the Design Fluency Test, the Tower Test, the Trail Making Test, and the Verbal Fluency Test from Delis-Kaplan Executive Function System (D-KEFS), the Wechsler Adult Intelligence Scale (WAIS version III or IV; all subtests except the Object Assembly were administered), the Continuous Performance Test II, and the Rey Complex Figure Test (REY). To reduce the workload for study participants, certain tests were omitted in the SBP-G cohort (at baseline and follow-up) and for follow-up examinations in SBP-S. Table 3 details the administered neuropsychological tests to the patient and control groups across visits.

Blood and CSF sampling were performed on the same day. Patients were in a stable euthymic mood. Participants fasted overnight before the blood collection but were served breakfast before the CSF collection. Blood samples were drawn between 8 and 9 a.m. and CSF samples between 9 and 11 a.m. An identical procedure was performed for the controls.

Blood samples were allowed to clot in room temperature for 30–60 min before centrifugation (10 min at 1,700 g). In Stockholm, the supernatant was kept in low temperature (<5°C) pending direct transport to the biobank within 4 hours for long-term storage at −70°C. In Gothenburg, the supernatant was immediately stored in a local −70°C freezer, awaiting bulk transport to the biobank.

For CSF sampling via lumbar puncture, the spinal needle was inserted into either the L3/L4 or L4/L5 interspace. A total volume of 12 mL of the CSF was collected, gently inverted to prevent gradient effects, and divided into 1.0–1.6 mL aliquots that were stored at −80°C pending analysis. In SBP-S, baseline samples were not centrifugated up to May 2011, but in SBP-G all samples were centrifugated. The procedure for CSF and blood sampling remained identical at the 7-year follow-up visit. At the 14-year follow-up visit, blood but not CSF was collected. Blood and CSF samples are stored at the Karolinska Institutet Biobank.

At the time of MRI scanning, patients were in a mood stable state. MRI scans for the SBP-S cohort were conducted at the MR Research Centre, Karolinska University Hospital, Stockholm. All subjects were examined in the same scanner for the baseline assessment and the first follow-up on a 1.5-Tesla MRI scanner (General Electric Signa Excite 1.5T) equipped with an eight-channel head coil. At second follow-up, the MRI scanner has been changed to a 3-Tesla General Electric Discovery MR750. Coronal 3D T1-weighted images were obtained using a spoiled gradient echo recall sequence (3D-SPGR), with the following parameters: repetition time (TR) of 21.0 ms, echo time (TE) of 6 ms, field of view (FOV) of 18 cm, flip angle of 30°, acquisition matrix of 256 × 256, 128 slices, and voxel size of 0.7 × 0.7 × 1.8 mm³. Axial fluid attenuation inversion recovery T2-weighted scans were acquired for examination by a senior radiologist. This assessment aimed to exclude clinically significant anatomical abnormalities or neuropathology. During the study, the calibration filter for the scanner was changed. This led to variations in the intensity distribution among subjects scanned with different filters. At the first follow-up, the scanning protocol was repeated, and two additional sequences were added: a resting-state functional magnetic resonance imaging (fMRI) sequence and a diffusion tensor imaging (DTI) scan. The resting-state fMRI sequence lasted 7.5 min and included the acquisition of 180 volumes. Each volume consisted of 39 BOLD-sensitive T2-weighted axial echo-planar images each. The imaging parameters were as follows: resolution of 3.79 mm × 3.79 mm, slice thickness 4 mm covering the whole brain, TR = 2.5 s, TE = 40 ms, FOV = 243 mm, and flip angle of 85°. The DTI protocol included TR = 7.39 s, TE = 85 ms, FOV = 220 mm, flip angle 90°, acquisition matrix 96 × 96, voxel size 0.94 × 0.94 mm, slice thickness 2.3 mm.

For the SBP-G cohort, MRI scanning was initially conducted on a Philips Achieva dStream 3T scanner using a 32-channel head coil. The scanner was replaced with a Philips M7700 3T scanner in January 2022. We conducted a structural T1-weighted 3D turbo field echo HR sequence (TR = 6 ms, TE = 2.8 ms, FOV 240 mm × 240 mm, flip angle 8°, acquisition matrix 242 × 232, 180 slices, voxel size 1 × 1 × 1 mm³), a resting-state fMRI sequence (resolution of 3.06 mm × 3.06 mm, slice thickness 3 mm, TR = 2.47 s, TE = 30 ms, FOV = 220 mm × 220 mm, flip angle = 75°), and a DTI sequence (TR = 7.62 s, TE = 92.46 ms, FOV 224 mm × 224 mm, flip angle 90°, acquisition matrix 96 × 98, voxel size 2.33 mm × 2.29 mm, slice thickness 2.3 mm).

Table 4 shows the number of individuals with bipolar disorder and controls who have undergone clinical interview, collection of biological samples, MRI brain scan, and neuropsychological assessment stratified by center and time point. While outcomes such as relapse rates and treatment responses have not yet been analyzed as primary outcomes, pending the completion of the 14-year follow-ups at both centers, they have been utilized as dependent variables in publications referenced below.

We have shown lower performance on neuropsychological tests across multiple cognitive domains in patients with bipolar disorder compared to controls [41, 42]. Importantly, however, we found that clinically significant impairment was restricted to a subgroup of patients, suggesting that cognitive impairments are not a general trait in bipolar disorder [43]. In a subsequent investigation, bipolar disorder individuals with or without comorbid ADHD were compared. Interestingly, no disparities in neuropsychological profiles were observed, except for poorer working memory performance among those with comorbid ADHD [44]. The importance of cognitive functioning was demonstrated in a study investigating the association between cognitive performance and occupational functioning among the individuals with bipolar disorder. Executive functioning emerged as a key predictor of occupational status in the population [45]. To shed light on whether cognitive functions worsen over time in bipolar disorder, we compared cognitive test results obtained at the 7-year follow-up visit with baseline performance. Interestingly, results revealed that the trajectory of cognitive performance remained similar between patients with bipolar disorder and controls [46].

Building upon the MRI brain scans conducted with study participants, we were able to show a significant association between lower executive functioning and deep white matter hypointensities in bipolar disorder [47]. Similarly, we found CSF biomarkers of neurodegeneration and neuroinflammation to be associated with cognitive impairment in individuals with bipolar disorder [48, 49]. With respect to personality traits, we found that individuals with bipolar disorder scored higher on measures of neuroticism, aggressiveness, and disinhibition compared with controls, but these personality traits were neither associated with worse outcome in bipolar disorder [50], nor with the presence of trigger factors for affective episodes [51].

To shed light on the pathophysiology of bipolar disorder, we have undertaken a series of studies comparing potential CSF biomarkers between individuals with bipolar disorder and controls. We found that individuals with bipolar disorder had altered concentrations of amyloid precursor protein metabolites [52], micrometer-sized particles in CSF [53], tryptophan metabolites [54], monoamine metabolites [55], secretogranin II [56], neurofilament light chain [57], amino acids related to glutamate signaling [58], interleukin 1β [59], monocyte and microglia markers [60], interleukin-8 [61], and blood-CSF barrier dysfunction [62]. CSF metabolomics [63] and proteomics studies [64, 65] have also revealed differences between individuals with bipolar disorder and controls in the St. Göran project.

Fewer studies have revealed associations between CSF biomarkers and disease severity or outcome. We did, however, found low CSF concentration of neuropeptide Y to be associated with higher risk of suicide attempts [66] and low secretogranin concentrations to be associated with more severe bipolar disorder [56]. Further, we found higher concentrations of CSF kynurenic acid to be associated with manic and psychotic features [67] and lower concentration of CSF HVA and 5-HIAA to be associated with a history of childhood ADHD in bipolar disorder type 1 [68].

By comparing blood samples from individuals with bipolar disorder and controls, we have revealed altered concentrations of brain-derived neurotrophic factor [69], zinc [70], and lipids [71, 72]. Further, blood metabolomics [73] and proteomics studies [74] have also found differences between individuals with bipolar disorder and controls.

We have found structural brain differences between individuals with bipolar disorder and controls [75]. Further, data from patients and controls in the SBP have been included in the Enhancing Neuro Imaging Genetics through Meta-Analysis (ENIGMA) consortium, thereby contributing to large-scale collaborative brain imaging studies [76‒83]. In addition, we have demonstrated associations between brain structure, executive function [84], and psychomotor function [85, 86]. We have explored the association between psychotic symptoms and grey matter volume [87] and how manic episodes impact brain structure [88, 89]. Furthermore, we found that the polygenic risk for bipolar disorder and schizophrenia was related to a thinner ventromedial prefrontal cortex [90]. Finally, cortical brain structure in relation to sexual orientation has been studied in females with bipolar disorder and ADHD [91].

Using functional MRI data, differences in functional connectivity measures between individuals with bipolar disorder and controls were shown in the default mode network and in the functional connectivity between the default mode network, the frontoparietal network, and the somatomotor network [90, 92]. Additionally, recent work has shown differences also in the somatomotor network in individuals with bipolar disorder as compared to controls [93].

In the SBP, both candidate gene studies and genome-wide analyses have identified genetic variants associated with blood or CSF concentrations of biomarkers [94‒98], brain structure and function [88, 90, 99, 100], and cognitive function [84, 101, 102]. Further, genome-wide studies of CSF biomarkers have identified biological pathways associated with bipolar disorder [64, 103].

The two key strengths of the SBP are the detailed and multimodal information collected at each study visit and the extended follow-up period. The wealth of detailed information gathered allows for the exploration of a range of questions, drawing from diverse combinations of data sources. The longitudinal nature of the study enables the investigation of prognostic factors and the trajectory of the illness over time, offering valuable insights into the course of bipolar disorder. Moreover, the availability of two separate cohorts using the same study protocol enhances the reproducibility of findings. Lack of replication poses a significant challenge problem in research and causes of non-replication can be difficult to identify. By using a consistent study protocol across both cohorts in the SBP, several sources of error can be effectively controlled for, bolstering the reliability and validity of study results.

Our patient enrollment strategy was built on the integration of the research project with outpatient care units, rather than advertising for participants. This naturalistic setting of the study is both a strength and a weakness. On the one hand, findings are more likely to be applicable to individuals with bipolar disorder in real-world scenarios. However, the variability in completeness of collected information is substantial for some modalities and dropout rates between baseline and follow-up study visits can introduce selection bias. Various strategies, such as censoring of data and multiple imputation, have been employed in published work to deal with missing data. Nonetheless, the lack of data completeness and study dropout remain potential sources of bias to be considered in the interpretation of results.

Even though the control population in the study was randomly selected from the general population, it may not be fully representative as recruitment was limited to individuals who actively responded to contact. This could introduce bias for comparisons between the patient and control groups.

Finally, since the study was primarily designed to provide a resource for research rather than to address a specific research question, the amount of information may not always be adequate for certain research questions. Ancillary information or biological samples that could be relevant are not always available. For example, published work on cardiometabolic risk factors [71, 72] would have benefited from more detailed information on diet and physical activity that were not available. And while MRI brain scans, blood, and CSF are available, the study lack, for instance, samples collected for mRNA expression analyses.

In summary, the SBP has generated many original findings since its inception, leveraging the advantages of longitudinal follow-ups and the presence of a replication cohort within the study framework. While data collection is approaching completion, there remain ample opportunities to address new research questions using the wealth of collected data. Further, SBP has served as a valuable platform for training junior researchers, involving numerous doctoral and post-doctoral researchers. The project provides an active research environment, robust research infrastructure, and access to a rich dataset, contributing to the advancement of knowledge in the field of bipolar disorder. Future studies should build on longitudinal data sampling and be guided by initiatives like the SBP to address critical unmet needs, including the identification of diagnostic and prognostic biomarkers, predictors of treatment response, and advancing understanding about individual differences in the trajectory of bipolar disorder.

We are deeply grateful for the participation of all individuals contributing to this research and to the staff at the bipolar affective disorder unit (Affektivt centrum) at Northern Stockholm Psychiatry and the Bipolar Clinic at the Sahlgrenska University Hospital in Gothenburg (Bipolärmottagning). We further thank study nurses Agneta Carlswärd-Kjellin, Annika Blom, Benita Gezelius, Lena Lundberg, Stefanie Unger, Stina Stadler, and Therese Turesson, study coordinator Martina Wennberg, test psychologist Andreas Aspholmer, and data manager Haydeh Olofsson. We thank Björn Hultman and Pascal Borgström for designing the neuropsychological test battery. We thank Rouslan Sitnikov for help with the MRI-protocol. Yngve Hallström is acknowledged for performing lumbar punctures in the SBP-S. We finally wish to thank the BBMRI.se and KI Biobank at Karolinska Institutet for professional biobank service.

SBP is an observational clinical study registered on Researchweb (https://www.researchweb.org/is/vgr/project/39411) on January 1st, 2009 (project ID: 39411). All study participants provided oral and written informed consent to participate in the study. The ethical review board in Stockholm approved the study in 2005 (dnr 2005/3:7) and several amendments have been added to the original ethical approval (dnr 2008/1931-32, 2009/1221-32, 2011/1700-32, 2012/559-32, 2012/762-39, 2012/1488-32, 2013/43-32, 2013/1418-32, 2014/678-32, 2014/1789-32, 2015/618-32, and 2021-00326).

Mikael Landén declares that he has received lecture honoraria from Lundbeck pharmaceuticals. Christoph Abé is an employee of Quantify Research. No other author has any conflict of interest to declare.

The SBP was supported by grants from the Swedish Research Council (2008-14647-06-3, 2010-21569, 2010-21568, 2018-02653, 2022-01643), Hjärnfonden/the Swedish Brain foundation (FO2020-0261, FO2022-0217), the Swedish foundation for Strategic Research (KF10-0039), and the Swedish Government under the LUA/ALF agreement (ALF 20060378, ALFGBG-142041, ALFGBG-426721, ALFGBG-965444).

E.P. drafted the manuscript and prepared figures. E.P. and M.K. prepared the tables. M.L., L.J., A.L.K., M.K., A.G., C.A., A.A., B.L., A.P., T.S., and E.P. edited and reviewed the manuscript and approved the final version.

Access and use of research data are regulated by specific national legislation in addition to general data protection legislation. Research data in the SBP are controlled by the University of Gothenburg, which handles questions on access to and use of data. For researchers, please contact the corresponding author for more information on data sharing and access to meta-data for the SBP.