Ascertainment Bias in a Historic Cohort Study of Residents in an Asbestos Manufacturing Community

Our plain-English summary of Ascertainment Bias in a Historic Cohort Study of Residents in an Asbestos Manufacturing Community, by J.D. Wortzel, D.J. Wiebe, S. Elahi, et al (International Journal of Environmental Research and Public Health, 2021).

The study

• Authors: J.D. Wortzel, D.J. Wiebe, S. Elahi, A. Agawu, F.K. Barg, E.A. Emmett
• Journal: International Journal of Environmental Research and Public Health
• Published: 2021;18(5)
• Cited by: 7 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

This paper describes follow-up for a cohort of 4530 residents living in the asbestos manufacturing community of Ambler, PA, U.S. in 1930. Using re-identified census data, cause and date of death data obtained from the genealogic website Ancestry.com, along with geospatial analysis, we explored relationships among demographic characteristics, occupational, paraoccupational and environmental asbestos exposures. We identified death data for 2430/4530 individuals. Exposure differed significantly according to race, gender, age, and recency of immigration to the U.S. Notably, there was a significant difference in the availability of year of death information for non-white vs. white individuals (odds ratio (OR) = 0.62 p-value < 0.001), females (OR = 0.53, p-value < 0.001), first-generation immigrants (OR = 0.67, p-value = 0.001), second-generation immigrants (OR = 0.31, p-value < 0.001) vs. non-immigrants, individuals aged less than 20 (OR = 0.31 p-value < 0.001) and individuals aged 20 to 59 (OR = 0.63, p-value < 0.001) vs. older individuals. Similarly, the cause of death was less often available for non-white individuals (OR = 0.42, p-value <0.001), first-generation immigrants and (OR = 0.71, p-value = 0.009), second-generation immigrants (OR = 0.49, p-value < 0.001), individuals aged less than 20 (OR = 0.028 p-value < 0.001), and individuals aged 20 to 59 (OR = 0.26, p-value < 0.001). These results identified ascertainment bias that is important to consider in analyses that investigate occupational, para-occupational and environmental asbestos exposure as risk factors for mortality in this historic cohort. While this study attempts to describe methods for assessing itemized asbestos exposure profiles for a community in 1930 using Ancestry.com and other publicly accessible databases, it also highlights how historic cohort studies likely underestimate the impact of asbestos exposure on vulnerable populations. Future work will aim to assess mortality patterns in this cohort.

In plain English

This historic cohort study assembled 4530 people listed in the 1930 census as living in Ambler, Pennsylvania, where asbestos products were made, mainly from chrysotile. From census records, maps, wind and flood data, and interviews, the authors classified each person's likely exposure pathway: factory work (474 people), a household member working there (1638), or where they lived. 201 people were in both of the first two groups. In Ancestry.com and other public records, they found death records for 2430 people (53.6%). The text says death certificates with cause of death covered 22.4% of the cohort, but the analysis table counts cause of death for only 567 of 4530. Year of death was less often found for non-white than white residents (odds ratio 0.62, 95% confidence interval 0.52 to 0.74), women, immigrants, and people under 60 versus over 60. Cause of death was found less often for those under 20 in 1930 than those over 60 (odds ratio 0.028, 95% confidence interval 0.019 to 0.042). The authors, who had not yet analyzed mortality, concluded that historic cohort studies likely underestimate the impact of asbestos exposure on vulnerable populations.

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: Historic (retrospective) cohort study, combined with qualitative interviews and geospatial mapping. This paper describes how exposure was classified and how often death information could be found. It does not include a mortality analysis.
• Who was studied: 4530 residents of Ambler Borough and West Ambler, Pennsylvania, U.S., identified from the 1930 census. In addition, 56 current and former residents were interviewed (7 past asbestos factory workers, 15 family members of workers, 34 other residents) in 2013–2014 and 2015–2016.
• Exposure: Asbestos from the Keasby and Mattison (K&M) asbestos products factory, which mainly used chrysotile, with minor amounts of amphiboles. Each resident's likely primary exposure pathway was classified as occupational (listed in the 1930 census as working at the factory), paraoccupational (living in a household with a factory worker), or environmental. Environmental exposure meant living closer than 1852.5 ft to the South Ambler waste piles, living downwind of the piles, or living in a flood zone. No exposure measurements were available.
• Main results:
  • Of the 4530 cohort members, 474 (10.4%) were occupationally exposed as K&M workers and 1638 (36.2%) had paraoccupational exposure. 201 (4.44%) were classified as both occupationally and paraoccupationally exposed. 3661 (80.8%) had some form of environmental exposure.
  • Exposure differed by race. Among African Americans, 15.7% were occupationally exposed, compared with 9.7% of whites.
  • Records of death were found for 2430 of the 4530 cohort members (53.6%). Vital status was unknown for the rest.
  • The text states that death certificates with cause of death were available for 22.4% of the cohort. Table 3 lists 1014 (22.4) death certificates as a death data source.
  • In Table 4, the ascertainment analysis table, cause of death is counted as present for 567 of the 4530 cohort members and absent for 3963. This is fewer than the 22.4% stated in the text.
  • Year of death was less often available for non-white vs. white individuals (OR 0.62; table: 95% CI 0.52 to 0.74). It was also less often available for females (OR 0.53; 95% CI 0.47 to 0.60). For cause of death, females did not differ significantly (OR 0.86; 95% CI 0.72 to 1.03; p = 0.106). Cause of death was less often available for non-white individuals (OR 0.42; 95% CI 0.29 to 0.60).
  • Compared with people over 60 in 1930, those aged 19 or younger were much less likely to have a recorded cause of death (OR 0.028; 95% CI 0.019 to 0.042) or year of death (OR 0.31; 95% CI 0.25 to 0.39). Those aged 20 to 59 were also less likely: year of death OR 0.63 (0.51 to 0.79), cause of death OR 0.26 (0.21 to 0.32).
  • Compared with non-immigrants, first-generation immigrants were less likely to have a recorded year of death (OR 0.67; 95% CI 0.57 to 0.79) and cause of death (OR 0.71; 95% CI 0.55 to 0.92). Second-generation immigrants were also less likely: cause of death OR 0.49 (95% CI 0.37 to 0.64). For their year of death, the table gives OR 0.59 (95% CI 0.51 to 0.69), while the text gives 0.31.
  • Compared with the entire cohort, paraoccupationally exposed individuals were less likely to have a recorded year of death (OR 0.59; 95% CI 0.53 to 0.67) and cause of death (OR 0.41; 95% CI 0.33 to 0.50).
  • People living in close proximity to the waste piles were more likely to have records for year of death (OR 1.25; 95% CI 1.04 to 1.49) and cause of death (OR 2.37; 95% CI 1.89 to 2.97). Occupational exposure did not significantly affect whether year of death (OR 1.02; 95% CI 0.85 to 1.24) or cause of death (OR 1.12; 95% CI 0.85 to 1.48) was found.
  • Interviewees reported that playing on the asbestos waste piles in childhood and adolescence was common. Almost all who were children during that period reported it.
• The authors' conclusion: The authors concluded that ascertainment bias is important to consider when studying asbestos exposure as a risk factor for mortality in this and other historic cohorts. They also concluded that historic cohort studies likely underestimate the impact of asbestos exposure on more vulnerable populations.
• Limitations the authors state:
  • Losses to follow-up and lack of death information were major limitations.
  • No historical occupational information from the factory was available. This limited how finely occupational exposure could be divided.
  • Exposure was estimated at only a single point in time (1930), so it gives no information on how long people were exposed.
  • There were no measurements of environmental asbestos levels in Ambler around the 1930s, so exposure could not be quantified.
  • None of the databases could confirm interviewees' descriptions of lifestyle-related exposures, which may have led to underestimating exposure.
  • The long latency of mesothelioma is difficult to account for. The relationship between mortality and exposure has not yet been analyzed.

Topics

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

Medical terms in this study, explained in our Lexicon: Cohort Study, Asbestos Exposure, Odds Ratio, P-Value.

Related studies

• Increased risk of lung cancer mortality among residents near an asbestos product manufacturing plant — S. Kumagai, N. Kurumatani, et al. (2010)
• Cohort study of workers employed in an Italian tire manufacturing plant, 1962-2004 — D. Mirabelli, A.M. Cacciatore, et al. (2012)
• Survival analysis under imperfect record linkage using historic census data — A.K. Marks-Anglin, F.K. Barg, et al. (2024)
• Modeling community asbestos exposure near a vermiculite processing facility: Impact of human activities on cumulative exposure — J.L. Adgate, S.J. Cho, et al. (2011)
• Use of death certificates in epidemiological studies, including occupational hazards: variations in discordance of different asbestos-associated diseases on best evidence ascertainment — I.J. Selikoff, H. Seidman (1992)
• Incidence of Cancer and Asbestos-Related Diseases among Residents Living near Abandoned Asbestos Mines in South Korea: A Retrospective Cohort Study Using National Health Insurance Database — K. Kwak, K.E. Zoh, et al. (2021)

Browse more research in International Journal of Environmental Research and Public Health, or return to medical research.