Abstract

The objective of this nationwide study was to assess the association between cellular phone use and development of parotid gland tumors (PGTs). The methods were based on the international INTERPHONE study that aimed to evaluate possible adverse effects of cellular phone use. The study included 402 benign and 58 malignant incident cases of PGTs diagnosed in Israel at age 18 years or more, in 2001–2003, and 1,266 population individually matched controls. For the entire group, no increased risk of PGTs was observed for ever having been a regular cellular phone user (odds ratio = 0.87; p = 0.3) or for any other measure of exposure investigated. However, analysis restricted to regular users or to conditions that may yield higher levels of exposure (e.g., heavy use in rural areas) showed consistently elevated risks. For ipsilateral use, the odds ratios in the highest category of cumulative number of calls and call time without use of hands-free devices were 1.58 (95% confidence interval: 1.11, 2.24) and 1.49 (95% confidence interval: 1.05, 2.13), respectively. The risk for contralateral use was not significantly different from 1. A positive dose-response trend was found for these measurements. Based on the largest number of benign PGT patients reported to date, our results suggest an association between cellular phone use and PGTs.

Since the mid-1990s when cellular phones became widespread in most Western countries, there has been concern about the possible carcinogenic effects of the electromagnetic radiofrequency fields thereby emitted (1, 2). Numerous studies addressing this issue have been published recently (3–14). Most focused on brain tumors that occur in an anatomic location where a substantial amount of the power is absorbed. The vast majority did not show an association between cellular phone use and the development of such tumors. Most studies, however, included few long-term users, reducing the chance of finding any association if one exists because of an assumed long latency time.

In accordance with the recommendations of several expert committees, an international series of case-control studies (known as the INTERPHONE study) on the relation between cellular phone use and several types of head tumors was conducted in 13 countries, including Israel (15). Several individual reports from this collaboration have been published, each reporting on part of the collaborative study population (8–14).

The Israeli study included meningioma, glioma, acoustic neuroma, and benign and malignant parotid gland tumors (PGTs).

The Israeli population is characterized by high levels of use as expressed by prevalence of phone use and cumulative number and duration of calls (16) (unpublished data).

The absorption of radiofrequency energy emitted by cellular phones is attenuated by more than 90 percent within 40–50 mm from the exposure source (3). The anatomic location of the parotid gland (at the anterior border of the external ear and between the mandibular ramus and the sternocleidomastoid muscle, 4–10 mm deep in the skin surface) makes these tumors plausible candidates for being influenced by exposure to cellular phones, on the side of the head where the cellular phone is held. This tumor occurs in the relatively young (mean age at diagnosis: 43 years for benign and 55 years for malignant tumors) (17), an age group likely to include a substantial number of long-term cellular phone users (15).

This report, from a nationwide population-based study, presents results of analyses of risk of benign and malignant PGTs in relation to cellular phone use.

MATERIALS AND METHODS

The Israeli study followed the core protocol of the INTERPHONE study. Additional national funding was allocated to extend the age range in the Israeli component from 30 to 59 years to include all persons aged 18 years or more. All 22 otolaryngology departments throughout the country participated in the study. Eligible cases were adult Jewish individuals with histologically or cytologically confirmed benign or malignant PGT, diagnosed between January 2001 and December 2003. Cases were identified through a periodic review of pathology/cytology reports of all relevant institutions. All diagnoses were validated by a single physician (S. D.).

Controls for the entire Israeli INTERPHONE study were randomly selected from the National Population Registry and were individually matched to the original case (meningioma/glioma/acoustic neuroma/PGT). Controls were not eligible if they had left the country or were mentally disabled. To increase the statistical power of the PGT comparison, we tried to include all the participating controls in the analysis by a post hoc matching. We individually assigned up to seven controls to each case, using a hierarchical algorithm that included the matching variables in the following order: gender, interview date (preferably within 1 year), age (preferably within 1 year and up to 5 years), and continent of birth.

Data were obtained by a personal interview including information on demographic variables, cellular phone use, and other possible risk factors. Participants were asked whether they were “regular users” of a cellular phone, defined as making or receiving more than one call per week for at least 6 months. Regular users were asked to identify in an album all of the cellular phones they had ever used and to recall their history of use of each phone, namely, dates of starting and stopping use, number of calls made or received, average duration of calls, changes in the pattern of use over a period of more than 6 months, use of headsets and hands-free devices in vehicles, the side of the head on which the phone was predominantly held, handedness, and main area of use (urban/rural/both). Exposures were considered up to 1 year prior to the reference date, which was defined as the date of diagnosis for cases and as the same date as their matched case for controls.

To investigate the possibility that response rates were influenced by cellular phone use, we conducted a short telephone interview of cellular phone use with subjects refusing to participate in the study.

Medical records were used to define the clinical characteristics of the tumor. The study was approved by the ethics committees of all hospitals and by the Clinical Trial Unit, Pharmaceutical Department, Ministry of Health. Informed consent was obtained from all participants.

For descriptive purposes pertaining to statistical analysis, distributions among controls were weighted by the inverse number of controls in matched sets, because the number of controls per case varied with some of the risk factors, particularly age. Such weighting effectively produces the distribution among controls that would have been seen if each case had the same number of controls. Case-control comparisons were performed by use of the weighted chi-square test.

The main analyses were based on conditional logistic regression; adjustment for smoking was also performed. Because of the possibility of a consistent downward bias in odds ratios caused by selective participation on the basis of cellular phone use, we repeated the analyses among regular users using the low-exposure level as reference.

Cumulative use was analyzed with modification for reported use of hands-free devices as follows: It was reduced by 100, 75, 50, and 25 percent when participants reported use of such devices all of, most of, half of, or less than half of the time, respectively.

“Ipsilateral” exposure was defined as phone use on the same side as the tumor, “contralateral” as phone use on the opposite side, and “both” as no preferred side of phone use or presence of tumor on both sides. The association between laterality of tumor and side of cellular phone use was assessed by two approaches:

  • A case-only analysis (Inskip et al. (4)), including only cases who were regular users reporting on a single dominant side of use. The relative risk was estimated as (1 + the square root of the odds ratio)/2.

  • Case-control analysis stratified to ipsilateral and contralateral use in which laterality for controls was defined as the anatomic side of the tumor of their matched case. Exposure was defined as “ipsilateral” or “both” for the ipsilateral analysis and as “contralateral” only for the contralateral analysis; “nonregular users” were considered as the reference category.

To avoid exclusion of large numbers of matched sets, we based laterality analyses on unconditional logistic regression adjusted for age, gender, and year of interview. Results with conditional and unconditional logistic regression were similar.

For trend tests, categorical variables were assigned scores of 0 and 1–4 (nonregular users and quartiles) according to their ordered categories and were treated in the model as continuous variables.

RESULTS

Study population

During the study period, 531 eligible cases were identified. Of them, 87 percent agreed to participate, 11 percent refused, and 2 percent could not be traced (table 1). Response rates did not differ significantly by gender; by place of birth; among recent (<10 years) immigrants, earlier (≥10 years) immigrants, and the Israeli born; or by tumor classification. Participants tended to be younger than nonparticipants (p = 0.003), and a larger proportion of the recent immigrants could not be traced (p = 0.001).

TABLE 1.

Comparison of selected variables between participants and nonparticipants by study group and reason for nonparticipation, Israel, 2001–2003

 Cases Controls 
 No. eligible (n = 531) Participants (n = 460) Refused (n = 60) Unable to be traced* (n = 11) p value† No. eligible (n = 1,920) Participants (n = 1,266) Refused (n = 454) Unable to be traced (n = 200) p value† 
Total (%)  86.6 11.3 2.1   65.9 23.7 10.4  
Gender (%)           
    Males 288 88.2 10.1 1.7 0.3 806 68.4 20.7 10.9 0.06 
    Females 243 84.8 12.8 2.5  1,114 64.2 25.8 10.1  
Mean age (years) at recruitment  53.3 (16.6)‡ 59.8 (18.0) 58.7 (13.9) <0.01  59.3 (14.9) 59.0 (14.1) 60.8 (15.5) 0.7 
Place of birth (%)           
    Asia 71 85.9 14.1  0.2 260 63.1 30.0 6.9 0.03 
    North Africa 64 93.8 6.3   252 67.1 24.6 8.3  
    Europe-United States 178 83.2 13.5 3.4  877 63.5 21.0 15.5  
    Israel 218 87.6 10.1 2.3  531 70.8 24.5 4.7  
Time from immigration to first contact (%)§           
    <10 years 43 79.1 11.6 9.3¶ 0.3 56 51.8 19.6 28.6 <0.01 
    ≥10 years 270 87.0 12.2 0.7  1,326 64.9 23.2 11.9  
    Born in Israel 218 87.6 10.1 2.3  531 70.8 24.5 4.7  
Tumor classification (%)           
    Benign 462 87.0 11.0 1.9 0.6      
    Malignant 69 84.1 13.0 2.9       
 Cases Controls 
 No. eligible (n = 531) Participants (n = 460) Refused (n = 60) Unable to be traced* (n = 11) p value† No. eligible (n = 1,920) Participants (n = 1,266) Refused (n = 454) Unable to be traced (n = 200) p value† 
Total (%)  86.6 11.3 2.1   65.9 23.7 10.4  
Gender (%)           
    Males 288 88.2 10.1 1.7 0.3 806 68.4 20.7 10.9 0.06 
    Females 243 84.8 12.8 2.5  1,114 64.2 25.8 10.1  
Mean age (years) at recruitment  53.3 (16.6)‡ 59.8 (18.0) 58.7 (13.9) <0.01  59.3 (14.9) 59.0 (14.1) 60.8 (15.5) 0.7 
Place of birth (%)           
    Asia 71 85.9 14.1  0.2 260 63.1 30.0 6.9 0.03 
    North Africa 64 93.8 6.3   252 67.1 24.6 8.3  
    Europe-United States 178 83.2 13.5 3.4  877 63.5 21.0 15.5  
    Israel 218 87.6 10.1 2.3  531 70.8 24.5 4.7  
Time from immigration to first contact (%)§           
    <10 years 43 79.1 11.6 9.3¶ 0.3 56 51.8 19.6 28.6 <0.01 
    ≥10 years 270 87.0 12.2 0.7  1,326 64.9 23.2 11.9  
    Born in Israel 218 87.6 10.1 2.3  531 70.8 24.5 4.7  
Tumor classification (%)           
    Benign 462 87.0 11.0 1.9 0.6      
    Malignant 69 84.1 13.0 2.9       
*

Including one case who was not interviewed because of language difficulties.

p value for comparison of participants and nonparticipants.

Numbers in parentheses, standard deviation.

§

Seven controls with missing data.

Untraced versus traced: p = 0.001.

In total, 1,920 eligible controls were identified; 66 percent were interviewed, 24 percent refused, and 10 percent could not be traced. Age at recruitment was similar in participants and nonparticipants, the response rate was influenced by time since immigration, and it was higher in males than females (p = 0.06) and in the Israeli born compared with the non-Israeli born (p = 0.01) (table 1).

All cases had at least one matched control; 75 percent and 42 percent had at least two and three controls, respectively; 87 percent of the controls were matched to cases within 2 years of age. The time difference between the interviews of cases and controls was within 1 year for 98 percent. Proxy interviews (spouse or offspring) were conducted for 18 cases and eight controls, and telephone interviews were conducted for 19 cases and 49 controls.

The participants' cooperation and apparent recall of cellular phone use were classified by their interviewers as very or fairly good for 95 percent of the interviewees, with no difference between cases and controls.

A short telephone interview was completed by 78 percent of the refusal controls. A significantly lower rate of regular cellular phone use was reported by the refusal control group compared with the participant group (p < 0.001). Among regular users, 57 percent of the refusal controls reported starting cellular phone use before 2001, compared with 80 percent in the participant controls (p < 0.001). No significant differences in regular cellular phone use were observed between the cases who participated in the study and those who refused but participated in the refusal questionnaire (66 percent of the 60 refusal cases).

Our analysis included 460 cases (58 malignant, 264 pleomorphic adenoma, 117 Warthin's tumor, and 21 others) and 1,266 controls (table 2). The male/female ratio among cases was 1.2, and their mean age at diagnosis was about 52 (range: 18–98) years. About 40 percent of the cases were Israeli born, more than 70 percent were married, 14 percent had only a primary school level of education, and 30 percent were university or technical school graduates. There were no significant differences in the distribution of these variables between cases and controls. Smoking was significantly more prevalent among cases compared with controls (p = 0.001).

TABLE 2.

Distribution of demographic characteristics by study group, Israel, 2001–2003

 Cases (n = 460) Controls (n = 1,266) Weighted* controls (%) p value† 
 No. No. 
Gender       
    Males 254 55.2 551 43.5 55.2 1.0 
    Females 206 44.8 715 56.5 44.8  
Age (years) at reference date       
    <30 56 12.2 69 5.5 12.6 1.0 
    30–39 60 13.0 88 7.0 12.0  
    40–49 88 19.1 201 15.9 19.1  
    50–59 113 24.6 350 27.6 24.8  
    60–69 67 14.6 256 20.2 14.8  
    ≥70 76 16.5 302 23.8 16.7  
Origin       
    Asia 60 13.0 164 13.0 12.8 0.9 
    North Africa 60 13.0 169 13.4 13.0  
    Europe-United States 149 32.4 557 44.0 32.7  
    Israel 191 41.5 376 29.7 41.5  
Marital status       
    Married 332 72.2 912 72.0 73.4 0.8 
    Single 52 11.3 63 5.0 9.3  
    Divorced/separated 35 7.6 100 7.9 7.8  
    Widowed 40 8.7 184 14.5 8.7  
    Unknown 0.2 0.6 0.8  
Education       
    Primary school 66 14.4 177 14.0 11.5 0.4 
    Secondary/high school 164 35.7 425 33.6 33.6  
    Medium-level school‡ 87 18.9 249 19.6 20.4  
    University/high-level technical school 140 30.4 408 32.2 33.8  
    Unknown 0.6 0.6 0.8  
Cigarette smoking       
    No 183 39.8 680 53.7 50.6 0.001 
    Yes 286 60.0 586 46.3 49.4  
    Unknown 0.2     
 Cases (n = 460) Controls (n = 1,266) Weighted* controls (%) p value† 
 No. No. 
Gender       
    Males 254 55.2 551 43.5 55.2 1.0 
    Females 206 44.8 715 56.5 44.8  
Age (years) at reference date       
    <30 56 12.2 69 5.5 12.6 1.0 
    30–39 60 13.0 88 7.0 12.0  
    40–49 88 19.1 201 15.9 19.1  
    50–59 113 24.6 350 27.6 24.8  
    60–69 67 14.6 256 20.2 14.8  
    ≥70 76 16.5 302 23.8 16.7  
Origin       
    Asia 60 13.0 164 13.0 12.8 0.9 
    North Africa 60 13.0 169 13.4 13.0  
    Europe-United States 149 32.4 557 44.0 32.7  
    Israel 191 41.5 376 29.7 41.5  
Marital status       
    Married 332 72.2 912 72.0 73.4 0.8 
    Single 52 11.3 63 5.0 9.3  
    Divorced/separated 35 7.6 100 7.9 7.8  
    Widowed 40 8.7 184 14.5 8.7  
    Unknown 0.2 0.6 0.8  
Education       
    Primary school 66 14.4 177 14.0 11.5 0.4 
    Secondary/high school 164 35.7 425 33.6 33.6  
    Medium-level school‡ 87 18.9 249 19.6 20.4  
    University/high-level technical school 140 30.4 408 32.2 33.8  
    Unknown 0.6 0.6 0.8  
Cigarette smoking       
    No 183 39.8 680 53.7 50.6 0.001 
    Yes 286 60.0 586 46.3 49.4  
    Unknown 0.2     
*

Weighted (inversely) to the number of controls for each case.

Cases versus weighted controls.

Technical/professional/“Yeshiva” graduate school.

Cellular phone analysis

The weighted proportions of regular users among controls were 64 percent, 30 percent, and 3 percent at 1, 5, and 10 years before the reference date, respectively (table 3). No increased risk of PGT was seen for any of the exposure measures tested (regular use, time since start, duration of use, cumulative number of calls, and cumulative call time) for the total group, the malignant group, and the benign group. These results were substantially unchanged after controlling for smoking.

TABLE 3.

Levels of cellular phone use and risk of parotid gland tumors by tumor type, Israel, 2001–2003

 Total Benign Malignant 
 Cases (no.) Controls (no.) Odds ratio* 95% confidence interval Cases (no.) Controls (no.) Odds ratio* 95% confidence interval Cases (no.) Controls (no.) Odds ratio* 95% confidence interval 
Regular user             
    No, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    Yes 285 691 0.87 0.68, 1.13 252 603 0.85 0.64, 1.12 33 88 1.06 0.54, 2.10 
Regular user 5 years in the past             
    No 319 969 1.0†  273 808 1.0†  46 161 1.0†  
    Yes 141 297 1.01 0.77, 1.32 129 264 1.04 0.78, 1.38 12 33 0.79 0.36, 1.76 
Regular user 10 years in the past             
    No 447 1,240 1.0†  390 1,050 1.0†  57 190 1.0†  
    Yes 13 26 0.93 0.47, 1.86 12 22 1.02 0.49, 2.12 0.45 0.05, 4.28 
Time since start use (years)             
    Never, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    1–4.9 138 389 0.82 0.61, 1.10 117 335 0.77 0.56, 1.06 21 54 1.25 0.58, 2.68 
    5–9.9 134 276 0.95 0.70, 1.30 123 246 0.95 0.68, 1.32 11 30 0.92 0.37, 2.27 
    ≥10 13 26 0.86 0.42, 1.77 12 22 0.93 0.44, 1.98 0.47 0.05, 4.51 
Duration of use (years)             
    Never, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    1–4.9 148 405 0.84 0.63, 1.12 127 351 0.79 0.54, 1.08 21 54 1.25 0.58, 2.68 
    5–9.9 124 264 0.92 0.67, 1.27 113 234 0.92 0.65, 1.29 11 30 0.92 0.37, 2.27 
    ≥10 13 22 1.0 0.48, 2.09 12 18 1.11 0.50, 2.44 0.47 0.05, 4.51 
Cumulative no. of calls (with no hands-free devices)‡             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    ≤5,479 117 382 0.79 0.59, 1.05 99 328 0.74 0.54, 1.02 18 54 1.12 0.53, 2.35 
    5,480–18,996 86 157 1.13 0.79, 1.61 77 141 1.10 0.76, 1.61 16 1.33 0.48, 3.71 
    ≥18,997 81 140 1.06 0.73, 1.54 75 123 1.09 0.74, 1.62 17 0.72 0.22, 2.39 
Cumulative call time (hours)(with no hands-free devices)‡             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    ≤266.3 121 390 0.82 0.62, 1.09 103 336 0.78 0.57, 1.06 18 54 1.21 0.58, 2.53 
    266.4–1,034.9 80 155 1.03 0.72, 1.47 75 139 1.05 0.72, 1.53 16 0.67 0.19, 2.38 
    ≥1,035 83 134 1.09 0.75, 1.60 73 117 1.08 0.72, 1.62 10 17 1.22 0.43, 3.48 
Cumulative call time (hours) (with no hands-free devices)‡ by time since start use (years)             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    <5 years             
     ≤266.3 91 294 0.82 0.60, 1.12 76 248 0.77 0.55, 1.09 15 46 1.17 0.53, 2.56 
     >266.3 47 88 1.03 0.66, 1.62 41 81 0.96 0.60, 1.54 2.03 0.47, 8.80 
    ≥5 years             
        ≤266.3 30 96 0.83 0.52, 1.31 27 88 0.79 0.48, 1.27 1.36 0.31, 5.92 
        >266.3 116 201 1.07 0.76, 1.50 107 175 1.11 0.77, 1.59 10 26 0.77 0.29, 2.06 
Cumulative no. of calls (with no hands-free devices)‡ by time since start use (years)             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    <5 years             
        ≤5,479 91 300 0.79 0.57, 1.08 75 252 0.73 0.52, 1.03 16 48 1.19 0.54, 2.60 
        >5,479 47 82 1.16 0.74, 1.82 42 77 1.09 0.68, 1.74 2.26 0.49, 10.40 
    ≥5 years             
        ≤5,479 26 82 0.79 0.48, 1.31 24 76 0.77 0.45, 1.30 1.06 0.20, 5.57 
        >5,479 120 215 1.08 0.77, 1.50 110 187 1.10 0.78, 1.57 10 28 0.85 0.33, 2.22 
 Total Benign Malignant 
 Cases (no.) Controls (no.) Odds ratio* 95% confidence interval Cases (no.) Controls (no.) Odds ratio* 95% confidence interval Cases (no.) Controls (no.) Odds ratio* 95% confidence interval 
Regular user             
    No, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    Yes 285 691 0.87 0.68, 1.13 252 603 0.85 0.64, 1.12 33 88 1.06 0.54, 2.10 
Regular user 5 years in the past             
    No 319 969 1.0†  273 808 1.0†  46 161 1.0†  
    Yes 141 297 1.01 0.77, 1.32 129 264 1.04 0.78, 1.38 12 33 0.79 0.36, 1.76 
Regular user 10 years in the past             
    No 447 1,240 1.0†  390 1,050 1.0†  57 190 1.0†  
    Yes 13 26 0.93 0.47, 1.86 12 22 1.02 0.49, 2.12 0.45 0.05, 4.28 
Time since start use (years)             
    Never, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    1–4.9 138 389 0.82 0.61, 1.10 117 335 0.77 0.56, 1.06 21 54 1.25 0.58, 2.68 
    5–9.9 134 276 0.95 0.70, 1.30 123 246 0.95 0.68, 1.32 11 30 0.92 0.37, 2.27 
    ≥10 13 26 0.86 0.42, 1.77 12 22 0.93 0.44, 1.98 0.47 0.05, 4.51 
Duration of use (years)             
    Never, <1 year 175 575 1.0†  150 469 1.0†  25 106 1.0†  
    1–4.9 148 405 0.84 0.63, 1.12 127 351 0.79 0.54, 1.08 21 54 1.25 0.58, 2.68 
    5–9.9 124 264 0.92 0.67, 1.27 113 234 0.92 0.65, 1.29 11 30 0.92 0.37, 2.27 
    ≥10 13 22 1.0 0.48, 2.09 12 18 1.11 0.50, 2.44 0.47 0.05, 4.51 
Cumulative no. of calls (with no hands-free devices)‡             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    ≤5,479 117 382 0.79 0.59, 1.05 99 328 0.74 0.54, 1.02 18 54 1.12 0.53, 2.35 
    5,480–18,996 86 157 1.13 0.79, 1.61 77 141 1.10 0.76, 1.61 16 1.33 0.48, 3.71 
    ≥18,997 81 140 1.06 0.73, 1.54 75 123 1.09 0.74, 1.62 17 0.72 0.22, 2.39 
Cumulative call time (hours)(with no hands-free devices)‡             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    ≤266.3 121 390 0.82 0.62, 1.09 103 336 0.78 0.57, 1.06 18 54 1.21 0.58, 2.53 
    266.4–1,034.9 80 155 1.03 0.72, 1.47 75 139 1.05 0.72, 1.53 16 0.67 0.19, 2.38 
    ≥1,035 83 134 1.09 0.75, 1.60 73 117 1.08 0.72, 1.62 10 17 1.22 0.43, 3.48 
Cumulative call time (hours) (with no hands-free devices)‡ by time since start use (years)             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    <5 years             
     ≤266.3 91 294 0.82 0.60, 1.12 76 248 0.77 0.55, 1.09 15 46 1.17 0.53, 2.56 
     >266.3 47 88 1.03 0.66, 1.62 41 81 0.96 0.60, 1.54 2.03 0.47, 8.80 
    ≥5 years             
        ≤266.3 30 96 0.83 0.52, 1.31 27 88 0.79 0.48, 1.27 1.36 0.31, 5.92 
        >266.3 116 201 1.07 0.76, 1.50 107 175 1.11 0.77, 1.59 10 26 0.77 0.29, 2.06 
Cumulative no. of calls (with no hands-free devices)‡ by time since start use (years)             
    Nonusers, <1 year 176 587 1.0†  151 480 1.0†  25 107 1.0†  
    <5 years             
        ≤5,479 91 300 0.79 0.57, 1.08 75 252 0.73 0.52, 1.03 16 48 1.19 0.54, 2.60 
        >5,479 47 82 1.16 0.74, 1.82 42 77 1.09 0.68, 1.74 2.26 0.49, 10.40 
    ≥5 years             
        ≤5,479 26 82 0.79 0.48, 1.31 24 76 0.77 0.45, 1.30 1.06 0.20, 5.57 
        >5,479 120 215 1.08 0.77, 1.50 110 187 1.10 0.78, 1.57 10 28 0.85 0.33, 2.22 
*

The odds ratios were substantially unchanged after controlling for cigarette smoking; therefore, the unadjusted odds ratios are presented here.

Referent.

Based on the distribution of users' controls (weighted) and divided into the following categories: ≤median, >median–≤third quartile, and >third quartile.

When analyses were restricted to regular users only, increased odds ratios were seen for start of use 5 years or more in the past, as well as for cumulative number of calls and cumulative call time (Appendix table 1).

Among the 284 cases who were regular users, 155 (54.6 percent) reported ipsilateral use, 101 (35.6 percent) contralateral use, and 27 (9.5 percent) use on both sides (table 4). Based on Inskip's method, the relative risk for ipsilateral compared with contralateral use was 1.32 (p = 0.001).

TABLE 4.

Side of cellular phone use and side of the tumor for regular users among the study cases, Israel, 2001–2003*

Side of tumor Side of regular cellular phone use† 
Left Right Both 
Left 44 78 10 
Right 23 111 16 
Left and right 
    Total 67 190 27 
Side of tumor Side of regular cellular phone use† 
Left Right Both 
Left 44 78 10 
Right 23 111 16 
Left and right 
    Total 67 190 27 
*

Inskip test: odds ratio (OR) = 2.72, 95% confidence interval: 1.47, 5.08; relative risk = 1.32; p = 0.001. Note: the relative risk was calculated as (1 + √OR)/2.

One user with missing data.

The proportions of cases and controls whose handedness corresponded to their side of phone use were similar (69 percent and 70 percent, respectively) (p = 0.86) (table 5).

TABLE 5.

Handedness and side of cellular phone use among regular users by study group, Israel, 2001–2003*

 Cases Controls 
 No. No. % (weighted) 
Same side 197 69.4 470 69.7 
Opposite side 57 20.0 132 18.4 
No preferred side 27 9.5 64 10.2 
Ambidextrous 1.0 13 1.8 
 Cases Controls 
 No. No. % (weighted) 
Same side 197 69.4 470 69.7 
Opposite side 57 20.0 132 18.4 
No preferred side 27 9.5 64 10.2 
Ambidextrous 1.0 13 1.8 
*

p = 0.86 for comparison of cases and weighted controls.

Laterality analyses are presented in table 6. Elevated odds ratios were found for ipsilateral use 5 years (odds ratio (OR) = 1.34; p = 0.06) and 10 years (OR = 1.69; p = 0.2) in the past.

TABLE 6.

Risk of parotid tumor for different levels of cellular phone use by ipsilateral and contralateral use, Israel, 2001–2003*

 Ipsilateral and both Contralateral 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No, <1 year 175 575 1.0‡  175 575 1.0‡  
    Yes 183 390 1.01 0.75, 1.35 101 294 0.87 0.63, 1.20 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 96 166 1.34 0.99, 1.83 45 128 0.89 0.61, 1.30 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240 1.0‡  
    Yes 10 16 1.69 0.74, 3.83 10 0.59 0.15, 2.26 
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 84 220 0.88 0.63, 1.24 53 166 0.82 0.56, 1.21 
    5–9.9 89 154 1.14 0.79, 1.65 45 118 0.96 0.63, 1.46 
    ≥10 10 16 1.60 0.68, 3.72 10 0.58 0.15, 2.32 
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 90 229 0.88 0.64, 1.24 57 171 0.86 0.59, 1.26 
    5–9.9 83 148 1.13 0.78, 1.64 41 114 0.91 0.59, 1.40 
    ≥10 10 13 1.89 0.79, 4.57 0.61 0.15, 2.47 
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    ≤5,479 61 223 0.72 0.51, 1.03 55 156 0.98 0.67, 1.43 
    >5,479 121 159 1.58 1.11, 2.24 46 135 0.78 0.51, 1.19 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    ≤266.3 67 224 0.79 0.56, 1.11 53 162 0.92 0.63, 1.34 
    >266.3 115 158 1.49 1.05, 2.13 48 129 0.84 0.55, 1.28 
Cumulative call time (hours) (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    <5 years         
        ≤266.3 50 171 0.77 0.52, 1.13 40 121 0.91 0.59, 1.39 
        >266.3 34 45 1.56 0.91, 2.67 13 42 0.67 0.33, 1.35 
    ≥5 years         
        ≤266.3 17 53 0.84 0.46, 1.53 13 41 0.95 0.49, 1.85 
        >266.3 81 113 1.47 0.99, 2.17 35 87 0.92 0.57, 1.47 
Cumulative no. of calls (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    <5 years         
        ≤5,479 49 176 0.73 0.50, 1.08 41 122 0.91 0.60, 1.40 
        >5,479 35 40 1.80 1.05, 3.10 12 41 0.63 0.31, 1.30 
    ≥5 years         
        ≤5,479 12 47 0.69 0.35, 1.36 14 34 1.19 0.61, 2.32 
        >5,479 86 119 1.50 1.03, 2.20 34 94 0.84 0.52, 1.34 
 Ipsilateral and both Contralateral 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No, <1 year 175 575 1.0‡  175 575 1.0‡  
    Yes 183 390 1.01 0.75, 1.35 101 294 0.87 0.63, 1.20 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 96 166 1.34 0.99, 1.83 45 128 0.89 0.61, 1.30 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240 1.0‡  
    Yes 10 16 1.69 0.74, 3.83 10 0.59 0.15, 2.26 
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 84 220 0.88 0.63, 1.24 53 166 0.82 0.56, 1.21 
    5–9.9 89 154 1.14 0.79, 1.65 45 118 0.96 0.63, 1.46 
    ≥10 10 16 1.60 0.68, 3.72 10 0.58 0.15, 2.32 
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 90 229 0.88 0.64, 1.24 57 171 0.86 0.59, 1.26 
    5–9.9 83 148 1.13 0.78, 1.64 41 114 0.91 0.59, 1.40 
    ≥10 10 13 1.89 0.79, 4.57 0.61 0.15, 2.47 
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    ≤5,479 61 223 0.72 0.51, 1.03 55 156 0.98 0.67, 1.43 
    >5,479 121 159 1.58 1.11, 2.24 46 135 0.78 0.51, 1.19 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    ≤266.3 67 224 0.79 0.56, 1.11 53 162 0.92 0.63, 1.34 
    >266.3 115 158 1.49 1.05, 2.13 48 129 0.84 0.55, 1.28 
Cumulative call time (hours) (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    <5 years         
        ≤266.3 50 171 0.77 0.52, 1.13 40 121 0.91 0.59, 1.39 
        >266.3 34 45 1.56 0.91, 2.67 13 42 0.67 0.33, 1.35 
    ≥5 years         
        ≤266.3 17 53 0.84 0.46, 1.53 13 41 0.95 0.49, 1.85 
        >266.3 81 113 1.47 0.99, 2.17 35 87 0.92 0.57, 1.47 
Cumulative no. of calls (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 583 1.0‡  175 578 1.0‡  
    <5 years         
        ≤5,479 49 176 0.73 0.50, 1.08 41 122 0.91 0.60, 1.40 
        >5,479 35 40 1.80 1.05, 3.10 12 41 0.63 0.31, 1.30 
    ≥5 years         
        ≤5,479 12 47 0.69 0.35, 1.36 14 34 1.19 0.61, 2.32 
        >5,479 86 119 1.50 1.03, 2.20 34 94 0.84 0.52, 1.34 
*

Unconditional logistic regression.

Adjusted for age, sex, and year of interview; adjustment for ethnic origin did not influence the results and was not included in the final statistical model.

Referent.

§

Based on the distribution of users' controls (weighted) and divided into ≤median and >median.

For all exposure measures studied, the odds ratios for above-median ipsilateral use were elevated and statistically significant: 1.58 (95 percent confidence interval (CI): 1.11, 2.24) for cumulative number of calls and 1.49 (95 percent CI: 1.05, 2.13) for cumulative call time. Significant trends were seen for ipsilateral use for cumulative number of calls (p = 0.02) and for cumulative call time (p = 0.03). For contralateral use, odds ratios were less than 1.0 but were not statistically significant. Both the ipsilateral group and the “both side users” group showed elevated odds ratios for most cellular phone use parameters in the highest categories of use (Appendix table 2). The odds ratios for the ipsilateral analysis remained elevated after stratifying for malignant and benign tumors, as well as when stratifying for pleomorphic adenoma and Warthin's tumor. However, statistical significance was reached only for the largest group, that is, all benign tumors. In this group, the odds ratios for above-median ipsilateral use for cumulative number of calls and cumulative call time were 1.52 (95 percent CI: 1.05, 2.20) and 1.44 (95 percent CI: 0.99, 2.09), respectively (not shown).

Among the 690 regular users' controls, 505 (73 percent) reported their use mainly in urban areas, 43 (6 percent) mainly in rural areas, and 142 (21 percent) in both (table 7). No statistical differences were noted in the distribution of areas of use between cases and controls after weighting (p = 0.7). For cellular phone use in rural and both areas, increased odds ratios were found for most measures of exposure. The odds ratios for the highest quartile of cumulative number of calls and cumulative call time were 1.81 (95 percent CI: 1.04, 3.14) and 1.96 (95 percent CI: 1.11, 3.44) (ptrend = 0.06 and 0.04), respectively. This pattern was not seen for mainly urban cellular phone use.

TABLE 7.

Risk of parotid tumor for different levels of cellular phone use by main area of use, Israel, 2001–2003*

 Mainly rural or mixed rural/urban Mainly urban 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No, <1 year 175 575 1.0‡  175 575 1.0‡  
    Yes 88 185 1.12 0.79, 1.61 197 505 0.90 0.68, 1.18 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 52 90 1.43 0.97, 2.11 89 206 1.01 0.75, 1.37 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240 1.0‡  
    Yes 1.37 0.40, 4.68 18 1.16 0.50, 2.67 
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 34 95 0.84 0.53, 1.35 104 294 0.86 0.63, 1.18 
    5–9.9 50 82 1.44 0.92, 2.25 84 193 0.93 0.66, 1.32 
    ≥10 1.39 0.40, 4.84 18 1.14 0.48, 2.70 
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 37 100 0.87 0.55, 1.38 111 305 0.87 0.64, 1.19 
    5–9.9 47 78 1.43 0.91, 2.26 77 185 0.91 0.63, 1.30 
    ≥10 1.52 0.42, 5.44 15 1.31 0.54, 3.18 
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 579 1.0‡  175 583 1.0‡  
    ≤5,479 33 94 0.89 0.56, 1.42 84 288 0.81 0.59, 1.11 
    5,480–18,996 22 46 1.16 0.64, 2.08 64 110 1.24 0.83, 1.85 
    ≥18,997 32 41 1.81 1.04, 3.14 49 99 1.0 0.65, 1.55 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 579 1.0‡  175 583 1.0‡  
    ≤266.3 28 100 0.72 0.44, 1.17 93 289 0.88 0.65, 1.21 
    266.4–1,034.9 27 43 1.57 0.90, 2.74 53 112 1.0 0.66, 1.51 
    ≥1,035 32 38 1.96 1.11, 3.44 51 96 1.02 0.67, 1.58 
 Mainly rural or mixed rural/urban Mainly urban 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No, <1 year 175 575 1.0‡  175 575 1.0‡  
    Yes 88 185 1.12 0.79, 1.61 197 505 0.90 0.68, 1.18 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 52 90 1.43 0.97, 2.11 89 206 1.01 0.75, 1.37 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240 1.0‡  
    Yes 1.37 0.40, 4.68 18 1.16 0.50, 2.67 
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 34 95 0.84 0.53, 1.35 104 294 0.86 0.63, 1.18 
    5–9.9 50 82 1.44 0.92, 2.25 84 193 0.93 0.66, 1.32 
    ≥10 1.39 0.40, 4.84 18 1.14 0.48, 2.70 
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 37 100 0.87 0.55, 1.38 111 305 0.87 0.64, 1.19 
    5–9.9 47 78 1.43 0.91, 2.26 77 185 0.91 0.63, 1.30 
    ≥10 1.52 0.42, 5.44 15 1.31 0.54, 3.18 
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 579 1.0‡  175 583 1.0‡  
    ≤5,479 33 94 0.89 0.56, 1.42 84 288 0.81 0.59, 1.11 
    5,480–18,996 22 46 1.16 0.64, 2.08 64 110 1.24 0.83, 1.85 
    ≥18,997 32 41 1.81 1.04, 3.14 49 99 1.0 0.65, 1.55 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 579 1.0‡  175 583 1.0‡  
    ≤266.3 28 100 0.72 0.44, 1.17 93 289 0.88 0.65, 1.21 
    266.4–1,034.9 27 43 1.57 0.90, 2.74 53 112 1.0 0.66, 1.51 
    ≥1,035 32 38 1.96 1.11, 3.44 51 96 1.02 0.67, 1.58 
*

Unconditional logistic regression.

Adjusted for age, sex, and year of interview; adjustment for ethnic origin did not influence the results and was not included in the final statistical model.

Referent.

§

Based on the distribution of users' controls (weighted) and divided into the following categories: ≤median, >median–≤third quartile, and >third quartile.

DISCUSSION

Our results suggest a relation between long-term and heavy cellular phone use and PGTs. This association was seen in analyses restricted to regular users, analyses of laterality of phone use, and analyses of area of main use.

Increased risk estimates were found for ipsilateral regular use 5 and 10 years in the past, although the latter was based on small numbers. Significantly elevated odds ratios were observed consistently in the highest category of each of the measures of cellular phone use on the ipsilateral side, supporting a dose-response association. This association between side of use and PGTs was also seen by a case-only analysis. In addition, a positive association was seen for cellular phone use in rural areas, which was not shown for use mainly in urban areas.

The association between cellular phone use and PGTs has been studied in Finland, Sweden, and Denmark. One case-control study (6) and one cohort study (5) used cellular phone subscription as a measure of exposure. No association was found for salivary gland cancers in either study. Apart from the small sample size (34 cases) and a short latent period in the case-control study, major limitations of these studies were the misclassification between subscribers and users and the limited information about exposure. A case-control study conducted in several areas of Sweden during 1994–2000, comprising 267 cases (199 PGTs) and 1,053 controls, yielded odds ratios close to 1 (18). No association was found with tumor side or latent period. As only six cases had a latency of more than 10 years, no conclusion about long-term use could be drawn.

Recently, a combined analysis of PGTs from the Swedish and Danish components of the INTERPHONE study, including 60 malignant and 112 benign PGTs, was published (14). The estimated odds ratios for any regular use were 0.7 and 0.9, respectively. No association was seen with time since first use, duration of use, cumulative number of calls, or call time. Only two persons with malignant and seven individuals with benign PGTs had used cellular phones for 10 years or more. For benign tumors and ipsilateral exposure, however, the odds ratio for regular use was 1.4 (95 percent CI: 0.9, 2.2) and, for time since first use more than 10 years ago, it was 2.6 (95 percent CI: 0.9, 7.9). The odds ratio for regular use on the contralateral side was 0.7 (95 percent CI: 0.4, 1.1).

An advantage of studying patients with PGTs is that, unlike those with brain tumors, they generally do not suffer from cognitive dysfunction that would compromise the quality of the reported data. Indeed, nearly all (95 percent) cases were classified by their interviewers as very or fairly cooperative, and proxy interviews were needed for only 18 cases. Furthermore, unlike with acoustic neuroma, hearing loss, which could induce switching use of the phone to the opposite side, is not a symptom of this disease.

Our active procedure for ascertaining cases allowed us to assemble a population-based group of patients with incident malignant and benign PGTs. Moreover, our study is the first to investigate a group of benign Warthin's tumors.

The controls in our study were population based; however, while compliance among cases was rather high (87 percent), only 65 percent of the eligible controls were interviewed.

A bias that controls who were users preferentially self-select themselves as participants is an important possibility given that the study was presented as a “cellular phone study.” Indeed, nonparticipants who responded to the short telephone questionnaire (78 percent of refusals) reported a significantly lower rate of regular use than did control participants and, among regular users, a later start of use. This phenomenon was also seen in other components of the INTERPHONE study (9, 13). In the presence of a real association, such a differential bias would tend to diminish the odds ratio, possibly resulting in risk estimates below unity (particularly for low-to-moderate exposure level); it would not, however, create spuriously increased odds ratios when no risk exists (19, 20).

Sensitivity analyses restricted to regular users and to cases only (which are less affected by such bias) were conducted by use of the assumption that the overall decreased risks may be related to such a selection bias; both suggest an association with cellular phone use.

Another issue of concern when interpreting the results is the possibility of differential recall bias, with cases being more likely to overestimate their past use of cellular phones and to overreport their predominant side of use as the side of the head where their tumor occurred. A recent simulation paper within the INTERPHONE study (20) considered a number of scenarios of recall error. In the presence of the large random errors in recall such as those found in validation studies (16), differential errors in reporting were found to have very little additional impact. Laterality recall bias, however, would be expected to yield elevated risks for ipsilateral use and reduced risk for contralateral use. Such a reduction, although not statistically significant, was observed and may indicate the existence of such a bias. The observation of a similar relation in cases and controls between reported side of cellular phone use and handedness (a variable not expected to be affected by recall bias), however, argues against differential laterality bias. Because radiofrequency energy absorption is very localized, if radiofrequency exposure increases the risk of PGTs, any increase will be seen on the side of the head where the phone is usually held, and no effect will be observed on the opposite side. Our finding of an association in analyses restricted to ipsilateral tumors is consistent with similar findings in studies of cellular phone use and acoustic neuromas (9, 12), gliomas (7, 21, 22), and benign PGTs (14). Nevertheless, a reporting bias regarding laterality of use cannot be ruled out.

Israeli cellular phone users are exceptionally heavy users (16). Among our controls, the 75th percentiles for cumulative call time and cumulative number of calls were 1,348 hours and 26,100 calls, respectively, compared with 450 hours and 7,350 calls in a Swedish and Danish report (14) and 534 hours and 8,000 calls in a pooled analysis of north European INTERPHONE study countries (12). Thus, the exposures to cellular phone use considered in this study were higher than those in previous studies. In the ipsilateral, urban/rural, and regular-user analyses, our data showed increased odds ratios in the highest categories of use consistently over the different exposure measures supporting a dose-response relation.

Although most studies of cellular phone use have failed to find an overall association with risk of tumors, several suggest an increased risk among those who started use 10 years or more in the past, particularly for ipsilateral use. A Swedish study reported a significantly increased risk for acoustic neuroma for 10 years or more of ipsilateral use (OR = 3.9, 95 percent CI: 1.6, 9.5) (9) and nonsignificantly increased risks for glioma and meningioma (10), on the basis of small numbers of exposed cases (14, 15, and five, respectively). A pooled analysis of acoustic neuromas found an odds ratio of 1.8 (95 percent CI: 1.1, 3.1) based on 23 cases for 10 years or more of ipsilateral use (12). Our study also included a limited number of long-term users (13 cases with >10-year latency and 10 users on the ipsilateral side). Significant increases were found in relation to heavy use of cellular phones, even among shorter-term use in analyses of ipsilateral and rural use and in analyses restricted to regular users only.

The electromagnetic fields emitted from cellular phones do not have enough energy to break chemical bonds or damage DNA and, hence, are unlikely to act in the initiation of a tumor (5). If radiofrequency radiation acts as a tumor promoter, a clinical effect could be expected more rapidly. The high level of cellular phone use in the Israeli population may explain why we found an association between ipsilateral use and PGTs despite the relatively short latency period studied.

An important determinant of output power is distance from base stations. Lönn et al. (23) have shown that, in rural areas where base stations are located far apart, the average output power of phones tends to be higher than that in urban areas. Thus, classification of cellular phone use by area (urban/rural) may be another indicator of exposure. Our results showed that the increased risk from cellular phone use was confined mostly to use involving rural areas. Although this urban-rural difference was not seen in the study of Lönn et al. on brain tumors, it is compatible with that seen in another Swedish study (24).

In conclusion, based on the largest group of benign PGT patients reported to date, a number of complementary analyses suggest a positive association between cellular phone use and PGTs. Results from a single epidemiologic study do not, however, form a strong-enough basis to assume causality, and additional investigations of this association, with longer latency periods and large numbers of heavy users, are needed to confirm our findings. Until more evidence becomes available, we believe that the precautionary approach currently adopted by most scientific committees and applied by many governments should continue to be used.

APPENDIX

APPENDIX TABLE 1.

Risk of parotid gland tumors in relation to cellular phone use, with analyses restricted to regular users only, Israel, 2001–2003*

 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Time since start use (years)     
    1–4.9 138 389 1.0‡  
    5–9.9 134 276 1.40 1.03, 1.90 
    ≥10 13 26 1.45 0.82, 2.57 
Duration of use (years)     
    1–4.9 148 405 1.0‡  
    5–9.9 124 264 1.19 0.88, 1.59 
    ≥10 13 22 1.65 0.78, 3.45 
Cumulative no. of calls (with no hands-free devices)§     
    ≤5,479 117 382 1.0‡  
    5,480–18,996 86 157 1.48 1.05, 2.10 
    ≥18,997 81 140 1.51 1.05, 2.17 
Cumulative call time (hours) (with no hands-free devices)§     
    ≤266.3 121 390 1.0‡  
    266.4–1,034.9 80 155 1.37 0.97, 1.95 
    ≥1,035 83 134 1.50 1.04, 2.16 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Time since start use (years)     
    1–4.9 138 389 1.0‡  
    5–9.9 134 276 1.40 1.03, 1.90 
    ≥10 13 26 1.45 0.82, 2.57 
Duration of use (years)     
    1–4.9 148 405 1.0‡  
    5–9.9 124 264 1.19 0.88, 1.59 
    ≥10 13 22 1.65 0.78, 3.45 
Cumulative no. of calls (with no hands-free devices)§     
    ≤5,479 117 382 1.0‡  
    5,480–18,996 86 157 1.48 1.05, 2.10 
    ≥18,997 81 140 1.51 1.05, 2.17 
Cumulative call time (hours) (with no hands-free devices)§     
    ≤266.3 121 390 1.0‡  
    266.4–1,034.9 80 155 1.37 0.97, 1.95 
    ≥1,035 83 134 1.50 1.04, 2.16 
*

Unconditional logistic regression.

Adjusted for age, sex, and year of interview.

Referent.

§

Based on the distribution of users' controls (weighted) and divided into the following categories: ≤median, >median–≤third quartile, and >third quartile.

APPENDIX TABLE 2.

Risk of benign parotid tumor for different levels of cellular phone use by ipsilateral and both sides use, Israel, 2001–2003*

 Ipsilateral Both 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No 175 575 1.0‡  175 575 1.0‡  
    Yes 155 322 1.0 0.74, 1.35 28 68 1.23 0.71, 2.11 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 82 139 1.31 0.95, 1.82 14 27 1.56 0.78, 3.14 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240   
    Yes 10 12 2.17 0.91, 5.18   
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 71 180 0.89 0.62, 1.28 13 40 0.87 0.43, 1.77 
    5–9.9 74 130 1.06 0.72, 1.56 15 24 1.90 0.91, 3.98 
    ≥10 10 12 1.97 0.81, 4.85   
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 74 187 0.89 0.62, 1.27 16 42 1.0 0.51, 1.94 
    5–9.9 71 126 1.06 0.72, 1.57 12 22 1.72 0.78, 3.78 
    ≥10 10 2.53 0.97, 6.58   
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    ≤5,479 50 181 0.71 0.48, 1.04 11 42 0.79 0.38, 1.65 
    >5,479 104 136 1.49 1.03, 2.14 17 23 2.42 1.14, 5.11 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    ≤266.3 56 186 0.76 0.52, 1.10 11 38 0.94 0.45, 1.96 
    >266.3 98 131 1.45 1.0, 2.10 17 27 1.88 0.90, 3.92 
Cumulative call time (hours) (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    <5 years         
        ≤266.3 43 140 0.78 0.52, 1.17 31 0.69 0.29, 1.66 
        >266.3 28 38 1.51 0.85, 2.69 1.87 0.54, 6.52 
    ≥5 years         
        ≤266.3 13 46 0.72 0.37, 1.40 2.37 0.63, 8.93 
        >266.3 70 93 1.42 0.94, 2.15 11 20 1.96 0.84, 4.58 
Cumulative no. of calls (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    <5 years         
        ≤5,479 43 142 0.78 0.52, 1.18 34 0.48 0.19, 1.23 
        >5,479 28 36 1.51 0.84, 2.71 5.0 1.31, 19.05 
    ≥5 years         
        ≤5,479 39 0.46 0.20, 1.08 2.67 0.79, 8.99 
        >5,479 76 100 1.48 0.99, 2.20 10 19 1.90 0.79, 4.57 
 Ipsilateral Both 
 Cases (no.) Controls (no.) Odds ratio† 95% confidence interval Cases (no.) Controls (no.) Odds ratio† 95% confidence interval 
Regular user         
    No 175 575 1.0‡  175 575 1.0‡  
    Yes 155 322 1.0 0.74, 1.35 28 68 1.23 0.71, 2.11 
Regular user 5 years in the past         
    No 319 969 1.0‡  319 969 1.0‡  
    Yes 82 139 1.31 0.95, 1.82 14 27 1.56 0.78, 3.14 
Regular user 10 years in the past         
    No 447 1,240 1.0‡  447 1,240   
    Yes 10 12 2.17 0.91, 5.18   
Time since start use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 71 180 0.89 0.62, 1.28 13 40 0.87 0.43, 1.77 
    5–9.9 74 130 1.06 0.72, 1.56 15 24 1.90 0.91, 3.98 
    ≥10 10 12 1.97 0.81, 4.85   
Duration of use (years)         
    Never, <1 year 175 575 1.0‡  175 575 1.0‡  
    1–4.9 74 187 0.89 0.62, 1.27 16 42 1.0 0.51, 1.94 
    5–9.9 71 126 1.06 0.72, 1.57 12 22 1.72 0.78, 3.78 
    ≥10 10 2.53 0.97, 6.58   
Cumulative no. of calls (with no hands-free devices)§         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    ≤5,479 50 181 0.71 0.48, 1.04 11 42 0.79 0.38, 1.65 
    >5,479 104 136 1.49 1.03, 2.14 17 23 2.42 1.14, 5.11 
Cumulative call time (hours) (with no hands-free devices)§         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    ≤266.3 56 186 0.76 0.52, 1.10 11 38 0.94 0.45, 1.96 
    >266.3 98 131 1.45 1.0, 2.10 17 27 1.88 0.90, 3.92 
Cumulative call time (hours) (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    <5 years         
        ≤266.3 43 140 0.78 0.52, 1.17 31 0.69 0.29, 1.66 
        >266.3 28 38 1.51 0.85, 2.69 1.87 0.54, 6.52 
    ≥5 years         
        ≤266.3 13 46 0.72 0.37, 1.40 2.37 0.63, 8.93 
        >266.3 70 93 1.42 0.94, 2.15 11 20 1.96 0.84, 4.58 
Cumulative no. of calls (with no hands-free devices)§ by time since start use (years)         
    Nonusers, <1 year 176 580 1.0‡  175 578 1.0‡  
    <5 years         
        ≤5,479 43 142 0.78 0.52, 1.18 34 0.48 0.19, 1.23 
        >5,479 28 36 1.51 0.84, 2.71 5.0 1.31, 19.05 
    ≥5 years         
        ≤5,479 39 0.46 0.20, 1.08 2.67 0.79, 8.99 
        >5,479 76 100 1.48 0.99, 2.20 10 19 1.90 0.79, 4.57 
*

Unconditional logistic regression.

Adjusted for age, sex, and year of interview.

Referent.

§

Based on the distribution of users' controls (weighted) and divided into ≤median and >median.

This study was supported by the European Commission Fifth Framework Program—Quality of Life and Management of Living Resources; a grant from Union Internationale Contre le Cancer (UICC); and a supplement grant from the Israel Cancer Association.

The authors acknowledge the late Professor Baruch Modan for his significant contribution at the start of this study and who, unfortunately, did not live long enough to see the fruits of his labors. They also thank Dr. Lori Mandelzweig (Cancer and Radiation Epidemiology Unit, Gertner Institute, Chaim Sheba Medical Center, Tel Hashomer) for her contribution and assistance in editing this paper. They are also deeply indebted to the following otolaryngology–head and neck surgeons for their assistance with patient enrollment: Dr. N. Bartal, Poria Medical Center; Dr. I. Braverman, Hillel Yaffe Medical Center; Dr. D. Cohen, Shaare Zedek Medical Center; Dr. J. Elidan, Hadassah Medical Organization; Dr. M. Englander, Barzilai Medical Center; Dr. R. Feinmesser, Rabin Medical Center; Dr. D. Fliss, Tel Aviv Sourasky Medical Center; Dr. A. Goltz, Rambam Medical Center; Dr. E. Greenberg, Hacarmel Medical Center; Dr. D. Halperin, Kaplan Medical Center; Dr. A. Liberman, Soroka Medical Center; Dr. M. Luntz, Bnei Zion Medical Center; Dr. G. Marshak, Assuta Medical Center; Dr. O. Nachlieli, Barzilai Medical Center; Dr. D. Ophir, Meir Hospital–Sapir Medical Center; Dr. J. Rakover, Haemek Medical Center; Dr. Y. Roth, Edith Wolfson Medical Center; Dr. S. Segal, Assaf-Harofeh Medical Center, Ashkelon; and Dr. A. Shapira, Barzilai Medical Center. Finally, the authors thank the International Agency for Research on Cancer (IARC) team for their input in the INTERPHONE study.

The UICC received funds for this purpose from the Mobile Manufacturers' Forum and the GSM Association. Provisions of funds to the INTERPHONE study investigators via the UICC were governed by agreements that guaranteed INTERPHONE's complete scientific independence (http://www.iarc.fr/ENG/Units/RCAd.html).

Abbreviations

    Abbreviations
  • CI

    confidence interval

  • OR

    odds ratio

  • PGT

    parotid gland tumor

  • UICC

    Union Internationale Contre le Cancer

References

1.
Repacholi
MH
Low level exposure to radiofrequency electromagnetic fields: health effects and research needs
Bioelectromagnetics
 , 
1998
, vol. 
19
 (pg. 
1
-
9
)
2.
McKinlay
A
Possible health effects related to the use of radiotelephones
Radiol Prot Bull
 , 
1997
, vol. 
187
 (pg. 
9
-
16
)
3.
Rothman
KJ
Loughlin
JE
Funch
DP
, et al.  . 
Overall mortality of cellular telephone customers
Epidemiology
 , 
1996
, vol. 
7
 (pg. 
303
-
5
)
4.
Inskip
PD
Tarone
RE
Hatch
EE
, et al.  . 
Cellular-telephone use and brain tumors
N Engl J Med
 , 
2001
, vol. 
344
 (pg. 
79
-
86
)
5.
Schuz
J
Jacobsen
R
Olsen
JH
, et al.  . 
Cellular telephone use and cancer risk: update of a nationwide Danish cohort
J Natl Cancer Inst
 , 
2006
, vol. 
98
 (pg. 
1707
-
13
)
6.
Auvinen
A
Hietanen
M
Luukkonen
R
, et al.  . 
Brain tumors and salivary gland cancers among cellular telephone users
Epidemiology
 , 
2002
, vol. 
13
 (pg. 
356
-
9
)
7.
Hardell
L
Carlberg
M
Hansson Mild
K
Pooled analysis of two case-control studies on use of cellular and cordless telephones and the risk for malignant brain tumours diagnosed in 1997 –2003
Int Arch Occup Environ Health
 , 
2006
, vol. 
79
 (pg. 
630
-
9
)
8.
Christensen
HC
Schüz
J
Kosteljanetz
M
, et al.  . 
Cellular telephone use and risk of acoustic neuroma
Am J Epidemiol
 , 
2004
, vol. 
159
 (pg. 
277
-
83
)
9.
Lönn
S
Ahlbom
A
Hall
P
, et al.  . 
Mobile phone use and the risk of acoustic neuroma
Epidemiology
 , 
2004
, vol. 
15
 (pg. 
653
-
9
)
10.
Lönn
S
Ahlbom
A
Hall
P
, et al.  . 
Long-term mobile phone use and brain tumor risk
Am J Epidemiol
 , 
2005
, vol. 
161
 (pg. 
526
-
35
)
11.
Christensen
HC
Schüz
J
Kosteljanetz
M
, et al.  . 
Cellular telephones and risk for brain tumors
Neurology
 , 
2005
, vol. 
64
 (pg. 
1189
-
95
)
12.
Schoemaker
MJ
Swerdlow
AZ
Ahlbom
A
, et al.  . 
Mobile phone use and risk of acoustic neuroma: results of the Interphone case-control study in five North European countries
Br J Cancer
 , 
2005
, vol. 
93
 (pg. 
842
-
8
)
13.
Schuz
J
Bohler
E
Berg
G
, et al.  . 
Cellular phones, cordless phones, and the risks of glioma and meningioma (Interphone Study Group, Germany)
Am J Epidemiol
 , 
2006
, vol. 
163
 (pg. 
512
-
20
)
14.
Lönn
S
Ahlbom
A
Christensen
HC
, et al.  . 
Mobile phone use and risk of parotid gland tumor
Am J Epidemiol
 , 
2006
, vol. 
164
 (pg. 
637
-
43
)
15.
Cardis
E
Kilkenny
M
International case-control study of adult brain, head and neck tumors: results of the feasibility study
Radiat Prot Dosimetry
 , 
1999
, vol. 
83
 (pg. 
179
-
83
)
16.
Vrijheid
M
Cardis
E
Armstrong
BK
, et al.  . 
Validation of short term recall of mobile phone use for the Interphone study
Occup Environ Med
 , 
2006
, vol. 
63
 (pg. 
237
-
43
)
17.
Nagler
RM
Laufer
D
Tumors of the major and minor salivary glands: review of 25 years of experience
Anticancer Res
 , 
1997
, vol. 
17
 (pg. 
701
-
8
)
18.
Hardell
L
Hallquist
A
Hansson Mild
K
, et al.  . 
No association between the use of cellular or cordless telephone and salivary gland tumors
Occup Environ Med
 , 
2004
, vol. 
61
 (pg. 
675
-
9
)
19.
Lahkola
A
Salminen
T
Auvinen
A
Selection bias due to differential participation in a case-control study of mobile phone use and brain tumors
Ann Epidemiol
 , 
2005
, vol. 
15
 (pg. 
321
-
5
)
20.
Vrijheid
M
Deltour
I
Krewski
D
, et al.  . 
The effects of recall errors and of selection bias in epidemiologic studies of mobile phone use and cancer risk
J Expo Sci Environ Epidemiol
 , 
2006
, vol. 
16
 (pg. 
371
-
84
)
21.
Lahkola
A
Auvinen
A
Raitanen
J
, et al.  . 
Mobile phone use and risk of glioma in 5 North European countries
Int J Cancer
 , 
2007
, vol. 
120
 (pg. 
1769
-
75
)
22.
Hardell
L
Mild
KH
Carlberg
M
Further aspects on cellular and cordless telephones and brain tumours
Int J Oncol
 , 
2003
, vol. 
22
 (pg. 
399
-
407
)
23.
Lönn
S
Forssen
U
Vecchia
P
, et al.  . 
Output power levels from mobile phones in different geographical areas; implications for exposure assessment
Occup Environ Med
 , 
2004
, vol. 
61
 (pg. 
769
-
72
)
24.
Hardell
L
Carlberg
M
Mild
KH
Case-control study of the association between the use of cellular and cordless telephones and malignant brain tumors diagnosed during 2000 –2003
Environ Res
 , 
2006
, vol. 
100
 (pg. 
232
-
41
)