The Timing, the Treatment, the Question: Comparison of Epidemiologic Approaches to Minimize Immortal Time Bias in Real-World Data Using a Surgical Oncology Example.
Journal
Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology
ISSN: 1538-7755
Titre abrégé: Cancer Epidemiol Biomarkers Prev
Pays: United States
ID NLM: 9200608
Informations de publication
Date de publication:
02 11 2022
02 11 2022
Historique:
received:
02
05
2022
revised:
01
07
2022
accepted:
17
08
2022
pubmed:
20
8
2022
medline:
4
11
2022
entrez:
19
8
2022
Statut:
ppublish
Résumé
Studies evaluating the effects of cancer treatments are prone to immortal time bias that, if unaddressed, can lead to treatments appearing more beneficial than they are. To demonstrate the impact of immortal time bias, we compared results across several analytic approaches (dichotomous exposure, dichotomous exposure excluding immortal time, time-varying exposure, landmark analysis, clone-censor-weight method), using surgical resection among women with metastatic breast cancer as an example. All adult women diagnosed with incident metastatic breast cancer from 2013-2016 in the National Cancer Database were included. To quantify immortal time bias, we also conducted a simulation study where the "true" relationship between surgical resection and mortality was known. 24,329 women (median age 61, IQR 51-71) were included, and 24% underwent surgical resection. The largest association between resection and mortality was observed when using a dichotomized exposure [HR, 0.54; 95% confidence interval (CI), 0.51-0.57], followed by dichotomous with exclusion of immortal time (HR, 0.62; 95% CI, 0.59-0.65). Results from the time-varying exposure, landmark, and clone-censor-weight method analyses were closer to the null (HR, 0.67-0.84). Results from the plasmode simulation found that the time-varying exposure, landmark, and clone-censor-weight method models all produced unbiased HRs (bias -0.003 to 0.016). Both standard dichotomous exposure (HR, 0.84; bias, -0.177) and dichotomous with exclusion of immortal time (HR, 0.93; bias, -0.074) produced meaningfully biased estimates. Researchers should use time-varying exposures with a treatment assessment window or the clone-censor-weight method when immortal time is present. Using methods that appropriately account for immortal time will improve evidence and decision-making from research using real-world data.
Sections du résumé
BACKGROUND
Studies evaluating the effects of cancer treatments are prone to immortal time bias that, if unaddressed, can lead to treatments appearing more beneficial than they are.
METHODS
To demonstrate the impact of immortal time bias, we compared results across several analytic approaches (dichotomous exposure, dichotomous exposure excluding immortal time, time-varying exposure, landmark analysis, clone-censor-weight method), using surgical resection among women with metastatic breast cancer as an example. All adult women diagnosed with incident metastatic breast cancer from 2013-2016 in the National Cancer Database were included. To quantify immortal time bias, we also conducted a simulation study where the "true" relationship between surgical resection and mortality was known.
RESULTS
24,329 women (median age 61, IQR 51-71) were included, and 24% underwent surgical resection. The largest association between resection and mortality was observed when using a dichotomized exposure [HR, 0.54; 95% confidence interval (CI), 0.51-0.57], followed by dichotomous with exclusion of immortal time (HR, 0.62; 95% CI, 0.59-0.65). Results from the time-varying exposure, landmark, and clone-censor-weight method analyses were closer to the null (HR, 0.67-0.84). Results from the plasmode simulation found that the time-varying exposure, landmark, and clone-censor-weight method models all produced unbiased HRs (bias -0.003 to 0.016). Both standard dichotomous exposure (HR, 0.84; bias, -0.177) and dichotomous with exclusion of immortal time (HR, 0.93; bias, -0.074) produced meaningfully biased estimates.
CONCLUSIONS
Researchers should use time-varying exposures with a treatment assessment window or the clone-censor-weight method when immortal time is present.
IMPACT
Using methods that appropriately account for immortal time will improve evidence and decision-making from research using real-world data.
Identifiants
pubmed: 35984990
pii: 710023
doi: 10.1158/1055-9965.EPI-22-0495
pmc: PMC9627261
mid: NIHMS1832724
doi:
Types de publication
Journal Article
Research Support, N.I.H., Extramural
Research Support, N.I.H., Intramural
Langues
eng
Sous-ensembles de citation
IM
Pagination
2079-2086Subventions
Organisme : NCI NIH HHS
ID : T32 CA116339
Pays : United States
Organisme : Intramural NIH HHS
ID : Z99 MD999999
Pays : United States
Informations de copyright
©2022 The Authors; Published by the American Association for Cancer Research.
Références
Am J Epidemiol. 2008 Feb 15;167(4):492-9
pubmed: 18056625
Surgery. 2002 Oct;132(4):620-6; discussion 626-7
pubmed: 12407345
JAMA Surg. 2016 May 1;151(5):424-31
pubmed: 26629881
J Clin Oncol. 1983 Nov;1(11):710-9
pubmed: 6668489
J Clin Oncol. 2016 Jul 10;34(20):2359-65
pubmed: 27001590
Ann Intern Med. 2006 Sep 5;145(5):361-3; discussion 392
pubmed: 16954361
Pharmacoepidemiol Drug Saf. 2008 Oct;17(10):1036
pubmed: 18816873
Multivariate Behav Res. 2011 May;46(3):399-424
pubmed: 21818162
Stat Med. 1992 Oct-Nov;11(14-15):1871-9
pubmed: 1480879
Stat Med. 2009 May 30;28(12):1725-38
pubmed: 19347843
Int J Epidemiol. 2020 Oct 1;49(5):1719-1729
pubmed: 32386426
Am J Epidemiol. 2019 Feb 1;188(2):438-443
pubmed: 30299451
Clin Breast Cancer. 2020 Apr;20(2):e200-e213
pubmed: 32089454
Breast. 2019 Jun;45:104-112
pubmed: 30928762
Surg Oncol. 2021 Jun;37:101539
pubmed: 33706057
Comput Stat Data Anal. 2014 Apr;72:219-226
pubmed: 24587587
JAMA. 2021 Feb 16;325(7):686-687
pubmed: 33591334
J Clin Oncol. 2006 Jun 20;24(18):2694-6
pubmed: 16702578
Epidemiology. 2014 Nov;25(6):889-97
pubmed: 25140837
J Intern Med. 2014 Jun;275(6):570-80
pubmed: 24520806
Ann Intern Med. 2007 Oct 16;147(8):W163-94
pubmed: 17938389
PLoS Med. 2015 Oct 06;12(10):e1001885
pubmed: 26440803
Ann Intern Med. 2007 Oct 16;147(8):573-7
pubmed: 17938396
Am J Epidemiol. 2008 Sep 15;168(6):656-64
pubmed: 18682488
Ann Surg Oncol. 2008 Mar;15(3):683-90
pubmed: 18183467
Am J Epidemiol. 2016 Apr 15;183(8):758-64
pubmed: 26994063
Pharmacoepidemiol Drug Saf. 2007 Mar;16(3):241-9
pubmed: 17252614
JAMA. 2018 Sep 4;320(9):867-868
pubmed: 30105359
Ann Surg Oncol. 2018 Oct;25(11):3141-3149
pubmed: 29777404
JAMA Oncol. 2018 Jan 01;4(1):63-70
pubmed: 28822996
Breast J. 2019 Jul;25(4):644-653
pubmed: 31087448
Circ Cardiovasc Qual Outcomes. 2011 May;4(3):363-71
pubmed: 21586725
JAMA Intern Med. 2021 Apr 1;181(4):569-570
pubmed: 33464281
N Engl J Med. 2020 Jun 25;382(26):2582
pubmed: 32501665
J Clin Epidemiol. 2016 Nov;79:70-75
pubmed: 27237061
Ann Surg. 2019 Mar;269(3):537-544
pubmed: 29227346
Ann Intern Med. 2001 May 15;134(10):955-62
pubmed: 11352696
Ann Surg Oncol. 2020 Aug;27(8):2711-2720
pubmed: 32157524
BMJ. 2018 Nov 14;363:k3532
pubmed: 30429167
Stat Med. 2002 Nov 30;21(22):3493-510
pubmed: 12407686
Epidemiology. 2000 Sep;11(5):550-60
pubmed: 10955408
Ann Surg Oncol. 2007 Aug;14(8):2187-94
pubmed: 17522944
Lancet Oncol. 2015 Oct;16(13):1380-8
pubmed: 26363985
Lancet. 2020 Jun 13;395(10240):1820
pubmed: 32511943
Ann Surg. 2016 Jun;263(6):1188-98
pubmed: 26943635
Clin Epidemiol. 2020 Dec 21;12:1403-1420
pubmed: 33376409
Ann Surg Oncol. 2007 Dec;14(12):3285-7
pubmed: 17891444
JAMA Intern Med. 2021 Apr 1;181(4):568-569
pubmed: 33464282
N Engl J Med. 2016 Dec 8;375(23):2293-2297
pubmed: 27959688