Wind Turbine Noise and Health Study: Summary of Comments Received

Part 1 - Study Design Related

Participation

Reasons for Non-participation

Comments were received which expressed concern that the study would be incomplete insofar as many subjects will not be able to participate. The reasons given in these comments were that subjects receiving payment from wind turbine operators are under "gag" orders and unable to divulge any adverse health impacts they may have. Another reason noted for non-participation was that there are apparently many individuals who have suffered so badly that they have abandoned their homes and would therefore not be recruited to participate. In addition to these reasons, some claimed that communities would be under pressure by neighbours to not participate.

Expert Committee Response

In any epidemiological study there is less than a 100% participation rate and there are a variety of reasons that could explain one's motivation to decline to participate in any given study. Health Canada will be unable to compel participation by specific individuals, as doing so would introduce a sampling bias into the design. And, if Statistics Canada specifically designed the sampling to recruit subjects that have relocated from their homes, for whatever reason, this would introduce bias into the design. A study could be designed to specifically recruit subjects that have relocated, but this would represent a different design from the one proposed by Health Canada/Statistics Canada. Members of the Expert Committee are not aware of other community noise studies e.g., studies of individuals living near high traffic roads or airports, that have included residents who have relocated for any reason.

The study will consider introducing a non-response analysis to support comparisons between responders and non-responders. Nevertheless, it has been noted by the Expert Committee that some individuals will be unwilling to cooperate and Health Canada/Statistics Canada may not know the reasons why. Health Canada has no formal knowledge that people receiving financial gain are under pressures that would preclude their ability to participate. It should be noted that this is an anonymous survey and no data at the individual level will be published.

Participation

"Most surveys - when they keep coming back to the people who refused earlier invitations to participate - find dramatic differences between the last to participate and those who accepted earlier invitations. In this case we'd be concerned that individuals who had symptoms or were annoyed by turbines might accept their first invitations to be studied, whereas those with no symptoms and no annoyance would be less likely to take part. Unless the latter, more reluctant individuals joined the study, the survey would generate an incorrect ("biased") higher rate of complaints than truly existed. Accordingly, the "standard" participation rate required before survey results are "credible" (likely to be believed) is >75%; and in many serious health surveys used to make serious decisions, it is demanded to be >90%."

Expert Committee Response

Initial refusals that are successfully converted by Statistics Canada can be compared to those who initially agree to participate. Systematic differences can be assessed. The Expert Committee agrees that the lower the response rate, the more difficult it will be to interpret the meaningfulness of the results.

International Participation

Questions were received with respect to whether individuals near wind turbines (WT) in other countries will be interviewed for this study?

Expert Committee Response

The study is being conducted in Canada only and individuals from other countries will not be interviewed. However investigators have been consulting with researchers in other countries with regards to the study design and questionnaire. Collaboration with international partners involved in WT research by the Health Canada research team is ongoing.

Control Group

Lack of Control Group

Comments were received suggesting that the study design is flawed because it does not include a control group.

Expert Committee Response

In an experimental design, a control group is a group that does not receive the treatment, but is similar to the group that does receive the treatment (treatment group) in every other aspect. The wind turbine noise and health study is not an experimental design, however if viewed as an experimental design the "treatment" in this case would be exposure to sound from wind turbines. A proxy to this is distance to wind turbines. Therefore, in the study, one could view the "control" group (or reference group) as the category or group which includes subjects that live in dwellings where wind turbine sound can no longer be reliably detected due to background sound levels. Because the study includes an assessment of low frequencies, which propagate further than higher frequencies, the "low frequency unexposed" group can be further away than the group that is considered to no longer be exposed to higher frequencies. Similarly, since the size of the wind farms in this study will vary, so too will the classification of the reference groups (i.e. for larger wind farms they will be at further distances).

A primary objective of this study is to apply statistical modeling to assess relationships between the sound/distance from wind turbines and both self-reported and measured endpoints. The propagation of sound from wind turbines is influenced by meteorological conditions and topography. For this reason, it would not be appropriate to recruit a remote sample of subjects as a "control" group because they would not necessarily be exposed to these same conditions. Such a group would also come at the expense of reducing the sample size of the exposed group. A much more robust approach is to assess how the prevalence of self-reported and measured endpoints change as a function of sound levels and distance from wind turbines. Once all data is collected and analyzed, the analysis will be adjusted statistically to account for socio-demographic factors and meteorological conditions, should there be differences in any of these factors or conditions between groups of subjects who live close to or far from wind turbines.

Expert Committee Composition

Expertise and Bias

Comments were received that suggested the Expert Committee established to lead this study does not have adequate expertise, and/or includes members with demonstrated bias.

Expert Committee Response

As with all research conducted by Health Canada, this study is impartial in nature and will adhere to strict ethical standards, and undergo a peer review process to strengthen scientific rigour. Efforts have been taken to ensure a thorough, transparent selection process and an Expert Committee with international representation, appropriate expertise and a declared lack of bias.

The collective mandate of the Expert Committee was to design a robust study that would assess public concerns that exposure to the sound from wind turbines is having a negative impact on human health. In the initial planning stages of this study it was clear that the Expert Committee would need to include members that had demonstrated expertise in multiple areas. Any member would however need to be without any perceived or actual conflict of interest. In an area of specialization where expertise is limited, it is widely understood that such individual experts may be called upon to provide advice to any number of bodies. Accordingly, it is important to note that provision of advice to a particular organization does not necessarily constitute a conflict of interest. Confidentiality statements were required from Committee members, and are common in such research, to protect against disclosure of certain aspects of the study design that could compromise the integrity of the results (e.g. the questionnaire content, study location, timing, etc.). Members also could not have any preconceived judgment regarding exposure to sound from wind turbines and health impacts. A final evaluation of members' declaration of activities and affiliations was carried out prior to acceptance of members to the Expert Committee for the purpose of determining whether any bias or conflict with the mandate of the Expert Committee was present.

The areas of expertise identified included:

  • Acoustics, with specific knowledge regarding low frequency sound and sound propagation, Sound Measurement and Modeling and a thorough understanding of the acoustical models/standards that pertain to environmental sound levels in general and sound from wind turbines, in particular;
  • Epidemiology, with specific knowledge in study design, sampling protocols, including the strengths and weaknesses of various approaches;
  • Statistics, including multiple regression modeling;
  • Health impacts of environmental noise in general and in response to sound from wind turbines, in particular;
  • Clinical medicine, including expertise in sleep and the health impacts of sleep disturbance;
  • Psychiatry, including expertise in mental health
  • Questionnaire design, development and testing.

The nature of the Expert Committee is such that additional individuals may be consulted or added as members in accordance with the needs of the Committee. For example, Health Canada is consulting with experts in meteorology and infrasound measurement to enquire as to the feasibility of introducing changes in design as a result of feedback obtained. Should continued engagement be required, these individuals will be requested to sign declarations of affiliations and interests and take part as members of the Expert Committee as necessary. It should be noted that several international members of the Expert Committee do have experience in wind turbine research and other international researchers not on the Committee have been consulted.

Abbreviated biographies for current Expert Committee members are now available on the Health Canada website.

Additional Members

Comments were received that suggested the Expert Committee should have representation from particular individuals with noted positions and particular experience on the issue of WTN and health. Similarly, there have been calls to have representatives from the public and the wind turbine industry on the Expert Committee.

Expert Committee Response

As with all research conducted by Health Canada, this study is impartial in nature and will adhere to strict ethical standards and undergo a peer review process to strengthen scientific rigour. Efforts have been taken to ensure a thorough, transparent process and composition of the Expert Committee with international representation, appropriate expertise and a declared lack of bias. The Expert Committee is not intended to be representative of the broad community at large, but rather a group possessing the necessary expertise to advance the study and support its achieving of objectives. The 60-day public consultation provided an opportunity for individuals outside the Expert Committee to provide input and as such to contribute to finalization of the design. Expert Committee members suggested that if a particular area of expertise was deemed necessary by the Principal Investigator and Expert Committee because the expertise was lacking from the current membership, additional representation should be considered. Overall, the Expert Committee felt that the individuals recommended in submissions during the consultation period did not possess unique expertise considered essential and therefore do not warrant inclusion.

Inclusion on the Expert Committee does not preclude the ability for Health Canada to seek opinions of other individuals. Furthermore the Committee has considered the literature by others in the development of the overall study.

Bias

Health Canada Bias

Comments were received that Health Canada should not be leading this study because of a perception that the Government of Canada is biased because it has supported wind generated power. Some comments suggest that Health Canada is biased because of statements made in published manuscripts that question the health impacts associated with WTN exposure. Some individuals/groups that have submitted these comments have expressed a desire for Health Canada to turn the study over to the Canadian Institutes of Health Research (CIHR).

Expert Committee Response

The 2011 Speech from the Throne emphasized Canada's support for "new clean energy projects of national or regional significance". With early indication of community concerns related to possible health impacts, Health Canada seeks to investigate any potential related adverse effects on the health of Canadians. Health Canada's role is to protect the health and safety of Canadians which would include seeking a clear understanding of any risks that may be posed to the health of Canadians by wind turbines.

Health Canada has developed expertise in acoustical measurement and health impact assessment through its responsibilities under the Radiation Emitting Devices Act (REDA) and the Canadian Environmental Assessment Act (CEAA), by providing advice to responsible authorities on sound measurement and the potential health effects of noise relating to sound levels, including that from wind turbines.

Advice provided by Health Canada officials is based on the best available science at that time. This advice is based on ongoing reviews of the scientific literature regarding sound caused by wind turbines and its effects on human health, World Health Organization guidelines, international standards, and knowledge gained from environmental assessment reviews.

There is currently limited scientific evidence to conclude whether there is a relationship between exposure to sound from wind turbines and harm to human health and there is a need to further investigate and support research in this area. All members of the research Expert Committee are either bound by the federal government Values and Ethics code (federal public servants) or have completed a declaration of affiliations and interests (external members) to declare the existence of bias that might in any way influence study outcomes.

The study design has been reviewed by the Health Canada Science Advisory Board and the World Health Organization Noise Committee. Advice or guidance provided by these organizations has been considered and, when appropriate, integrated into the design by the Expert Committee. Health Canada has also undertaken a transparent process to consult with the public on the study design and has engaged key affected stakeholders including advocacy groups and wind industry officials at various stages to discuss the study and its information requirements. All groups received identical information.

See also response to 'CIHR' under "Wind Turbine Noise and Health Study: Summary of Comments Received Part 2 - Not-Specific to Study Design".

Selection, Self-selection and Awareness Bias

Comments were received that suggested the study will be flawed because of selection, self-selection and awareness bias.

Expert Committee Response

Selection Bias includes any issues associated with sample selection that could influence the results of a study. This is discussed in more detail now in section 8.1 of the revised design paper. The most common concern in this regard is sampling bias, whereby subjects in a study are not randomly sampled. Self-selection bias occurs when subjects select themselves into a study for reasons that cannot always be known to investigators. People that self-select themselves to participate may have their own beliefs about what the study is evaluating, which makes it difficult to interpret the results. Awareness bias occurs when subjects in a study self-report more (or less) illness because of concerns arising from proximity to a perceived hazard. In the absence of any measurable biological effects, this type of bias can make it very difficult to understand the true health impacts associated with any given exposure. Even a study with a response rate of 100% can still be impacted by awareness bias.

The Expert Committee acknowledges that this study, like any epidemiological study, cannot completely eliminate all sources of bias. However, in many cases, it is possible to take steps to minimize bias and possibly to identify the presence of a bias and evaluate its impact on the results. With studies of this nature that rely in part on self-reported data, the possibility for contamination of results exists due to participation and awareness bias. Such an occurrence would make interpretation/evaluation of results difficult to carry out.

Some group members have concern that the study may not attain a 70-75% response rate. However this value is based on Statistics Canada's experience with similar types of field studies. A response rate of between 70-75% for the self-reported measures would effectively minimize the impact of self-selection bias. While the impact of self-selection bias may be minimized, it will not necessarily be eliminated. Those who have sleep disturbance, for example, may be less likely to agree to participate in sleep actimetry resulting in a self-selection bias whereby more subjects without sleep disturbance participate in the study. Similarly, those with compromised health conditions may be less likely to agree to participate, resulting in a self-selection bias.

Selection bias in this study will be reduced to the extent that Statistics Canada will target all dwellings within the highest wind turbine sound categories and employ random sampling methods for the dwellings at further set-back distances. Furthermore, in all cases, the one subject per home that participates in the study will be randomly selected. The participation rate in this study may see a decrease for the measured endpoints, especially hair cortisol. Statistical analyses after the study can assess any systematic differences that may exist in subjects that participate fully, partially or not at all. For example, an analysis by distance to the closest turbine can be carried out to reveal a potential selection bias in the sample. Moreover, subgroup analyses according to whether participants are receiving any financial compensation by the wind turbine industry and according to participant's perceptions towards wind turbines can be made in order to identify any potential self-selection bias issues.

Another type of sampling bias includes the timing of the sampling. For example, conducting sample recruitment at the same time of the day could lead to a bias because the sample would only represent those who are home at this time of the day. To minimize bias associated with time of sampling, Statistics Canada will visit homes at all times of day, weekdays and weekend days included. As mentioned above, only one person will be randomly selected per dwelling. While the possibility of sampling more than one adult per household was discussed, it was decided that this would lead to problems of multiple associations in the clustered data and was therefore not recommended. On conclusion of the study, should it be noticed that specific demographic groups are over-represented, statistical adjustments can be made to account for this.

Awareness bias will be difficult to fully eliminate in this study because there has been a long history of widespread claims that exposure to wind turbine sound causes adverse health impacts, which range from headaches, to nausea and even cancer. Most recently there have been more than 125 symptoms ascribed to wind turbines. The media and internet can therefore generate an awareness bias concerning the associations between exposure to sound from wind turbines and self-reported health effects. For example, it may lead to the situation where people are more aware and therefore tend to over-report symptoms or ill health compared to people who are less aware (i.e., live further away). The same issues can lead to situations whereby people under-report health symptoms. Accordingly, more emphasis will be placed on the measured endpoints since they are less sensitive to awareness bias and reflect the most serious potential health effects (i.e. stress and sleep impacts).

Nocebo Effect

Some comments expressed concern that the results of this study would not be reliable because of the "nocebo effect".

Expert Committee Response

The "nocebo" effect describes a phenomenon whereby people start to develop health effects because they believe they should, that is, they have preconceptions about what will cause health effects and develop symptoms based on what they believe to be true. This is in contrast to the well-known placebo effect whereby people can show relief from an ailment because they believe they are part of a treatment that is supposed to improve their health. In the context of exposure to sound from wind turbines there have been claims in the media and on the Internet that link this exposure to a number of adverse health outcomes. When people learn of these claims they may start to attribute what they may be feeling to the sound from wind turbines. Furthermore, people can become so concerned about these claims that they begin to worrying a great deal about their own health and this could potentially, in itself, have adverse consequences on their well-being, in addition to sleep quality. Some may also worry that even if they do not hear the wind turbines, the inaudible infrasound will impair their health.

While such expectations may influence an individual's health, the HC study is investigating health effects in relation to persons living in proximity to a wind farm, i.e., one's actual experience in relation to exposure is of primary importance, versus one's expectations.

It is the view of the Expert Committee that the possibility of a nocebo effect influencing the results could be assessed through the study design and with carefully worded content in the questionnaire. Health Canada/Statistics Canada are collecting both subjective or self-reported data through a face-to-face questionnaire and gathering objective endpoints such as blood pressure measures, hair cortisol analysis and sleep actimetry. Study findings will include analysis of self-reported endpoints and in addition, self-reported data will be correlated with the measured objective endpoints. It is possible that measured outcomes may not be consistent with self-reported endpoints; however, the prevalence of self-reported health impacts will be reported and shared. Even with these measures, if the nocebo effect does exist, it may not be entirely eliminated.

WTN Characterization

Modeling versus Measurement

Some of the feedback received commented on how Health Canada plans to characterize WTN and expressed specific concern that noise measurements would not be carried out in every home in the study. Many of these same comments also indicated that modeling was a completely inappropriate approach because of the uncertainties that exist with noise modeling. There was general concern that WTN is influenced by weather and that Health Canada would not be able to account for this through modeling. Some comments suggested the reverse; that is, because measurements are susceptible to transient events, they would produce much less valuable information than modeling which could account for variables that short term measurements could not.

Expert Committee Response

An ideal study would be one that included continuous long-term measurements indoors and outdoors at every dwelling. Ideally, these measurements would capture all imaginable variables that can influence sound from wind turbines and they would be carried out across a broad spectrum including infrasound and audible frequencies at every planned and operational wind farm in Canada. A study of this nature would be exponentially more expensive and theoretically take many years to complete. Time and resource constraints make measurement at every dwelling impossible.

This study will use measurements as a way of reducing the uncertainty associated with wind turbine sound modeling and not for the purpose of characterizing wind turbine sound pressure levels at each dwelling. Indeed, it is the opinion of the acoustic experts on the Expert Committee that there are limitations to doing direct measurements, and that modeling can actually be a superior approach (factors to account for wind classes, seasons, in situ conditions can be more easily accounted for in modeling). Spot checks will be done to ensure modeling is sufficiently accurate. The acoustic experts state that modeling is a common approach in socio-acoustic surveys, and also in the peer-reviewed wind turbine socio-acoustical surveys. It is well recognized that modeling is an effective and reliable method of measurement and is known for good predictability and prediction compared to direct measurements. Furthermore, modeling allows for acquisition of knowledge which can be used in predicting for future wind projects in Canada.

In this study, modeling will be A-weighted so that comparisons can be made to existing research, however the Expert Committee recognizes that there are concerns over the utilization of A-weighting for sources that contain low frequencies. For this reason, other weightings (C and G) can also be modeled. This study will rely upon sophisticated software models/algorithms, such as NORD 2000, CadnaA/Harmonoise and ISO 9613 in 1/3 octave bands.

Information will also be requested from wind turbine operators since they may have mitigation measures (i.e. turning off the turbine, different operational modes) or other features (see next comment) that can reduce sound levels. This information is required to reduce the uncertainty in modeling.

Tones, Impulsiveness and Amplitude Modulation

Some of the comments received expressed concern that WTN is tonal, impulsive and modulated, but Health Canada's approach to WTN characterization would not be able to account for these features.

Expert Committee Response

Obtaining technical data post study from operators may support an analysis of these features. Health Canada has initiated discussions with operators to obtain information on available data that operators collect. The questionnaire will, to the extent possible, probe the subjective evaluation of wind turbine sound characteristics, however this is challenging because of the ambiguity that exists in how people describe sound sources. Statistics Canada cautions against the use of terminology in a survey that people cannot easily describe (e.g. swooshing, beating, impulsive, etc.). Local temporal effects that are not easily characterized drive amplitude modulation. This is a dynamic effect so relevant measurements are beyond the scope of this study. Health Canada may draw upon the results of new work in Japan and the ongoing UK study on amplitude modulation, which is currently unavailable.

Accounting for Other Noise Sources

There was concern expressed that this study will not be able to distinguish between other sources of environmental noise and WTN. Some of the identified sources of noise that could contaminate WTN were road traffic, farming equipment, rail noise and aircraft noise.

Expert Committee Response

The study will use models that already account for these phenomena, using CadnaA, traffic noise-modeling, modeling for rail (by hand or CadnaA). Furthermore, the questionnaire will seek information on individuals' sensitivities to certain types of noise, for the purposes of relative comparison, allowing investigators to assess the relative response to sound from wind turbines and other sources within the same subjects. Furthermore, because of the exposure-response design it would be statistically unlikely that any associations found with wind turbine sound levels would be due to some other source.

Infrasound

Suggestion to include infrasound

Some research "on guinea pigs shows that the outer hair cells of the auditory system can respond to infrasonic stimuli that are below the threshold of audibility. The contention is that this stimulation may account for the adverse impacts that communities make reference to. This, in addition to the research on human exposure to infrasound conducted in the past has led to the suggestion that this study needs to include an assessment of infrasound."

Expert Committee Response

There is ample scientific evidence that high levels of infrasound exposure can produce adverse responses in humans. However, the research on infrasound and human response shows that the sound pressure levels need to be much higher than anything that has been reported from modern day wind turbines.

Infrasound in the environment is everywhere, but is exceedingly difficult to measure and model. The Expert Committee agreed that this study should include an assessment of infrasound from wind turbines. At the present time Health Canada has decided to pursue infrasound measurements and modeling. The infrasound measurements will take place at 4 setback distances (up to 10KM) over the course of an entire year. At certain points in the measurements, the wind turbines will be turned off so that there can be certainty as to the source of measured infrasound. Representative data will be collected for between 1- 3 weeks in each season and this data will be used, in combination with time matched meteorological data, to refine predictions of infrasound transmission from wind turbines. The parabolic equation will be used to predict propagation of infrasound. The intent of the infrasound sub-study is to capture frequencies as low as 0.5Hz using specialized equipment designed for these types of measurements. In addition, short-term sound level measurements in the higher frequencies will be made with unattended sound level meters.

Upon completion of the study, wind turbine operators will be asked to provide historical sound power output data to allow Health Canada to model sound from the wind turbines. Since this research study deals with health impacts that represent different time periods (1 year, 30-days, 7-days, etc.) the historical data will need to capture these time references used in both the questionnaire and measured sleep actimetry.

Modeling WTN

CadnaA model

Concern was expressed over the choice of the CadnaA model to predict sound levels at receptors.

"It is supposed to follow ISO9613, however it violates the height limitation of 30 meters required for ISO9613 applicability to ground based noise sources. The large Industrial Wind Turbines (IWT) today approach heights of 150 meters, which is the elevation of aircraft on final approach. ISO9613 is specifically designed for prediction of noise from ground level noise sources. Today's IWT'S do not fit the ISO9613 design criteria."

Expert Committee Response

Subjects' wind turbine sound pressure level exposures will be based on modeled values. Constraints of the research make measurement at every dwelling and turbine impossible and not necessarily even desirable. The study will use a sub-sample of measurements as a way of reducing the uncertainty associated with modeling and not to characterize wind turbine sound pressure levels at each dwelling.

The study will use a range of prediction models, from those that can be implemented by the general public to those models that can only be implemented by a few experts. All models will be evaluated by their ability to explain variance in the survey results. At the conclusion of the study there may be opportunity to follow up on responses that do not conform to the general trend.

Due to the widespread use of ISO9613 for wind turbines and other sources it must be included in the study.

Simultaneous sound level measurement

Concern was expressed with respect to research design criteria where indoor sound levels would not occur simultaneously with individuals wearing actimeters. The respondent advised that measured sound levels were needed to correlate to sleep disturbance and reliance on model-based noise calculations for this correlation would be an oversight.

Expert Committee Response

Local wind turbine sound pressure levels will be monitored in some areas during actimetry. Indoor sound pressure levels will not be measured because these measurements are difficult to interpret and poorly correlated with outdoor noise levels (10 to 20m from dwelling). In addition, relevant measurements do not distinguish between sleep disturbance caused by snoring of the subject, versus disturbances from snoring of the subject's partner. Overall, A-weighting cannot distinguish between the sound of a lullaby and an alarming sound; criteria for this subtle distinction have never been developed.

B&K PULSE System

The protocol prescribes the use of B&K PULSE systems for measurement of both source and recipient noise. This sophisticated instrument may not be the most adequate for the purpose. The PULSE system is designed to measure sound and vibrations within a very high dynamic range, such as car crashes. This is done by combining two high resolution A/D converters with different preamplifier gains in such a way that when the high resolution low level A/D converter is close to saturation, the digital output is switched to the lower resolution high level A/D converter. The converters are continuously calibrated to ensure a seamless transition. The consequence is that high level inputs are converted with considerably lower resolution than low levels, and therefore low level signals of one frequency range may not be detected if high level frequencies of another range are present.

Expert Committee Response

Wind turbine sound levels, even underneath the turbine itself, should not normally exceed 100 dB at any frequency. For this reason, the B&K PULSE should be more than adequate as a measurement system in the study. However, should the available dynamic range on the PULSE system pose a problem, we will use the Bruel & Kjaer 2270, which also has a 24 bit digitiser, but does not employ the Dyn-X technology used in PULSE.

Wind Filters

"Since noise signals from modern >1.5MW wind turbines in the given measurement situation have fairly limited dynamic ranges, and wide spectra of limited variation, high resolution conversion of the entire spectrum can be achieved with a single 24 bit Sigma-Delta converter, a distance-dependent attenuator and one or two well-specified and accurate pre-emphasis analog filters (similar to, but not identical to A-weighting). Two filters are proposed to take into account that higher frequencies are absorbed much more in the atmosphere than low frequencies. To assess the pulsating pattern of WTN, wide band recordings with very low limiting frequency is required. 0.1Hz or less should be possible with precision 1" microphones, however the sensitivity to influence from wind turbulence becomes an issue and carefully designed windscreens are required."

Expert Committee Response

This type of measurement is possible using a microbarometer with a large radius wind filter constructed from porous (soaker) hose. This equipment will be used in the study to validate the low frequency noise and infrasound propagation modeling. It should be noted that infrasound propagation, including the effects of temperature and wind speed gradients, is being investigated as a separate part of the study (see above).

Propagation

"It is known from previous investigations, that the propagation of sound in the lower atmosphere can be nonlinear (van den Berg and others). The assessments of temperature gradients are therefore important, but not trivial to measure in the context of this study. It is proposed that nonlinear propagation properties of low frequency sound in the atmosphere are recognized as a potential important phenomenon that should be investigated in a separate study. To be practical, the noise emitting properties of a wind turbine should be characterized in such a way that it is possible to predict the emitted noise spectrum with good accuracy for any relevant wind speed, based on a source noise measurements (ISO) at a few low and high wind speeds combined with the certified power/wind-speed curve."

Expert Committee Response

Dr. van den Berg (who is a member of the Expert Committee) indicated that the flaw in modeling that was identified in his thesis was not due to incorrect noise propagation modeling, but to a wrong understanding of wind behaviour. It is also important to note that the decibel decrease in sound pressure level per doubling of distance is not a constant in any of the models that will be used in the study.

The term non-linear propagation is generally only associated with sound pressures levels of the order of 140 dB or higher. Our proposed method of sound modeling measurement will be used whenever possible. In addition, measurement spot checks will be made to validate these data at each location.

Data integrity

"It is recommended, that the digital data recording devices are equipped with a GPS receiver to ensure proper location tagging and time stamping. The recording header should also contain information of the settings of attenuator and filter. Once the noise properties from the wind turbines in a park have been determined, the long-term exposure at a given location can be calculated with reasonable accuracy for low frequencies based on atmospheric data."

Expert Committee Response

The recommendation is consistent with good practice and will be followed in principle.

Seasonal Differences

Clarification was requested on how seasonal issues will be taken into account.

Expert Committee Response

In the noise modeling, seasonal issues will be taken into account using meteorological propagation classes and weather statistics.

Topography

A comment was received that computer models that wind companies use are all based on flat land and do not consider land topography.

Concerns were received that the study will not be able to relate symptoms and outcomes to WTN and that turbine power output is a poor proxy for noise levels at an individual dwelling. Noise levels will depend upon wind direction, wind speed, atmospheric stability (wind shear), ground conditions and topography as well as the structure of the house itself. Noise measurement within the property was therefore felt to be essential.

Additional questions were received inquiring whether the impact of wind turbines on ridge lines would be considered as well as the impact of varied geological conditions i.e. soil foundation versus rock.

Expert Committee Response

Noise modeling will incorporate wind direction, wind speed, atmospheric stability (wind shear), ground conditions, topography, the structure of the house, as well as the turbine power. The modeling will be validated by spot check measurements within the property.

As per ISO9613, the sound pressure level predictions must be interpreted as representing the 'most likely' levels, provided that propagation parameters in the model i.e., wind, temperature, topography, reflect the local conditions. This assertion is supported by extended measurements of WT sound.

It should be noted that Canadian wind power projects typically use ISO9613 which does account for topography in propagation. However, wind turbine manufacturers using IEC Standard 61400-11 routinely characterise wind turbine sound levels on flat land. Health Canada's study will include spot checks of wind turbine sound levels. These spot checks will help identify consistent deviations from the manufacturer's specifications.

Assessment of the impact of ridge lines will depend on whether any of the sites studied have WTs built on ridge lines. As indicated above, topography and different geological conditions will be factored into the noise modeling methodology being used.

Wake-induced Noise

A submission was received that addressed the impact of wake on noise (wake-induced noise) production from wind turbines. This information was provided so that the study can consider the impact that wind farm layouts can have on the sound characteristics and ultimately on community response.

Expert Committee Response

Wake could influence the sound production of a second WT by increasing turbulence inflow noise. This study will consider the impact of wind farm layout by including parameters related to wake effect, such as relative spacing between turbines, for each WT farm.

Sample Size

There were some comments that indicated that the sample size is too limited to statistically reveal differences in health endpoints that are relatively rare. The comment was that if we do not show an association with WTN that our sample size was simply too small.

Another comment pertaining to sample size and proposed setback distances was that the study needs to ensure there is an adequate sample close to the turbines in order to reveal any statistical differences (or lack thereof) on health endpoints. There was concern that by having too many subjects at very far setbacks would wash out any effects for those that live closer.

Expert Committee Response

Research such as that undertaken by Health Canada on the sound emitted by wind turbines is costly. The budget for this study permits an initial sampling target of 2000 dwellings within various distances from 8 to 12 wind farms in Canada. The sleep literature indicates that rates of sleep disturbance in the general population are approximately 10%. According to Statistics Canada, the anticipated occupancy rate in rural Canadian communities is approximately 80% and the anticipated response rate for the questionnaire is between 70-75%, among which approximately 20% will be within the closest distances to wind turbines. This will result in a total sample size of approximately 1200 subjects. These calculations indicate that this study should have sufficient statistical power to detect a 7% difference with 80% power and a 5% false positive rate (Type 1 error). There is a measure of uncertainty in this power assessment as this is the first study in this area to implement both measured and self-reported endpoints. A recent paper by Nissenbaum et al (2012) reports dose-response relationships for multiple measures of sleep disturbance and distance to wind turbines with a sample size of only 79 subjects from 56 different homes. The Health Canada study represents a much larger and more comprehensive investigation than any other study in this area to date. Ultimately, these research findings will contribute to the body of international peer-reviewed scientific research examining the community impact of wind turbine sound levels, and may also lead the way for supporting future studies examining this complex issue.

The Expert Committee has considered the potential impact on the participation rates for the measured endpoints due to the perceived inconvenience to the subjects. There is a possibility that the response rates could be as low as 60% for the measured endpoints making interpretation of results more difficult. This possibility is however mitigated by the high level of motivation expressed by communities requesting involvement. Additionally Statistics Canada surveyors are trained to convert initial refusals. For these reasons, and the added value of having measured endpoints, the Expert Committee felt strongly that the study will be more informative if it includes these measures.

The revised, more detailed, design paper provides additional information on sample selection. The conclusions drawn from the study will apply only to wind farms included in the study and cannot be generalized to all Canadian wind farms. An in-depth investigation including random selection from all types of wind farms from every province in Canada would need to be undertaken before such generalizations could be made.

Design (General)

Cross-Sectional Design

Some comments pertained to the actual study design as being inappropriate to meet the desired research objectives. Some suggestions were that the study should be longitudinal over several years to track the development of adverse health impacts before and at several points following operation of the wind turbines.

Expert Committee Response

There are strengths and weaknesses to any epidemiological design. The current cross-sectional study is an observation study at one point in time among a sample of subjects living various distances from wind turbines. It does not have the strength of an experimental design that would permit inferences about causality. The strength of the current design is that it can capture a relatively large sample size, examine multiple outcomes, and provide timely answers using a well-accepted methodology that includes random sampling. In the case of sleep actimetry, the collection of data over 7 days represents a repeated measures design for this aspect of the study. Multiple measures of sleep data from actimeters worn by subjects will be analyzed in relation to time-matched wind turbine operational data. Therefore, assumptions should be possible and will allow for generation of hypotheses regarding the possible associations between calculated wind turbine sound pressure levels and the actimetry measured endpoints. It is also important to consider that results may be affected by uncontrolled confounding factors, which is a limitation of this study design.

A longitudinal study would improve our ability to assess whether reactions to wind turbine sound change with one's history of exposure (habituation or sensitization). However such a design would still be subject to the same limitations of awareness bias and participation bias as well as the inherent difficulty in a longitudinal design of losing subjects due to moving or dropping out of study for other reasons (subject attrition).The costs of a longitudinal study would be far in excess of the current study and the timeframes would be lengthy. An approach suggested by the Expert Committee is to statistically account for the differences in operational times for the various wind farms selected in addition to the self-reported duration of residency.

Sampling

"The desired results will be achieved only if the distribution of the sample set is defined more clearly. For example, the following statement needs clarification: 'The sample will consist of 2000 dwellings at setback distances from less than 500 metres to greater than 5 kilometres from 8-12 wind turbine power plants'; or alternately 'The study will be conducted on a sample of 2000 dwellings randomly selected from those located near 8 to 12 WT installations in Canada'. Does this mean the dwellings will be related to installations of 8 to 12 wind turbines, only? (There are only 7 of such arrays identified by the existing listing of wind turbines in Canada). Or does it mean the dwellings will be related to 8 to 12 installations of wind turbines?"

Expert Committee Response

The updated design paper provides more details on sample selection. The results of this study will not be applicable to wind farms that are not part of this study. An analysis of how communities compare would be mandatory before such comparisons can be made.

Design (Suggestions)

Use of Medical/Death Records

Recommendations were received that the medical records of participants also be examined in order to establish other factors that may lead to variations in blood pressure or other symptoms. Controls for other measures of stress were recommended as well as use of different measured endpoints.

The study Expert Committee was also advised to access available public death records from all wind turbine installation areas to look for correlations.

Expert Committee Response

A study that included medical records was something that was considered at the early stages of this study. In some provinces the Longitudinal Health Administrative Database is available and if subjects consent to linking their health information by agreeing to provide their health card number, we would be able to have access to this data. This is being investigated as a possibility for this study but has not yet been confirmed. Generally, self-reported health impacts do positively correlate with measured health and subjects should be able to recall accurately if they have been diagnosed by a physician as having any of the particular health conditions asked about in the questionnaire developed for this study. It is also possible that some subjects may experience impacts on their health and well-being that they have not discussed with a physician, for a variety of reasons.

The study does not include collection of records on deaths from areas around wind turbines.

Affidavit

One respondent suggested the use of an affidavit to encourage subjects to be honest.

Expert Committee Response

Statistics Canada has extensive expertise in collecting data from interviews and the suggested approach of requesting Canadians to sign an affidavit is not part of their policy in conducting surveys and would also be considered inappropriate. Trained interviewers will be administering the questionnaire and physical health measurements in respondents' homes during which time a rapport and trust with respondents is established. Furthermore, Statistics Canada is legally authorised by the Statistics Act to collect data from respondents and they in turn guarantee that confidentiality will be preserved (according to the same Act). To increase participation, the Act is there to reassure the respondent that their personal responses are protected by law. Given the fact that Statistics Canada offers confidentiality protection, ensuring that the data is anonymous, respondents are more likely to be truthful in their responses to questions.

Community Health

"A significant concern to councillors and citizens or rural communities, is that any "health study" of the impact from wind turbines must look at the impact on community health and relationships posed by wind turbines. Wind turbines are driving a huge divide into rural Ontario, and causing immense harm to the fabric of community spirit. This impact on community health needs to be included in the Health Canada study."

Expert Committee Response

This study has a mandate to assess the potential health impacts associated with exposure to sound from wind turbines. Health Canada is aware of the concerns that have been expressed by some members of communities who report that wind turbine farms are causing a divide amongst residents. Although the main focus of the study is on how individuals are impacted by sound levels/characteristics, some items on the questionnaire do relate to one's perception of the community as a whole.

Guidelines compliance

Comments were received urging that the study must not only look for compliance with already published guidelines.

Expert Committee Response

It is not an objective of this study to look for compliance with already published guidelines.

Results Interpretation

Correlation versus causal effect

Comments were received that questioned the ability of the study to differentiate between correlation and a causal effect.

Expert Committee Response

The design of this epidemiology study permits conclusions with respect to associations between the various endpoints assessed. The study does not have the strength of a controlled laboratory study whereby the independent variable can be manipulated to evaluate the impacts on the dependent variables. It would be misleading to make causal inferences based on statistical associations found in this study. Conclusions about exposure to wind turbine sound are strengthened with respect to sleep actimetry because multiple measures of sleep will be captured over time which can be correlated to multiple measures of wind turbine operational data.

General

Some respondents felt that the study was flawed from an epidemiological standpoint.

"The investigators state that prevalence data on community reaction and self-reported health concerns 'could then be compared to the prevalence of other community health concerns and also to the prevalence of similar health concerns in communities that are not situated near WT installations'. However, the present proposal does not appear to be gathering data from such a control group. A participation rate of 70-75% could still lead to biased results. However the proposed study as set out is deeply flawed. It fundamentally fails to acknowledge the numerous deficiencies in the study design that are basic requirements of any epidemiological study. At best the results will provide even greater uncertainty to the scientific debate that surrounds the industry that has been promoted by vocal but scientifically ill-informed special interest groups. At worst, the study could produce misleading evidence that in turn could influence policy making at a national and international level. Should this in any way influence policy makers and energy companies to move away from renewable technologies to for example increased reliance on fossil fuels, the potential health (& safety) consequences (occupational and public) could be catastrophic."

Expert Committee Response

The Expert Committee has determined that it would be inappropriate to make comparisons observed in this study to those reported in other studies because there could be important differences in not only the sample, but in how the data was collected.

The Expert Committee notes there remains the possibility that the results could be biased even with a 100% participation rate. Careful measures will be implemented to support assessment of the possible impact that bias may have on the results, if evidence suggests it is present (see above comments on bias). Furthermore, the revised research design summary highlights in more detail the scope and limitations of the current design (see section 8.1).

Nonetheless, while the data obtained will contribute to the global knowledge of the relationship between exposure to sound from wind turbines and health, results will not provide a definitive answer on their own and no decisions should be made on the basis of one study. Results will support decision makers by strengthening the peer-reviewed scientific evidence base that supports decisions, advice and policies regarding wind turbine development proposals, installations and operations in Canada.

Contributions to Guidelines

"Will this study be able to provide guidelines for modifications to wind turbine operation? E.g., Curtailment under specific conditions (wind speed & direction, and other weather conditions), between certain hours? Will the study result in mitigation measures that can be used to lessen any impacts from wind turbines, if found? For example, earthworks and landscaping that can be done to reduce both the visual and sound related impacts of wind turbines? Or by reducing visibility by reducing visual contrast of wind turbines with the landscape? Or recommended changes in operating patterns?"

Expert Committee Response

Health Canada's research study will support decision makers by strengthening the evidence base of peer reviewed scientific research that ultimately supports decisions, advice and policies regarding wind power development proposals, installations and operations in Canada. As such the results of this study will be shared with all stakeholders who will decide, in the context of their respective jurisdictions, the extent to which results may be implemented into their decision making processes with respect to wind turbines.

Causal attribution

"It has already been determined through other research that some people who live near a wind farm are disturbed by the noise while others are not, and that some people report experiencing adverse health effects, while others do not. Will this study be able to help answer why this is the case?"

Expert Committee Response

The wind turbine noise and health study will provide additional scientific insight in the area based on significant associations. However the study design does not allow investigators to address individual characteristics of the subjects in the study and whether these characteristics are causing subjects to be disturbed by wind turbine sound or causing them to have an adverse health effect as a result of exposure to sound from wind turbines. The study will examine significant associations which may include associations between self-reported health characteristics and annoyance from wind turbine noise.

Risk factor identification

"Will the study be able to determine risk factors for an individual to develop health effects?"

Expert Committee Response

The study will be able to show which variables are statistically related to one another. As such, it will be possible to show, for example, if annoyance from wind turbine sound is a predictor of, elevated cortisol concentrations in hair. This is not however equivalent to identifying risk factors. This takes a tremendous amount of research, typically done over many years, across very divergent populations, under multiple conditions. An example of such research would be the identification of high blood pressure as a risk factor for heart disease and stroke.

Measured Endpoints

Cortisol in Hair

Comments were received that expressed concern that hair cortisol concentrations may not capture a person's true stress levels.

Expert Committee Response

Cortisol is only one of several biomarkers that respond to stressor exposure. Hair cortisol is not a new measure; however, it has not been used previously in environmental noise studies. Nevertheless, there is a growing reliance on hair cortisol among scientists interested in studying the impacts of long-term stressors. The advantage of relying upon hair cortisol results, over blood or saliva and even urine, is that cortisol accumulates in hair such that 1cm is approximately equivalent to 30 days. As such it can be more informative when the purpose of the study is to assess long-term impacts on a biomarker of stress. Hair cortisol is not as sensitive to transient stress responses, making it desirable for the purposes of the study. Hair cortisol provides a unique opportunity to assess the possibility that exposure to sound from wind turbines may be associated with long-term stress responses. While hair cortisol concentrations may not show differences among the sample if the magnitude of the stressor is not severe enough or chronic in nature, an outcome such as this would also be informative and would add to the knowledge base in this area.

Sleep Actimetry

Some comments expressed concern that because WTN is influenced by so many variables, measuring sleep impacts for only 7 days of sleep actimetry is insufficient. The suggestion is that the study would need to measure sleep for at least an entire year. It has also been expressed that a sleep pilot study needs to be done to evaluate the usefulness of this endpoint, including the need to ask those wearing the watch about their self-reported sleep quality.

Sleep assessment needs to include arousals that the sleeper may not be aware of.

"Actigraphy has advantages of simplicity and body movements are a reasonable proxy for awakenings and arousals. However, the variability in baseline nocturnal movement indices between individuals will make it very difficult to attribute any night-to-night changes to WTN. In my opinion, it is important that subjects have a true zero WTN exposure as well as the variable exposure of "normal" exposure. Thus subjects will have to be studied, ideally, in their own beds, with the turbines in situ but not turning. This can be achieved by requiring the operators to shut down operations for a minimum of three nights before and after the test period to provide a non-exposure baseline. Alternatively, a site that has been constructed but not yet operational could be studied before and after operations commence. Alternatively, subjects could be asked to sleep away from home to provide the non-exposure data although such an option is very much second best. Consideration should be given to the use of single channel EEG based devices such as the Zeo. While not as easy to use as actigraphy, they provide a hypnogram and a better analysis of sleep quality and arousals. They compare very well on cost. Limited studies on subjects exposed to WTN, which I have conducted, suggests that the Zeo shows a clear difference between exposed and non-exposed nights. Monitoring of wind speed and power output during the study period and comparison with historic records is mandatory to forestall any attempts to reduce noise output during the study period."

Expert Committee Response

The sleep study literature indicates that actigraphic studies be undertaken for a minimum of 3 consecutive 24-hour periods (Thorpy et al., 1995). For individuals with insomnia, a period of at least 1 week has been suggested as a time period which would provide reliable estimates (Van Someren, 2007). Sadeh (2011) recommends that 5 days or longer be undertaken in actigraphic studies as this period of time reduces inherent measurement errors in actigraphy and increases reliability. Therefore the 7 consecutive days, 24 hour per day time period selected for the present study is well supported in the literature and is more than adequate for evaluating sleep in our non-clinical study sample. A sleep pilot was conducted by Statistics Canada on an in-house sample of 27 subjects, 11 males and 16 females, ranging from 27 to 56 years of age. The purpose was to evaluate which sleep watch model was most effective for the purposes of our study and whether a sleep diary was needed. The results indicated that sleep actimetry does provide a valid measurement of sleep parameters (total sleep time, awakenings, sleep latency) and is associated with self-reported sleep as recorded in the sleep diaries that each subject was asked to complete on a daily basis over a 7 day period. It was therefore decided that sleep diaries were not required for this study. Specific sleep watch characteristics were evaluated and specific ones were established as necessary for selection of the sleep watch. These sleep watch characteristics were then specified during the competitive purchasing process.

Awakenings/movements that the sleeper may not be aware of can be detected by the sleep watch if there is an associated increase in physical arm movements. The sleep literature and the results of pilot testing indicate that sleep actimetry is a valid, reliable method of evaluating sleep parameters.

Consideration also needs to be given to the inconvenience imposed on participants in requesting them to wear the sleep watch 24 hours a day for even a 7 consecutive day period. This duration was partially determined in conjunction with Statistics Canada and their experience with compliance in subjects wearing activity monitors in pilot testing. It was felt that a longer timeframe would likely impact on compliance rates; and the 7-day time period was supported in the literature and was not overly burdensome as a 2 week time period or longer would be. However, the possibility exists that, for any given individual, the results obtained from the sleep actimetry may not represent their overall sleep response because their exposure to sound from wind turbines during the 7 days of testing may not represent their "typical" exposure. The most dramatic example of this would be a scenario where there is insufficient wind to make the wind turbines operational. However, as we plan to analyze the actigraphic data partly by correlating it with wind turbine sound measurements and operations, we will be able to examine the frequency of awakenings during higher noise times when we correlate the WT operational data with actigraphic data.

The Expert Committee indicated that concerns pertaining to the 7-day actimetry period not being a representative or typical sleep pattern would be addressed as the results will be averaged across the entire sample. Thus, the overall collection period should capture the "typical" operational conditions for the study sample. The strength of the sleep actimetry is that data will be collected over multiple days and time-matched with wind turbine power output. Therefore any changes in measured sleep parameters can be examined in the context of parallel changes in wind turbine operational conditions. Data from operators will only be requested upon completion of the study. This should eliminate the concerns raised by some comments that sound reduction procedures will be implemented during the data collection period.

Expert Committee sleep experts indicate that it would be better to study subjects in their own beds with and without the WT turning if this were feasible, however awareness bias could impact the results. The Expert Committee was not in agreement with the suggestion of comparing people sleeping away from home (with no WT) versus in their home (with WT) as there would be too many factors that may result in poor sleep when sleeping away from home. Furthermore, Expert Committee sleep experts noted that there is not enough data on the Zeo device to conclude that it compares well to actigraphy. One study compares Zeo to PSG in 29 subjects and another study of 5 subjects that show positive results. The larger study with 29 subjects was published by authors working for or with the Zeo sleep center. There is another study published recently which shows that the Zeo is not accurate in detecting wake episodes.

The Expert Committee agrees that monitoring of wind speed and power output from the WT during the study period is necessary and as such analysis will attempt to correlate real-time operational data from the WT to the actigraphic data. Furthermore, information on window position in sleeping quarters will be attained to improve the estimates of indoor sound pressure levels from wind turbines.

Blood Pressure and Heart Rate

Comments were received that question the value in measuring blood pressure and heart rate at a single point in time because so many variables influence blood pressure. Some have also suggested that we need to ensure that subjects are looking at the wind turbines during the measurements.

The design must be specific as to whether "taking blood pressure following a standardized protocol" involves a single point measurement, or tracking blood pressure as turbine conditions change.

Expert Committee Response

The Expert Committee agreed that it would be inappropriate to take measurements in situations that could temporarily inflate blood pressure and that it is important to measure in a neutral environment following standardized protocols. The purpose of this study is not to measure blood pressure changes in response to operational wind turbines, but it is to examine whether there is evidence that living in areas with higher levels of wind turbine sound is associated with changes in average diastolic and systolic blood pressure and heart rate. The reason for this is that the study is assessing changes in blood pressure as a function of chronic exposure to sound from wind turbines, controlling for confounding variables. For this reason the Expert Committee agreed that there is value in measuring average blood pressure (diastolic and systolic) and heart rate in this study. The Expert Committee agreed however that it is important to account for variables that are known to influence blood pressure readings (e.g. smoking, caffeine, food consumption, activity level, age, etc.). These questions are being posed during the face to face questionnaire.

The blood pressure measurement protocol is in accordance with published best practices, and the standardized protocol that is outlined for the automated oscillometric device being used in this study: the BpTRU™ BPM-100 (BpTRU™ Medical Devices Ltd., Coquitlam, British Columbia). The BpTRU™ meets the Association for the Advancement of Medical Instrumentation standard and the British Hypertension Society protocol. In this study, subjects will have been seated in a relaxed position for a period of 30-min (while completing a questionnaire) prior to their blood pressure being taken. In a quiet room, subjects will have their blood pressure and heart rate measured using an automated measurement system. A total of six measures will be taken that are separated by 1-min intervals. The first measurement will be discarded and the remaining five will be used to provide averages for diastolic, systolic and heart rate. The physical health measures specialist will only be visible to the respondent for the first measurement to ensure that the device is functioning correctly; for the subsequent blood pressure measurements, the specialist will be seated behind the respondent. The respondent is briefed beforehand regarding this protocol. The device will be positioned in such a way that they will not be able to see their results. Subjects will be instructed to not speak during measurements. Expert Committee physicians agree that examining average blood pressure and heart rate in relation to modeled wind turbine sound levels/distance is a reasonable and feasible methodology and analysis plan.

Stress Levels

"Obviously you will have to conduct this study in some area that already has the wind turbines installed in it. Is it not entirely possible that the earliest effects of abnormally high levels of stress (such as sleep loss) could have passed by the time you study these human subjects, and that that stress, or other factors relating to the new disturbances, could eventually give rise to unanticipated disease processes?"

Expert Committee Response

It is true that an individual may adapt to an early disturbance and/or stress level presented in their environment. However if sleep disturbance is a persistent problem for individuals living near wind turbines, this will be seen following the analysis of actigraphic data. This actigraphic data will be compared to actigraphic data from individuals living far from wind turbines which will provide information as to whether sleep quality or quantity is significantly different in these groups, while controlling for other factors that may affect sleep. The sleep parameters that will be compared using actimetry are: number of awakenings, sleep/wake times and sleep latency. The Committee's sleep experts indicated that once the period of high stress and sleep loss has passed, it is not valid to argue that there would be ongoing problems. People pay off sleep debt rapidly even after several days of total sleep loss. Sleep debt as measured electrophysiologically and neuropsychologically is nearly fully paid back after 2 night's sleep.

Audiometric testing

Comments received suggested the required components of the epidemiological data collection should include a full "hearing test" for all parties from whom epidemiologic data will be collected in order to ensure that those with any physical hearing conditions that is known to be exacerbated by WTN can be added to a high risk profile sub set of data that might be used as a permission determinate triggering a "tighter" set of noise parameters for the location of parties falling into that high risk sub set.

Expert Committee Response

While there may be value in conducting a thorough audiometric assessment of all subjects in this study, this would preclude the ability to collect other measures and also reduce the amount of content that could be asked during the interview. For audiometric assessments to be valid they need to be conducted in a very quiet ambient environment. Health Canada has done this research in the past using a modular sound booth. However, the Expert Committee has noted that audiometric thresholds obtained in quiet environments are poor predictors of real-life hearing abilities in "common" noisy situations. Furthermore, there is no evidence that hearing conditions are exacerbated by exposure to wind turbine sound. This data is being collected on a sub-sample of subjects in a large-scale epidemiology study currently being conducted in Japan on communities living near wind turbines and therefore Health Canada may be able to draw upon the results of the Japan study.

Other Measured Endpoints

Additional Endpoints

Comments received recommended other possible endpoints be considered.

"A further application for the wind turbine noise recording could be to test for balance disturbance/vertigo/nausea in any subject reporting such an effect, by using electronystagmography to measure, amongst other vestibular involvements, a measurable nystagmographic Tullio-like effect from the WTN recording stimulation. Include a whole-body vibration study, changes in fatigue and mood. The study should include the Beck Depression inventory and other tools."

Expert Committee Response

The Expert Committee acknowledges that there are many possible endpoints that could be justifiable. In addition to the high costs associated with some of the suggested measures, there is a need to minimize responder burden as much as possible. Asking too much from study participants can effectively reduce the response rate and make the results difficult to interpret. Some of the suggested measures are more appropriately assessed in a controlled laboratory setting.

Questionnaire/Survey Design

Secrecy

Some comments expressed concern that the study questionnaire was not revealed as part of the public consultation process.

"At present, the secrecy that identifies that study questions cannot be revealed until after the study is complete lends itself to the suspicion that the survey does not trust that those surveyed will be honest. This must be rectified."

Expert Committee Response

To reveal the self-reported instrument before the completion of the data collection would severely compromise the value of the obtained results introducing bias to the point where interpreting the data would be seriously flawed. To support transparency to the extent possible, the general modules that will be evaluated in this study have been shared publicly. The public will be fully informed on all aspects of the study when the data has been collected.

Questionnaire Order

"It is not clear how questions are constructed and if as such they may lead the participant in a certain direction of thought. If so, this would lead to biased results. Furthermore, additional measurements on individuals will be taken. However, for example, blood pressure would be related to factors such as age, body mass, occupation and socio-economic status (it is not stated that such data would be noted). These additional data would also need to be recorded for full analysis in a statistical model.

Responses to questions will be biased by order in which they are asked. Health and quality of life questions should come before questions related to wind turbine (or other environmental) noise."

Expert Committee Response

Statistics Canada questionnaire development personnel and the epidemiologists on the Expert Committee have extensive expertise in questionnaire design. Issues related to how questions are presented have been carefully considered and all questions that are not part of Statistics Canada's bank of previously tested questions, undergo rigorous scrutiny by committees dedicated to the process of questionnaire development and must be qualitatively tested prior to utilization on representative subjects. In addition, the Expert Committee is well aware of the variables that can influence the measured endpoints over and above the possible impacts of wind turbine sound. These have been identified and will be accounted for in this study.

Survey Design

"People who are being studied should be interviewed in-depth, in person, rather than rely on forms and surveys. An internationally accepted survey IALS indicates that 42.5% of adult Canadians are functionally illiterate (granted a harsh term) and unable to process accurately written text they are exposed to. Rates of functional illiteracy rise to 62.5% in some areas all across Canada."

Expert Committee Response

This study does include an in-depth (in-person) interview. The "computer assisted personal interview" simply means that when the subjects in the study are being interviewed by Statistics Canada the questions will be read to them by an interviewer and their responses will be entered into the computer.

Statistics

Modelling and Statistical Power

The comments submitted that pertained to the statistical approach to this study wanted Health Canada to provide more details on what type of modeling will be done to assess the various associations explored in the study. Furthermore, some comments pertained to the level of statistical power and level of significance that would be used to assess statistical significance.

Expert Committee Response

The updated revised methodology paper now provides much more detail in this regard.

An exposure response model is being used to study the health effects of exposure to wind turbine sound in local communities. Initially, dwellings will be classified into 5 dB(A) intervals according to A-weighted sound emission level due only to wind turbine sound i.e., <30, 30-35, 36-40, 41-45, and >45.

The endpoints of primary interest are the measured endpoints and their occurrence among individuals living at varying "sound bins" or distances from wind turbines. Self-reported questionnaire data will also be analyzed and correlated to measured outcomes in some cases in order to account for variables that could influence measured endpoints. These outcomes will also be analyzed with respect to their relationship to self-reported measures including validated scales incorporated in the questionnaire. Many of the endpoints of interest that would be measured are subjective self-reported impacts on health and well-being as reported in the questionnaire. The statistical analysis would include, but is not limited to the following:

  • Chi-square test of independence to compare proportions of individuals in the various sound level groups of those annoyed/not annoyed by wind turbine sound conditional on other factors (such as (but not limited to) wind turbine sound pressure level, economic benefits, visibility of wind turbine, etc.)
  • Analysis of variance (ANOVA) to compare wind turbine sound pressure levels groups for interval data, such as cortisol levels, blood pressure/heart rate and sleep actimetry.
  • Correlations between ordinal and interval scale variables (ordinal scale variables being those where participants grade their responses on a scale from 1-5, whereas interval data are variables that are measured such as cortisol concentrations and blood pressure).
  • Logistic multivariate exposure response regression model will be developed for dichotomous response variables (e.g. annoyed/not annoyed) to compare the characteristics of these groups with respect to a continuous scale of the exposure variables and other environmental and moderating variables.
  • Multinomial or ordinal logistic regression models will be developed when the outcome is categorical or ordinal (several outcome categories for the measured endpoints).
  • Multivariate linear regression exposure response models will be developed for interval data (cortisol concentrations in hair, blood pressure, sleep patterns) to compare the characteristics of these groups with respect to a continuous scale of the exposure variables and other environmental and moderating variables.
  • Simultaneous regression models to compare wind turbine noise disturbance to other noise sources as determined by the questionnaire.
  • Principal components to generate the dependence structure between subjective response variables.

Scope

Expansion of scope

Comments were received suggesting the scope should be expanded to consider a broader scope than WTN, including visual impacts, property value impacts or shadow flicker.

Expert Committee Response

The Expert Committee's opinion was that even though the primary focus of this study is on evaluation of possible health impacts resulting from sound levels generated by wind turbines, the study should still consider other features associated with wind turbines including those mentioned here. It is important that this study is designed in such a way that it is possible to isolate (as much as possible) the extent to which sound from wind turbines may be associated with self-reported and measured endpoints. In order to accomplish this, the study needs to account for other variables that are related to the community response to wind turbines. Therefore, questions about non-acoustic variables will be included in the study.

Community Engagement

Comments received suggested the study needs to consider issues related to community engagement. Some comments suggested that it adverse community reaction started because wind turbines were forced upon them without adequate consultation.

Expert Committee Response

The Expert Committee is aware of the concerns that have been expressed by some communities who report that wind turbine farms are causing a divide amongst residents. Although the main focus of the study is on how individuals are impacted by wind turbine sound, the Expert Committee agreed to amend the questionnaire to add additional items probing extent of involvement of respondents in the process of introducing wind turbines into their community. Other items on the questionnaire are related to respondent's perception of property values and the extent to which they were consulted by the operators prior to the installation of wind turbines in their community.

Electromagnetic fields (EMF) from wind turbines and dirty electricity

Recommendation to include in study design scope.

Expert Committee Response

Electric and magnetic fields (EMF) are invisible forces that surround electrical equipment, power cords, and wires that carry electricity, including outdoor power lines. Electric and magnetic fields are strongest when close to their source. As you move away from the source, the strength of the fields fades rapidly. This means you are exposed to stronger electric and magnetic fields when standing close to a source (e.g., right beside a transformer box or under a high voltage power line), and you are exposed to weaker fields as you move away. When you are indoors at home, the magnetic fields from high voltage power lines and transformer boxes are often weaker than those from household electrical appliances. There are no established adverse health effects caused by Extremely Low Frequency (ELF) EMF exposures at levels found within or around Canadian homes, offices and schools, including those located just outside the boundaries of power line corridors and, therefore, will not be included in the study's design scope.

The term 'dirty electricity' is referred to in some literature as "electrostatic fields that vary rapidly in a random or noise-like pattern" whereby "the power distribution wires are the antennas and grounds that couple these noise-like signals to humans". In essence, 'dirty electricity' refers to unintended frequency components on electrical power lines. Some people are concerned that these additional frequency components may have adverse health effects on people.

Measurements of so-called 'dirty electricity' conducted by Health Canada have found these distortions on the electrical power supply to be of very low magnitude in comparison to the main (line) voltage and the aggregate human exposure to EMF from these distortions would represent an extremely small fraction of that received from exposure to the main (60 Hz, 120 V) voltage. Since there are no established adverse health effects caused by ELF EMF exposures at levels found within or around Canadian homes, offices and schools (including those located just outside the boundaries of power line corridors), Health Canada considers that 'dirty electricity' on electrical power lines would have no adverse impact on health.

Stray voltage refers to electrical energy that is conducted (directly), rather than radiated, from an energized source (such as power lines). Health Canada only establishes human exposure limits for radiated electromagnetic energy. Concerns about health effects from 'stray voltage' should be directed to provincial bodies which regulate the safety of electrical power distribution.

Other Energy Sources

Comments received suggested that the study needs to include an interpretation of the results associated with wind turbines in the context of the health impacts that have been found for other energy sources.

Expert Committee Response

The Expert Committee felt that a comparison with other energy sources would be outside the scope of this study.

Wildlife & other animals

Comments received suggested that the study needs to consider the effects that sound from wind turbines may have on livestock including bats, earthworms, horses, dogs, cats, etc. because animals are not susceptible to media, placebo effects, social pressures and other variables that make humans undesirable subjects.

Expert Committee Response

The Expert Committee feels that an assessment of wildlife, livestock and impacts on other animals would be outside the scope of the study.

Transformer Substations

Comments were received suggesting the study needs to consider noise complaints as a result of the operation of transformer substations.

Expert Committee Response

While outside the scope of the study, transformer sound emissions will be included in the sound modeling.

Defining Wind Turbines

Comments were received suggesting that the research design paper needs to better define "wind turbine" with respect to size and characteristics (including the number of turbines to be studied and the placement of the turbines in relation to each other).

Expert Committee Response

The turbines examined will typically be on the order of 1.5MW and greater; however, the number of wind turbines to be included in this study and their specific placement in relation to one another is currently unknown. It should be noted that the layout of the wind farm, type and number of wind turbines contained within it will be documented upon completion of the study. Such factors will be included in the analysis and considered in the noise modeling methodology being used for the study.

Citations

Citations are limited

Health Canada's study design references a limited number of scientific papers that have been published on the issue of wind energy and health and includes citations for "grey literature" that have not been peer reviewed nor published (or have been self-published). As such it is recommended that the list of citations be expanded and should give much greater weight to publications in indexed scientific journals that have benefited from the scholarly peer review process.

Expert Committee Response

The updated research design paper now includes a more extensive list of references. Nevertheless, the list of selected references is not intended to be exhaustive, nor does the inclusion of any given reference imply support of its conclusions by Health Canada. An extensive review of all literature in this area is planned to coincide with the completion of this study.

Duty of Care

Follow-Up with Participants

"In the event that individuals or families living near wind turbines report adverse health outcomes, either attributable to wind turbines or not, how will Health Canada follow-up with these people? Is there a study protocol for more rigorous or in-depth scientific and medical examinations that will be conducted to ascertain the root cause of the individuals' health problem, and will Health Canada work with these individuals to attempt the mitigation of these symptoms?"

Health Canada will not be conducting investigations into the origins of subjects' health problems. This is outside the scope of the present study. However, Statistics Canada's trained interviewers will provide respondents with their blood pressure results in written form with a chart to help them interpret their measurement. Respondents with dangerously high blood pressure readings will skip the rest of the interview and be told to see a doctor or go to a hospital Emergency Room as soon as possible. All respondents will also receive a report of their hair cortisol levels and sleep results about six to seven months after the study, unless they do not consent to receiving these results.

Page details

Date modified: