Table 2

Synopsis of epidemiological studies attributed as “positive” with respect to a possible association between mobile phone use and cancer

Hardell et al (1999, 2000, 2001)Stang et al (2001)Auvinen et al (2002)Hardell et al (2002a,b)
*In addition to age and gender.
†Based on the assumption of a relative risk of 2.
TypePopulation based case-controlHospital and population based case-controlPopulation based case-controlPopulation based case-control
Endpoint(s)Brain tumoursUveal melanomaBrain tumours, salivary gland cancerBrain tumours
No. cases/controls; size of cohorts136 malignant; 62 benign tumours; 425 controls37 population cases; 81 hospital cases; 327 population controls; 148 hospital controls398 brain tumours; 34 salivary gland cancer; 2160 controls529 malignant; 774 benign tumours; 1303 controls
Exposure assessmentQuestionnaire and telephone interviewInterview, expert ratingCompany recordsQuestionnaire and telephone interview
Outcome assessmentHistopathologyReviewed by pathologistFinnish Cancer RegistryHistopathology
Telephone type(s)43% analogue; 34% digital (23% both)Not specified61% analogue; 32% digital (6% both)Overall 16% analogue, 30% digital and 28% cordless phones (combinations not specified)
Duration of follow up/duration of phone use37% users in cases, of these 44% more than 5 y5% probable/certain heavy occupational mobile phone use, 49% of these more than 5 y13% users in brain tumour cases, 12% in salivary gland cancer cases, in analogue type users 43% more than 2 y7 y av. analogue phone, 3 y av. digital phone, 5 y av. cordless phone use
Confounders considered*Occupations with increased risk, x ray examinationGeographic area, socioeconomic status, hair and eye colourUrban residence, socioeconomic status, occupationSocioeconomic status
FindingsOverall no increased OR. For ipsilateral use OR = 2.62, significant in multivariate analysisSignificant OR of 4.2 for probable/certain heavy occupational mobile phone use, increase to 4.9 if latency of 5 y is consideredSignificant OR for all brain tumours and analogue phone use (1.6) and for glioma (2.1). Significant trend for years of use. No increased risk for salivary gland cancerSignificantly increased overall OR (1.3) for analogue phones. Increase with latency to 1.8 for 10 y. OR = 2.5 for ipsilateral use of phone. Highest risk for acoustic neurinoma (OR = 3.5)
Evaluation
Selection of participantsOnly prevalent cases. Response rate: 90% cases, 91% controlsOnly prevalent cases. Response rate: 86% cases, 48% population controls, 79% hospital controlsStudy solely based on registry dataOnly prevalent cases. Response rate: 88% cases, 91% controls
Power†98% overall, 68% for lateralityOverall 48%Overall brain tumours 99%; salivary gland cancer 32%Overall 99%, 93% for laterality
Exposure assessmentInterviewer blinded to case status, not only mobile phone use but a variety of conditions investigatedOnly heavy occupational use. Different types of portable radio transmitting devices discriminated only by expert ratingsExposure misclassification possible, corporate users excluded. Due to sole use of registry data no analysis of laterality possibleInterviewer blinded to case status
Outcome assessmentState of the artState of the artNoneState of the art
Confounding and biasRecall bias possibleRecall bias unlikely. UV exposure and welding not considered as possible confounders. Bias from expert ratings possible but unlikely to affect outcomeBias due to exposure misclassification possible (especially by leaving out corporate customers). Response or recall bias impossible because subject were not contactedComprehensive analysis of possible recall and response bias (only 2 cases reported thinking of mobile phone use as a potential cause of their disease)
Latency considered1, 5, and 10 y5 yNo1, 5, and 10 y (1–6, 6 y for malignant tumours)
Statistical methodsStandardStandardStandardStandard