Different statistical techniques dealing with confounding in observational research: measuring the effect of breast-conserving therapy and mastectomy on survival

Marissa C. van Maaren, Saskia le Cessie, Luc J.A. Strobbe, Catharina G.M. Groothuis-Oudshoorn, Philip M.P. Poortmans, Sabine Siesling

Research output: Contribution to journalArticleAcademicpeer-review

1 Downloads (Pure)

Abstract

Purpose: Propensity trimming, hierarchical modelling and instrumental variable (IV) analysis are statistical techniques dealing with confounding, cluster-related variation or confounding by severity. This study aimed to explain (dis)advantages of these techniques in estimating the effect of breast-conserving therapy (BCT) and mastectomy on 10-year distant metastasis-free survival (DMFS). Methods: All women diagnosed in 2005 with primary T1-2N0-1 breast cancer treated with BCT or mastectomy were selected from the Netherlands Cancer Registry. We used multivariable Cox regression to correct for confounding. Propensity trimming was used to create a more homogeneous population for which the treatment choice was not self-evident. Hospital of surgery was used as hierarchical level to handle hospital-related variation, and as IV to deal with unmeasured confounding. Results: Multivariable Cox regression showed higher 10-year DMFS for BCT than mastectomy [HR 0.70 (95% CI 0.60–82)]. Propensity trimming on the 10–90th and the 20–80th percentile of the propensity score distribution and hierarchical modelling showed similar HRs. IV analysis showed no significant difference between BCT and mastectomy. Conclusion: Unmeasured confounding is very difficult to eliminate in observational research. We cannot conclude that BCT or mastectomy has a causal relationship with 10-year DMFS. It is crucial to critically evaluate all model’s assumptions, and to be careful in drawing firm conclusions.

Original languageEnglish
Pages (from-to)1485-1493
Number of pages9
JournalJournal of Cancer Research and Clinical Oncology
Volume145
Issue number6
Early online date24 Apr 2019
DOIs
Publication statusPublished - 1 Jun 2019

Fingerprint

Mastectomy
Breast
Survival
Research
Neoplasm Metastasis
Therapeutics
Propensity Score
Netherlands
Registries
Breast Neoplasms
Population
Neoplasms

Keywords

  • UT-Hybrid-D
  • Breast-conserving therapy
  • Hierarchical modelling
  • Instrumental variable
  • Mastectomy
  • Propensity scores
  • Breast cancer

Cite this

@article{2b3891bb84be4ec9a512b89cb8d84e57,
title = "Different statistical techniques dealing with confounding in observational research: measuring the effect of breast-conserving therapy and mastectomy on survival",
abstract = "Purpose: Propensity trimming, hierarchical modelling and instrumental variable (IV) analysis are statistical techniques dealing with confounding, cluster-related variation or confounding by severity. This study aimed to explain (dis)advantages of these techniques in estimating the effect of breast-conserving therapy (BCT) and mastectomy on 10-year distant metastasis-free survival (DMFS). Methods: All women diagnosed in 2005 with primary T1-2N0-1 breast cancer treated with BCT or mastectomy were selected from the Netherlands Cancer Registry. We used multivariable Cox regression to correct for confounding. Propensity trimming was used to create a more homogeneous population for which the treatment choice was not self-evident. Hospital of surgery was used as hierarchical level to handle hospital-related variation, and as IV to deal with unmeasured confounding. Results: Multivariable Cox regression showed higher 10-year DMFS for BCT than mastectomy [HR 0.70 (95{\%} CI 0.60–82)]. Propensity trimming on the 10–90th and the 20–80th percentile of the propensity score distribution and hierarchical modelling showed similar HRs. IV analysis showed no significant difference between BCT and mastectomy. Conclusion: Unmeasured confounding is very difficult to eliminate in observational research. We cannot conclude that BCT or mastectomy has a causal relationship with 10-year DMFS. It is crucial to critically evaluate all model’s assumptions, and to be careful in drawing firm conclusions.",
keywords = "UT-Hybrid-D, Breast-conserving therapy, Hierarchical modelling, Instrumental variable, Mastectomy, Propensity scores, Breast cancer",
author = "{van Maaren}, {Marissa C.} and {le Cessie}, Saskia and Strobbe, {Luc J.A.} and Groothuis-Oudshoorn, {Catharina G.M.} and Poortmans, {Philip M.P.} and Sabine Siesling",
note = "Springer deal",
year = "2019",
month = "6",
day = "1",
doi = "10.1007/s00432-019-02919-x",
language = "English",
volume = "145",
pages = "1485--1493",
journal = "Journal of Cancer Research and Clinical Oncology",
issn = "0171-5216",
publisher = "Springer",
number = "6",

}

Different statistical techniques dealing with confounding in observational research : measuring the effect of breast-conserving therapy and mastectomy on survival. / van Maaren, Marissa C.; le Cessie, Saskia; Strobbe, Luc J.A.; Groothuis-Oudshoorn, Catharina G.M.; Poortmans, Philip M.P.; Siesling, Sabine.

In: Journal of Cancer Research and Clinical Oncology, Vol. 145, No. 6, 01.06.2019, p. 1485-1493.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Different statistical techniques dealing with confounding in observational research

T2 - measuring the effect of breast-conserving therapy and mastectomy on survival

AU - van Maaren, Marissa C.

AU - le Cessie, Saskia

AU - Strobbe, Luc J.A.

AU - Groothuis-Oudshoorn, Catharina G.M.

AU - Poortmans, Philip M.P.

AU - Siesling, Sabine

N1 - Springer deal

PY - 2019/6/1

Y1 - 2019/6/1

N2 - Purpose: Propensity trimming, hierarchical modelling and instrumental variable (IV) analysis are statistical techniques dealing with confounding, cluster-related variation or confounding by severity. This study aimed to explain (dis)advantages of these techniques in estimating the effect of breast-conserving therapy (BCT) and mastectomy on 10-year distant metastasis-free survival (DMFS). Methods: All women diagnosed in 2005 with primary T1-2N0-1 breast cancer treated with BCT or mastectomy were selected from the Netherlands Cancer Registry. We used multivariable Cox regression to correct for confounding. Propensity trimming was used to create a more homogeneous population for which the treatment choice was not self-evident. Hospital of surgery was used as hierarchical level to handle hospital-related variation, and as IV to deal with unmeasured confounding. Results: Multivariable Cox regression showed higher 10-year DMFS for BCT than mastectomy [HR 0.70 (95% CI 0.60–82)]. Propensity trimming on the 10–90th and the 20–80th percentile of the propensity score distribution and hierarchical modelling showed similar HRs. IV analysis showed no significant difference between BCT and mastectomy. Conclusion: Unmeasured confounding is very difficult to eliminate in observational research. We cannot conclude that BCT or mastectomy has a causal relationship with 10-year DMFS. It is crucial to critically evaluate all model’s assumptions, and to be careful in drawing firm conclusions.

AB - Purpose: Propensity trimming, hierarchical modelling and instrumental variable (IV) analysis are statistical techniques dealing with confounding, cluster-related variation or confounding by severity. This study aimed to explain (dis)advantages of these techniques in estimating the effect of breast-conserving therapy (BCT) and mastectomy on 10-year distant metastasis-free survival (DMFS). Methods: All women diagnosed in 2005 with primary T1-2N0-1 breast cancer treated with BCT or mastectomy were selected from the Netherlands Cancer Registry. We used multivariable Cox regression to correct for confounding. Propensity trimming was used to create a more homogeneous population for which the treatment choice was not self-evident. Hospital of surgery was used as hierarchical level to handle hospital-related variation, and as IV to deal with unmeasured confounding. Results: Multivariable Cox regression showed higher 10-year DMFS for BCT than mastectomy [HR 0.70 (95% CI 0.60–82)]. Propensity trimming on the 10–90th and the 20–80th percentile of the propensity score distribution and hierarchical modelling showed similar HRs. IV analysis showed no significant difference between BCT and mastectomy. Conclusion: Unmeasured confounding is very difficult to eliminate in observational research. We cannot conclude that BCT or mastectomy has a causal relationship with 10-year DMFS. It is crucial to critically evaluate all model’s assumptions, and to be careful in drawing firm conclusions.

KW - UT-Hybrid-D

KW - Breast-conserving therapy

KW - Hierarchical modelling

KW - Instrumental variable

KW - Mastectomy

KW - Propensity scores

KW - Breast cancer

UR - http://www.scopus.com/inward/record.url?scp=85064946984&partnerID=8YFLogxK

U2 - 10.1007/s00432-019-02919-x

DO - 10.1007/s00432-019-02919-x

M3 - Article

VL - 145

SP - 1485

EP - 1493

JO - Journal of Cancer Research and Clinical Oncology

JF - Journal of Cancer Research and Clinical Oncology

SN - 0171-5216

IS - 6

ER -