The Use of Personal and Indoor Air Pollution Monitors in Reproductive Epidemiology Studies

Abstract

Background.

Personal and indoor air pollution monitors represent two ways to assess acute air pollution exposures; however, few reproductive epidemiology studies have incorporated these tools.

Objective.

To provide an overview of the unique challenges and opportunities that arise when measuring acute exposure to air pollution in two ongoing reproductive epidemiology studies.

Methods.

The Air Pollution, In Vitro Fertilization (IVF), and Reproductive Outcomes (AIR) Study recruits women undergoing IVF to wear a personal particulate matter (PM) air pollution monitor (AirBeam2©) for the 72-hour period following the start of controlled ovarian stimulation. The Reproductive Effects of Chemicals and Air Pollutants (RECAP) Study recruits men across the US to place an air pollution monitor (emmET) in their home for 3 months, use a smartphone application, and provide a semen sample. We highlight the key issues identified in implementing exposure assessment for both studies.

Results.

The main advantages of using the AirBeam2© personal monitor are: 1) the low cost, 2) the ability to collect multiple size fractions of PM data every second, 3) the portability, 4) its capability to track GPS location and 5) the ability for the participant to observe their real-time exposure information. The limited battery life, incompatibility with iOS-based smartphones, and frequent connection issues that arise between the AirBeam2© and smartphone are the main disadvantages. The main advantages of the emmET are the ability to measure multiple air pollutants at a high level of accuracy, collect data for a long period of time without burdening the participant, and the ability to ship monitors to participants around the country without the need for in-person setup by trained technicians; however, the monitor only measures the indoor home environment.

Conclusions.

Novel methods can be utilized to characterize short-term air pollution exposure in reproductive epidemiology studies and represent an exciting area for future research.

Background.

Exposure to air pollution has been associated with several adverse reproductive and perinatal outcomes, including diminished semen quality¹ and increased risks of menstrual irregularities,² spontaneous abortion,³ hypertensive disorders of pregnancy,⁴ intrauterine growth restriction,⁵ preterm birth,⁶ and low birth weight.⁶ However, characterizing a person’s unique exposure to air pollution in studies of human health outcomes remains a challenging endeavor. Historically, most studies have relied on an indirect approach where exposure levels are estimated based on a combination of a person’s location, ambient air pollution measures, and exposure modelling. Yet, even the most sophisticated versions of these indirect methods have limitations due to the significant temporal and spatial variability of outdoor air pollution concentrations.⁷ This is because actual exposure levels are determined not only by the pollutant concentration in the environment but also on the amount of time spent by an individual in that environment. Moreover, in most industrialized countries, people spend >90% of their time indoors,⁸ where concentrations may not be accurately reflected by ambient levels. Thus, directly measuring pollution exposures of interest via personal samples is hypothesized to improve our ability to determine the true relationship between air pollution and health effects, as it would be a more accurate measure of what an individual breathes.

In reproductive and perinatal epidemiology, the need for personal exposure assessment may be even greater given the relatively short time windows of exposure susceptibility. For example, folliculogenesis, the time period during which an ovarian follicle matures, is estimated to take between 2–6 months,⁹ and spermatogenesis, the process by which spermatozoa develops, is estimated to take ~74 days.¹⁰ Ambient air pollution levels generally correlate well with personal exposure levels over the long-term but this varies greatly by pollutant and may not be true in the short-term.^11–13 This reinforces the need to not only understand the spatial and temporal variability of urban air pollution levels but also the main sources of indoor air pollution and an individual’s time-activity pattern to better characterize personal exposure.

The most obvious way to characterize short-term exposure to air pollution is through personal sampling which quantifies the real exposure values for the individuals. The drawback of this approach, however, is the high cost of implementation and high participant burden that comes with personal sampler wearing protocols. Additionally, choosing the appropriate personal monitor is a practice in balancing instrument validity with more practical concerns such as cost, weight, usability, reliability, and battery life.

To try and address many of these issues, low cost personal air pollution monitors are becoming increasingly common and represent an emerging potential tool to enhance personal exposure assessments in research studies, as long as they can be appropriately calibrated.^14–22 Many of these monitors also have the capacity to link with a person’s smartphone, further enhancing data collection possibilities including GPS tracking and momentary ecological assessment. The Air Pollution, In Vitro Fertilization, and Reproductive Outcomes (AIR) Study is one example of an ongoing study, which is utilizing personal monitors to characterize short-term exposure to air pollution. This study will be used to highlight the unique challenges and opportunities that arise when utilizing a low-cost personal sensor and the best practices for future research in this area.

Outside of personal monitoring using a wearable sensor, an alternate way to improve air pollution exposure assessment is to incorporate an indoor air pollution monitor, especially in studies that aim to assess exposures to multiple air pollutants simultaneously. The advantages of this method include reduced participant burden relative to a personal sampler, fewer worries regarding battery life or weight, and the ability to sample for longer durations of time. Moreover, when combined with GPS tracking and/or time activity diaries, information on the person’s full day of air pollution exposure can be reconstructed with the help of air pollution prediction models throughout the study period. The Reproductive Effects of Chemicals and Air Pollutants (RECAP) study, is an example of a study that uses this approach to assess exposures to multiple pollutants for a 90 day sampling period in relation to male fertility endpoints.

The AIR Study: A Case Study in Using Personal Air Pollution Monitors.

Study Overview.

The AIR Study is nested within a larger cohort study- the Environment and Reproductive Health (EARTH) Study (2004-present)²³ - which recruits couples presenting at the Massachusetts General Hospital (MGH) Fertility Center for infertility treatment and evaluation. Participants are followed from study entry through each fertility treatment cycle until a live birth, discontinuation of treatment, or withdrawal from the study. The study prospectively collects biological samples, questionnaire data, and medical information abstracted from medical records. The main goal of the EARTH Study is to investigate the relationship between environmental agents on reproductive health. The AIR Study specifically focuses on exposure to particulate matter (PM) air pollution and its impact on reproductive outcomes following in vitro fertilization (IVF).

Women participating in the EARTH Study who are undergoing a fresh, autogulous IVF cycle are eligible for the AIR Study. Exclusion criteria includes women who are egg donors or egg donor recipients, women undergoing a freeze-all IVF cycle, and women planning to do pre-implantation genetic diagnosis/screening. Once an interested woman has signed the consent form, the personal air pollution monitor, accompanying Android smartphone, and 3-day activity diary are given to them when they come in for their baseline ultrasound, the first day of controlled ovarian stimulation (Figure 1). Women are instructed to wear the air pollution monitor for the following 72-hour period either hooked onto the external strap of a purse or backpack (via a carabineer located on the top of the device) or on their belt (via a clip on the back of the device). When at home or at work, the participants are instructed to place the device upright on a table, desk, or counter top near where they are located.. The study materials are returned to EARTH Study staff at their next visit to MGH, which usually occurs 3 days later. During this monitoring visit, a blood sample is collected from participants for clinical and research purposes. Upon receipt of the study materials, the air pollution data is downloaded from the smartphone as a zipped csv file, and participants are given a $50 Amazon gift card.

Figure 1.

Overview of the timeline of the Air Pollution, In Vitro Fertilization, and Reproductive Outcomes (AIR) Study.

The AirBeam2© Monitors.

The AirBeam2© is a personal, palm-sized PM monitor that uses a light scattering method to measure three size fractions of PM (PM₁₀, PM_2.5, and PM₁) mass (μg/m³) and count (hundred particles/ft³). It has been validated by the EPA²⁴, the SCAQMD²⁵, and outside investigators^26–28 against EPA reference methods with r² ranging from 0.68–0.76 for 12–24 hour average PM_2.5 with excellent reliability and data recovery. The monitor also measures relative humidity and temperature, noise, and GPS location. The AirBeam2© monitor syncs with an Android phone via Bluetooth and measurements are communicated every second to the AirCasting Android application.

Initial Results.

Enrollment into the AIR Study began in late Fall 2018 and is ongoing. To date, six women have enrolled and successfully completed the study protocol. The first participant collected data across three full days of monitoring, providing over 137,000 individual PM measurements. Based on her time-activity diary, it was determined that over the course of three days, she spent 90% of her time indoors in her home. An overview of her 1-minute averaged personal exposure to PM₁₀ (Figure 2) PM_2.5 and PM₁ (Supplemental Figure 1), and relative humidity and temperature (Supplemental Figure 2) are shown during the study. As evidenced by the lack of data from 23:00 Saturday to 9:00 Sunday and 22:00 Sunday to 7:00 Monday, the participant had issues with the AirBeam2© monitor disconnecting from the smartphone overnight that were not discovered until waking. There were also some short periods during the day (e.g. on Saturday from 11:30 to 12:30) where no data was collected because the participant experienced difficulties keeping the AirBeam2© connected. Overall though, this data nicely highlights the large variation in exposure to PM (from <1 to 215 µg/m³) that this woman experienced during just three days of personal monitoring. A notable trend was that her peak exposure concentrations often occurred during mealtimes (e.g. mid-day and evening) and may be the result of cooking-related exposures.

Figure 2.

Sample data on personal exposure to particulate matter <10 µm (PM₁₀) from one participant in the Air Pollution, In Vitro Fertilization, and Reproductive Outcomes (AIR) Study collected over three consecutive days: Saturday (Panel A), Sunday (Panel B), and Monday (Panel C). The data presented are minute-averaged concentrations. The horizontal dashed line represents the EPA National Ambient Air Quality Standard for PM₁₀ (150 µg/m³).

The following two participants in the AIR Study were not as successful with completing the entire protocol. Both women were given the study materials at MGH, collected PM data for the first 10 hours, and then the AirBeam2© died and it was never re-charged. This led to a reworking of our study protocol including purchasing external batteries for women to extend the life of the AirBeam2©, particularly when they are away from electrical outlets for extended periods of time. We also had a meeting with our study staff to reinforce the importance of explaining the entire protocol to participants and going through some common issues, such as how to avoid battery life problems and how to reconnect the AirBeam2© after a disconnection, before the women leaves the clinic. We also send an email to participants 24 hrs after they start the study to check-in on how things are going and troubleshoot problems, if necessary.

Lessons Learned.

The main advantages of the AirBeam2© monitor are the cost (only $250 per device), the ability to collect PM air pollution exposure data every second, the portability (only weighs 5 oz), and its capability to track GPS location and measure temperature, humidity, and noise (in addition to PM₁₀, PM_2.5, and PM₁). In addition, the ability for women to observe their real-time PM exposure data on the AirCasting application was an added incentive for participation. For instance, women can visually observe how their one-minute averaged PM exposures vary throughout the day by viewing a graph showing their peak and average exposure levels as well as a graph showing how their PM exposures vary across different geolocations.

However, the device does have notable disadvantages. The first being that there is currently no iOS version of the AirCasting application and thus the AirBeam2© monitors can only sync and communicate with Android phones. Since 40% or more of the US population utilizes iOS based phones,²⁹ this warranted us having to purchase Android phones to give out with the monitors. The second disadvantage is that the battery life of the AirBeam2© at full charge is only 10 hours and the monitor gives no indication of how much battery life is left. Since our participants are using these monitors for 72-hours this warrants them having to charge the monitors (or keep them plugged into an outlet) a couple hours out of the day. It has also resulted in many monitoring sessions discontinuing (and having to be restarted) due to the AirBeam2© running out of battery. To circumvent this issue, we now dispense small external batteries for our participants to use for charging if the battery is low and they cannot access an electrical outlet; however, this is an added burden to carry. Third, the AirCasting application will randomly freeze and disconnect from the phone without obvious warning. This requires the participants to regularly check on the application to make sure data is being collected and displayed. If not, they are asked to restart the application and device and initiate a new monitoring session, which can be time consuming for participants. Finally, the AirBeam2© is only capable of measuring personal exposure to PM and does not measure other gaseous pollutants. Moreover, because the AirBeam2© devices use a light scattering method to measure fine particulate matter exposure, rather than a filter, it is not possible to determine the elemental make-up of the fine particulate matter women are being exposed to.

As with every piece of emerging technology, the strengths and weaknesses of the device must be weighed in the context of each individual study. In our case, since a huge barrier was cost, and not being able to afford PM monitors such as the RTI microPEM, the AirBeam2© was an obvious choice since it was one of the few low-cost PM sensors that had external validation data. Yet it has been our experience that with the AirBeam2© monitors, given their aforementioned limitations, the more tech-savvy and interested a participant is in the research, the better her compliance and the better her data will look after 3 days of monitoring. Thus, in settings where participants are not accustomed to dealing with smartphone applications or are not actively invested in understanding their exposure to air pollution, the deployment of AirBeam2©s may be less successful.

The RECAP Study: A Case Study in Using in Home Air Pollution Monitors and a Smartphone Application.

Study Overview.

The RECAP study is also nested within a larger cohort study – the Growing Up Today Study (GUTS), a prospective nationwide cohort of children of women participating in the Nurses’ Health Study II (NHSII) cohort.³⁰ Briefly, children 9 to 14 years of age were enrolled in 1994 and 2004, and have been followed by questionnaire annually or biennially. At present, participants in the GUTS cohort are between the ages of 24 and 38 years. The main goal of GUTS is to examine factors throughout the life course that are associated with chronic disease risks.

Men participating in GUTS who indicated on a follow-up questionnaire that they are interested in providing semen samples for research form the recruitment pool for the RECAP study. These eligible men are then randomly selected to receive an email inviting them to participate in the RECAP study, and if interested, are directed to a secure online screening questionnaire. Exclusion criteria include residence outside of the conterminous United States, previous vasectomy or other procedure that may impact the ability to produce semen, active chemotherapy or anabolic steroid use, as well as participants who do not have a smartphone on which they can use the smartphone application. Men who are eligible electronically sign a consent form, and are provided with instructions on how to download the study smartphone application.³¹

In the smartphone application, men are asked if the next 3 months will be typical for them. Men who answer yes are immediately eligible and a package of study supplies is rapidly shipped to their preferred address (Figure 3). Upon receipt of the package, men are asked to place the air pollution sampler (emmET, described below) in the room where they spend the most time (other than the kitchen), and attach it to their home WiFi. This allows the sampler to determine the correct time zone, and allows the research team to see that the sampler has been activated. Each participant also puts on the first of three passive sampling wristbands (24hourwristbands.com, Houston, Texas), to collect 1 month of integrated exposure to over 1,500 chemicals.^32–34 After setting up the sampler and putting on the wristband, the participant competes a survey in the smartphone application to alert the study team that all Day 1 items have been successfully completed. There is a direct call button in the app to allow participants to contact the study team with any questions throughout the study.

Figure 3.

Overview of the timeline of the Reproductive Effects of Chemicals and Air Pollutants (RECAP) Study.

Air pollution sampling continues for all 90 days of the sampling period, and participants switch wristbands at Days 30 and 60, completing short questionnaires in the app each time. At Day 90, participants are asked to produce a semen sample via masturbation, and to collect the sample in a sterile collection cup. They are directed to aliquot a small amount of sample onto a provided Leja slide, and to use a study provided smartphone with a custom 3D printed attachment to analyze their sample. The smartphone attachment uses the camera of the Android smartphone, along with lenses and a light, to magnify the sample. A custom application then records the sample, and via machine learning, calculates sperm count, concentration, motility, and morphology.³⁵ The participant is alerted when the sample has been successfully analyzed, and is then prompted to package the remaining semen sample into a cooler box with ice packs for immediate return via express mail, and the remaining exposure sampler materials into the original shipping box for ground service.

After return, the remaining semen sample is tested in an infertility lab for traditional measures of semen quality (except motility), and the remaining sample is aliquoted for storage, and assessment of the sperm DNA methylome. After all materials have been returned, participants are compensated $100 for their participation.

The emmET.

The environmental multi-pollutant monitor for extended time periods (emmET, Figure 4) is a lunchbox sized monitor (approximately 30cm wide by 23cm long and 10cm thick) that measures PM_2.5, NO₂, CO₂, temperature, relative humidity, and noise, that costs approximately $3,300 (excluding personnel time and consumables such as filters). The emmET is a modification of the EMMA (Environmental Multi-pollutant Monitoring Assembly),³⁶ to allow samplers to be shipped to participants (as opposed to set up in each home by trained staff members) and to extend collection periods up to three months. PM_2.5 is monitored with an Alphasense OPC-N3 particle monitor, along with three gravimetric filters for calibration, NO₂ is monitored with an Alphasense NO2B43F, which is calibrated before and after each sampling period, and temperature, relative humidity, noise, and CO₂ are monitored with a Netatmo weather station mounted on the outside of the emmET. Each box has an internal data logger, as well as a cloud data logger that can be used by study team members to determine if the box is active and working correctly. The pumps and external cases were chosen to reduce the noise from the internal pumps (the average noise level is 58.5dB), and to date, we have not received any participant complaints. Details of the calibration procedures and selection of the specific sensors chosen are described in detailed in Gillooly et al, which also discusses the true cost of these low-cost sensors, and the need for rigorous quality assurance protocols. Participants are shipped the emmET and provided with an instruction sheet on where to place the emmET and how to activate the sampler.

Figure 4:

Details of the environmental multi-pollutant monitor for extended time periods (emmET).

Initial Results:

Enrollment into the RECAP study began in March of 2019 and is ongoing. As of June of 2019, 25 men (out of the anticipated 200) have been enrolled and are actively on the study protocol. The first month of data for NO₂, CO₂, temperature, relative humidity, and noise from the first participant is shown in Figure 5 and Supplemental Figure 3. As expected from an indoor environment, there were not large swings in temperature or humidity over the month, although there was evidence of diurnal variability in temperature. The average (standard deviation, SD) temperature was 20.2 °C (1.1) and humidity was 36.6% (3.8). There was higher variability in levels of noise (average=51.0 dB, SD=3.2 dB, minimum=45 dB, maximum=66 dB) and CO₂ (average=766.6 ppm, SD=196.1, minimum=400 ppm, maximum=1461.9 ppm). The average NO₂ over the month was 22.5 (7.8) ppb, and the maximum observed concentration was 85.0 ppb. Peaks in NO₂ were observed most mornings 6–8am and evenings 4–7pm, which could be associated with exposures from cooking or local traffic. When paired with GPS data from the smartphone application, we will be able to assign each participant exposures while he is home for each day of the study. Additionally, we have the ability to match the non-home GPS locations with air pollution prediction models to construct exposures using the full activity space of each individual throughout the study period. To date, there has been no loss of exposure data for the initial round of participants.

Figure 5.

Sample data from one participant in the Reproductive Effects of Chemicals and Air Pollutants (RECAP) Study. Panels show (top row) Nitrogen Dioxide (NO₂) concentrations (ppb), bottom row Noise (dB) levels. The left-hand panel shows data from one month (3/5/19 to 4/4/19) and the right-hand panels show daily data for a selected 24-hour period (3/23/19).

Lessons Learned:

The main advantages of the emmET are: 1) the ability to measure multiple air pollutants and other exposures inside the homes of our participants at a high level of accuracy, 2) to be able to collect data at a fine time scale for long periods of time without burdening the participant, 3) the ability to determine over the 3 month sampling period if the sampler is on and functioning properly, and 4) the ability to ship samplers to participants around the country without the need for trained technicians to set up each sampler. Participants have also noted that the sampler was easy to set up and that it has not been burdensome to have an emmET in the main living area of their home.

There is one main disadvantage to using the emmET. Due to its weight (3.6kg) and power requirements, we are unable to ask participants to carry the device for the 90-day sampling period to obtain personal exposure measures. However, we were willing to make this tradeoff to have high quality data on multiple pollutants for the substantial portion of time we anticipate participants will spend in their homes, and to reduce participant burden to enhance compliance. We will be able to use the non-home GPS coordinates and spatial prediction surfaces of air pollution, temperature, and noise to estimate exposures at other locations to more accurately estimate personal exposure levels for each participant. Another disadvantage to using the emmET is the volume of data received. For each participant, over 90 days of sampling data lead to approximately 420 GB of raw exposure data. This requires thoughtful assessments of storage limitations, and, depending on the time resolution of data needed for a project, algorithms to covert data from multiple sensors to a single, common, time interval.

Conclusions.

The rapid technological advancement of air pollution monitors over the past decade has changed the landscape of air pollution monitoring for epidemiologic studies by greatly enhancing researchers’ ability to characterize short-term personal exposure to air pollutants. In reproductive epidemiology, this is particularly exciting as many reproductive and perinatal outcomes are suspected to be linked to acute air pollution exposures. Given the high penetrance of smartphones among men and women 18–40 years of age and their general openness to utilizing newer technologies, this reproductive aged group represents an optimal population for the successful implementation of personal and/or indoor air pollution monitors. As illustrated in the two case studies, it is possible to successfully recruit study participants and deploy these types of air pollution monitors in studies that are both clinic and population based. Moreover, based on the particular outcome of interest and the corresponding relevant window of air pollution exposure, these air pollution monitors can be given to participants for time spans ranging from a handful of days up to many months.

While choosing between a personal or indoor air pollution monitor will depend on many aspects of the study design- most notably cost, participant burden, study timeline, and whether data is desired on more than one pollutant- in either case investigators should prioritize using validated and calibrated instruments to enhance the quality of data being collected. Unlike more expensive monitors with comprehensive regulatory standards and processes for evaluation and certification, few standards and no certifications exist for low-cost air pollution monitors.³⁷ There is a rapidly growing body of literature (and government agency guidance) on methods to calibrate and assess the performance of low-cost sensors that is beyond the scope of this article to review.^{14–22, 36} However, at a minimum, researchers should determine if the instruments they are considering need to be calibrated against more traditional measures of the exposure of interest (and at what frequency), if there is information regarding the performance of the sensor in environments similar to those proposed, if collection of other information (e.g. temperature and relative humidity, filter based particle measures) is needed to calibrate the sensor readings, and perform colocation of all instruments at set intervals to determine if individual units require calibration or replacement. These steps at minimum are necessary, especially with newly developed sensors. Despite these concerns, the potential use of personal and indoor air pollution monitors in reproductive epidemiology studies remains an exciting avenue for future research and when applied appropriately, will greatly enhance our understanding of how air pollution affects human health.

Social Media Quote.

Novel methods to monitor personal and indoor air pollution can be utilized to characterize short-term exposure in reproductive epidemiology studies and represent an exciting area for future research

Synopsis.

Study question.

How can personal and indoor air pollution monitors be integrated into reproductive epidemiology studies?

What’s already known.

Exposure to air pollution has been associated with several adverse reproductive and perinatal outcomes; however, characterizing a person’s unique exposure to air pollution in reproductive epidemiology studies remains a challenging endeavor.

What this study adds.

This study uses two ongoing studies- Air Pollution, In Vitro Fertilization (IVF), and Reproductive Outcomes (AIR) Study and the Reproductive Effects of Chemicals and Air Pollutants (RECAP) Study- to provide an overview of the unique challenges and opportunities that arise when measuring personal and indoor air pollution concentrations.