Abstract
Polypharmacy is a result of multimorbidity and may limit treatment options in patients with lymphoid cancer (LC). To investigate the clinical impact of polypharmacy, we gathered a prediagnostic 1‐year medication history from the Danish prescription register for 46,803 newly diagnosed patients with six lymphoma subtypes, chronic lymphocytic leukemia, and multiple myeloma (MM). Medication was grouped into individual drug classes (n = 79), polypharmacy, and the number of prescription medications and correlated with overall survival (OS), hospitalization, and severe infection. Adjusting for age, sex, and multiple testing, we demonstrated associations between 39 drug classes and OS, 29 drug classes and hospitalization, and 27 drug classes and severe infection. Although polypharmacy (<5 vs. ≥5) was associated with adverse outcomes (hazard ratio [HR] 1.4 for OS, hospitalization, and severe infection; P < 0.001), the number of medications (0–3 vs. 4–7 vs. 8–11 vs. >11) gradually stratified OS (HR 1.0 vs. 1.2 vs. 1.4 vs. 1.9, respectively; P < 0.001), hospitalization (HR 1.0 vs. 1.3 vs. 1.4 vs. 1.9, respectively; P < 0.001), and severe infection (HR 1.0 vs. 1.2 vs. 1.5 vs. 1.9, respectively; P < 0.001) in multivariable analyses adjusted for age, sex, comorbidity, and prognostic indices. Lastly, the time to next treatment for Hodgkin lymphoma, mantle cell lymphoma, and MM was gradually shorter with an increasing number of medications (P < 0.001). In conclusion, a 1‐year medication history summarized as the number of medications is a strong, independent prognostic and predictive marker that should be considered as a key baseline characteristic in randomized clinical trials and in clinical practice for LC patients.
INTRODUCTION
Comorbidity is common among older patients and may limit treatment options for patients with lymphoid cancers (LCs) such as lymphoma, chronic lymphocytic leukemia (CLL), and multiple myeloma (MM). 1 Antineoplastic treatment in LC is mostly determined by a limited number of factors such as age, stage, comorbidity, and genetics/cytogenetics. 2 , 3 , 4 For instance, pulmonary disease is considered a relative contraindication for bleomycin and cardiac disease for doxorubicin and ibrutinib. 5 Further, frailty impacts treatment options and dose intensity and may be assessed by information on activities of daily living. This, however, is rarely performed systematically in routine clinical practice. 6 , 7 By contrast, methods for assessing comorbidity using the Charlson comorbidity index (CCI) score by International Classification of Diseases version 10 (ICD‐10) codes are robust and have been both automatized and validated. 8
Utilizing the Danish Lymphoid Cancer Research (DALY‐CARE) data resource, 9 we recently suggested to include information on prescription medication on top of diagnosis codes to better define certain comorbidities, where the indication for a medication is narrow and specific for the underlying medical condition (e.g., antidiabetics for type 2 diabetes and antimicrobials against infections). 10 , 11 , 12 , 13 Similarly, routine biochemistry may accurately define conditions such as anemia, hypogammaglobulinemia, and acute kidney injury. 14 Most often, clinically relevant comorbidity is accompanied by medication, leading to polypharmacy in multimorbid patients. Thus, polypharmacy acts as a proxy for multimorbidity. Without a clear consensus definition, polypharmacy is most often defined as the use of five or more different medications and is associated with mortality, adverse drug reactions, and longer hospitalization. 1 , 15 Medication is grouped using World Health Organization anatomical therapeutic chemical (ATC) codes according to organ system (e.g., antiemetics) or mechanism of action (e.g., calcium channel blockers), and the ATC level may define specific drugs or their more broad applications (e.g., phenoxymethylpenicillin [J01CE02], penicillins [J01C], or systemic antibacterials [J01]). In Denmark, almost all medication requires a prescription, and over‐the‐counter medication is limited to vitamins and small packages of paracetamol, ibuprofen, and third‐generation antihistamines. 16
Prescription medication has previously been used to define comorbidity. For instance, the Nordic multimorbidity index (NMI) was recently developed and validated in more than half a million individuals selecting the top 50 predictors of 5‐year survival from both ICD‐10 categories and ATC codes captured within nationwide health registers. 17 Likewise, the CLL Comorbidity Index (CLL‐CI) uses a combination of ICD10, ATC, and surgery codes to define three comorbidity categories with prognostic impact on overall survival (OS) and event‐free survival. 10 , 18
We recently demonstrated a prognostic value of data on prediagnostic antimicrobial and antidiabetic prescriptions in patients with LC, 11 , 12 and we thus propose that other specific drug classes and polypharmacy per se or as a proxy for multimorbidity may impact OS, hospitalization, and risk of severe infection in patients with LC. Whereas comorbidity is difficult to eliminate, critical revision of polypharmacy through geriatric assessment, 19 may offer an attractive approach to optimize treatment for patients with LC, while the potential benefits of some classes of medications are largely unstudied in LC.
METHODS
We retrieved nationwide data on all Danish patients registered with eight common LC (i.e., classical Hodgkin lymphoma [cHL], diffuse large B‐cell lymphoma [DLBCL], follicular lymphoma, marginal zone lymphoma [MZL], mantle cell lymphoma [MCL], CLL, lymphoplasmacytic lymphoma [LPL], and MM) from the DALY‐CARE data resource covering the period between January 2002 and March 31, 2023. 9 Information on medication prescribed within the year up until LC diagnosis (i.e., days −365 to 0) was retrieved, 16 and the number of distinct ATC codes (using the seventh ATC level) was used to define polypharmacy as at least five medications (<5 vs. ≥5 medications) considering a change from one drug to another (e.g., metformin to a combination drug such as metformin with sitagliptin or penicillin to azithromycin) as two distinct medications. 1 We stress that over‐the‐counter medication is not recorded but mainly includes paracetamol, ibuprofen, antihistamines, and vitamins. 16
Available covariates considered to be associated with polypharmacy and/or confound clinical outcomes included age at diagnosis, biological sex, international prognostic indices (IPIs), and comorbidity. Age and sex were available for all patients. To account for comorbidity, we calculated CCI scores from hospital‐acquired ICD‐10 diagnoses before the LC diagnosis. 8 , 9 , 20 For all patients, a hematological CCI score of 2 was added. NMI and CLL‐CI scores were calculated as described elsewhere. 17 , 18 To calculate guideline‐recommended subtype‐specific IPI scores such as the revised IPI (R‐IPI) and the revised international staging system (R‐ISS), 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 we only used patients from quality registers with information on stage. 29 , 30 , 31 Date of last follow‐up was November 15, 2023.
We investigated the prognostic impact of (1) individual prediagnostic drug classes using the third ATC level (e.g., J01 systemic antibacterial), (2) polypharmacy (5 vs. ≥5 medications), and (3) the number of prediagnostic medications (0–3 vs. 4–7 vs. 8–11 vs. >11) on (A) OS, (B) hospitalization, and (C) severe infection. We used Bonferroni to correct for multiple testing in analyses of 79 distinct drug classes on the three different outcomes. OS was calculated from the time of diagnosis until death or the end of follow‐up. LC‐related mortality was assessed using cause‐specific Cox regression for patients diagnosed before the date of extraction (i.e., December 31, 2022: 99.99% of patients) of the Danish cause of death register. 32 Deaths registered with an ICD‐10 LC diagnosis (i.e., C81.0–C91.9) were considered LC‐related, whereas other causes of death were censored at the time of death. Patients alive by December 31, 2022, were censored, and follow‐up time was truncated accordingly. Hospitalization was defined as in‐hospital visits for at least 24 h. Severe infection was defined as use of intravenous antimicrobials (IVAM) for at least 24 h, and each infection was separated by more than 7 days without IVAM. 33 To avoid immortality bias and because hospitalizations and IVAM were only recorded within the electronic health record (EHR) system, cause‐specific time to first event (i.e., hospital admission and severe infection) was calculated from time of EHR system deployment (May 2016, November 2016, and November 2017 at three different sites) or date of LC diagnosis, whichever came last, until date of first event or death or end of follow‐up (i.e., March 1, 2023, for EHR data), whichever came first. Time to next treatment was calculated from the date of first‐line to second‐line treatment in four LC subtypes requiring initial antineoplastic therapy, that is, cHL, DLBCL, MCL, and MM.
Multivariable Cox regression analysis (MVA) was equally adjusted for age, sex, subtype‐specific IPI, CCI, and polypharmacy. MVA was performed on all LC patients and stratified for each LC subtype. In stratified analyses, age and sex were only included for LC subtypes if unaccounted for in the subtype‐specific IPIs (e.g., IPS and R‐ISS). Pairwise Fisher's exact tests adjusted for multiple testing by false detection rate (FDR) investigated correlations between comorbidity and drug class (i.e., third ATC level) pairs.
Approvals
This study was approved by the Danish Health Data Authority and National Ethics Committee (approvals P‐2020‐561 and 1804410, respectively). According to Danish legislation, register studies do not require written informed consent.
Data sharing
Data may be shared on a collaborative basis within the DALY‐CARE data resource based on a data processing agreement. 9 Statistical analyses were performed in R software. 34 The base code is available at github (https://github.com/RH-CLL-LAB/PolyRX).
RESULTS
Patient characteristics
We identified 47,543 patients with an LC of whom 46,803 (98.4%) were diagnosed before extraction of prescription data (Cohort 1 with available prescription data), and 26,857 patients had complete subtype‐specific IPI scores (Cohort 2). These two cohorts were used for OS analyses. We next identified 12,668 patients with EHR data (Cohort 3) for whom 9798 (77.3%) had complete subtype‐specific IPI scores (Cohort 4). The EHR cohorts were used for analyses of hospitalization and severe infection (Figure 1). Patient characteristics and common comorbidities are provided in Tables 1 and S1, respectively. Compared with the frequency of hospital diagnoses used to calculate CCI scores, prescriptions were highly prevalent in the year leading up to LC diagnosis in Cohort 1: only 4081 (8.7%) patients had no prescription medication in the year leading up to LC diagnosis. The median number of distinct medications was 6 (interquartile range [IQR] 3; 10), whereas 15,369 (32.8%), 14,294 (30.5%), 9305 (19.9%), and 7835 (16.7%) patients had been prescribed 0–3, 4–7, 8–11, and >11 drug classes during the year before LC diagnosis, respectively. As a result, 27,489 (58.7%) patients had prediagnostic polypharmacy with ≥5 distinct medications. Patients with an increasing number of medications were also gradually older, with gradually higher female predominance, higher subtype‐specific IPI, and higher CCI scores (Table 2; P < 0.0001). Grouping medication according to the third ATC level (total of 79 distinct drug classes), patients had most frequently been prescribed antibacterial (J01; 44.6%), analgesic (N02; 42.7%), renin‐angiotensin system (C09; 30.3%), anti‐inflammatory/rheumatic (M01; 28.4%), antithrombotic (B01; 27.6%), dyspepsia related (A02; 25.9%), diuretic (C03; 25.8%), and lipid modifying (C10; 24.9%) medication. To investigate potential interaction and test whether ATC drug classes were associated with Charlson comorbidities subgroups, 8 we performed pairwise Fisher's exact tests. After adjustment for multiple testing (790 distinct third‐level ATC:Charlson comorbidity category pairs), there was surprisingly not a single combination correlating with a third ATC level and a Charlson comorbidity subgroup (FDR > 0.14; data not shown) indicating poor correlation between ICD10 codes and prescription medicine.
Figure 1.
CONSORT diagram of patients with lymphoid cancer (LC) included in the study. All patients with prescription (Rx) data were included in univariable analyses (UVAs) of overall survival (OS; Cohort 1), whereas patients with information on subtype‐specific international prognostic index (IPI) were included in multivariable analyses (MVAs) of OS (Cohort 2). Only patients with electronic health record (EHR) data (Cohort 3) were included in the UVA and MVA of time to hospitalization and severe infection. Abbreviations: cHL, classical Hodgkin lymphoma; CLL, chronic lymphocytic lymphoma; DLBCL, diffuse large B‐cell lymphoma; FL, follicular lymphoma; MCL, mantle cell lymphoma; MZL, marginal zone lymphoma; LPL, lymphoplasmacytic lymphoma; MM, multiple myeloma.
Table 1.
Baseline characteristics stratified by lymphoid cancer subtype.
| Variable | Unit | cHL (n = 3290) | DLBCL (n = 9436) | FL (n = 5195) | MZL (n = 2208) | MCL (n = 1435) | CLL (n = 11,065) | LPL (n = 3291) | MM (n = 10,883) | Total (46,803) |
|---|---|---|---|---|---|---|---|---|---|---|
| Age, years | Median [IQR] | 50.8 [31.4, 67.1] | 69.6 [59.4, 77.6] | 65.8 [56.9, 73.9] | 70.7 [61.5, 77.9] | 70.7 [62.6, 78.1] | 71.6 [63.6, 79.0] | 73 [66.1, 79.0] | 71.6 [63.4, 78.8] | 70 [60.5, 77.7] |
| Sex, n (%) | F | 1340 (40.7) | 4098 (43.4) | 2650 (51.0) | 1199 (54.3) | 449 (31.3) | 4470 (40.4) | 1317 (40.0) | 4787 (44.0) | 20,310 (43.4) |
| M | 1950 (59.3) | 5338 (56.6) | 2545 (49.0) | 1009 (45.7) | 986 (68.7) | 6595 (59.6) | 1974 (60.0) | 6096 (56.0) | 26,493 (56.6) | |
| Subtype‐specific IPI, n (%) | Low | 1055 (57.5) | 499 (7.1) | 575 (16.5) | 387 (23.9) | 186 (16.2) | 1884 (53.3) | 387 (27.7) | 1803 (26.6) | 6776 (25.2) |
| Intermediate | 0 (0.0) | 3256 (46.1) | 2209 (63.6) | 658 (40.6) | 372 (32.5) | 1003 (28.4) | 690 (49.3) | 4231 (62.4) | 12,419 (46.2) | |
| Higha | 780 (42.5) | 3307 (46.8) | 692 (19.9) | 575 (35.5) | 588 (51.3) | 648 (18.3) | 322 (23.0) | 750 (11.1) | 7662 (28.5) | |
| Missing | 1455 | 2374 | 1719 | 588 | 289 | 7530 | 1892 | 4099 | 19,946 | |
| No. of medications, n (%) | 0–3 | 1587 (48.2) | 2881 (30.5) | 2261 (43.5) | 685 (31.0) | 495 (34.5) | 3996 (36.1) | 995 (30.2) | 2469 (22.7) | 15,369 (32.8) |
| 4–7 | 890 (27.1) | 2916 (30.9) | 1505 (29.0) | 741 (33.6) | 450 (31.4) | 3455 (31.2) | 1034 (31.4) | 3303 (30.4) | 14,294 (30.5) | |
| 8–11 | 449 (13.6) | 2006 (21.3) | 835 (16.1) | 436 (19.7) | 287 (20.0) | 2013 (18.2) | 693 (21.1) | 2586 (23.8) | 9305 (19.9) | |
| >11 | 364 (11.1) | 1633 (17.3) | 594 (11.4) | 346 (15.7) | 203 (14.1) | 1601 (14.5) | 569 (17.3) | 2525 (23.2) | 7835 (16.7) | |
| CCI, n (%) | 0–2 | 2282 (69.4) | 5535 (58.7) | 3491 (67.2) | 1355 (61.4) | 941 (65.6) | 7513 (67.9) | 2016 (61.3) | 6341 (58.3) | 29,474 (63.0) |
| 2–4 | 578 (17.6) | 2562 (27.2) | 1049 (20.2) | 657 (29.8) | 323 (22.5) | 2765 (25.0) | 1002 (30.4) | 3213 (29.5) | 12,149 (26.0) | |
| 5–6 | 95 (2.9) | 446 (4.7) | 165 (3.2) | 98 (4.4) | 60 (4.2) | 486 (4.4) | 174 (5.3) | 742 (6.8) | 2266 (4.8) | |
| >6 | 335 (10.2) | 893 (9.5) | 490 (9.4) | 98 (4.4) | 111 (7.7) | 301 (2.7) | 99 (3.0) | 587 (5.4) | 2914 (6.2) | |
| CCI score | Median [IQR] | 2 [2, 2] | 2 [2, 3] | 2 [2, 2] | 2 [2, 2] | 2 [2, 2] | 2 [2, 2] | 2 [2, 3] | 2 [2, 3] | 2 [2, 2] |
| NMI score | Median [IQR] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] |
| CLL‐CI score | Median [IQR] | 0 [0, 1] | 1 [0, 1] | 0 [0, 1] | 1 [0, 1] | 1 [0, 1] | 1 [0, 1] | 1 [0, 1] | 1 [0, 1] | 1 [0, 1] |
Abbreviations: CCI, Charlson comorbidity index; cHL, classical Hodgkin lymphoma; CLL, chronic lymphocytic lymphoma; CLL‐CI, CLL comorbidity index; DLBCL, diffuse large B‐cell lymphoma; FL, follicular lymphoma; IPI, international prognostic index; IQR, interquartile range; LPL, lymphoplasmacytic lymphoma; MCL, mantle cell lymphoma; MM, multiple myeloma; MZL, marginal zone lymphoma; NMI, Nordic multimorbidity index.
High (n = 523) and very high risk (n = 125) CLL‐IPI were combined as high risk.
Table 2.
Baseline characteristics stratified by the number of medications.
| Variable | Unit | 0–3 (n = 15,369) | 4–7 (n = 14,294) | 8–11 (n = 9305) | >11 (n = 7835) | Total (N = 46,803) |
|---|---|---|---|---|---|---|
| Age, years | Median [IQR] | 64.1 [54.0, 72.6] | 69.7 [60.9, 77.2] | 73.3 [65.8, 80.1] | 75.6 [68.5, 81.7] | 70 [60.5, 77.7] |
| Sex, n (%) | F | 5804 (37.8) | 6254 (43.8) | 4288 (46.1) | 3964 (50.6) | 20,310 (43.4) |
| M | 9565 (62.2) | 8040 (56.2) | 5017 (53.9) | 3871 (49.4) | 26,493 (56.6) | |
| Subtype‐specific IPI, n (%) | Low | 2811 (33.4) | 2168 (25.5) | 1096 (19.8) | 701 (15.9) | 6776 (25.2) |
| Intermediate | 3633 (43.1) | 3943 (46.4) | 2640 (47.8) | 2203 (49.8) | 12,419 (46.2) | |
| High | 1980 (23.5) | 2380 (28.0) | 1786 (32.3) | 1516 (34.3) | 7662 (28.5) | |
| Missing | 6945 | 5803 | 3783 | 3415 | 19,946 | |
| CCI intervals, n (%) | 0–2 | 12,970 (84.4) | 9666 (67.6) | 4560 (49.0) | 2278 (29.1) | 29,474 (63.0) |
| 2–4 | 1513 (9.8) | 3582 (25.1) | 3511 (37.7) | 3543 (45.2) | 12,149 (26.0) | |
| 5–6 | 77 (0.5) | 348 (2.4) | 659 (7.1) | 1182 (15.1) | 2266 (4.8) | |
| >6 | 809 (5.3) | 698 (4.9) | 575 (6.2) | 832 (10.6) | 2914 (6.2) | |
| CCI score | Median [IQR] | 2 [2, 2] | 2 [2, 2] | 2 [2, 3] | 3 [2, 3] | 2 [2, 2] |
| NMI score | Median [IQR] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] | 0 [0, 1] |
| CLL‐CI score | Median [IQR] | 0 [0, 0] | 1 [0, 1] | 1 [0, 2] | 2 [1, 2] | 1 [0, 1] |
Abbreviations: CCI, Charlson comorbidity index; CLL‐CI, CLL comorbidity index; IPI, international prognostic index; IQR, interquartile range; NMI, Nordic multimorbidity index.
Overall survival
Patients in Cohort 1 were followed for a median of 9.3 years (IQR 5.4; 14.4). In Bonferroni corrected univariate analyses, 43 of 79 ATC drug classes were associated with shorter OS, whereas six were associated with longer OS (Table S2). Of note, “other gynecologicals” (G02) containing mainly contraceptives were associated with the largest decrease in mortality (hazard ratio [HR] 0.16, 95% confidence interval [CI] 0.09; 0.30), and most drug classes associated with longer OS (D10, G01, G02, G03, and P02) indicated clear confounding by age and sex. We thus adjusted for age, sex, and multiple testing to demonstrate that 39 of 49 were independently associated OS (Table 3); anti‐acne drugs (D10), gynecologicals (G01), other gynecologicals (G02), and antihelmintics (P02) were no longer independent, whereas sex hormones (G03) retained independence (90.4% estradiol; median age 68.9 [IQR 58.7; 77.0] years). Although prescribed vaccines (J07; excluding those in the Danish childhood vaccination program as well as against seasonal influenza and COVID) within the year up to a LC diagnosis were not associated with longer OS, a post hoc analysis including vaccines since 2002 (J07; mostly against viral hepatitis [44.4%], Pneumococci [27.8%], human papillomavirus [9.3%], and typhus fevers [8.5%]) demonstrated longer OS in vaccinated patients (HR 0.69, 95% CI 0.64; 0.74). Among these 2542 (5.4%) vaccinated patients, the latest vaccine was given at a median of 3.9 years (IQR 8.2; 1.6) before LC diagnosis.
Table 3.
Associations of drug classes and overall survival (OS), hospitalization, and severe infection adjusted for age, sex, and multiple testing.
| ATC | Drug class name | n | OS | Hospitalization | Severe infection | |||
|---|---|---|---|---|---|---|---|---|
| HR | P‐value | HR | P‐value | HR | P‐value | |||
| L03 | Immunostimulants | <5 | 28.75 [4.05; 204.22] | 0.001 | NA | NA | NA | NA |
| B05 | Blood substitutes and perfusion solutions | 9 | 12.02 [6.25; 23.12] | <0.001 | NA | NA | NA | NA |
| A09 | Digestives, incl. enzymes | 72 | 1.98 [1.52; 2.59] | <0.001 | NA | NA | 3.73 [2.00; 6.93] | <0.001 |
| B02 | Antihemorrhagics | 119 | 1.90 [1.52; 2.37] | <0.001 | NA | NA | NA | NA |
| A04 | Antiemetics and antinauseants | 226 | 1.66 [1.38; 1.99] | <0.001 | 2.04 [1.61; 2.58] | <0.001 | 2.20 [1.69; 2.87] | <0.001 |
| A12 | Mineral supplements | 5371 | 1.65 [1.60; 1.71] | <0.001 | 1.38 [1.28; 1.49] | <0.001 | 1.66 [1.52; 1.81] | <0.001 |
| A06 | Drugs for constipation | 3133 | 1.63 [1.56; 1.70] | <0.001 | 1.49 [1.36; 1.63] | <0.001 | 1.72 [1.56; 1.90] | 0.001 |
| A03 | Drugs for functional gastrointestinal disorders | 2058 | 1.56 [1.48; 1.65] | <0.001 | 1.31 [1.17; 1.47] | <0.001 | 1.33 [1.17; 1.52] | <0.001 |
| N07 | Other nervous system drugs | 752 | 1.48 [1.34; 1.62] | <0.001 | NA | NA | 1.39 [1.15; 1.68] | 0.001 |
| A11 | Vitamins | 192 | 1.44 [1.22; 1.71] | <0.001 | NA | NA | NA | NA |
| N03 | Antiepileptics | 931 | 1.43 [1.31; 1.55] | <0.001 | 1.38 [1.18; 1.62] | <0.001 | 1.39 [1.16; 1.68] | <0.001 |
| C03 | Diuretics | 12,089 | 1.43 [1.39; 1.47] | <0.001 | 1.17 [1.11; 1.23] | <0.001 | 1.28 [1.20; 1.36] | <0.001 |
| N02 | Analgesics | 19,978 | 1.40 [1.36; 1.44] | <0.001 | 1.48 [1.42; 1.55] | <0.001 | 1.59 [1.51; 1.68] | <0.001 |
| M04 | Antigout preparations | 1662 | 1.40 [1.32; 1.49] | <0.001 | 1.24 [1.11; 1.39] | <0.001 | 1.38 [1.22; 1.56] | <0.001 |
| N06 | Psychoanaleptics | 5796 | 1.39 [1.34; 1.44] | <0.001 | 1.25 [1.17; 1.34] | <0.001 | 1.25 [1.16; 1.36] | <0.001 |
| C01 | Cardiac therapy | 3454 | 1.37 [1.31; 1.42] | <0.001 | 1.25 [1.14; 1.36] | <0.001 | 1.25 [1.13; 1.39] | <0.001 |
| M03 | Muscle relaxants | 1327 | 1.36 [1.26; 1.47] | <0.001 | 1.75 [1.59; 1.92] | <0.001 | 1.98 [1.79; 2.19] | <0.001 |
| N04 | Anti‐Parkinson drugs | 780 | 1.36 [1.24; 1.48] | <0.001 | 1.36 [1.15; 1.61] | <0.001 | NA | NA |
| H02 | Corticosteroids for systemic use | 4216 | 1.35 [1.29; 1.40] | <0.001 | 1.23 [1.14; 1.34] | <0.001 | 1.34 [1.22; 1.46] | <0.001 |
| B03 | Antianemic preparations | 4120 | 1.35 [1.30; 1.41] | <0.001 | 1.23 [1.14; 1.33] | <0.001 | 1.44 [1.32; 1.58] | <0.001 |
| A07 | Antidiarrheals, intestinal anti‐inflammatory agents | 1455 | 1.34 [1.26; 1.43] | <0.001 | 1.27 [1.12; 1.45] | <0.001 | 1.41 [1.22; 1.63] | 0.001 |
| R03 | Drugs for obstructive airway diseases | 6081 | 1.31 [1.27; 1.36] | <0.001 | 1.25 [1.18; 1.34] | <0.001 | 1.35 [1.26; 1.45] | <0.001 |
| N05 | Psycholeptics | 8659 | 1.31 [1.27; 1.35] | <0.001 | 1.12 [1.05; 1.18] | <0.001 | 1.12 [1.05; 1.20] | 0.001 |
| J02 | Antimycotics for systemic use | 1153 | 1.29 [1.19; 1.40] | <0.001 | NA | NA | NA | NA |
| M02 | Topical products for joint and muscular pain | 281 | 1.29 [1.11; 1.48] | 0.001 | NA | NA | NA | NA |
| A10 | Drugs used in diabetes | 4380 | 1.28 [1.23; 1.33] | <0.001 | 1.23 [1.15; 1.32] | <0.001 | 1.35 [1.24; 1.46] | <0.001 |
| R05 | Cough and cold preparations | 4382 | 1.26 [1.21; 1.31] | <0.001 | NA | NA | NA | NA |
| B01 | Antithrombotic agents | 12,919 | 1.24 [1.20; 1.27] | <0.001 | 1.25 [1.19; 1.32] | <0.001 | 1.28 [1.20; 1.35] | <0.001 |
| L04 | Immunosuppressants | 871 | 1.22 [1.12; 1.34] | <0.001 | NA | NA | NA | NA |
| A02 | Drugs for acid‐related disorders | 12,121 | 1.22 [1.19; 1.26] | <0.001 | 1.25 [1.19; 1.31] | <0.001 | 1.28 [1.21; 1.35] | <0.001 |
| M05 | Drugs for treatment of bone diseases | 2718 | 1.22 [1.16; 1.28] | <0.001 | NA | NA | NA | NA |
| C07 | Beta blocking agents | 8602 | 1.21 [1.18; 1.25] | <0.001 | 1.22 [1.15; 1.29] | <0.001 | 1.27 [1.19; 1.35] | <0.001 |
| J01 | Antibacterials for systemic use | 20,873 | 1.19 [1.16; 1.22] | <0.001 | 1.17 [1.12; 1.22] | <0.001 | 1.13 [1.07; 1.19] | <0.001 |
| A01 | Stomatological preparations | 1145 | 1.19 [1.10; 1.28] | <0.001 | NA | NA | NA | NA |
| M01 | Anti‐inflammatory and antirheumatic products | 13,277 | 1.17 [1.14; 1.20] | <0.001 | 1.20 [1.15; 1.26] | <0.001 | 1.19 [1.12; 1.26] | <0.001 |
| H03 | Thyroid therapy | 2440 | 1.10 [1.04; 1.16] | 0.001 | NA | NA | NA | NA |
| C08 | Calcium channel blockers | 8847 | 1.08 [1.05; 1.12] | <0.001 | 1.18 [1.11; 1.24] | <0.001 | 1.25 [1.17; 1.33] | <0.001 |
| R01 | Nasal preparations | 3155 | 0.86 [0.81; 0.90] | <0.001 | NA | NA | NA | NA |
| G03 | Sex hormones and modulators of the genital system | 4296 | 0.83 [0.79; 0.88] | <0.001 | NA | NA | NA | NA |
| J07 | Vaccines | NA | NA | NA | 1.41 [1.19; 1.68] | <0.001 | NA | NA |
| R06 | Antihistamines for systemic use | NA | NA | NA | 1.17 [1.08; 1.27] | <0.001 | NA | NA |
| G04 | Urologicals | NA | NA | NA | 1.15 [1.07; 1.23] | <0.001 | NA | NA |
| C09 | Agents acting on the renin‐angiotensin system | NA | NA | NA | 1.12 [1.07; 1.18] | <0.001 | 1.15 [1.09; 1.22] | <0.001 |
| C10 | Lipid‐modifying agents | NA | NA | NA | 1.11 [1.05; 1.16] | <0.001 | 1.11 [1.05; 1.18] | <0.001 |
Note: Each multivariate analysis (adjusted for age and sex) compares patients who have been prescribed an ATC drug class in the year leading up to LC diagnoses as compared to those without the specified ATC drug class.
Abbreviations: ATC, anatomical therapeutic chemical; HR, hazard ratio; LC, lymphoid cancer.
We next investigated polypharmacy (<5 vs. ≥5 drugs) and the number of distinct medications (0–3 vs. 4–7 vs. 8–11 vs. >11 drugs). While polypharmacy was associated with a poor OS (HR 2.0, 95% CI 2.0; 2.1; Figure S1A), we demonstrated a negative dose–response effect on OS for patients with an increasing number of prescription drugs in univariate analyses, which could be seen in all LC subgroups (Figures 2A and S2). In MVA adjusted for age, sex, subtype‐specific IPI, and CCI score, polypharmacy was an independent prognostic marker of OS in all LC subtypes with a HR of 1.4 (95% CI 1.3; 1.4) for the entire cohort (Figure S3A). Even so, the number of medications (0–3 vs. 4–7 vs. 8–11 vs. >11 drugs) demonstrated a clear negative dose–response effect on OS independent of age, sex, subtype‐specific IPI, and CCI (Figure 3A), which could be demonstrated in all LC subtypes (Figure S4). To investigate LC‐related mortality, we repeated these MVAs using cause‐specific Cox regression for the 26,856 patients (99.99% of Cohort 2) diagnosed before the extraction of cause of death data. 32 The number of medications continued to independently impact LC‐related mortality with a negative dose–response effect on OS for all patients and most LC subtypes, except for MZL and LPL (Figure S5).
Figure 2.
Univariate analyses of (A) overall survival, (B) time to first admission, and (C) time to first infection stratified by the number of medications in the year before lymphoid cancer (LC) diagnosis. Time to first admission and infection analyses in (B) and (C) were performed using cause‐specific Kaplan–Meier statistics (i.e., censoring upon death and end follow‐up).
Figure 3.
Multivariable Cox regression analyses on (A) overall survival, (B) time to first admission, and (C) time to first severe infection equally adjusted for age, sex, subtype‐specific international prognostic index (IPI), and Charlson comorbidity index (CCI) score.
Risk of hospitalization
To assess the impact of polypharmacy on health care utilizations, we investigated hospitalizations among patients in eastern Denmark for whom EHR data were available in 12,668 patients (Cohort 3). Following patients from LC diagnosis for a median of 6.2 (IQR 4.1–7.2) years within EHR systems, we recorded 36,788 hospital admissions in 8330 (65.8%) patients after their LC diagnosis, which accounted for a median 14 in‐hospital days (IQR 6; 30) among admitted patients. After Bonferroni correction, time to first admission was significantly shorter for 33 of 79 specific ATC drug classes (Table S2). Among these 33 drug classes, 29 were independently associated with hospitalization after adjusting for age, sex, and multiple testing (Table 3). We next demonstrated a shorter time to first hospitalization for patients with an increasing number of medications among the 4410 (34.8%), 4052 (32.0%), 2420 (19.1%), and 1786 (14.1%) patients with 0–3, 4–7, 8–11, and >11 medications in the year before LC diagnosis, respectively (Figure 2B; P < 0.0001). In MVA adjusted for age, sex, IPI, and CCI score, the number of medications maintained its independence with a dose–response effect on time to first hospitalization (Figure 3B; P < 0.0001). These associations were also found within the eight LC subtypes, although with less clear separation and dose–response effect among patients with MCL, LPL, and MM (Figures S6 and S7). We further demonstrated a gradually increasing median number of admissions per patient‐year (0.2 [IQR 0.0; 1.1], 0.6 [IQR 0.0; 1.6], 0.9 [IQR 0.0; 2.2], and 1.3 [IQR 0.3; 3.0], respectively) and an increasing number of days in‐hospital per patient (11 [IQR 5; 25], 13 [IQR 6; 28], 16 [IQR 7; 32], and 19 [IQR 8; 38] days, respectively) in patients with 0–3, 4–7, 8–11, and >11 medications (Figure 4A,B; P < 0.0001).
Figure 4.
Health care utilizations among patients followed in eastern Denmark. (A) The median number of hospital admissions per patient‐year, (B) median days in‐hospital per patient, (C) the median number of infections per patient‐year, and (D) the median number of days on antimicrobials per patient among patients with infections gradually increased with the number of medications in the year before lymphoid cancer diagnosis (P < 0.0001).
Risk of severe infections
Next, we analyzed the time to severe infection. For specific drug classes (i.e., using the third ATC level), 32 of 79 drug classes were significantly associated with a shorter time to first infection after Bonferroni correction (Table S2). Among these, 27 were independently associated with severe infection after adjusting for age, sex, and multiple testing (Table 3).
From the time of LC diagnosis, we also demonstrated a gradually increased risk of severe infection with an incremental dose–response increase by the number of medications (Figure 2C; P < 0.0001). In MVA adjusting for age, sex, IPI, and CCI score, the number of medications remained an independent prognostic marker of time to first severe infection (Figure 3C; P < 0.0001). These associations were seen in all eight LC subtypes, although with less clear separation and dose–response effect among patients with MCL, LPL, and MM (Figures S8 and S9). The infection burden summarized as the mean number of infections per patient‐year (0.2 [standard deviation (sd) 1.7], 0.4 [sd 2.5], 0.6 [sd 3.0], and 1.0 [sd 4.3]) and the median number of days on IVAM treatment per patient (9 [IQR 4; 19], 9 [IQR 4; 20], 10 [IQR 4; 22], and 11 [IQR 5; 24] days) were gradually increasing for patients with 0–3, 4–7, 8–11, and >11 medications, respectively (Figure 4C,D; P < 0.0001). To test the impact of the COVID pandemic, we performed sensitivity analyses specifically excluding remdesivir as a proxy for infection and showed similar results (Figure S10).
Sensitivity analyses of medication and comorbidity
To investigate the impact of drugs prescribed during the diagnostic workup, we repeated the MVAs excluding medication prescribed during the month leading up to the LC diagnosis. In this sensitivity analysis, the number of medications (between days −365 and −30) remained an independent prognostic marker with a dose–response effect on all clinical outcomes (Figure S11; P < 0.001). As the NMI and CLL‐CI also incorporate medication as a proxy for comorbidity, 17 , 18 we next replaced CCI with both NMI and CLL‐CI scores in two separate MVAs. Here, we likewise demonstrated a similar independent dose–response effect for the number of medications on all clinical outcomes (Figures S12 and S13).
Time to next treatment in aggressive lymphoid cancer
Finally, we analyzed time to next treatment stratified on the number of medications in the year before diagnosis in patients with cHL, DLBCL, MCL, and MM, who require antineoplastic therapy from the time of diagnosis. We demonstrated a gradually shorter time to next treatment with increasing number of medications for patients with cHL, MCL, and MM (Figure 5A, C, and D; P ≤ 0.004), but not with DLBCL (Figure 5B; P = 0.14).
Figure 5.
Time to next‐line treatment from end of first‐line treatment (EoFLT) stratified on number of medications in patients with (A) classical Hodgkin lymphoma (cHL), (B) diffuse large B‐cell lymphoma (DLBCL), (C) mantle cell lymphoma (MCL), and (D) multiple myeloma (MM).
DISCUSSION
In this study, we utilized the DALY‐CARE data resource to analyze a wide range of electronic health data in a large cohort of patients with eight different LCs. Specific drug classes prescribed in the year before diagnosis were associated with OS, time to severe infection, and time to hospitalization. Polypharmacy (≥5) was correlated with unfavorable clinical outcomes, and we could further demonstrate that an increasing number of prediagnostic medications was associated with gradually shorter OS, time to first hospitalization, and time to severe infection, while also predictive of duration of response in cHL, MCL, and MM. Importantly, the number of medications retained an independent dose–response effect on all three clinical outcomes after adjusting for age, sex, subtype‐specific IPI, and three different comorbidity scores in the entire cohort and in subgroup analyses of LCs.
As anticipated, we show that many drug classes are associated with poor clinical outcomes, which likely reflect comorbidity, older age, and possibly more aggressive disease. For instance, immunostimulants (L03: filgrastim) and blood substitutes (B05: mostly electrolytes and solutions for parenteral nutrition) had a marked impact on OS likely representing patients who had started supportive therapy before securing an LC diagnosis, and those with very poor health conditions. On the other hand, gynecologicals (G02: mostly contraceptives), sex hormones (G03), and gynecological antiinfectives (G01) were initially associated with favorable OS, likely representing younger female patients. Interestingly, after adjusting for age and sex, sex hormones (G03) retained independent positive prognostic impact OS eluding to further studies of menopausal women with LC. 35 Taken together, these results indicate that patients being prescribed certain medication most likely represent patient selection as the key factor of outcome rather than a causal effect of the medication itself, although we underscore that this epidemiological study does not report any causal relationships. In agreement with previous work, systemic antibacterials (J01) and antimycotics (J01) were associated with unfavorable OS, 12 whereas vaccines (J07) were associated with hospitalization, which may likely represent immunodeficient patients. Notably, vaccines (J07) prescribed within a longer prediagnostic period (covering up to 19.9 years) were associated with longer OS even though most vaccines were mostly against Hepatitis. Others have found a decreased risk of lymphoma in individuals vaccinated against Influenza, yellow fever, Hepatitis A and B, 36 whereas previous vaccination is not associated with an increased risk of monoclonal B‐cell lymphocytosis in another study. 37 Thus, further research into the general effects of vaccination as a modifiable risk factor for LC subtypes as well as for LC in general is warranted. 38
Interestingly, cardiac drugs and antihypertensives (including C01, C03, C07, C08, and C09) seemed to dominate the drug classes associated with severe infection. At the same time, infections have been associated with cardiovascular disease, 39 , 40 , 41 and we speculate whether these patients are more likely to contract infections and experience neutropenia. Another plausible hypothesis is that hypotension in patients with fever might affect the decision to initiate IVAMs. Thus, the causal relationship remains to be investigated. For instance, we wonder whether patients with hypertension without cardiac or chronic kidney disease might potentially benefit from pausing their antihypertensive treatment during antineoplastic therapy. Likewise, the large group of patients without registered malabsorption or alcohol abuse disorders treated with minerals (A11, A12), patients without cerebrovascular disease or congestive heart failure treated with lipid modifying drugs (C10), and those without anxiety or cerebral palsy treated with muscle relaxants (M03), might benefit from cessation of inappropriate medication. 19 Trials of geriatric assessment before starting antineoplastic therapy are ongoing and show that changes to routine medication are common. Lastly, nervous system‐directed drugs such as analgesics (N01), antiepileptics (N03), anti‐Parkinson drugs (N04), psycholeptics (N05), and psychoanaleptics (N07) associated with poor clinical outcomes in this study likely represent underlying comorbidity in agreement with previous studies. 13 , 42
We here demonstrate that polypharmacy is a significant prognostic marker of clinical outcomes independent of age, sex, subtype‐specific IPI, and CCI score. On top of this, we also provide validation for a range of subtype‐specific IPIs in a large real‐world cohort. Importantly, we could demonstrate an independent dose–response effect for the number of mediations that divided patients into clinically meaningful proportional groups. Thus, the number of medications—and not polypharmacy per se—predicted OS, hospitalization, and severe infection to the same extent as guideline‐recommended IPIs. Although polypharmacy (and the number of medications) is most likely a result of an underlying comorbidity informing the prognosis further than register‐based CCI scores, we were unable to correlate a single drug class with a Charlson‐defined comorbidity. As such, robust definitions and proxies for specific medical conditions from laboratory and medication data are warranted. 10 , 11 , 14 , 17 This also emphasizes the granularity of data that is attained from prescribed medication. Importantly, the number of medications retained its independent prognostic value when adjusting for other comorbidity scores defined by prescription medication. 17 , 18 Not only prognostic, but also predictive of first‐line treatment, we showed a gradual shorter time to next treatment in cHL, MCL, and MM. Others have found higher relapse rates in IPI high‐risk lymphoma, 43 and in this study, high‐risk IPI was also correlated with an increasing number of medications, which may in part explain this finding. Of particular interest, further studies investigating the number of medications and antineoplastic dose‐intensity therapy are warranted. Finally, our results indicate that consistent collection of data on both comorbidity and polypharmacy on top of prognostic index variables at baseline may improve prognostication and treatment prediction; however, such data are largely missing in clinical trials of LC. These easily accessible and non‐costly data should be collected and considered at baseline and are of particular importance when extrapolating clinical trial results to clinical‐trial‐ineligible populations. 44 Based on these results, a 30‐day medical history has already been implemented as a baseline variable in ongoing clinical trials of CLL within the HOVON and German CLL study group.
In conclusion, a complete 1‐year medication history, summarized as the number of medications before LC diagnosis, significantly and independently impacts OS, hospitalization, and risk of severe infections, as well as duration of response for cHL and MCL. We thus recommend that information on prescription medication in the year before diagnosis and treatment initiation is included for clinical trials as essential baseline information, and taken into account with assessment of frailty and comorbidity when interpreting and extrapolating clinical trial results for clinical decisions.
AUTHOR CONTRIBUTIONS
Christian Brieghel: Conceptualization; investigation; methodology; visualization; formal analysis; writing—original draft; writing—review and editing. Thomas Lacoppidan: Software; methodology; data curation; writing—review and editing. Esben Packness: Methodology; software; data curation; writing—review and editing. Casper Møller Frederiksen: Data curation; resources; project administration. Mikkel Werling: Data curation; software. Michael Asger Andersen: Conceptualization; methodology; writing—review and editing. Christian Bjørn Poulsen: Writing—review and editing. Emelie Curovic Rotbain: Methodology; writing—review and editing. Carsten Utoft Niemann: Conceptualization; writing—review and editing; resources; supervision; funding acquisition.
CONFLICT OF INTEREST STATEMENT
C.B. received travel grants from AbbVie and Octapharma outside this study. T.L. received travel grants from AbbVie outside this study. C.M.F. received funding from Octapharma. E.C.R. received consultancy fees and/or travel grants from AbbVie, Janssen, and AstraZeneca outside of this work. C.U.N. received research funding and/or consultancy fees from AstraZeneca, Janssen, AbbVie, BeiGene, Genmab, CSL Behring, Octapharma, Takeda, Eli Lilly, MSD, and Novo Nordisk Foundation. All other authors declare no competing interests to disclose. All other authors declare no conflicts of interest.
FUNDING
This study was funded by the Alfred Benzon Foundation, the Danish Cancer Society (grant R269‐A15924), and the CLL‐CLUE project funded by the European Union.
Supporting information
Supporting Information.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
REFERENCES
- 1. Masnoon N, Shakib S, Kalisch‐Ellett L, Caughey GE. What is polypharmacy? A systematic review of definitions. BMC Geriatr. 2017;17(1):230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Eichhorst B, Ghia P, Niemann CU, et al. ESMO Clinical Practice Guideline interim update on new targeted therapies in the first line and at relapse of chronic lymphocytic leukaemia. Ann Oncol. 2024;35(9):762‐768. [DOI] [PubMed] [Google Scholar]
- 3. Dimopoulos MA, Moreau P, Terpos E, et al. Multiple myeloma: EHA‐ESMO Clinical Practice Guidelines for diagnosis, treatment and follow‐up. Ann Oncol. 2021;32(3):309‐322. [DOI] [PubMed] [Google Scholar]
- 4. Tilly H, Gomes da Silva M, Vitolo U, et al. Diffuse large B‐cell lymphoma (DLBCL): ESMO Clinical Practice Guidelines for diagnosis, treatment and follow‐up. Ann Oncol. 2015;26:v116‐v125. [DOI] [PubMed] [Google Scholar]
- 5. Lyon AR, López‐Fernández T, Couch LS, et al. 2022 ESC Guidelines on cardio‐oncology developed in collaboration with the European Hematology Association (EHA), the European Society for Therapeutic Radiology and Oncology (ESTRO) and the International Cardio‐Oncology Society (IC‐OS). Eur Heart J. 2022;43(41):4229‐4361. [DOI] [PubMed] [Google Scholar]
- 6. Kim DH, Rockwood K. Frailty in older adults. N Engl J Med. 2024;391(6):538‐548. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Cook G, Larocca A, Facon T, Zweegman S, Engelhardt M. Defining the vulnerable patient with myeloma—a frailty position paper of the European Myeloma Network. Leukemia. 2020;34(9):2285‐2294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Quan H, Li B, Couris CM, et al. Updating and validating the Charlson comorbidity index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011;173(6):676‐682. [DOI] [PubMed] [Google Scholar]
- 9. Brieghel C, Werling M, Frederiksen C, et al. The Danish Lymphoid Cancer Research (DALY‐CARE) data resource: the basis for developing data‐driven hematology. Clin Epidemiol. 2025;17:131‐145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Rotbain EC, Gordon MJ, Vainer N, et al. The CLL comorbidity index in a population‐based cohort: a tool for clinical care and research. Blood Adv. 2022;6(8):2701‐2706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Rotbain EC, Allmer C, Rostgaard K, et al. Impact of type 2 diabetes on mortality, cause of death, and treatment in chronic lymphocytic leukemia. Am J Hematol. 2023;98(8):1236‐1245. [DOI] [PubMed] [Google Scholar]
- 12. Packness E, Davidsson OB, Rostgaard K, et al. Infections and their prognostic significance before diagnosis of chronic lymphocytic leukemia, non‐Hodgkin lymphoma, or multiple myeloma. Br J Cancer. 2024;131:1186‐1194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Rotbain EC, Niemann CU, Rostgaard K, da Cunha‐Bang C, Hjalgrim H, Frederiksen H. Mapping comorbidity in chronic lymphocytic leukemia: impact of individual comorbidities on treatment, mortality, and causes of death. Leukemia. 2021;35(9):2570‐2580. [DOI] [PubMed] [Google Scholar]
- 14. Sawhney S, Ball W, Bell S, et al. Recovery of kidney function after acute kidney disease—a multi‐cohort analysis. Nephrol Dial Transplant. 2024;39(3):426‐435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Li Y, Zhang X, Yang L, et al. Association between polypharmacy and mortality in the older adults: a systematic review and meta‐analysis. Arch Gerontol Geriat. 2022;100:104630. [DOI] [PubMed] [Google Scholar]
- 16. Johannesdottir SA, Horváth‐Puhó E, Ehrenstein V, Schmidt M, Pedersen L, Sørensen H. Existing data sources for clinical epidemiology: the Danish National Database of Reimbursed Prescriptions. Clin Epidemiol. 2012;4:303‐313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kristensen KB, Lund LC, Jensen PB, et al. Development and validation of a Nordic multimorbidity index based on hospital diagnoses and filled prescriptions. Clin Epidemiol. 2022;14:567‐579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Gordon MJ, Kaempf A, Sitlinger A, et al. The chronic lymphocytic leukemia comorbidity index (CLL‐CI): a three‐factor comorbidity model. Clin Cancer Res. 2021;27(17):4814‐4824. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Soo WK, King MT, Pope A, Parente P, Dārziņš P, Davis ID. Integrated Geriatric Assessment and Treatment Effectiveness (INTEGERATE) in older people with cancer starting systemic anticancer treatment in Australia: a multicentre, open‐label, randomised controlled trial. Lancet Health Longev. 2022;3(9):e617‐e627. [DOI] [PubMed] [Google Scholar]
- 20. Schmidt M, Schmidt SAJ, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol. 2015;7:449‐490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Bahlo J, Kutsch N, Bauer K, et al. An international prognostic index for patients with chronic lymphocytic leukaemia (CLL‐IPI): a meta‐analysis of individual patient data. Lancet Oncol. 2016;17(6):779‐790. [DOI] [PubMed] [Google Scholar]
- 22. Hasenclever D, Diehl V, Armitage JO, et al. A prognostic score for advanced Hodgkin's disease. N Engl J Med. 1998;339(21):1506‐1514. [DOI] [PubMed] [Google Scholar]
- 23. Sehn LH, Berry B, Chhanabhai M, et al. The revised International Prognostic Index (R‐IPI) is a better predictor of outcome than the standard IPI for patients with diffuse large B‐cell lymphoma treated with R‐CHOP. Blood. 2007;109(5):1857‐1861. [DOI] [PubMed] [Google Scholar]
- 24. Federico M, Bellei M, Marcheselli L, et al. Follicular lymphoma international prognostic index 2: a new prognostic index for follicular lymphoma developed by the international follicular lymphoma prognostic factor project. J Clin Oncol. 2009;27(27):4555‐4562. [DOI] [PubMed] [Google Scholar]
- 25. Hoster E, Dreyling M, Klapper W, et al. A new prognostic index (MIPI) for patients with advanced‐stage mantle cell lymphoma. Blood. 2008;111(2):558‐565. [DOI] [PubMed] [Google Scholar]
- 26. Kastritis E, Morel P, Duhamel A, et al. A revised international prognostic score system for Waldenström's macroglobulinemia. Leukemia. 2019;33(11):2654‐2661. [DOI] [PubMed] [Google Scholar]
- 27. Thieblemont C, Cascione L, Conconi A, et al. A MALT lymphoma prognostic index. Blood. 2017;130(12):1409‐1417. [DOI] [PubMed] [Google Scholar]
- 28. Palumbo A, Avet‐Loiseau H, Oliva S, et al. Revised international staging system for multiple myeloma: a report from International Myeloma Working Group. J Clin Oncol. 2015;33(26):2863‐2869. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. da Cunha‐Bang C, Geisler C, Enggaard L, et al. The Danish National Chronic Lymphocytic Leukemia Registry. Clin Epidemiol. 2016;8(1):561‐565. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Arboe B, Josefsson P, Jørgensen J, et al. Danish National Lymphoma Registry. Clin Epidemiol. 2016;8:577‐581. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Gimsing P, Holmstrøm M, Wirenfelt Klausen T, et al. The Danish National Multiple Myeloma Registry. Clin Epidemiol. 2016;8:583‐587. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Helweg‐Larsen K. The Danish register of causes of death. Scand J Public Health. 2011;39(7 suppl):26‐29. [DOI] [PubMed] [Google Scholar]
- 33. Lacoppidan T, Brieghel C, Frederiksen CM, Niemann C. Infection burden upon treatment of patients with lymphoid‐lineage cancers. In: 2024 Madrid. EHA Annual Meeting, 2024:P1581. https://library.ehaweb.org/eha/2024/eha2024-congress/419668/thomas.lacoppidan.infection.burden.upon.treatment.of.patients.with.html?f=listing%3D0%2Abrowseby%3D8%2Asortby%3D1%2Asearch%3Dinfection+burden [Google Scholar]
- 34. R Core Team . R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; 2022. https://www.R-project.org/ [Google Scholar]
- 35. Menopause research is globally underfunded. It's time to change that. Nature. 2025;637(8047):763. [DOI] [PubMed] [Google Scholar]
- 36. Kleinstern G, Larson MC, Ansell SM, et al. Vaccination history and risk of lymphoma and its major subtypes. Cancer Epidemiol Biomarkers Prevent. 2022;31(2):461‐470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Boddicker NJ, Achenbach SJ, Parikh SA, et al. Associations of history of vaccination and hospitalization due to infection with risk of monoclonal B‐cell lymphocytosis. Leukemia. 2022;36(5):1404‐1407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Niemann CU, Wiestner A. B‐cell receptor signaling as a driver of lymphoma development and evolution. Sem Cancer Biol. 2013;23(6):410‐421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Kwong JC, Schwartz KL, Campitelli MA, et al. Acute myocardial infarction after laboratory‐confirmed influenza infection. N Engl J Med. 2018;378(4):345‐353. [DOI] [PubMed] [Google Scholar]
- 40. Sipilä PN, Lindbohm JV, Batty GD, et al. Severe infection and risk of cardiovascular disease: a multicohort study. Circulation. 2023;147(21):1582‐1593. [DOI] [PubMed] [Google Scholar]
- 41. Parviz M, Agius R, Rotbain EC, Vainer N, Aarup K, Niemann CU. Identifying CLL patients at high risk of atrial fibrillation on treatment using machine learning. Leuk Lymphoma. 2024;65(4):449‐459. [DOI] [PubMed] [Google Scholar]
- 42. Kisely S, Crowe E, Lawrence D. Cancer‐related mortality in people with mental illness. JAMA Psychiatry. 2013;70(2):209‐217. [DOI] [PubMed] [Google Scholar]
- 43. Ekberg S, Crowther M, Harrysson S, Jerkeman M, Eloranta S. Patient trajectories after diagnosis of diffuse large B‐cell lymphoma—a multistate modelling approach to estimate the chance of lasting remission. Br J Cancer. 2022;127(9):1642‐1649. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Klausen TW, Gregersen H, Abildgaard N, et al. The majority of newly diagnosed myeloma patients do not fulfill the inclusion criteria in clinical phase III trials. Leukemia. 2019;33(2):546‐549. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.