Ebea, Erice (Sicily), 29/03/2008 Fixed site transmitter studies

Size: px

Start display at page:

Download "Ebea, Erice (Sicily), 29/03/2008 Fixed site transmitter studies"

Laura Barton
6 years ago
Views:

1 Ebea, Erice (Sicily), 29/03/2008 Fixed site transmitter studies M. Röösli, PhD University of Bern Institute of Social and Preventive Medicine

2 Content > Introduction > Methodological challenges of studying long term effects of EMF (focus on symptoms) > Overview and critical discussion of symptom studies > Hard outcomes such as leukemia > Conclusions 2

3 Human studies > Intervention studies in the laboratory : controlled exposure conditions sophisticated outcome measurements short term effects no serious diseases due to ethical constraints small sample size > Intervention studies in the everyday environment > Observational studies: real life exposure, large sample size long term effects, serious diseases uncertain exposure assessment outcome measurement restricted Bias: selection bias, information bias, confounding 3

4 Long term effects > For studying medium to long term effects, we need: > Intervention studies in the everyday environment > Observational studies 4

5 Methodological challenges of fixed site transmitter studies > Exposure assessment > Selection bias > Information bias > Limited possibilities for objective outcome measurements > Reverse causality > Confounding by lifestyle > Nocebo effects Neubauer et al., Bioelectromagnetics 2007, 28(3):

6 Exposure assessment > Ubiquitous EMF in our everyday environment is a challenge for the exposure assessment. > Close to fixed site transmitters, RF-EMF can be considerably increased. > However, for mobile phone base station, distance is a poor proxy. > Self estimated distance is vulnerable to bias. QUEBEB-study (Berg, GMDS-Meeting Leipzig, 2006): Subjectively perceived mobile phone base station at the living place correlated with symptoms. Objectively measured (geo-coded) presence of a mobile phone base station did not correlate with symptoms. 6

7 Selection bias > Selection bias can occur in case-control studies if controls are not completely representative for the source (general) population. > Selection bias can occur in cross-sectional studies if participants are not selected randomly: in a France study, participants were recruited using newspapers articles. Individuals attributing their symptoms to base station have to live in the vicinity of a base station. They are likely to have more symptoms than the rest of the population. 7

8 Selection bias Allg. Bevölkerung general population 8

9 Selection bias Allg. Bevölkerung general population Betroffene concerned people with symptoms 9

10 Information bias Major Challenge: Subjective reporting of symptoms Knowledge about exposure > Information bias concerns both, exposure and outcome measure. It is of particular concern for self reported health outcomes or exposure status: Awareness of exposure status may affect outcome reporting. Study participants around transmitters may know their exposure status. (May also be considered as confounding by concern.) Health status may affect exposure reporting. Ill persons may be more motivated to report exposures than healthy persons. 10

11 Objective exposure and outcome measures > If both, exposure and outcome measures, are selfreported, bias is very likely. > Objective exposure measures can be obtained by measurements, modelling or from routine data sources (e.g. mobile phone operator data). > Objective outcome measurements for symptoms are often limited; e.g. EEG, cognitive functions, etc. > validated questionnaire guarantee a certain degree of objectivity. Nevertheless, they can be vulnerable to information bias. 11

12 A) Can we avoid information bias for the outcome? observable psychological effect building not known physical effect transmission starts time These approaches have to be evaluated first. B) In our everyday environment with a lot of RF-EMF sources, correlation, between actual exposure and self-estimated exposure is expected to be poor. This allows comparing truly exposed and unexposed individuals, by taking into account whether they perceive themselves as being exposed or not. 12

13 Reverse causality > Differences in exposure are a result of the health status (and not vice versa). > Example 1: Self declared EHS study participants may tend to avoid exposure and to report more symptoms: -> would result in a negative correlation between exposure and health status > Example 2: Morbid persons spend more time at home and may thus be less exposed. > Of concern in pure observational studies, in particular cross-sectional studies. 13

14 Confounding EXPOSURE? HEALTH EFFECT Confounding > Confounding by lifestyle, concerns and socio economic status is most crucial in this context. > Lifestyle related EMF sources: mobile phone, W-LAN, etc. 14

15 Nocebo > contrary to placebo > development of symptoms due to expectation (e.g. concern) > This can occur after a fixed site transmitter has been installed. 15

16 Cognitive, sympotms and sleep studies conducted in the everyday environment > Cross-sectional survey on cognitive functions and use of mobile phone (Lee et al. 2001) > Schwarzenburg study (Abelin et al. 2005; Altepeter, et al., 2006) > Austrian cross-sectional survey (Hutter et al., 2006) > Cyprus survey (Preece et al. 2007) > QUEBEB (Berg, 2007) > Bavarian UMTS intervention study (Heinrichs et al., 2007) > Symptoms and personal RF-EMF exposure (Radon) > Shielding sleep study (Leitgeb) > Sleep field study (Dorn) > QUALIFEX (Röösli) 16

17 Lee et al, 2001 > Cross-sectional comparison > Three measures of attention were administered to 72 teenagers, 37 of whom were mobile phone users. > Mobile phone users performed better on one of the three measures of attention than the nonmobile phone users. 17

18 Lee et al, 2001 > Cross-sectional comparison Exposure assessment > Three measures Selection of bias attention were administered to 72 teenagers, Information 37 bias of whom were mobile phone users. Objective outcome measurements > Mobile phone Reverse users causality? performed better on one of the three measures Confounding of by attention lifestyle? than the nonmobile phone users. Nocebo 18

19 Schwarzenburg study (CH) Reference Study design Collective Association observed Abelin, 2005 cross-sectional in residents of three differently exposed areas around shortwave transmitter (participation rate: 60%) sleep disturbances, general weakness and tiredness, nervousness, restlessness, limb and joint pain Abelin, 2005 Panel: interrupted during 3 days 65 residents of a shortwave transmitter (participation rate: 64%) Decrease of awakening after transmission stopped Abelin, 2005 cross-sectional survey in residents of four differently exposed areas around the shortwave transmitter (participation rate: 77%) Sleep disturbances, nervousness, restlessness Altpeter, 2006 Panel study in 1998: permanent shut down 54 residents of a shortwave transmitter Self rated sleep quality Melatonin excretion in poor sleepers Motivation for the study: 1990 petition to the Swiss Federal Government to shut down the transmitter 19

20 Exposure characteristics > Broadcasting in blocks of 105 minutes > Elevation of the beam by 11 degrees > Power: 3 x 150 kw > Frequency: 6.1 to 21.8 MHz > Amplitude modulation 20

21 Exposure levels H-field [ma/m] Exposure threshold according IRPA guidelines R R R R Distance [m] 21

22 Objectives of the shut down study > Possible chronic effects of short-wave exposure on sleep quality and salivary melatonin excretion in humans: comparing outcome measurements before shut down with respect to the extent of exposure (between subject analysis). > Possible acute effects of a sudden reduction in the short-wave exposure on sleep quality and salivary melatonin excretion in humans: comparing outcome measurements before shut down with measurements after shut down (within subject analysis). 22

23 Methods > Exposure assessment calculation of magnetic field for each participant s house assuming free space propagation. Sample measurements were used to validate the calculations. > Sleep quality sleep diary records of duration and freshness/tiredness (visual analogue scale) > Salivary melatonin levels 5 double saliva samples per day (before breakfast, at noon, tea time, dinner time, before bed) the 24h melatonin excretion curves were estimated from the 5 daily samples using a complex-cosinor-model 23

24 Study participants n (%) Total number of participants 54 (100) Sex - male 21 (39) - female 33 (61) Occupation - general farming 15 (28) - household work 19 (35) - other 20 (37) mean (sd) Age 52.8 (12.3) Magnetic field strength [ma/m] 1.5 (1.5) Distance to transmitter [km] 1.88 (1.32) Years of education (without kindergarten) 12.9 (2.6) Socioeconomic status (magnitude prestige scale) 39.2 (34.9) 24

25 Chronic effects on sleep quality Below median exposure p=0.001* Above median exposure * linear median regression model adjusted for age and sex 25

26 Chronic effects on melatonin excretion Below median exposure p=0.48* Above median exposure * linear median regression model adjusted for age and sex 26

27 Acute effects on sleep quality EMF effect: p=0.04* Change in magnetic field level [ma/m] Change in sleep quality Moving average * random effects model adjusted for age, sex and baseline sleep quality 27

28 Acute effects on melatonin excretion Ratio AUC EMF effect: p=0.11* Change in magnetic field level [ma/m] Ratio melatonin excretion Moving average * random effects model adjusted for age, sex and baseline value 28

29 Acute effects stratified for baseline sleep quality sleep quality Coefficient * 95%-Confidence interval Morning poor ; -1.7 Freshness/Tiredness good ; 3.9 AUC [ratio]** poor ; 1.47 good ; 1.39 Peak time [min] poor ; 19.3 good ; 18.0 * random effects model adjusted for age, sex and baseline value ** due to the logarithm transformation the coefficient refers to a change in the ratio 29

30 Schwarzenburg study Reference Abelin, 2005 Abelin, 2005 Abelin, 2005 Altpeter, 2006 Study design cross-sectional in 1992 Panel: interrupted during 3 days cross-sectional survey in 1996 Panel study in 1998: permanent shut down Collective Exposure assessment 404 residents of three differently exposed areas around shortwave transmitter (participation rate: 60%) Selection bias? Information bias? 65 residents of a shortwave transmitter (participation rate: 64%) 399 residents of four differently exposed areas around the shortwave transmitter (participation rate: 77%) Association observed Objective outcome measurements? Reverse causality 54 residents of a shortwave transmitter Confounding by lifestyle Nocebo? sleep disturbances, general weakness and tiredness, nervousness, restlessness, limb and joint pain Decrease of awakening after transmission stopped Sleep disturbances, nervousness, restlessness Self rated sleep quality Melatonin excretion in poor sleepers 30

31 Austrian cross-sectional survey (Hutter et al., 2006). > 365 random sample residents of mobile phone base station > Participation rate: 60% > Outcome: cognitive functions, Zerssen symptom scale, Pittsburgh Sleep Quality Index > Measurements yielded field values in the high frequency range from 0.01 to 0.75 V/m > 3 of 17 symptoms associated with exposure: Headache, cold hands or feet, difficulties to concentrate 31

32 Austrian cross-sectional survey (Hutter et al., 2006). > 365 random Exposure sample assessment? residents of mobile phone base station Selection bias? > Participation Information rate: 60% bias? > Outcome: Objective cognitive outcome functions, measurements? Zerssen symptom scale, Pittsburgh Reverse causality Sleep Quality Index > Measurements Confounding yielded by lifestyle field values in the high frequency Nocebo? range from 0.01 to 0.75 V/m > 3 of 17 symptoms associated with exposure: Headache, cold hands or feet, difficulties to concentrate 32

33 Cyprus survey (Preece et al. 2007) > 1870 inhabitants from three differently exposed villages around a short wave military antenna system (participation rate: 87%) > Participation rate: 87% > Exposure levels: 0.57 V/m; 0.46 V/m and <0.01 V/m (most from mobile phone base stations) > Observed associations: migraine, headache, dizziness, depression, SF-36-scores 33

34 Cyprus survey (Preece et al. 2007) > 1870 inhabitants Exposure from assessment? three differently exposed villages around Selection a bias short wave military antenna system (participation rate: 87%) Information bias? > Participation Objective rate: outcome 87% measurements? > Exposure Reverse levels: causality 0.57 V/m; 0.46 V/m and <0.01 V/m (most Confounding from mobile by lifestyle phone base stations) > Observed Nocebo? associations: migraine, headache, dizziness, depression, SF-36-scores 34

35 QUEBEB (Berg, 2007) > Base survey of individuals in Germany about concerns, symptoms and self-estimated exposure. > Modelled exposure using geo-coded data from base station site. > Nested cross-sectional survey in individuals. > Acute symptoms and exposure investigated in individuals > Positive correlation between symptoms and self-estimated exposure (n=30 000). > No association between measured EMF and 5 symptoms: sleep disturbations, headache, somatic symptoms, somatic and psychic health related quality of life (n=1 300) 35

36 QUEBEB (Berg, 2007) > Base survey of individuals in Germany about concerns, symptoms Exposure and assessment? self-estimated exposure. > Modelled exposure Selection using bias geo-coded data from base station site. Information bias > Nested cross-sectional survey in individuals. Objective outcome measurements > Acute symptoms and exposure investigated in individuals Reverse causality? > Positive correlation Confounding between by symptoms lifestyle and self-estimated exposure (n=30 000). Nocebo > No association between measured EMF and 5 symptoms: sleep disturbations, headache, somatic symptoms, somatic and psychic health related quality of life (n=1 300) 36

37 Bavarian intervention study (Heinrich et al. 2007) > A newly installed UMTS mobile phone base station on an office building was randomly turned on and off over a period of 70 working days. > Software was modified in a way that the operating state could not be retrieved from the own UMTS mobile phone. > Maximum field level 0.53 V/m. > 95 workers in the building filled in a symptom questionnaire every morning and evening. > Slight tendency of an increase of self reported complaints on days when the mobile phone base station was operating (p=0.08). > Evidence for nocebo phenomenom (p<0.0001). 37

38 Bavarian intervention study (Heinrich et al. 2007) > A newly installed UMTS mobile phone base station on an Exposure assessment? office building was randomly turned on and off over a period of 70 working Selection days. bias > Software was Information modified in bias? a way that the operating state could not be retrieved from the own UMTS mobile phone. Objective outcome measurements? > Maximum field level 0.53 V/m. Reverse causality > 95 workers in the building filled in a symptom questionnaire every morning Confounding and evening. by lifestyle > Slight tendency Nocebo of an increase of self reported complaints on days when the mobile phone base station was operating (p=0.08). > Evidence for nocebo phenomenom. 38

Leukaemia and broadcast transmitters from Environ Health Persp (2004),112, 1741ff Park et al. 2004 AM radio <2 km (approximately) Leukemia 55 1.70 (0.84-3.

39 Leukaemia and broadcast transmitters from Environ Health Persp (2004),112, 1741ff Park et al AM radio <2 km (approximately) Leukemia ( ) Korea Definition of exposed area not clear Ha et al AM radio 4th quartile vs. 1st quartile Child. lymph Leuk. Child. myel Leuk ( ) 1.40 ( ) 0.55 ( ) 0.63 ( ) Korea Mean exposure Peak exposure Mean exposure Peak exposure In Germany a study is ongoing (Schüz). 39

40 Leukaemia and broadcast transmitters > Texas sharpshooter fallacy? An eye-catching increase of leukaemia cases can trigger the conduct of a study. In contrast, lack of cases is not expected to trigger the conduct of a study. Clusters are expected due to chance. > Surveillance bias? Case identification/registration in the viccinity of suspicious sources can be more complete than in the rest of the country. > Publication bias? Studies that found an effect are more likely to be published. This is of particular concern for small studies. > Real risk increase? 40

41 Summary fixed site transmitter studies > The conduct of studies around fixed site transmitters is methodologically challenging. > The conduct of studies around fixed site transmitters may facilitate exposure assessment. > Available fixed site transmitter studies on unspecific symptoms of ill health are not do not allow drawing firm conclusions. > Observed increased leukaemia risks around broadcast transmitters may be due to the Texas sharpshooter fallacy, surveillance or publication bias. 41

42 Research Needs > Perception of low level EMF: individual focused familiar environments > Symptoms - immediate to short term effects: laboratory studies: longer exposure duration, higher levels, different signal characteristics > Electromagnetic hypersensitivity: understanding the mechanism therapeutic options > Symptoms - medium to long term effects : more sophisticated observational studies 42

Section 12. Symptoms Attributed to Radiofrequency Electromagnetic Fields

Section 12. Symptoms Attributed to Radiofrequency Electromagnetic Fields Section 12 Symptoms Attributed to Radiofrequency Electromagnetic Fields Table of Contents Summary... 306 12.1 Introduction... 307 12.2 Methods... 307 12.3 Is Electromagnetic Hypersensitivity a Clinical