Conditions for valid empirical estimates of cancer overdiagnosis in randomized trials and population studies

Roman Gulati, Eric J. Feuer, Ruth Etzioni

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Cancer overdiagnosis is frequently estimated using the excess incidence in a screened group relative to that in an unscreened group. However, conditions for unbiased estimation are poorly understood. We developed a mathematical framework to project the effects of screening on the incidence of relevant cancers - that is, cancers that would present clinically without screening. Screening advances the date of diagnosis for a fraction of preclinical relevant cancers. Which diagnoses are advanced and by how much depends on the preclinical detectable period, test sensitivity, and screening patterns. Using the model, we projected incidence in common trial designs and population settings and compared excess incidence with true overdiagnosis. In trials with no control arm screening, unbiased estimates are available using cumulative incidence if the screen arm stops screening and using annual incidence if the screen arm continues screening. In both designs, unbiased estimation requires waiting until screening stabilizes plus the maximum preclinical period. In continued-screen trials and population settings, excess cumulative incidence is persistently biased. We investigated this bias in published estimates from the European Randomized Study of Screening for Prostate Cancer after 9-13 years. In conclusion, no trial or population setting automatically permits unbiased estimation of overdiagnosis; sufficient follow-up and appropriate analysis remain crucial.

Original languageEnglish (US)
Pages (from-to)140-147
Number of pages8
JournalAmerican Journal of Epidemiology
Volume184
Issue number2
DOIs
StatePublished - Jul 15 2016
Externally publishedYes

Fingerprint

Incidence
Population
Neoplasms
Medical Overuse
Prostatic Neoplasms

Keywords

  • bias
  • early detection of cancer
  • mass screening
  • mathematical model
  • overdiagnosis
  • randomized clinical trial

ASJC Scopus subject areas

  • Epidemiology

Cite this

Conditions for valid empirical estimates of cancer overdiagnosis in randomized trials and population studies. / Gulati, Roman; Feuer, Eric J.; Etzioni, Ruth.

In: American Journal of Epidemiology, Vol. 184, No. 2, 15.07.2016, p. 140-147.

Research output: Contribution to journalArticle

Gulati, Roman ; Feuer, Eric J. ; Etzioni, Ruth. / Conditions for valid empirical estimates of cancer overdiagnosis in randomized trials and population studies. In: American Journal of Epidemiology. 2016 ; Vol. 184, No. 2. pp. 140-147.
@article{c818311c10be4898bbb1a8b72b672ded,
title = "Conditions for valid empirical estimates of cancer overdiagnosis in randomized trials and population studies",
abstract = "Cancer overdiagnosis is frequently estimated using the excess incidence in a screened group relative to that in an unscreened group. However, conditions for unbiased estimation are poorly understood. We developed a mathematical framework to project the effects of screening on the incidence of relevant cancers - that is, cancers that would present clinically without screening. Screening advances the date of diagnosis for a fraction of preclinical relevant cancers. Which diagnoses are advanced and by how much depends on the preclinical detectable period, test sensitivity, and screening patterns. Using the model, we projected incidence in common trial designs and population settings and compared excess incidence with true overdiagnosis. In trials with no control arm screening, unbiased estimates are available using cumulative incidence if the screen arm stops screening and using annual incidence if the screen arm continues screening. In both designs, unbiased estimation requires waiting until screening stabilizes plus the maximum preclinical period. In continued-screen trials and population settings, excess cumulative incidence is persistently biased. We investigated this bias in published estimates from the European Randomized Study of Screening for Prostate Cancer after 9-13 years. In conclusion, no trial or population setting automatically permits unbiased estimation of overdiagnosis; sufficient follow-up and appropriate analysis remain crucial.",
keywords = "bias, early detection of cancer, mass screening, mathematical model, overdiagnosis, randomized clinical trial",
author = "Roman Gulati and Feuer, {Eric J.} and Ruth Etzioni",
year = "2016",
month = "7",
day = "15",
doi = "10.1093/aje/kwv342",
language = "English (US)",
volume = "184",
pages = "140--147",
journal = "American Journal of Epidemiology",
issn = "0002-9262",
publisher = "Oxford University Press",
number = "2",

}

TY - JOUR

T1 - Conditions for valid empirical estimates of cancer overdiagnosis in randomized trials and population studies

AU - Gulati, Roman

AU - Feuer, Eric J.

AU - Etzioni, Ruth

PY - 2016/7/15

Y1 - 2016/7/15

N2 - Cancer overdiagnosis is frequently estimated using the excess incidence in a screened group relative to that in an unscreened group. However, conditions for unbiased estimation are poorly understood. We developed a mathematical framework to project the effects of screening on the incidence of relevant cancers - that is, cancers that would present clinically without screening. Screening advances the date of diagnosis for a fraction of preclinical relevant cancers. Which diagnoses are advanced and by how much depends on the preclinical detectable period, test sensitivity, and screening patterns. Using the model, we projected incidence in common trial designs and population settings and compared excess incidence with true overdiagnosis. In trials with no control arm screening, unbiased estimates are available using cumulative incidence if the screen arm stops screening and using annual incidence if the screen arm continues screening. In both designs, unbiased estimation requires waiting until screening stabilizes plus the maximum preclinical period. In continued-screen trials and population settings, excess cumulative incidence is persistently biased. We investigated this bias in published estimates from the European Randomized Study of Screening for Prostate Cancer after 9-13 years. In conclusion, no trial or population setting automatically permits unbiased estimation of overdiagnosis; sufficient follow-up and appropriate analysis remain crucial.

AB - Cancer overdiagnosis is frequently estimated using the excess incidence in a screened group relative to that in an unscreened group. However, conditions for unbiased estimation are poorly understood. We developed a mathematical framework to project the effects of screening on the incidence of relevant cancers - that is, cancers that would present clinically without screening. Screening advances the date of diagnosis for a fraction of preclinical relevant cancers. Which diagnoses are advanced and by how much depends on the preclinical detectable period, test sensitivity, and screening patterns. Using the model, we projected incidence in common trial designs and population settings and compared excess incidence with true overdiagnosis. In trials with no control arm screening, unbiased estimates are available using cumulative incidence if the screen arm stops screening and using annual incidence if the screen arm continues screening. In both designs, unbiased estimation requires waiting until screening stabilizes plus the maximum preclinical period. In continued-screen trials and population settings, excess cumulative incidence is persistently biased. We investigated this bias in published estimates from the European Randomized Study of Screening for Prostate Cancer after 9-13 years. In conclusion, no trial or population setting automatically permits unbiased estimation of overdiagnosis; sufficient follow-up and appropriate analysis remain crucial.

KW - bias

KW - early detection of cancer

KW - mass screening

KW - mathematical model

KW - overdiagnosis

KW - randomized clinical trial

UR - http://www.scopus.com/inward/record.url?scp=84980417582&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84980417582&partnerID=8YFLogxK

U2 - 10.1093/aje/kwv342

DO - 10.1093/aje/kwv342

M3 - Article

C2 - 27358266

AN - SCOPUS:84980417582

VL - 184

SP - 140

EP - 147

JO - American Journal of Epidemiology

JF - American Journal of Epidemiology

SN - 0002-9262

IS - 2

ER -