Isabelle Deltour

Publications of Isabelle Deltour

• Mobile Phone Use and Incidence of Glioma in the Nordic Countries 1979-2008: Consistency Check.

  Authors: Isabelle Deltour, Anssi Auvinen, Maria Feychting, Christoffer Johansen, Lars Klaeboe, Risto Sankila, Joachim Schüz

  Epidemiology (Cambridge, Mass.). 03/2012; 23(2):301-307.

  BACKGROUND:: Some case-control studies have reported increased risks of glioma associated with mobile phone use. If true, this would ultimately affect the time trends for incidence rates (IRs).BACKGROUND:: Some case-control studies have reported increased risks of glioma associated with mobile phone use. If true, this would ultimately affect the time trends for incidence rates (IRs). Correspondingly, lack of change in IRs would exclude certain magnitudes of risk. We investigated glioma IR trends in the Nordic countries, and compared the observed with expected incidence rates under various risk scenarios. METHODS:: We analyzed annual age-standardized incidence rates in men and women aged 20 to 79 years during 1979-2008 using joinpoint regression (35,250 glioma cases). Probabilities of detecting various levels of relative risk were computed using simulations. RESULTS:: For the period 1979 through 2008, the annual percent change in incidence rates was 0.4% (95% confidence interval = 0.1% to 0.6%) among men and 0.3% (0.1% to 0.5%) among women. Incidence rates have decreased in young men (20-39 years) since 1987, remained stable in middle-aged men (40-59 years) throughout the 30-year study period, and increased slightly in older men (60-79 years). In simulations, assumed relative risks for all users of 2.0 for an induction time of up to 15 years, 1.5 for up to 10 years, and 1.2 for up to 5 years were incompatible with observed incidence time trends. For heavy users of mobile phones, risks of 2.0 for up to 5 years' induction were also incompatible. CONCLUSION:: No clear trend change in glioma incidence rates was observed. Several of the risk increases seen in case-control studies appear to be incompatible with the observed lack of incidence rate increase in middle-aged men. This suggests longer induction periods than currently investigated, lower risks than reported from some case-control studies, or the absence of any association.

• Psychological distress among women with newly diagnosed breast cancer.

  Authors: Birgitte Goldschmidt Mertz, Pernille Envold Bistrup, Christoffer Johansen, Susanne Oksbjerg Dalton, Isabelle Deltour, Henrik Kehlet, Niels Kroman

  European journal of oncology nursing : the official journal of European Oncology Nursing Society. 10/2011;

  PURPOSE: Psychological distress is common in the cancer continuum. Our objectives were to determine the prevalence of distress and to investigate the related problems and the characteristics of womenPURPOSE: Psychological distress is common in the cancer continuum. Our objectives were to determine the prevalence of distress and to investigate the related problems and the characteristics of women with breast cancer who experienced psychological distress at the time of diagnosis. METHODS: We used cross-sectional data from a questionnaire study. Women with newly diagnosed breast cancer were consecutively invited to respond before breast surgery. Between October 2008 and October 2009, a total of 357 responded out of 426 (84%) invited. Among these, 343 patients completed the 'distress thermometer' to measure psychological distress and the accompanying 'problem list' to identify related problems. Logistic regression models with 95% confidence intervals were used to estimate the associations between psychological distress, age, social support and domains on the problem list. RESULTS: With a cut-off of 3 on the distress thermometer, 77% of women with breast cancer reported distress, whereas when the cut-off was 7, 43% were distressed. The mean distress score was 5.4 (SD, 3.1). The most frequently reported problems were worry (77%) and nervousness (71%). Distress was significantly associated with the total score and three domains on the problem list. Younger women (<50 years) reported higher levels of distress than older (≥50 years). We found no significant association between distress and having a partner or someone outside the family to rely on. CONCLUSIONS: Distress was reported by 77% of patients. Age and problem list were significantly associated with distress. No significant association between psychological distress and social support was observed.

• Analysis of three-dimensional SAR distributions emitted by mobile phones in an epidemiological perspective.

  Authors: Isabelle Deltour, Joe Wiart, Masao Taki, Kanako Wake, Nadège Varsier, Simon Mann, Joachim Schüz, Elisabeth Cardis

  Bioelectromagnetics. 06/2011; 32(8):634-43.

  The three-dimensional distribution of the specific absorption rate of energy (SAR) in phantom models was analysed to detect clusters of mobile phones producing similar spatial deposition of energy inThe three-dimensional distribution of the specific absorption rate of energy (SAR) in phantom models was analysed to detect clusters of mobile phones producing similar spatial deposition of energy in the head. The clusters' characteristics were described from the phones external features, frequency band and communication protocol. Compliance measurements with phones in cheek and tilt positions, and on the left and right side of a physical phantom were used. Phones used the Personal Digital Cellular (PDC), Code division multiple access One (CdmaOne), Global System for Mobile Communications (GSM) and Nordic Mobile Telephony (NMT) communication systems, in the 800, 900, 1500 and 1800 MHz bands. Each phone's measurements were summarised by the half-ellipsoid in which the SAR values were above half the maximum value. Cluster analysis used the Partitioning Around Medoids algorithm. The dissimilarity measure was based on the overlap of the ellipsoids, and the Manhattan distance was used for robustness analysis. Within the 800 MHz frequency band, and in part within the 900 MHz and the 1800 MHz frequency bands, weak clustering was obtained for the handset shape (bar phone, flip with top and flip with central antennas), but only in specific positions (tilt or cheek). On measurements of 120 phones, the three-dimensional distribution of SAR in phantom models did not appear to be related to particular external phone characteristics or measurement characteristics, which could be used for refining the assessment of exposure to radiofrequency energy within the brain in epidemiological studies such as the Interphone.

• Factors associated with the prescription of antidepressive medication to breast cancer patients.

  Authors: Nis P Suppli, Isabelle Deltour, Lars H Damkjaer, Jane Christensen, Anders B Jensen, Niels T Kroman, Christoffer Johansen, Susanne O Dalton

  Acta oncologica (Stockholm, Sweden). 02/2011; 50(2):243-51.

  We evaluated factors associated with use of antidepressant medication subsequent to a diagnosis of breast cancer. We also evaluated the effect of participation in a cancer rehabilitation program onWe evaluated factors associated with use of antidepressant medication subsequent to a diagnosis of breast cancer. We also evaluated the effect of participation in a cancer rehabilitation program on use of antidepressants. We conducted a register-based cohort study of 1 247 women with breast cancer diagnosed between 1998 and 2006 who attended a week-long rehabilitation program and a comparison group of 2 903 women who did not attend the program matched through the registers of the Danish Breast Cancer Cooperative Group. The associations between breast cancer-related, treatment-related, and sociodemographic factors and use of antidepressants were evaluated in multivariate Cox proportional hazard models separated on use of antidepressants before diagnosis of breast cancer. The mean follow-up for the 4 150 women in the study was 3.3 years (5-95% range, 0.3-7.0 years) and 1 020 (25%) were users of antidepressants after diagnosis of breast cancer. Among women who had not used antidepressants before their breast cancer, the diagnosis of a new primary cancer increased the adjusted hazard ratio (HR) to 3.34 (95% CI, 1.50-7.76), and recurrence of breast cancer increased the HR for first use of antidepressants to 2.56 (95% CI, 1.86-3.52). Unemployment was associated significantly with use of antidepressants, whereas having no children living at home, lower income, and the number of tumor-positive axillary lymph nodes were of borderline significance. No effect of the rehabilitation program was observed on first use of antidepressants after breast cancer. Diagnosis of a new cancer or recurrence of breast cancer considerably increased the rate of use of antidepressants. Sociodemographic rather than disease- or treatment-related characteristics at the time of diagnosis were associated with first use of antidepressants after a breast cancer diagnosis.

• Breast cancer and early retirement: Associations with disease characteristics, treatment, comorbidity, social position and participation in a six-day rehabilitation course in a register-based study in Denmark.

  Authors: Lars Henrik Damkjaer, Isabelle Deltour, Nis Palm Suppli, Jane Christensen, Niels T Kroman, Christoffer Johansen, Susanne Oksbjerg Dalton

  Acta oncologica (Stockholm, Sweden). 02/2011; 50(2):274-81.

  The purpose of this register-based study was to identify factors related to disease, treatment, sociodemographics and comorbidity associated with taking early retirement among women treated forThe purpose of this register-based study was to identify factors related to disease, treatment, sociodemographics and comorbidity associated with taking early retirement among women treated for breast cancer, and to evaluate the risk for taking early retirement among breast cancer survivors who attended a six-day rehabilitation course. The study population consisted of 856 women who attended the rehabilitation course and a comparison group of 1 805 women who did not attend the course identified through the Danish Breast Cancer Cooperative Group. We obtained information on receipt of unemployment benefits, sickness benefits and early retirement pension for each of the years 1996-2007. Multivariate cox-regression models were used to analyze disease-specific, treatment-related, comorbidity and sociodemographics factors associated with early retirement after breast cancer and to evaluate the effect of attending a rehabilitation course on taking early retirement. The rate of retirement was higher for women with somatic comorbidity (hazard ratio [HR], 1.91; 95% CI, 1.3; 2.9 for score 1, and HR 1.42; 95% CI, 0.7; 2.7 for score ≥2), previous depression (HR, 2.29; 95% CI, 1.7; 3.2) or having received sickness benefits in the year before their breast cancer diagnosis (HR, 3.75; 95% CI, 1.8; 7.8). Living with a partner was associated with a reduced hazard ratio for taking early retirement (HR, 0.70; 95% CI, 0.5-0.9). Having received chemotherapy, alone or in combination with anti-hormone treatment, reduced the hazard ratio (HR, 0.49; 95% CI, 0.3; 0.8 and HR, 0.5; 95% CI, 0.3; 0.8, respectively). The rate of retirement was higher for women the year after attending the rehabilitation course but returned to unity by three years. The results of this study contribute to the identification of at-risk women and point to the need for tailored rehabilitation to avoid unnecessary marginalization of breast cancer survivors due to permanent labor market withdrawal.

• Accuracy of the Danish version of the 'distress thermometer'

  Authors: Pernille Envold Bidstrup, Birgitte Goldschmidt Mertz, Susanne Oksbjerg Dalton, Isabelle Deltour, Niels Kroman, Henrik Kehlet, Nina Rottmann, Rune Gärtner, Alex J Mitchell, Christoffer Johansen

  Psycho-oncology. 01/2011;

  Objective: Short screening instruments have been suggested to improve the detection of psychological symptoms. We examined the accuracy of the Danish version of the 'Distress Thermometer'. Methods:Objective: Short screening instruments have been suggested to improve the detection of psychological symptoms. We examined the accuracy of the Danish version of the 'Distress Thermometer'. Methods: Between October 2008 and October 2009, 426 women with newly diagnosed primary breast cancer who were operated at the Breast Surgery Clinic of the Rigshospitalet, Copenhagen, were eligible for this study. Of these, 357 participated (84%) and 333 completed a questionnaire. The distress thermometer was evaluated against the 'hospital anxiety and depression scale' (HADS). We also examined the women's wish for referral for psychological support. Results: A cut-off score of 6 vs 7 (low:⩽6, high:⩾7) on the distress thermometer was optimal for confirming distress, with a sensitivity of 42%, a specificity of 93%, a positive predictive value (PPV) of 78% and a negative predictive value (NPV) of 73%. A cut-off score of 2 vs 3 was optimal for screening, with a sensitivity of 99%, a specificity of 36%, a PPV of 47% and a NPV of 99%. Of those who were distressed using the cut-off score of 2 vs 3 on the distress thermometer, 17% (n = 41) wished to be referred for psychological support and 57% (n = 140) potentially wanted a later referral. Conclusion: The distress thermometer performed satisfactorily relative to the HADS in detecting distress in our study. A screening procedure in which application of the distress thermometer is a first step could be useful for identifying persons in need of support. Copyright © 2011 John Wiley & Sons, Ltd.

• Time Trends in Brain Tumor Incidence Rates in Denmark, Finland, Norway, and Sweden, 1974-2003.

  Authors: Isabelle Deltour, Christoffer Johansen, Anssi Auvinen, Maria Feychting, Lars Klaeboe, Joachim Schüz

  Journal of the National Cancer Institute. 12/2009;

  In Denmark, Finland, Norway, and Sweden, the use of mobile phones increased sharply in the mid-1990s; thus, time trends in brain tumor incidence after 1998 may provide information about possibleIn Denmark, Finland, Norway, and Sweden, the use of mobile phones increased sharply in the mid-1990s; thus, time trends in brain tumor incidence after 1998 may provide information about possible tumor risks associated with mobile phone use. We investigated time trends in the incidence of glioma and meningioma in Denmark, Finland, Norway, and Sweden from 1974 to 2003, using data from national cancer registries. We used joinpoint regression models to analyze the annual incidence rates of glioma and meningioma. During this period, 59 984 men and women aged 20-79 years were diagnosed with brain tumors in a population of 16 million adults. All statistical tests were two-sided. From 1974 to 2003, the incidence rate of glioma increased by 0.5% per year (95% confidence interval [CI] = 0.2% to 0.8%) among men and by 0.2% per year (95% CI = -0.1% to 0.5%) among women and that of meningioma increased by 0.8% per year (95% CI = 0.4% to 1.3%) among men, and after the early 1990s, by 3.8% per year (95% CI = 3.2% to 4.4%) among women. No change in incidence trends were observed from 1998 to 2003, the time when possible associations between mobile phone use and cancer risk would be informative about an induction period of 5-10 years.

• The estimation of 3D SAR distributions in the human head from mobile phone compliance testing data for epidemiological studies.

  Authors: Kanako Wake, Nadège Varsier, Soichi Watanabe, Masao Taki, Joe Wiart, Simon Mann, Isabelle Deltour, Elisabeth Cardis

  Physics in medicine and biology. 10/2009; 54(19):5695-5706.

  A worldwide epidemiological study called 'INTERPHONE' has been conducted to estimate the hypothetical relationship between brain tumors and mobile phone use. In this study, we proposed a method toA worldwide epidemiological study called 'INTERPHONE' has been conducted to estimate the hypothetical relationship between brain tumors and mobile phone use. In this study, we proposed a method to estimate 3D distribution of the specific absorption rate (SAR) in the human head due to mobile phone use to provide the exposure gradient for epidemiological studies. 3D SAR distributions due to exposure to an electromagnetic field from mobile phones are estimated from mobile phone compliance testing data for actual devices. The data for compliance testing are measured only on the surface in the region near the device and in a small 3D region around the maximum on the surface in a homogeneous phantom with a specific shape. The method includes an interpolation/extrapolation and a head shape conversion. With the interpolation/extrapolation, SAR distributions in the whole head are estimated from the limited measured data. 3D SAR distributions in the numerical head models, where the tumor location is identified in the epidemiological studies, are obtained from measured SAR data with the head shape conversion by projection. Validation of the proposed method was performed experimentally and numerically. It was confirmed that the proposed method provided good estimation of 3D SAR distribution in the head, especially in the brain, which is the tissue of major interest in epidemiological studies. We conclude that it is possible to estimate 3D SAR distributions in a realistic head model from the data obtained by compliance testing measurements to provide a measure for the exposure gradient in specific locations of the brain for the purpose of exposure assessment in epidemiological studies. The proposed method has been used in several studies in the INTERPHONE.

• Can loud noise cause acoustic neuroma? Analysis of the INTERPHONE study in France.

  Authors: Martine Hours, Marlene Bernard, Marie T Arslan, Lucile Montestrucq, Lesley Richardson, Isabelle Deltour, Elisabeth Cardis

  Occupational and environmental medicine. 04/2009;

  OBJECTIVES: To investigate possible associations between risk of acoustic neuroma and exposure to loud noise in leisure and occupational settings. METHODS: A case-control study was conducted inOBJECTIVES: To investigate possible associations between risk of acoustic neuroma and exposure to loud noise in leisure and occupational settings. METHODS: A case-control study was conducted in France within the international INTERPHONE study. The cases were the 108 subjects diagnosed with acoustic neuroma between June 1st 2000 and August 31st 2003. Two controls per case were selected from the electoral rolls and individually matched for gender, age (+/- 5 years) and area (local authority district) of residence at the time of the case diagnosis. Multivariate analyses were conducted using conditional logistic regression. Adjustment was made for socio-economic status. RESULTS: Acoustic neuroma was found to be associated with loud noise exposure [Odds-Ratio (OR) = 2.55; 95% Confidence Interval (CI) = 1.35-4.82], both in leisure settings, particularly when listening to loud music [OR = 3.88; 95% CI = 1.48-10.17] and at work [OR = 2.26; 95% CI = 1.08-4.72]. This risk increased with exposure duration (> 6 years leisure exposure: OR = 3.15; 95% CI = 1.07-9.24). Risk varied according to the type of noise (continuous or explosive, versus intermittent). CONCLUSION: The present results agree with other recent reports implicating loud noise in the risk of acoustic neuroma.

• Quantifying the Impact of Selection Bias Caused by Nonparticipation in a Case-Control Study of Mobile Phone Use.

  Authors: Martine Vrijheid, Lesley Richardson, Bruce K Armstrong, Anssi Auvinen, Gabriele Berg, Matthew Carroll, Angela Chetrit, Isabelle Deltour, Maria Feychting, Graham G Giles [......] Siegal Sadetzki, Tina Salminen, Marie Sanchez, Birgitte Schlehofer, Joachim Schüz, Jack Siemiatycki, Tore Tynes, Alistair Woodward, Naohito Yamaguchi, Elisabeth Cardis

  Annals of epidemiology. 02/2009; 19(1):33-41.e1.

  PURPOSE: To quantitatively assess the impact of selection bias caused by nonparticipation in a multinational case-control study of mobile phone use and brain tumor. METHODS: Non-responsePURPOSE: To quantitatively assess the impact of selection bias caused by nonparticipation in a multinational case-control study of mobile phone use and brain tumor. METHODS: Non-response questionnaires (NRQ) were completed by a sub-set of nonparticipants. Selection bias factors were calculated based on the prevalence of mobile phone use reported by nonparticipants with NRQ data, and on scenarios of hypothetical exposure prevalence for other nonparticipants. RESULTS: Regular mobile phone use was reported less frequently by controls and cases who completed the NRQ (controls, 56%; cases, 50%) than by those who completed the full interview (controls, 69%; cases, 66%). This relationship was consistent across study centers, sex, and age groups. Lower education and more recent start of mobile phone use were associated with refusal to participate. Bias factors varied between 0.87 and 0.92 in the most plausible scenarios. CONCLUSIONS: Refusal to participate in brain tumor case-control studies seems to be related to less prevalent use of mobile phones, and this could result in a downward bias of around 10% in odds ratios for regular mobile phone use. The use of simple selection bias estimation methods in case-control studies can give important insights into the extent of any bias, even when nonparticipant information is incomplete.

• Recall bias in the assessment of exposure to mobile phones.

  Authors: Martine Vrijheid, Bruce K Armstrong, Daniel Bédard, Julianne Brown, Isabelle Deltour, Ivano Iavarone, Daniel Krewski, Susanna Lagorio, Stephen Moore, Lesley Richardson, Graham G Giles, Mary McBride, Marie-Elise Parent, Jack Siemiatycki, Elisabeth Cardis

  Journal of exposure science & environmental epidemiology. 06/2008;

  Most studies of mobile phone use are case-control studies that rely on participants' reports of past phone use for their exposure assessment. Differential errors in recalled phone use are a majorMost studies of mobile phone use are case-control studies that rely on participants' reports of past phone use for their exposure assessment. Differential errors in recalled phone use are a major concern in such studies. INTERPHONE, a multinational case-control study of brain tumour risk and mobile phone use, included validation studies to quantify such errors and evaluate the potential for recall bias. Mobile phone records of 212 cases and 296 controls were collected from network operators in three INTERPHONE countries over an average of 2 years, and compared with mobile phone use reported at interview. The ratio of reported to recorded phone use was analysed as measure of agreement. Mean ratios were virtually the same for cases and controls: both underestimated number of calls by a factor of 0.81 and overestimated call duration by a factor of 1.4. For cases, but not controls, ratios increased with increasing time before the interview; however, these trends were based on few subjects with long-term data. Ratios increased by level of use. Random recall errors were large. In conclusion, there was little evidence for differential recall errors overall or in recent time periods. However, apparent overestimation by cases in more distant time periods could cause positive bias in estimates of disease risk associated with mobile phone use.Journal of Exposure Science and Environmental Epidemiology advance online publication, 21 May 2008; doi:10.1038/jes.2008.27.

• A Monte Carlo maximum likelihood method for estimating uncertainty arising from shared errors in exposures in epidemiological studies of nuclear workers.

  Authors: Leslie Stayner, Martine Vrijheid, Elisabeth Cardis, Daniel O Stram, Isabelle Deltour, Stephen J Gilbert, Geoffrey Howe

  Radiation research. 01/2008; 168(6):757-63.

  Errors in the estimation of exposures or doses are a major source of uncertainty in epidemiological studies of cancer among nuclear workers. This paper presents a Monte Carlo maximum likelihoodErrors in the estimation of exposures or doses are a major source of uncertainty in epidemiological studies of cancer among nuclear workers. This paper presents a Monte Carlo maximum likelihood method that can be used for estimating a confidence interval that reflects both statistical sampling error and uncertainty in the measurement of exposures. The method is illustrated by application to an analysis of all cancer (excluding leukemia) mortality in a study of nuclear workers at the Oak Ridge National Laboratory (ORNL). Monte Carlo methods were used to generate 10,000 data sets with a simulated corrected dose estimate for each member of the cohort based on the estimated distribution of errors in doses. A Cox proportional hazards model was applied to each of these simulated data sets. A partial likelihood, averaged over all of the simulations, was generated; the central risk estimate and confidence interval were estimated from this partial likelihood. The conventional unsimulated analysis of the ORNL study yielded an excess relative risk (ERR) of 5.38 per Sv (90% confidence interval 0.54-12.58). The Monte Carlo maximum likelihood method yielded a slightly lower ERR (4.82 per Sv) and wider confidence interval (0.41-13.31).

• The INTERPHONE study: design, epidemiological methods, and description of the study population.

  Authors: Elisabeth Cardis, Lesley Richardson, Isabelle Deltour, Bruce Armstrong, Maria Feychting, Christoffer Johansen, Monique Kilkenny, Patricia McKinney, Baruch Modan, Siegal Sadetzki [......] Brigitte Schlehofer, Jack Siemiatycki, Masao Taki, Toru Takebayashi, Tore Tynes, Martie van Tongeren, Paolo Vecchia, Joe Wiart, Alistair Woodward, Naohito Yamaguchi

  European journal of epidemiology. 02/2007; 22(9):647-64.

  The very rapid worldwide increase in mobile phone use in the last decade has generated considerable interest in the possible health effects of exposure to radio frequency (RF) fields. A multinationalThe very rapid worldwide increase in mobile phone use in the last decade has generated considerable interest in the possible health effects of exposure to radio frequency (RF) fields. A multinational case-control study, INTERPHONE, was set-up to investigate whether mobile phone use increases the risk of cancer and, more specifically, whether the RF fields emitted by mobile phones are carcinogenic. The study focused on tumours arising in the tissues most exposed to RF fields from mobile phones: glioma, meningioma, acoustic neurinoma and parotid gland tumours. In addition to a detailed history of mobile phone use, information was collected on a number of known and potential risk factors for these tumours. The study was conducted in 13 countries. Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, New Zealand, Norway, Sweden, and the UK using a common core protocol. This paper describes the study design and methods and the main characteristics of the study population. INTERPHONE is the largest case-control study to date investigating risks related to mobile phone use and to other potential risk factors for the tumours of interest and includes 2,765 glioma, 2,425 meningioma, 1,121 acoustic neurinoma, 109 malignant parotid gland tumour cases and 7,658 controls. Particular attention was paid to estimating the amount and direction of potential recall and participation biases and their impact on the study results.

• The effects of recall errors and of selection bias in epidemiologic studies of mobile phone use and cancer risk.

  Authors: Martine Vrijheid, Isabelle Deltour, Daniel Krewski, Marie Sanchez, Elisabeth Cardis

  Journal of exposure science & environmental epidemiology. 08/2006; 16(4):371-84.

  This paper examines the effects of systematic and random errors in recall and of selection bias in case-control studies of mobile phone use and cancer. These sensitivity analyses are based onThis paper examines the effects of systematic and random errors in recall and of selection bias in case-control studies of mobile phone use and cancer. These sensitivity analyses are based on Monte-Carlo computer simulations and were carried out within the INTERPHONE Study, an international collaborative case-control study in 13 countries. Recall error scenarios simulated plausible values of random and systematic, non-differential and differential recall errors in amount of mobile phone use reported by study subjects. Plausible values for the recall error were obtained from validation studies. Selection bias scenarios assumed varying selection probabilities for cases and controls, mobile phone users, and non-users. Where possible these selection probabilities were based on existing information from non-respondents in INTERPHONE. Simulations used exposure distributions based on existing INTERPHONE data and assumed varying levels of the true risk of brain cancer related to mobile phone use. Results suggest that random recall errors of plausible levels can lead to a large underestimation in the risk of brain cancer associated with mobile phone use. Random errors were found to have larger impact than plausible systematic errors. Differential errors in recall had very little additional impact in the presence of large random errors. Selection bias resulting from underselection of unexposed controls led to J-shaped exposure-response patterns, with risk apparently decreasing at low to moderate exposure levels. The present results, in conjunction with those of the validation studies conducted within the INTERPHONE study, will play an important role in the interpretation of existing and future case-control studies of mobile phone use and cancer risk, including the INTERPHONE study.