Bidirectional case-crossover studies of air pollution: Bias from skewed and incomplete waves

Jong-Tae Lee, Ho Kim, Joel Schwartz

Research output: Contribution to journalArticle

26 Citations (Scopus)

Abstract

The case-crossover design compares exposures during the period of time of failure with one or more periods when failure did not occur and evaluates the potential excess risk using conditional logistic regression. In this simulation study, we applied several control sampling approaches to control for confounding by various temporal patterns of an exposure variable and evaluated the usefulness of symmetric bidirectional control strategies. We simulated true relative risks (RRs; true β = 0.001) of deaths of 1.051 per 50-ppb increase of sulfur dioxide and included confounding by right- or left-skewed seasonal waves, linear long-term time trends, or a combination of both. The range of the estimated RRs from symmetric bidirectional control sampling approaches was 1.044 ∼ 1.056 at either a long-term trend or any skewed seasonal wave of SO2 levels, which indicated the bidirectional control sampling methods would successfully control confounding by design. The simulations with bidirectional sampling, however, show that biases may occur if waves are incomplete (20-43% underestimated RRs). In conclusion, our simulations show that the symmetric bidirectional case-crossover design can substantially control for confounding by linear long-term trends and/or seasonality of an exposure variable by design as well. However, unidirectional control sampling would fail to control confounding by those variations of air pollution. Simulation results also show that even the bidirectional case-crossover design can be biased in a situation where the exposure variable shows incomplete cyclic waves, and therefore it cannot completely control for temporal confounding.

Original languageEnglish
Pages (from-to)1107-1111
Number of pages5
JournalEnvironmental Health Perspectives
Volume108
Issue number12
Publication statusPublished - 2000 Dec 1
Externally publishedYes

Fingerprint

Air Pollution
Air pollution
Cross-Over Studies
atmospheric pollution
sampling
simulation
Sampling
Sulfur Dioxide
sulfur dioxide
Logistic Models
seasonality
logistics
exposure
Logistics
long-term trend

Keywords

  • Air pollution
  • Case-crossover designs
  • Control sampling strategy
  • Epidemiologic methods

ASJC Scopus subject areas

  • Environmental Science(all)
  • Environmental Chemistry
  • Public Health, Environmental and Occupational Health

Cite this

Bidirectional case-crossover studies of air pollution : Bias from skewed and incomplete waves. / Lee, Jong-Tae; Kim, Ho; Schwartz, Joel.

In: Environmental Health Perspectives, Vol. 108, No. 12, 01.12.2000, p. 1107-1111.

Research output: Contribution to journalArticle

@article{1349994bdabc49ee9d38818fdd4b9e8e,
title = "Bidirectional case-crossover studies of air pollution: Bias from skewed and incomplete waves",
abstract = "The case-crossover design compares exposures during the period of time of failure with one or more periods when failure did not occur and evaluates the potential excess risk using conditional logistic regression. In this simulation study, we applied several control sampling approaches to control for confounding by various temporal patterns of an exposure variable and evaluated the usefulness of symmetric bidirectional control strategies. We simulated true relative risks (RRs; true β = 0.001) of deaths of 1.051 per 50-ppb increase of sulfur dioxide and included confounding by right- or left-skewed seasonal waves, linear long-term time trends, or a combination of both. The range of the estimated RRs from symmetric bidirectional control sampling approaches was 1.044 ∼ 1.056 at either a long-term trend or any skewed seasonal wave of SO2 levels, which indicated the bidirectional control sampling methods would successfully control confounding by design. The simulations with bidirectional sampling, however, show that biases may occur if waves are incomplete (20-43{\%} underestimated RRs). In conclusion, our simulations show that the symmetric bidirectional case-crossover design can substantially control for confounding by linear long-term trends and/or seasonality of an exposure variable by design as well. However, unidirectional control sampling would fail to control confounding by those variations of air pollution. Simulation results also show that even the bidirectional case-crossover design can be biased in a situation where the exposure variable shows incomplete cyclic waves, and therefore it cannot completely control for temporal confounding.",
keywords = "Air pollution, Case-crossover designs, Control sampling strategy, Epidemiologic methods",
author = "Jong-Tae Lee and Ho Kim and Joel Schwartz",
year = "2000",
month = "12",
day = "1",
language = "English",
volume = "108",
pages = "1107--1111",
journal = "Environmental Health Perspectives",
issn = "0091-6765",
publisher = "Public Health Services, US Dept of Health and Human Services",
number = "12",

}

TY - JOUR

T1 - Bidirectional case-crossover studies of air pollution

T2 - Bias from skewed and incomplete waves

AU - Lee, Jong-Tae

AU - Kim, Ho

AU - Schwartz, Joel

PY - 2000/12/1

Y1 - 2000/12/1

N2 - The case-crossover design compares exposures during the period of time of failure with one or more periods when failure did not occur and evaluates the potential excess risk using conditional logistic regression. In this simulation study, we applied several control sampling approaches to control for confounding by various temporal patterns of an exposure variable and evaluated the usefulness of symmetric bidirectional control strategies. We simulated true relative risks (RRs; true β = 0.001) of deaths of 1.051 per 50-ppb increase of sulfur dioxide and included confounding by right- or left-skewed seasonal waves, linear long-term time trends, or a combination of both. The range of the estimated RRs from symmetric bidirectional control sampling approaches was 1.044 ∼ 1.056 at either a long-term trend or any skewed seasonal wave of SO2 levels, which indicated the bidirectional control sampling methods would successfully control confounding by design. The simulations with bidirectional sampling, however, show that biases may occur if waves are incomplete (20-43% underestimated RRs). In conclusion, our simulations show that the symmetric bidirectional case-crossover design can substantially control for confounding by linear long-term trends and/or seasonality of an exposure variable by design as well. However, unidirectional control sampling would fail to control confounding by those variations of air pollution. Simulation results also show that even the bidirectional case-crossover design can be biased in a situation where the exposure variable shows incomplete cyclic waves, and therefore it cannot completely control for temporal confounding.

AB - The case-crossover design compares exposures during the period of time of failure with one or more periods when failure did not occur and evaluates the potential excess risk using conditional logistic regression. In this simulation study, we applied several control sampling approaches to control for confounding by various temporal patterns of an exposure variable and evaluated the usefulness of symmetric bidirectional control strategies. We simulated true relative risks (RRs; true β = 0.001) of deaths of 1.051 per 50-ppb increase of sulfur dioxide and included confounding by right- or left-skewed seasonal waves, linear long-term time trends, or a combination of both. The range of the estimated RRs from symmetric bidirectional control sampling approaches was 1.044 ∼ 1.056 at either a long-term trend or any skewed seasonal wave of SO2 levels, which indicated the bidirectional control sampling methods would successfully control confounding by design. The simulations with bidirectional sampling, however, show that biases may occur if waves are incomplete (20-43% underestimated RRs). In conclusion, our simulations show that the symmetric bidirectional case-crossover design can substantially control for confounding by linear long-term trends and/or seasonality of an exposure variable by design as well. However, unidirectional control sampling would fail to control confounding by those variations of air pollution. Simulation results also show that even the bidirectional case-crossover design can be biased in a situation where the exposure variable shows incomplete cyclic waves, and therefore it cannot completely control for temporal confounding.

KW - Air pollution

KW - Case-crossover designs

KW - Control sampling strategy

KW - Epidemiologic methods

UR - http://www.scopus.com/inward/record.url?scp=0034518484&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034518484&partnerID=8YFLogxK

M3 - Article

C2 - 11133389

AN - SCOPUS:0034518484

VL - 108

SP - 1107

EP - 1111

JO - Environmental Health Perspectives

JF - Environmental Health Perspectives

SN - 0091-6765

IS - 12

ER -