Population Attributable Risks of Subtypes of Esophageal and Gastric Cancers in the United States

Abstract

Introduction: To help target preventive strategies, we estimated US population attributable risks (PARs) of demographic and potentially modifiable risk factors for esophageal squamous cell carcinoma (ESCC), esophageal adenocarcinoma (EAC), gastric cardia adenocarcinoma (GCA), and gastric noncardia adenocarcinoma (GNCA).

Methods: We prospectively examined the associations for risk factors and these cancers in 490,605 people in the National Institutes of Health-the American Association of Retired Persons Diet and Health cohort Diet and Health Study cohort from 1995 to 2011. Exposures were obtained from the baseline questionnaire. Diagnoses of gastroesophageal reflux disease were extracted for a subset of eligible National Institutes of Health-the American Association of Retired Persons Diet and Health cohort subjects through linkage to Medicare and then multiply imputed for non-Medicare-eligible subjects. Hazard ratios were calculated using multivariable-adjusted Cox proportional hazards regression. Adjusted population attributable risks were calculated for the US population aged 50-71 years by combining the hazard ratios with the estimated joint distribution of risk factor prevalence from the 2015 National Health Interview Survey.

Results: Smoking remained the most important risk factor for ESCC and was estimated to cause more than 1/3 of EAC and GCA and 1/10 of GNCA. Obesity and gastroesophageal reflux disease were associated with more than 1/2 of EAC and 1/3 of GCA. Compared with each lowest-risk level category, common risk factors were estimated to be associated with 73.7% of ESCC (95% confidence interval [CI]: 62.1%-85.4%), 70.3% of EAC (95% CI: 64.4%-76.2%), 69.3% of GCA (95% CI: 61.0%-77.7%), and 33.6% of GNCA (95% CI: 21.7%-45.5%).

Discussion: These factors accounted for a large proportion of esophageal and gastric cancers in the United States, highlighting opportunities for education and intervention to reduce the burden of these highly fatal cancers.

Figure 1.

HRs and 95% CIs of different risk factors for esophageal cancer in the NIH-AARP Diet and Health Study cohort (1995–2011), treating potential risk factors as multilevel variables. CI, confidence interval; EAC, esophageal adenocarcinoma; ESCC, esophageal squamous cell carcinoma; HR, hazard ratio.

Figure 2.

HRs and 95% CIs of different risk factors for gastric adenocarcinoma in the NIH-AARP Diet and Health Study cohort (1995–2011), treating potential risk factors as multilevel variables. CI, confidence interval; GCA, gastric cardia adenocarcinoma; GNCA, gastric noncardia adenocarcinoma; HR, hazard ratio.