Dose-response in case-control studies

Our plain-English summary of Dose-response in case-control studies, by G. Berry (Journal of Epidemiology and Community Health, 1980).

The study

• Author: G. Berry
• Journal: Journal of Epidemiology and Community Health
• Published: 1980;34(3):217-22
• Cited by: 18 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text
• In J-M's library: Johns-Manville's company medical library held a copy (J-M no. 002028).

Abstract

The evidence provided by a case-control study on the association between a disease and some factor is strengthened if the extent of exposure to the factor is categorised into several groups or measured on a continuous scale. Then dose-response relationships can be estimated. The methods available are illustrated by application to data on lung cancer and chrysotile asbestos exposure from Quebec in which there were three matched controls for each case. Regression-type models were fitted assuming that the relative risk of lung cancer was linearly related to an exposure measure; a covariate, smoking, was also included in the analysis. The data were first analysed ignoring the matching and secondly taking account of the matching. The methodology for the latter analysis has only recently been developed; formerly, matched studies were of necessity analysed as unmatched. Although, in this particular example, the unmatched and matched analyses gave similar results, this is not always the case and it is argued that, now that the methodology is available, matched case-control studies should be analysed taking proper account of the matching.

In plain English

This 1980 paper set out statistical methods for estimating dose-response in case-control studies. It was not a new epidemiological study. As an illustration, the author reanalysed data from a study of men who mined and milled chrysotile asbestos in Quebec. The data covered 245 deaths from lung cancer, each with three controls matched for year of birth. Dust exposure, in millions of particles per cubic foot times years, was counted up to 9 years before the death of the case. The author fitted a model in which the relative risk of lung cancer rose in a straight line with exposure. Ignoring the matching, the estimated rise was 0.00119 per exposure unit, a highly significant trend. Taking the matching into account, the estimate was 0.00136 (approximate 95% confidence limits 0.00068 to 0.00274), and the two analyses gave similar results. In a subgroup with smoking information, an asbestos effect remained after allowing for smoking. The author stated that the purpose was to illustrate methods, not to present epidemiological results. The author concluded that matched case-control studies should be analysed taking the matching into account.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: statistical methods paper illustrating dose-response analysis of case-control data, using data from a matched case-control study within a cohort (McDonald et al.) for illustration only
• Who was studied: 245 men who died of lung cancer, from a cohort of men employed in chrysotile asbestos mining and milling in Quebec. Each case had three controls from the same population, matched for year of birth and still living when the case died (735 controls). Smoking habits were known for about 60% of the men, and 140 cases had known smoking habits and at least one control with known smoking habits
• Exposure: cumulative dust exposure in chrysotile mining and milling, measured in millions of particles per cubic foot times years and evaluated up to 9 years before the death of the case. Smoking (smoker or non-smoker) was also included as a covariate
• Main results:
  • Unmatched analysis with 10 exposure groups: fitting a linear relative risk term reduced the deviance by 19.61 (1 df), a highly significant effect. The estimate of b (increase in relative risk per exposure unit) was 0.00119, and the residual deviance of 3.43 (8 df) gave no evidence against a linear fit
  • Dividing the data into 6, 20 or 27 groups, or analysing individuals ungrouped, gave similar estimates of b (0.00113 to 0.00122)
  • Matched analysis: the estimate of b was 0.00136, with approximate 95% confidence limits of 0.00068 and 0.00274, and a test statistic of 21.37 (1 df). The test statistic for adding a quadratic term was 0.10 (1 df)
  • The matched and unmatched analyses gave similar results. Allowing for matching only slightly increased the sensitivity of the analysis
  • Smoking subgroup (matched analysis): fitting smoking alone gave an estimated relative risk due to smoking of 5.5, with a test statistic of 40.59 (1 df). Adding asbestos exposure to smoking gave a test statistic of 12.82 (1 df), as evidence of an asbestos effect after allowing for smoking
  • An additive model combining smoking and asbestos fitted better than a multiplicative one, but there was not much discrimination between the two. The relative likelihood of additive versus multiplicative was 3.3 to 1
• The authors' conclusion: The author concluded that, now that methods for analysing matched categorical data are available, there is little reason to continue ignoring the matching in the analysis of matched case-control studies
• Limitations the authors state:
  • Smoking habits were known for only 60% of the men. The author did not discuss the possible influence of this because the paper's purpose was to illustrate methods, not to present epidemiological results
  • The matched analysis gives no test of how well the fitted dose-response relationship fits the data

Topics

Medical terms in this study, explained in our Lexicon: Dose-response relationship, Case-Control Study, Lung Cancer, Chrysotile, Asbestos Exposure, Relative Risk.

Related studies

• Hierarchical latency models for dose-time-response associations — D.B. Richardson, R.F. MacLehose, et al. (2011)
• Dose-response modeling of NLRP3 inflammasome-mediated diseases: asbestos, lung cancer, and malignant mesothelioma as examples — L.A. Cox (2019)
• Pleural mesothelioma: dose-response relation at low levels of asbestos exposure in a French population-based case-control study — Y. Iwatsubo, J.C. Pairon, et al. (1998)
• A study of dose-response relationships for asbestos associated disease — M.M. Finkelstein (1985)
• Pulmonary function in asbestos cement workers: a dose-response study — M. Finkelstein (1986)
• Dose-response relationships for asbestos-related disease: implications for hygiene standards. Part II. Mortality — J. Peto (1979)

Browse return to medical research.