To the Editor,
Rees and colleagues (1) examined whether the pediatric cancer rate in the Northeast region is higher than in other regions of the United States. In June 2018, the Centers for Disease Control and Prevention (CDC) released a report indicating that New Hampshire's (NH) pediatric cancer rate is the highest in the United States (2). The NH Department of Health and Human Services determined that the pediatric incidence rate of two rare cancers was significantly higher than expected in a five-town area of NH (3). Understanding whether cancer rates are significantly higher in any particular region and/or state is an important component of identifying and modifying causal risk factors.
Rees and colleagues concluded that the pediatric cancer rate in the Northeast region is statistically significantly higher than in other regions. Furthermore, they conclude that the causal risk factor(s) for NH pediatric cancers are likely regional, not localized, deemphasizing environmental contributors while noting that NH and the northeast region have more United States Environmental Protection Agency National Priorities List sites per capita than other regions of the United States.
While we are grateful for the attention to the alarming CDC report, we take strong issue with the conclusions drawn by the authors. The hypothesis did not test whether the NH pediatric cancer rate is different than other states within the Northeast region or whether the regional cancer rate is driven primarily by NH. The conclusions in this article are an overextension of the analysis and are a disservice to the important task of identifying and modifying localized causal risk factors for cancer prevention.
A significantly elevated regional risk of pediatric cancer in the Northeast is a valuable finding and one piece of a larger puzzle of data analysis and interpretation. The high incidence rate in the Northeast and NH demands further investigation. There are many contextual effects of pediatric cancers to explore; there are also geographic community-level exposures that potentially contribute to an increased risk of cancer [e.g., perfluoroalkyl and polyfluoroalkyl substance (PFAS)-contaminated drinking water and air]. The authors have access to individual-level data which would facilitate analysis of possible community-level contextual effects contributing to elevated cancer risk. For example, we recently found an elevated risk of several types of cancer in a community with known PFAS contamination in groundwater (4).
Finally, it is worth noting that the authors used a Bonferroni correction to avoid type 1 error, in which the null hypothesis is rejected for spurious associations resulting from multiple comparisons. A conservative alpha of 0.01, rather than the standard alpha of 0.05, was used to assess statistical significance. Biostatisticians debate the necessity of conservative alphas, given large adjustments to avoid a type 1 error increase the likelihood of a type 2 error, where meaningful group differences are missed. We suggest a more prudent approach rather than one that risks an unwarranted trade-off in the likelihood of type 1 and 2 errors. In fact, the CDC used a 95% confidence interval when they determined that the pediatric cancer rate in NH is the highest in the nation.
A full and transparent assessment of the environmental and public health damage is an essential step in that process since most cancer diagnoses could be prevented by minimizing environmental causes (5). The field would be well served by using resources to inspect internal surveillance, policies, and regulations to prevent environmental exposure and focus attention and resources on prevention and mitigation. We sincerely hope that Rees and colleagues, and others, direct resources to identifying and modifying localized factors contributing to cancers while focusing on prevention. We owe it to NH children who experience the highest risk of pediatric cancer in the nation.
Authors' Disclosures
No disclosures were reported.