The association between mobile phones and the risk of brain cancer mortality: A 25-year cross-country analysis

2019 · Harm · Low evidence

← Research stream

HarmLow evidencePaper 971

Study snapshot

Publicationmanual · 2019

Study typeecological

Reported directionHarm

Evidence ratingLow

Abstract

The association between mobile phones and the risk of brain cancer mortality: A 25-year cross-country analysis Mialon HM, Nesson ET. The association between mobile phones and the risk of brain cancer mortality: A 25-year cross-country analysis. Contemporary Economic Policy. December 4, 2019. doi:10.1111/coep.12456. Abstract We investigate the relationship between cell phones and brain cancer using brain cancer death rates for 88 countries between 1990 and 2015 from the World Health Organization and country-level mobile phone subscription rates from the World Bank. We estimate difference-in-difference models including country and year fixed effects and time-varying country covariates. We find that mobile phone subscription rates are positively and statistically significantly associated with death rates from brain cancer 15–20 years later. In falsification tests, we find few positive associations between mobile phone subscription rates and deaths from rectal, pancreatic, stomach, breast or lung cancer or ischemic heart disease. Finally, differential effects models suggest that mobile phone subscription rates are associated with brain cancer deaths 15–19 years later relative to deaths from other causes. Excerpts The link between mobile phones and brain tumors has received much attention in the media (e.g., Hakim 2016; Knutson 2016; Storrs 2016), although previous evidence has not yielded conclusive results (Hardell and Carlberg 2014; Hardell, Carlberg, and Hansson Mild 2011; Hardell et al. 2013; M. Little et al. 2012; Myung et al. 2009). ...We find a positive association between mobile phone subscriptions and brain cancer death rates. Specifically, one more mobile phone subscription per 100 people is associated with 0.003 and 0.02 additional brain cancer deaths per 100,000 people 15 and 20 years later, respectively. This result is robust to a number of specification tests, although it is sensitive to the exclusion of country-specific linear time trends. We run a number of falsification tests where we examine the relationship between mobile phone subscription rates and mortality rates from five other common forms of cancer, rectal cancer, pancreatic cancer, stomach cancer, lung cancer, and breast cancer, as well as ischemic heart disease. We find much smaller coefficients with much larger standard errors for these other mortality outcomes, although we do find a few positive and statistically significant coefficients. Given that we find some sensitivity to country-specific linear trends and some significant effects of mobile phone subscriptions on other cancers, we estimate a differential effects model where we compare the association between mobile phone subscription rates and brain cancer mortality relative to other disease mortality. In these models, we additionally find statistically significant, positive coefficients, suggesting that mobile phone prevalence is associated with brain cancer mortality about 15–20 years later relative to other disease mortality. Our study is the first to analyze the potential link between brain cancer and mobile phone use on the aggregate, across multiple countries, and over a period spanning more than 20 years.... We collect information on annual mobile cellular subscriptions per 100 people from the World Bank (The World Bank 2016). The first mobile cellular phones were introduced in the Nordic countries in the early 1980s, and by 1990, Iceland, Norway, and Sweden had more than three mobile phone subscriptions per 100 people. Additionally, the United States, Hong Kong, Singapore, Kuwait, and New Zealand were relatively early adopters, with more than one subscription per 100 people by 1990. By 1995, the European Union had an average of over four subscriptions per 100 people, growing to over 53 subscriptions per 100 people in 2000. North American and high income Asian and Middle Eastern countries followed similar patterns. Table 1 shows summary statistics for all variables in our sample over the period from 1990 to 2015. On average, the annual brain cancer death rate is about 3.7 individuals per 100,000. There are about 62 mobile phone subscriptions per 100 people on average. However, there is a lot of temporal variation in mobile phone subscriptions, which we detail more below.... Results from Table 2: ... one more mobile phone subscription per 1,000 people is associated with 0.00012 more brain cancer deaths per 1,000 people 15 years later and 0.00073 more brain cancer deaths per 1,000 people 20 years later. In the first column of Table 4, we find very similar effects to Table 2, and we again find that mobile phone subscriptions are related to increases in brain cancer mortality starting about 15–25 years later. The next four columns show results from differential effects models. In these models, we again find a positive and statistically significant association between mobile phone subscription rates and brain cancer mortality 15–19 years later relative to mortality from the other diseases studied. Our results here are robust to specification changes, with the exception that we lose statistical significance when not using country weights. 3 However, we do note that the mobile phone coefficient is much larger for the 15–19 year lag compared to other lags. We find a statistically significant relationship between mobile phone subscriptions and brain cancer death rates 15–20 years later. Our estimated effect sizes are small but statistically significant. The implications of our findings are potentially very large. Almost 16,000 people died from brain cancer in the United States in 2014. If our findings are correct, the number of brain cancer deaths will continue to rise as the number of mobile phone users rose rapidly in the early part of the century. To err on the side of caution, individuals may want to make greater use of the speaker phone and/or texting options on their cell phones or investigate cell phone cases that may deflect radiation away from their ears. Some caveats to our analysis are warranted. First, we cannot make an unambiguous claim of causality based on our results. Although we control for other characteristics likely to affect brain cancer death rates and find limited statistically significant relationships between mobile phone subscriptions and other mortality outcomes, there may be yet other factors that are correlated with mobile phone subscriptions and are also drivers of changes in brain cancer death rates. For example, ionizing radiation from computed tomography (CT) scans is an established risk factor for brain cancer (Pearce et al. 2012), and if the use of CT scans is correlated with mobile phone use, our estimates may be biased. We include controls for health spending, which should control for the number of X-rays to some degree, and moreover, if X-rays and mobile phone adoption were truly correlated, we would expect mobile phones to have an effect on cancer mortality rates in other sites. But, omitted variables do remain a concern in our analysis. Second, our measure of mobile phone use, mobile phone subscriptions per 100 people, may not be perfectly accurate.... Third, long lags between mobile phone subscriptions and brain cancer death rates mean that, depending on the specified lag between mobile phone subscriptions and mortality, the large recent variations in mobile phone adoption may not be part of our analysis. Additionally, the recent innovations in mobile phone technology, for example switching from 2G to 3G and 4G have resulted in changes in the radiation emitted from mobile phones (Croft et al. 2010; Leung et al. 2011). Moreover, the transition of peoples' use of mobile phones as predominantly phones to the rise of texting and other forms of social media may also change the relationship between mobile phones and brain cancer in the future. onlinelibrary.wiley.com

Machine Enhanced Logic

AI evidence extraction

Structured fields below were generated from the paper’s abstract and metadata to make the evidence easier to inspect. Automated extraction may be incomplete or incorrect; verify important details against the original paper.

Study typeecological

Effect directionharm

Population88 countries (country-level mortality and subscription data)

Sample size88

ExposureRF · Country-level mobile phone subscriptions (per 100 people) with 15–25 year lags (1990–2015)

Evidence strengthLowConfidence: 74% · Peer reviewed: yes

Main findings

Using WHO brain cancer death rates and World Bank mobile subscription rates for 88 countries (1990–2015), the authors report that higher mobile phone subscription rates are positively and statistically significantly associated with higher brain cancer death rates about 15–20 years later. Falsification tests showed generally smaller and less consistent associations for several other cancers and ischemic heart disease, though a few positive statistically significant coefficients were observed. The reported association was described as robust to several specification tests but sensitive to inclusion/exclusion of country-specific linear time trends and to some modeling choices (e.g., weighting).

Outcomes measured

Brain cancer mortality (death rates)
Rectal cancer mortality
Pancreatic cancer mortality
Stomach cancer mortality
Breast cancer mortality
Lung cancer mortality
Ischemic heart disease mortality
All-cause/other-cause mortality comparisons (differential effects models)

Limitations

Ecological (country-level) exposure proxy (subscriptions per 100 people) may not reflect individual mobile phone use or RF exposure.
Causality cannot be established; potential omitted/confounding factors correlated with mobile adoption (e.g., CT scan use) may bias estimates.
Results are sensitive to model specification (e.g., exclusion of country-specific linear time trends; loss of significance without country weights in some models).
Long lag structure may exclude more recent adoption patterns and technology/use changes (2G/3G/4G transitions; shift toward texting/social media).
Some falsification outcomes showed a few positive statistically significant associations, raising concern about residual confounding or multiple comparisons.

View raw extracted JSON

{
  "publication_year": 2019,
  "study_type": "ecological",
  "exposure": {
    "band": "RF",
    "source": "mobile phone",
    "frequency_mhz": null,
    "sar_wkg": null,
    "duration": "Country-level mobile phone subscriptions (per 100 people) with 15–25 year lags (1990–2015)"
  },
  "population": "88 countries (country-level mortality and subscription data)",
  "sample_size": 88,
  "outcomes": [
    "Brain cancer mortality (death rates)",
    "Rectal cancer mortality",
    "Pancreatic cancer mortality",
    "Stomach cancer mortality",
    "Breast cancer mortality",
    "Lung cancer mortality",
    "Ischemic heart disease mortality",
    "All-cause/other-cause mortality comparisons (differential effects models)"
  ],
  "main_findings": "Using WHO brain cancer death rates and World Bank mobile subscription rates for 88 countries (1990–2015), the authors report that higher mobile phone subscription rates are positively and statistically significantly associated with higher brain cancer death rates about 15–20 years later. Falsification tests showed generally smaller and less consistent associations for several other cancers and ischemic heart disease, though a few positive statistically significant coefficients were observed. The reported association was described as robust to several specification tests but sensitive to inclusion/exclusion of country-specific linear time trends and to some modeling choices (e.g., weighting).",
  "effect_direction": "harm",
  "limitations": [
    "Ecological (country-level) exposure proxy (subscriptions per 100 people) may not reflect individual mobile phone use or RF exposure.",
    "Causality cannot be established; potential omitted/confounding factors correlated with mobile adoption (e.g., CT scan use) may bias estimates.",
    "Results are sensitive to model specification (e.g., exclusion of country-specific linear time trends; loss of significance without country weights in some models).",
    "Long lag structure may exclude more recent adoption patterns and technology/use changes (2G/3G/4G transitions; shift toward texting/social media).",
    "Some falsification outcomes showed a few positive statistically significant associations, raising concern about residual confounding or multiple comparisons."
  ],
  "evidence_strength": "low",
  "confidence": 0.74,
  "peer_reviewed_likely": "yes",
  "stance": "concern",
  "stance_confidence": 0.72,
  "summary": "This cross-country ecological analysis (88 countries, 1990–2015) examined whether national mobile phone subscription rates predict later brain cancer mortality. The study reports a positive, statistically significant association between subscription rates and brain cancer death rates about 15–20 years later, with smaller/less consistent associations for several other causes of death. The authors note important caveats, including inability to make causal claims and sensitivity to some model specifications.",
  "key_points": [
    "The study uses WHO brain cancer mortality data and World Bank mobile subscription rates across 88 countries from 1990 to 2015.",
    "Difference-in-differences models with country and year fixed effects were used, including time-varying country covariates.",
    "Mobile phone subscription rates were reported to be positively and statistically significantly associated with brain cancer death rates 15–20 years later.",
    "Falsification tests across other cancers and ischemic heart disease generally showed smaller coefficients and larger standard errors, though a few positive significant results occurred.",
    "Differential effects models suggested the association with brain cancer mortality was positive relative to mortality from other causes around 15–19 years later.",
    "The authors state they cannot make an unambiguous causal claim and highlight potential omitted-variable confounding (e.g., CT scan use).",
    "The exposure metric (subscriptions per 100 people) is an imperfect proxy for actual mobile phone use and RF exposure.",
    "The authors note that technology and usage changes over time (e.g., 2G to 4G, more texting) could affect future relationships."
  ],
  "categories": [
    "Epidemiology",
    "Cancer",
    "Mobile Phones",
    "RF Exposure",
    "Public Health"
  ],
  "tags": [
    "Brain Cancer Mortality",
    "Ecological Study",
    "Cross-Country Analysis",
    "Mobile Phone Subscriptions",
    "Difference-in-Differences",
    "Lagged Association",
    "Falsification Tests",
    "Fixed Effects Models",
    "Omitted Variable Bias",
    "Confounding",
    "Country-Level Data",
    "World Health Organization Data",
    "World Bank Data"
  ],
  "keywords": [
    "mobile phones",
    "cell phones",
    "brain cancer",
    "mortality",
    "subscriptions",
    "cross-country",
    "difference-in-differences",
    "fixed effects",
    "lag",
    "falsification test",
    "ecological"
  ],
  "suggested_hubs": [
    {
      "slug": "cell-phones",
      "weight": 0.95,
      "reason": "Study evaluates mobile phone (cell phone) prevalence in relation to brain cancer mortality."
    }
  ],
  "social": {
    "tweet": "Ecological study of 88 countries (1990–2015) reports higher mobile phone subscription rates are statistically associated with higher brain cancer mortality about 15–20 years later; authors note key limitations and no definitive causality.",
    "facebook": "A cross-country analysis using WHO mortality data and World Bank mobile subscription rates found a positive statistical association between national mobile phone subscriptions and brain cancer death rates about 15–20 years later. The authors emphasize limitations, including ecological exposure measurement and potential confounding.",
    "linkedin": "This 25-year cross-country ecological analysis (88 countries) reports a positive, statistically significant association between mobile phone subscription rates and brain cancer mortality with ~15–20 year lags, while noting sensitivity to some model specifications and the inability to infer causality."
  }
}

AI can be wrong. Always verify against the paper.

AI-extracted fields are generated from the abstract and metadata and may be incomplete or incorrect. This material is for informational purposes and is not medical advice.

Identifiers and source

DOI: 10.1111/coep.12456
PubMed: Not supplied

Read the original paper →

Related papers

← Back to the Research Library