Pulmonary Function Deficit and Clinical Associations in Childhood Acute Lymphoblastic Leukaemia Survivors: A National Retrospective ALL‐STAR Lungs Cohort Study

Authors: Sonja Izquierdo Riis Meyer, Birgitte Klug Albertsen, Mette Tiedemann Skipper, Ruta Tuckuviene, Peder Skov Wehner, Thomas Leth Frandsen, Kjeld Schmiegelow, Liv Andrés‐Jensen, Kim Gjerum Nielsen, Sune Leisgaard Mørck Rubak

Abstract

Background

Although survival for childhood acute lymphoblastic leukaemia (ALL) has improved, long‐term pulmonary function deficit remains a concern. We aimed to explore the prevalence of pulmonary function deficit and abnormal pulmonary function tests in childhood ALL survivors and clinical associations.

Methods

This national, retrospective cohort study (February 2019–May 2024) included eligible 5–17.9 year‐old survivors (N = 295) who performed a valid pulmonary function test ≥ 1 year after treatment (N = 185). Clinical associations included treatment characteristics, pulmonary diagnosis, and radiological findings from medical charts.

Results

Among 185 survivors, 37% (95% confidence interval (CI): [30, 45]) had pulmonary function deficit, with the highest prevalence (56%) among 15–17.9‐year‐olds. Abnormal test prevalence was 37% [27, 48] for diffusing capacity (carbon monoxide and/or nitric oxide), 29% [22, 38] for lung clearance index, 12% [8,18] for forced expiratory volume in the first second, and 11% [7, 17] for broncho‐dilator response. Bronchiolitis obliterans, stem cell transplantation and CT‐verified bronchiectasis were significant clinical associations of pulmonary function deficit (100%, 97.5% CI: [40, 100], 89% [76, 103], 73% [51, 96]) and abnormal diffusing capacity (100% (97.5% CI [29, 100]), 82% [59, 105], 75% [45, 105]), respectively. Bronchiectasis (88% [65, 110]) and transplantation (64% [39, 89]) were associated with a higher prevalence of abnormal lung clearance index. Bronchiolitis obliterans (75% [19, 99]) and transplantation (61% [39, 84]) were associated with a higher prevalence of abnormal forced expiratory volume in the first second.

Conclusions

Pulmonary function deficit was frequent in childhood ALL survivors, especially after stem cell transplantation or pulmonary disease. Tailored long‐term pulmonary monitoring, including small airway function and diffusing capacity, may aid in timely detection and intervention.

Introduction

Despite improved survival rates for childhood acute lymphoblastic leukemia (ALL) under the contemporary Nordic Society of Pediatric Hematology and Oncology (NOPHO) ALL2008 protocol [1], treatment‐related toxicities remain a substantial risk [2]. Severe toxicities, including acute pulmonary toxicities, occur in approximately 50% of pediatric ALL patients [2, 3, 4, 5]. However, persisting late pulmonary adverse effects (LPAE), including pulmonary function deficit (PFD), defined as abnormal pulmonary function test (PFT), and the association with pulmonary toxicities, remain poorly explored.

PFD ranks 9th in absolute cumulative burden, affecting 17% of 25‐year‐old ALL survivors, with prevalence increasing with age [6]. Previous studies on LPAE‐associated factors have primarily focused on (now obsolete) thoracic radiation [4, 7, 8] and certain pulmotoxic chemotherapeutics (e.g., bleomycin, lomustine, carmustine, busulfan) [9, 10, 11], limiting their relevance to childhood survivors treated under contemporary ALL protocols. Additionally, high‐risk chemotherapy followed by stem cell transplantation (HR‐SCT) significantly reduces pulmonary diffusing capacity for carbon monoxide (DLCO) compared to chemotherapy only [12]. Bronchiolitis obliterans (BO), primarily associated with graft‐versus‐host disease (GvHD), and BO syndrome, characterized by new‐onset airway obstruction on spirometry, are recognized chronic HR‐SCT complications [13].

Identifying additional factors associated with PFD in ALL survivors is crucial for improving long‐term follow‐up recommendations, including timely use of PFTs for detection and intervention. However, the impact of pulmonary‐specific associated factors, such as pulmonary disease, toxicity, and thoracic imaging findings during and after ALL treatment, remains unclear. We aimed to assess the prevalence of PFD, including specific abnormal PFTs in childhood ALL survivors, while identifying potential associated factors. We hypothesized that the prevalence of PFD is higher among survivors with a history of pulmonary disease, pulmonary toxicity, or pulmonary disease‐associated abnormalities in thoracic imaging from ALL diagnosis than among survivors without these conditions.

Materials and Methods

Study Design

This national, exploratory, retrospective cohort study was part of the Acute Lymphoblastic Leukemia Survivor Toxicity And Rehabilitation (ALL‐STAR) study, a Danish cross‐sectional follow‐up study of patient‐reported and objective treatment‐related morbidity in childhood and young adult ALL survivors following ALL cure at one of the four Danish pediatric hematology and oncology treatment centers [14]. Given the complex interplay between ALL treatment, clinical characteristics, and pulmonary outcomes, the present study was designed as an exploratory analysis to characterize pulmonary function deficits and their clinical associations rather than to establish causal or predictive relationships. Study recruitment and clinical examinations across all organ systems were performed at Copenhagen University Hospital and Aarhus University Hospital (February 2019 to May 2024). The reporting of this study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [15]. Data were managed using Research Electronic Data Capture (REDCap) tools hosted at The Capital Region of Denmark [16, 17].

Study Subjects

ALL survivors aged 5–17.9 years upon examination were identified in the NOPHO ALL2008 registry (Figure 1). Eligible survivors included relapse‐treated survivors diagnosed between July 2008 and October 2018, at least 1 year after treatment cessation under the NOPHO ALL2008 protocol in Denmark (Philadelphia negative B‐cell precursor (BCP) or T‐cell ALL) (N = 295) [14]. A NOPHO ALL2008 protocol overview is provided in E‐Figure S1 in E‐Appendix S1. Written informed consent was obtained from the legal guardians of participants aged < 18.

Pulmonary Outcomes

A blood sample (hemoglobin concentration for diffusing capacity) was collected on the examination day. Additional methodological details on PFT procedures, quality assessment, and device specifications are available in E‐Table S1a and the E‐Appendix S1. PFTs were performed once the same day as part of the ALL‐STAR examination and validated following European Respiratory Society (ERS) and the American Thoracic Society (ATS) recommendations. Only these PFTs were analyzed; no preexisting clinically indicated PFTs were included. They included nitrogen multiple breath washout test (N2MBW) measuring lung clearance index (LCI) [18, 19, 20, 21], spirometry measuring forced expiratory volume in the first second (FEV1) [22], impulse oscillometry (IOS) including measure of airway resistance at a frequency of 5 Hz (R5Hz) [23], DLCO [24] and nitric oxide (DLNO) [25] and FEV1 and R5Hz broncho dilator response (BDR) [22, 23]. An overview of the known mechanisms contributing to PFD and the role of various PFTs in detecting them is provided in Figure 2. Survivors with ≥ one valid PFT were included and categorized as survivors with PFD if ≥ one abnormal valid PFT. Primary outcomes were (1) PFD, (2) abnormal LCI, (3) abnormal FEV1, (4) abnormal R5Hz, (5) abnormal DLCO and/or DLNO (DLCO/DLNO), and (6) abnormal FEV1 BDR and/or R5Hz BDR. PFT was abnormal if the z‐score was > +1.645 from the predicted value for LCI [27, 28] and R5Hz [33] or < −1.645 from the predicted value for FEV1 [31, 32], DLCO [29], and DLNO [30]. Abnormal BDR (+BDR) was defined as FEV1 BDR > 10% of predicted increase [31, 32] and/or R5Hz BDR < −40% of baseline [23] (details in E‐Table S1b). At both study sites, examinations were conducted by a trained team of pediatric research nurses, research assistants, and a physician.

Associated Factors

Demographic characteristics were obtained from the NOPHO registry (age, time from ALL diagnosis), a physical examination (height and body mass index (BMI)), and a questionnaire (current smoking status) [14]. An additional complete list of demographic and outcome variables is provided in E‐Table S1a in E‐Appendix S1. Secondary outcomes included predefined factors associated factors with PFD and abnormal specific PFTs identified from ALL treatment characteristics, diagnosed pulmonary disease, including pneumotoxicity, and pulmonary disease‐associated abnormalities in thoracic imaging (defined below).

Clinical characteristics from ALL (obtained from the NOPHO registry) included (1) immunophenotype, (2) final treatment risk‐stratified subgroups (e.g. standard risk (SR), intermediate risk (IR), high risk (HR) chemotherapy), (3) HR followed by stem cell transplantation (HR‐SCT), (4) total body irradiation and (5) relapse (E‐Table S1a).

Pulmonary‐specific factors from diagnosed pulmonary disease included (1) any pulmonary diagnosis (including asthma, but excluding pneumotoxicity), including (1a) BO (both BO and BO syndrome) and (1b) other interstitial lung disease (ILD), and separately (2) any pneumotoxicity (any lower airway infection; viral, bacterial or fungal infection) (E‐Table S1a). Data on these pulmonary factors were extracted from medical charts using targeted keyword searches in physicians' diagnostic descriptions and Diagnosis‐Related Group (DRG) codes and cross‐referenced with microbiological test results, covering the period from each participant's ALL diagnosis to their examination date. Pre‐existing conditions prior to ALL diagnosis were not assessed.

Other pulmonary specific factors based exclusively on radiologist‐reported findings included any of the (a) infiltrate, (b) atelectasis, (c) pneumothorax, (d) pleural empyema, (e) bronchiectasis, (f) ground glass opacities, (g) fibrosis, (h) air trapping, (i) mosaic attenuation and (j) emphysema from all (1) thoracic X‐rays (clinically indicated) (a–e), (2) computed tomography (CT) scans (a–j), (3) magnetic resonance imaging (MRI) (b–g) and (4) ultrasounds (b–d) performed during the same period (E‐Table S1a). In cases with missing radiologist reports (105 thoracic X‐rays, 7 CT scans), the same experienced pediatric radiologist provided the necessary description to ensure complete data collection. Data from medical charts and radiologist reports were consistently collected by the first author.

Statistical Methods

Statistical analyses were two‐sided (significance 0.05) and conducted with a descriptive approach with unadjusted estimates using Stata/SE 18 (StataCorp LP, College Station, Texas, USA) [34]. Participant demographics were summarized by PFD status. Continuous variables were presented using means with standard deviations (± SD) or medians with interquartile ranges (IQR), depending on whether normality could be assumed [35]. Dichotomous variables were depicted as frequencies and percentages. Groups were compared using Fisher's exact test for dichotomous and categorical variables, unpaired t‐test or ANOVA for normally distributed continuous variables, and the Wilcoxon rank‐sum or Kruskal‐Wallis test for non‐normally distributed continuous variables [35, 36]. A binary regression model was used for each outcome to estimate risk and risk difference (RD) with 95% confidence interval [95% CI] for each associated factor separately [36]. RD was not calculated when fewer than five participants met the outcome criteria. Instead, Fisher's exact test was reported. Analyses were adjusted for age, gender, and height using PFT z‐score variables. Due to the exploratory design of the study and the relatively limited sample size, no formal adjustment for multiple comparisons was applied [37]. Results should therefore be interpreted as hypothesis‐generating. In sub‐analyses, demographics, risk group distribution, and frequency of thoracic imaging were compared for included survivors and survivors excluded due to invalid PFTs, no PFT performed, and ALL treated patients deceased or lost to follow‐up (LTFU).

Ethics

The ALL‐STAR study adhered to the declaration of Helsinki and was approved by the Regional Ethics Committee for the Capital Region in Denmark (H‐18035090/H‐20006359) and the Danish Data Protection Agency (VD‐2018–519).

Results

Participant Demographics

Out of 295 eligible survivors, we included 185 survivors with ≥ one valid PFT in the analyses and a corresponding number of survivors with a valid specific PFT: 126 with N2MBW, 175 with spirometry, 61 with IOS, 89 with DLCO and/or DLNO, and 153 with BDR (spirometry and/or IOS) (Figure 1). Included survivors were older (mean 11.92 (± 2.95) vs. 8.06 (± 2.12) years, p = 0.001), taller (mean 151.42 (± 17.20) vs. 132.93 (± 11.72) cm, p = 0.005) and less recently diagnosed with ALL (mean 7.15 (± 2.36) vs. 5.20 (± 1.50) years, p = 0.032) than survivors excluded due to invalid PFTs (See E‐Table S2 in E‐Appendix S1). Participating survivors resembled excluded and declined survivors, deceased ALL patients, and those LTFU regarding risk group distribution, gender, smoking status, imaging frequency, height z‐score, age at ALL diagnosis, and BMI. Additional sub‐analysis with demographic comparison of participants and non‐participants is provided in E‐Table S2 and E‐Appendix S1.

Pulmonary Function Deficit

ALL survivors with PFD (37%, 69/185, 95% CI [32, 38]) were older at examination (mean 12.57 (± 2.86) vs. 11.53 (± 2.94) years, p = 0.021), shorter (mean height z‐score −0.45 (± 1.15) vs. 0.02 (± 1.10), p = 0.006) and older at ALL diagnosis (median 5.30 (3.60, 7.70) vs. 3.75 (2.65, 5.25) years, p < 0.001) than those without PFD (Table 1 and Figure 3). Both groups were comparable regarding gender, smoking status, frequency of thoracic imaging (X‐rays, ultrasounds, and MRIs), time since ALL diagnosis (years), time since end of therapy (years), height (cm), and BMI (z‐score and kg/m^2^). We observed a trend towards higher frequency of CT scans performed in survivors with PFD (26/69 (37.7%) vs. 28/116 (24.1%), p = 0.066). Furthermore, the PFD prevalence was higher among 15–17.9‐year‐olds (15/27, 56%) (E‐Table S1b). Significant PFD associations included HR‐SCT (17/19, RD 0.58, 95% CI [0.43, 0.74]), BO (4/4, p = 0.018), and CT‐verified bronchiectasis (11/15, RD 0.35 [0.08, 0.62]) (Figure 3). Pneumotoxicity was prevalent in 71.4% (132/185). Pneumotoxicity (48/132, RD −0.03 [−0.19, 0.12]) was not associated with PFD. We found no prior MRIs for survivors with any PFT abnormalities (Figures 3, 4, 5 ). No associations with abnormal IOS were found.

Abnormal LCI

LCI was abnormal in 29% (37/126 [22, 39]) of survivors with a valid N2MBW (Figure 4). HR‐SCT (9/14, RD 0.39 [013, 0.66]), a pulmonary diagnosis (10/20, RD 0.25 [0.01, 0.48]), and CT‐verified bronchiectasis (7/8, RD 0.57 [0.28, 0.86]) were significantly associated with abnormal LCI. Among the 10 survivors with abnormal LCI and a pulmonary diagnosis, 6 had a history of asthma and/or ILD, BO, or bronchiectasis. Trends indicating an increased risk of abnormal LCI were observed for relapse (4/6, p = 0.061), BO (2/2, p = 0.085), and X‐ray verified bronchiectasis (3/4, p = 0.073) (Figure 4).

Abnormal FEV1

An abnormal FEV1 was observed in 12% (21/175 [8, 18]) of survivors with valid spirometry (Figure 4). Significant associations included relapse (4/6, *p * = 0.002), HR‐SCT (11/18, RD 0.55 [0.32, 0.78]), BO (3/4, p = 0.006), and X‐ray verified atelectasis (10/38, RD 0.19 [0.04, 0.34]). We observed a trend towards an increased risk of abnormal FEV1 for survivors with X‐ray verified bronchiectasis (2/4, p = 0.073). Four survivors were diagnosed with BO (three HR‐SCT, one IR), of whom three had abnormal FEV1 consistent with the diagnostic criteria for BO syndrome, while one exhibited only +BDR. A BO diagnosis was established in 15.8% (3/19) of HR‐SCT survivors.

Abnormal DLCO/DLNO

DLCO/DLNO abnormalities were present in 37% (33/89 [28, 40]) of survivors with a valid test (Figure 5). Significant associations with abnormal DLCO/DLNO included HR‐SCT (9/11, RD 0.51 [0.26, 0.76]), BO (3/3, p = 0.048) and CT‐verified bronchiectasis (6/8, RD 0.37 [0.00, 0.73]).

Abnormal Broncho Dilator Response (+BDR)

+BDR was observed in 11% (17/153 [7, 17]) of survivors with a valid BDR (Figure 5).

Discussion

In this exploratory study of a national cohort, the prevalence of PFD was common, and several clinical characteristics were associated with these findings. These results should be interpreted as descriptive and hypothesis‐generating. The present study confirmed that the prevalence of PFD in childhood ALL survivors was substantial (37%), particularly among survivors aged 15–17.9 years at the time of their examination (56%). We also confirmed that a history of a pulmonary diagnosis, pulmonary toxicity, or pulmonary disease‐associated abnormalities in thoracic imaging following ALL diagnosis was significantly associated with a higher PFD prevalence in survivors. PFD and the most prevalent abnormal PFTs ‐ abnormal LCI (29%) and abnormal DLCO/DLNO (37%) ‐ were associated with a history of pulmonary diagnosis, BO, CT‐verified bronchiectasis, and HR‐SCT. The PFD prevalence encompasses multiple PFT modalities, which may detect subclinical deficits that spirometry and/or static lung volumes alone would not capture. Previous reports on overall PFD prevalence in childhood ALL survivors are scarce. However, prior childhood cancer survivor (CCSs) studies report PFD prevalences from 44% to 84.1% (depending on screening criteria) [39, 41, 42]. The prevalence of impaired DLCO/DLNO (37%) and abnormal FEV1 (12%) aligns with the previous few reports on childhood ALL survivors. Impaired DLCO was found in 13%–58%, reduced FEV1 in 0%–9% [12, 43], emphasizing the clinical relevance of exploring further PFT modalities in this population. In contrast, IOS results were normal, which may reflect the relatively small subgroup tested at one site and a lower sensitivity of IOS compared with LCI and DLCO/DLNO for detecting subtle late effects. Therefore IOS results should be interpreted with caution. Our finding of impaired LCI in 29% aligns well with previous reports, which found a prevalence of 34% in long‐term survivors of pediatric SCT [44] and 23%–60% in a heterogeneously treated group of CCSs [45]. Reports on prevalences of +BDR were unavailable for comparison.

Small airway dysfunction, measured by LCI, has been associated with persisting respiratory symptoms following SCT [44] and chronic GvHD [38, 46, 47]. In the present study, only one‐fourth of survivors with impaired LCI had a known pulmonary diagnosis, primarily asthma and/or ILD, BO, or bronchiectasis. These conditions are known to be associated with potential late effects of chemotherapy and SCT, especially when asthma involves +BDR and is not related to alternative causes (e.g., eosinophilia, allergy). One could speculate whether the remaining survivors with impaired LCI may have had undiagnosed pulmonary conditions.

Reduced FEV1, indicating diminished lung function, has been similarly observed in ALL survivors [4] and associated with CCSs who experienced relapse after SCT [48]. Atelectasis, related to lower airway infection, explains FEV1 impairment rather than irreversible lung sequelae. The observed frequency of pneumotoxicity was 71.4%, which was higher than the previously reported 40% [3, 49]. This discrepancy may be explained by the extended registration period beyond ALL treatment cessation in the present study, unlike previous reports that were limited to the ALL treatment period.

Impaired DLCO was similarly associated with HR‐SCT, BO, and bronchiectasis in other ALL and SCT survivor studies [4, 12, 40]. These factors were also linked to impaired DLNO in the present study. The BO/BO syndrome frequency among HR‐SCT survivors (15.8%) surpassed the 2.7% and 5.5% reported in other mixed‐diagnosis SCT cohorts [50, 51], as we specifically assessed registered BO diagnoses in HR‐SCT survivors with valid PFTs. The BO‐PFD association reflects the progressive structural airway changes (caused by BO) that lead to functional impairments [52], measurable by declines in FEV1 [53], DLCO [54], and/or DLNO, as also observed in the present study. Airway allo‐reactivity has been reported following SCT and linked to GvHD [44, 46], suggesting underlying active airway hyperreactivity.

Distinct associated factors were identified for impaired LCI compared to those associated with impaired FEV1 and +BDR. Chronic conditions (e.g., pulmonary diagnosis and bronchiectasis) were associated with impaired LCI, while reversible causes like atelectasis were associated with impaired FEV1. However, these associated factors account for only approximately half of the abnormal LCI and FEV1 cases. Bronchiectasis is often caused by other pulmonary conditions affecting the small airways, leading to irreversible airway dilation and recurrent respiratory infections/inflammation. This, in turn, causes impaired alveolar gas exchange and declining lung function over time.

Importantly, ALL treatment may significantly contribute to pulmonary complications, particularly in HR‐SCT survivors, increasing the risk of interstitial lung sequelae, as indicated by impaired DLCO/DLNO. When considered alongside impaired LCI, these findings suggest significant PFDs. However, whether these effects arise solely from treatment‐related adverse effects or pre‐existing conditions remains unclear.

This is the first study to examine the prevalence of PFD, including novel PFTs (N2MBW, IOS, DLNO), alongside pulmonary‐specific associated factors in a national pediatric ALL survivor cohort. The findings highlight that observed pulmonary impairments, even in the absence of current symptoms and physical limitations, may predispose survivors to long‐term complications. While some pulmonary abnormalities may be clinically suspected, pulmonary function testing provides objective quantification of impairment and may identify subclinical physiological deficits not captured by routine clinical assessment. If left undetected, these impairments may progress, leading to physical function limitations and reduced quality of life, as seen in SCT survivors and other CCSs [44, 55]. Current national CCS follow‐up guidelines differ in their recommendations regarding the management and active surveillance of HR‐SCT survivors with pulmonary infections [56, 57, 58]. The recent comprehensive International Guideline Harmonization Group (IGHG) recommendations do not support routine PFTs for HR‐SCT‐treated survivors, primarily based on expert opinions and evidence lack for routine surveillance [59]. The present study addresses this evidence gap, aligning with the IGHG's call for future research. We used various PFTs in a sufficiently large cohort to ensure statistical power and enable stratified analysis, addressing the limitations of previous heterogeneous studies. The NOPHO ALL2008 protocol shares similarities with other treatment regimens, underscoring our findings' applicability and clinical relevance. Thus, we recommend implementing tailored risk‐stratified PFT monitoring beyond spirometry to detect PFD and early chronic pulmonary disease in HR‐SCT survivors and those with a history of pulmonary diagnosis, infection, or bronchiectasis. This approach should include (1) DLCO supplemented by (2) N2MBW (LCI) in survivors with pulmonary complications/diagnosis, (3) consideration of DLNO use. DLNO has demonstrated superior to DLCO in detecting gas exchange impairments in ALL survivors as reported in our recent study [60].

Strengths and Limitations

This study has several strengths, including a robust sample size and a national cohort design alongside consistent data collection across multiple centers, which enhanced the applicability of the results. Using standardized height, gender, and age‐adjusted PFT z‐scores improved reliability and comparability with other studies. The recruitment rate (185/295) was sufficient to ensure a representative sample, minimizing selection bias. However, selection bias may be present, as thoracic imaging and scans were performed based on clinical indications, possibly leading to overrepresentation of the most severely affected survivors, especially for CT scans and ultrasound (Figures 3, 4, 5). However, this reflects clinical practice and highlights relevant factors associated with PFD and abnormal PFTs, particularly in these patients. The relatively small sample size, in combination with multiple statistical comparisons, increases the risk of type I error, and findings should therefore be interpreted cautiously. The analyses were exploratory and not designed to establish predictive models. Owing to the exploratory design and overlapping factors, the results should be considered descriptive and hypothesis‐generating. Not all participants underwent (all) PFTs, which may have led to under‐ or overestimation of PFD and specific abnormal PFT prevalences. N2MBW a‐files were unavailable for reanalysis in the latest Spiroware version for a group of participants; thus, pooled data from versions 3.1.6 and 3.3.1/3.3.2 were used, which may have introduced minor variability in LCI outcomes. Furthermore, inaccuracies in pulmonary history may have arisen due to the retrospective design, though using doctors' descriptions and diagnosis codes limits this risk. The known and observed challenges of obtaining valid PFTs in younger survivors may restrict the generalizability of our findings.

Conclusions

We identified substantial prevalences of PFD, especially in adolescents, alongside frequent small airway dysfunction and alveolar gas inefficiency. ALL survivors with a history of pulmonary diagnosis or infections, BO, CT‐verified bronchiectasis, and/or HR‐SCT had an increased prevalence of PFD. As this was an exploratory study, future studies need to confirm our results in larger cohorts. To conclude, incorporating LCI and DLCO/DLNO in long‐term follow‐up may help identify survivors at risk of PFD for timely intervention and guide future surveillance strategies.

Author Contributions

Sonja Izquierdo Riis Meyer: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, resources. Birgitte Klug Albertsen: conceptualization, investigation, funding acquisition, writing – review and editing, methodology, project administration, supervision, resources. Mette Tiedemann Skipper: conceptualization, investigation, funding acquisition, writing – review and editing, project administration, supervision, methodology. Ruta Tuckuviene: investigation, writing – review and editing. Peder Skov Wehner: writing – review and editing, investigation. Thomas Leth Frandsen: conceptualization, funding acquisition, writing – review and editing, project administration, supervision, resources. Kjeld Schmiegelow: funding acquisition, conceptualization, writing – review and editing, project administration, supervision, resources. Liv Andrés‐Jensen: conceptualization, investigation, funding acquisition, writing – review and editing, methodology, project administration, data curation, supervision, resources. Kim Gjerum Nielsen: conceptualization, investigation, writing – review and editing, validation, project administration, data curation, supervision, resources, methodology. Sune Leisgaard Mørck Rubak: conceptualization, investigation, funding acquisition, writing – review and editing, methodology, validation, project administration, data curation, supervision, resources.

Ethics Statement

The Regional Ethics Committee, Capital Region, Denmark (H‐18035090/H‐20006359) and the Danish Data Protection Agency (VD‐2018–519).

Conflicts of Interest

Dr. Schmiegelow has received speaker and advisory honoraria from Jazz Pharmaceuticals and Servier (2023) and holds stock in Novo Nordisk (unrelated to this study). The other authors declare no conflicts of interest.

Supporting information