Small area estimation of incidence of cancer around a known source of exposure with fine resolution data

Our plain-English summary of Small area estimation of incidence of cancer around a known source of exposure with fine resolution data, by E. Kokki, J. Ranta, A. Penttinen, et al (Occupational and Environmental Medicine, 2001).

The study

• Authors: E. Kokki, J. Ranta, A. Penttinen, E. Pukkala, J. Pekkanen
• Journal: Occupational and Environmental Medicine
• Published: 2001;58(5):315-20
• Cited by: 4 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Objectives: To describe the small area system developed in Finland. To illustrate the use of the system with analyses of incidence of lung cancer around an asbestos mine. To compare the performance of different spatial statistical models when applied to sparse data.

Methods: In the small area system, cancer and population data are available by sex, age, and socioeconomic status in adjacent "pixels", squares of size 0.5 km x 0.5 km. The study area was partitioned into sub-areas based on estimated exposure. The original data at the pixel level were used in a spatial random field model. For comparison, standardised incidence ratios were estimated, and full bayesian and empirical bayesian models were fitted to aggregated data. Incidence of lung cancer around a former asbestos mine was used as an illustration.

Results: The spatial random field model, which has been used in former small area studies, did not converge with present fine resolution data. The number of neighbouring pixels used in smoothing had to be enlarged, and informative distributions for hyperparameters were used to stabilise the unobserved random field. The ordered spatial random field model gave lower estimates than the Poisson model. When one of the three effects of area were fixed, the model gave similar estimates with a narrower interval than the Poisson model.

Conclusions: The use of fine resolution data and socioeconomic status as a means of controlling for confounding related to lifestyle is useful when estimating risk of cancer around point sources. However, better statistical methods are needed for spatial modelling of fine resolution data.

In plain English

This study described SMASH, a Finnish system that counts cancer cases and residents by sex, age and socioeconomic status in squares 0.5 km by 0.5 km. It compared statistical methods for estimating cancer risk around a known source. As an illustration, the authors examined lung cancer diagnosed in 1981–97 among 19 825 people living, on 31 December 1980, in a 50 km by 50 km square around a former anthophyllite asbestos mine in Paakkila, eastern Finland. All rural areas of Finland were the comparison. There were 184 cases. The standardised incidence ratio was 3.80 (95% CI 1.39 to 8.28) in the 4 km by 4 km area near the mine (194 residents, 6 cases), 1.44 (0.92 to 2.13) along the main roads, and 1.20 (1.02 to 1.41) elsewhere. Estimated risk fell with distance from the mine. The spatial model would not converge without added restrictions. The authors suggested occupational exposure seemed the main cause. However, their broad socioeconomic categories had no indicator of former asbestos work, so they could not adjust for miners' occupational exposure. They concluded the system was useful but better methods were needed for such fine data.

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: Methodological study comparing statistical methods for small-area cancer risk estimation, illustrated with a geographic analysis of lung cancer incidence around a former asbestos mine
• Who was studied: 19 825 inhabitants (population data for 31 December 1980) of a 50 km by 50 km square around the former mine in eastern Finland, a rural area with three built-up areas. There were 184 lung cancer cases diagnosed 1981–97. All rural areas of Finland served as the reference area.
• Exposure: Residence near a former anthophyllite asbestos mine at Paakkila (mining 1918–75). The area was divided into three parts. A1 was a 4 km by 4 km square around the mine, placed off-centre because of prevailing south or southwest winds. A2 was a band beside the main roads used for asbestos transport. A3 was the rest of the study area.
• Main results:
  • Area A1, nearest the mine, had 194 inhabitants and 6 lung cancer cases
  • Standardised incidence ratios compared with rural Finland were: A1 3.80 (95% CI 1.39 to 8.28); A2 1.44 (0.92 to 2.13); A3 1.20 (1.02 to 1.41)
  • Even in the most distant area, A3, the standardised incidence ratio was significantly higher than in rural areas on average
  • The ordered spatial random field model gave clearly lower estimates: A1 2.32 (1.27 to 5.94); A2 1.31 (0.99 to 1.81); A3 0.98 (0.88 to 1.10)
  • In the full bayesian analyses, estimated lung cancer risk decreased with increasing distance from the mine
  • The spatial random field model needed restrictions to converge. Ordering the area effects did not work, but fixing one of the three area effects allowed the model to converge.
  • Comparing their estimates with an earlier study, the authors suggested occupational exposure seemed to be the main cause of the increased lung cancer risk near the mine
• The authors' conclusion: The authors concluded that SMASH is useful for estimating cancer risk in small areas because it uses fine resolution data and socioeconomic status. They also concluded that the spatial model used in earlier studies did not work straightforwardly with such sparse data, so better statistical methods are needed.
• Limitations the authors state:
  • Exposure was assessed only from place of residence at one point in time
  • Former miners' occupational asbestos exposure could not be used as a covariate or confounder, because the broad socioeconomic classification had no indicator of former asbestos work. The authors noted that linkage to the existing miners' cohort could have classified miners.
  • Area comparisons like these can give only preliminary results on cancer risk
  • The data may not best represent environmental exposure, because the main asbestos exposure was occupational
  • Because of low numbers, indirect standardisation and SIRs were the only feasible method

Mines named here

Asbestos mines this study is about. Each mine's page says who ran it, when it worked, and what it produced.

• Paakkila mine, Finland

All asbestos mines

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer.

Medical terms in this study, explained in our Lexicon: Lung Cancer, Standardized Incidence Ratio (SIR), Confounding.

Related studies

• Mesothelioma trends in the United States: an update based on Surveillance, Epidemiology, and End Results Program data for 1973 through 2003 — B. Price, A. Ware (2004)
• Upcoming epidemic of asbestos-related malignant pleural mesothelioma in Taiwan: A prediction of incidence in the next 30 years — R.T. Lin, Y.Y. Chang, et al. (2019)
• Newly discovered quick, non-invasive screening method of bone marrow malignancies including various leukemias, Hodgkin's lymphoma, non-Hodgkin's lymphoma, & multiple myeloma by abnormality of small rectangular area within bone marrow organ representation areas of the face — Y. Omura, B. O'Young, et al. (2012)
• Incidence and survival trends for malignant pleural and peritoneal mesothelioma, Australia, 1982-2009 — M.J. Soeberg, J. Leigh, et al. (2016)
• Epidemiological patterns of asbestos exposure and spatial clusters of incident cases of malignant mesothelioma from the Italian national registry — M. Corfiati, A. Scarselli, et al. (2015)
• Incidence of cancer among anthophyllite asbestos miners in Finland — L.O. Meurman, E. Pukkala, et al. (1994)

Browse more research in Occupational and Environmental Medicine, or return to medical research.