We detail the consistent, high-quality procedures used throughout air and biological sampling processes at Indian field sites during a large randomized controlled trial. Insights gathered from the oversight of applications of innovative technologies, adapted for exposure assessment in rural regions, enable better field data collection practices with more reliable outcomes.
Here, we present a visual representation of standard procedures to collect population-level data on personal exposures to household air pollution (HAP) from two different study sites in a resource-constrained setting of Tamil Nadu, India. Particulate matter PM2.5 (particles smaller than 2.5 microns in aerodynamic diameter), carbon monoxide (CO), and black carbon (BC) were measured in pregnant mothers (M), other adult women (OAW), and children (C) at various times over a 4 year period. In addition, stove usage monitoring (SUMs) with data-logging thermometers and ambient measurements of air pollution were carried out. Furthermore, the feasibility of collecting biological samples (urine and dried blood spots [DBSs]) from study participants at the field sites was successfully demonstrated. Based on findings from this and earlier studies, the methods used here have enhanced the data quality and avoided issues with household air pollution and biological sample collection in resource-constrained situations. The procedures established may be a valuable educational tool and resource for researchers conducting similar air pollution and health studies in India and other low- and middle-income countries (LMICs).
Globally, exposure to household air pollution (HAP), mostly from solid-fuel cooking, is a major cause of morbidity and mortality1,2,3. Cooking and heating with solid fuels (biomass-such as wood, dung, crop residues, and coal) is widespread in low- and middle-income countries (LMICs), posing various health, environmental, and economic issues. PM2.5 is a 'silent killer,' occurring both indoors and outdoors4,5. Indoor air quality in India is often considerably worse than outside air quality, and it has gained enough attention to be regarded as a major environmental health hazard4. A paucity of measurement-based quantitative exposure data has impeded global burden of disease (GBD) evaluations connected with HAP6,7.
Current research often ignores that the measurement of HAP exposures is complicated and varies depending on many factors, including fuel type, stove type, and a mixed use of many clean and unclean stoves, a phenomenon known as "stove stacking". Other influences on exposure include the quantity of fuel consumed, kitchen ventilation levels, the length of time spent in proximity to the cookstove, age, and gender8. The most widely measured and arguably the best indicator of exposure to HAP is PM2.5; however, due to a lack of affordable, user-friendly, and reliable instrumentation, measuring fine particulate matter (PM2.5) has been particularly difficult.
Various studies have reported measuring the level of either single or multiple air pollutants using different methods8,9,10,11,12. In recent years, relatively low-cost sensors that are able to measure these pollutants in indoor and ambient environments have been emerging. However, not all of these sensors are viable for fieldwork for various reasons, including maintenance costs, deployment challenges, issues of comparability with conventional measurement methods, limited human resources to validate these sensors against reference methods, the difficulty of regular data quality checks (through the cloud), and limited or no decentralized troubleshooting facilities. Many of the studies with these types of measurements have used them as a proxy for exposure or by combining environmental measurements with exposure reconstruction using time activity assessments8,9,12,13,14.
Personal monitoring-in which a monitor is carried on or by an individual through space and time-may better capture their 'true' total exposure. Studies that measure personal exposure often only briefly communicate their exact protocols, often in supplementary materials to scientific manuscripts9,12,13,14,15. Although the techniques detailed in these studies provide a solid general sense of sampling methodology, there is often an absence of the specifics of the field data collection stages12,16.
Numerous additional characteristics, in addition to pollutant concentrations, may be monitored in these residences. Stove use monitoring, a method of assessing the time and intensity of use of household energy appliances, is a major part of many recent impact and exposure assessments16,17,18,19. Many of these monitors focus on measuring the temperature at or near the point of combustion on cooking stoves. While thermocouples and thermistors are employed, there is a lack of operating protocols for the monitors, including how best to put them on cookstoves to capture variability in stove usage patterns.
Biomonitoring, similarly, is an effective tool for evaluating environmental exposures, though several factors influence the choice of an optimum biologic matrix20. Under ideal circumstances, sample collection must be non- or minimally invasive. The methods employed should ensure ease of handling, non-restrictive shipping and storage, a good match between the proposed biomarker and biological matrix, a relatively low cost, and no ethical concerns.
Urine sample collection has some major advantages for biomonitoring. As with other sample collection techniques, a range of potential methods exist. Collecting 24 hour void urine can be cumbersome for participants, leading to non-adherence of sample collection20,21. In such cases, spot samples, first morning voids, or other 'convenient' samplings are recommended. The volume of urine collected can be a major disadvantage when collecting spot samples, leading to variability in the concentrations of endogenous and exogenous chemicals. In this case, adjustment using urine creatinine concentrations is a commonly used method for dilution corrections22.
Another commonly collected biospecimen is venous blood. Venous blood samples are often difficult to obtain for biomonitoring; they are intrusive, fear-inducing, and require proper sample handling, storage, and transport. An alternative approach using dried blood spots (DBSs) can be useful for collecting samples in adults and children for biomonitoring23.
A substantial literature gap exists between the simple description of field methods and the publication of detailed, replicable instructions on monitor use and deployment that reflects the true complexity of field data collection of quality-assured samples24,25. Some studies have outlined standard operating procedures (SOP) for measuring air pollutants (indoor and ambient) and monitoring stove use.
However, the essential steps behind the field measurement, laboratory support, and transport of monitoring instruments and samples are very rarely described8,11,25. The challenges and limitations of field-based monitoring in both high- and low-resource settings may be properly captured through video, which could complement written operating procedures and provide a more direct method of showing how devices and sampling and analytical techniques are performed.
In the Household Air Pollution Intervention Network (HAPIN) randomized controlled trial, we used video and written protocols to describe the procedures for measuring three pollutants (PM2.5, CO, and BC), for stove use monitoring and for biospecimen collection. HAPIN involves using harmonized protocols that require strict adherence to SOPs to maximize data quality from samples collected across multiple time points at four study sites (in Peru, Rwanda, Guatemala, and India).
The criteria for study design, site selection, and recruitment are described earlier24,26. The HAPIN trial was conducted in four countries; Clasen et al. described the study settings in detail26. Each study site recruited 800 households (400 intervention and 400 control) with pregnant women between the ages of 18 and 35, who are 9 to 20 weeks gestation, use biomass for cooking at home, and are nonsmokers. In a subset of these households (~120 per country), other adult women were also enrolled in this study.
After recruitment, a total of eight visits were made. The first, at baseline (BL), occurred prior to randomization. The next seven were split up by before birth (at 24-28 weeks gestation [P1], 32-36 weeks gestation [P2]), at birth (B0), and after birth (3 months [B1], 6 months [B2], 9 months [B3], and 12 months [B4]). For M, there were three assessments (BL, P1, and P2), for OAWs, six assessments (BL, P1, P2, B1, B2, and B4), and for C, four assessments (B0, B1, B2, and B4) were performed. At B0, biomarker and health assessments were carried out, while only health assessments were carried out at the B3 visit.
All four countries followed identical protocols. In this manuscript, we describe steps followed in India. The study was performed at two locations in Tamil Nadu: Kallakurichi (KK) and Nagapattinam (NP). These sites are located between 250 and 500 kilometers from the core research facility at the Department of Environmental Health Engineering at Sri Ramachandra Institute of Higher Education and Research (SRIHER) in Chennai, India. The complexity of field data collection protocols requires the deployment of many personnel with varying levels of skills and backgrounds.
We present a written and visual depiction of the steps involved in estimating micro-environmental and personal exposure samples in pregnant mothers (M), other/older adult women (OAW), and children (C) to fine particulate matter, carbon monoxide (CO), and black carbon (BC). Field protocols for (1) monitoring ambient air quality with reference-grade monitors and low-cost sensors, (2) long-term stove use monitoring on conventional and liquefied petroleum gas stoves, and (3) biological sample collection (urine and DBSs) for biomonitoring are also presented. This includes methods for transporting, storing, and archiving environmental and biological samples.
The Institutional Ethics Committee at Sri Ramachandra Institute of Higher Education and Research (IEC-N1/16/JUL/54/49), Emory University Institutional Review Board (00089799), and the Indian Council of Medical Research-Health Ministry Screening Committee (5/8/4-30/(Env)/ Indo-US/2016-NCD-I) approved the HAPIN trial. The HAPIN trial is identified as NCT02944682 on clinicaltrials.gov. Written informed consents were collected from the study participants prior to their participation and the study was conducted according to ethical guidelines.
NOTE: The case report forms (CRF) administered during the sampling and data collection are available in the RedCap database, stored at Emory University, and are maintained with the data-sharing agreement between all the collaborators, which can be provided to the readers upon request.
1. Instruments and materials
2. Filter conditioning and weighing
3. Microenvironment/personal air sampling
NOTE: A detailed outline of the instrumentation and steps involved in microenvironment/personal air sampling is given in Supplementary Figure 2.
4. Stove use monitoring
5. Ambient monitoring
NOTE: The ambient PM2.5 instrument records real-time airborne PM2.5 and has an inbuilt 47 mm filter that can collect PM2.5 for gravimetric evaluation19,26,29. A detailed outline of the instrumentation and steps involved in ambient monitoring of data collection in the central lab, field lab, and field site activities are given in Supplementary Figure 6.
6. Biomonitoring
7. Chain-of-custody (COC) of sampled filters
Microenvironment/personal air sampling methodologies:
Figure 1Ai shows a pregnant mother wearing the customized vest during the 24 h sampling period. The vest includes the ECM, CO logger, and time and location logger with the power bank. It was ensured that the participants wore the vest throughout the sampling period, except while bathing and sleeping. The stand that was provided to hang the vest within the sleeping periphery is shown in Figure 1Aii.
Figure 1Bi shows cleaning of the ECM sampler during the pre- and post-sampling period and Figure 1Bii shows the calibration of ECM in the field laboratory. The flow rates were checked and the sampling runtime was set prior to sampling, and any deviations were checked in the post-sampling period.
Figure 1C shows the instrument placement (ECM, CO logger, and TLL) in the kitchen (1.5 m height from the ground) for area monitoring. The instruments were placed and installed 1 m away from the cooking source. When windows or doors were near the sources, the instruments were installed 1 m away from these openings. Figure 1D shows the instruments (ECM, CO logger, and TLL) connected with a power bank, installed (1.5 m height from the ground) outdoors in each participants' house. The results of microenvironmental and personal PM2.5 following these methodologies are already published24,35,36.
The data of CO logger calibration over 1 year showed low malfunctions, as shown in Supplementary Figure 8 (legend shown as a circle represents the measurement of the CO data logger reads beyond the calibration range of 0-50 ppm)35. Details about the difference between the blank and sample slot of BC measurement for loading the filter is given in the supplementary material (Annexure C).
Figure 1Ei illustrates damaged filters before the pre-weighing process. Damaged and dropped filters are flagged as invalid. Filters were rechecked at the field laboratory before loading into any air monitoring instruments. Similarly, sampled filters were examined for any damage, such as holes, tears, stretching, or dislodgment, as given in Figure 1Eii. If any such damage existed, it was weighed, but not considered as valid for subsequent analyses. PM2.5 mass for each sample was obtained by subtracting the sample mass from the median field blank mass. The final PM2.5 concentration was estimated by dividing the blank corrected filter masses by the amount of air sampled by the pump during the measurement interval. The derived threshold criteria for valid PM2.5 and CO sampling is given in Table 1. Data within the threshold criteria are considered valid and taken up for the analysis.
Stove use monitoring
Figure 3A shows the calibration of temperature loggers, which was done initially on ice and then in hot water with the help of a standard temperature monitor. Figure 3Bi shows the regular pattern of peaks which are considered valid and identified by characteristic peak color (orange) when the stove is used. As shown in Figure 3B, the irregular pattern of peaks, such as (ii) probe problem (recording high temperatures and measuring beyond the temperature range), (iii) technical error (baseline shift with negative values), and (iv) thermocouple problem, was characterized as invalid sampling (no records of temperature measurement; baseline shift together with negative values). The temperature logger installed in different cookstoves is illustrated in Supplementary Figure 5. The results of stove use monitoring following this methodology is already published18,19,36.
Ambient air sampling
With US EPA guidelines (https://www.epa.gov/environmental-topics/air-topics), the ambient samplers were installed at the top of one of the selected households in each representative HAPIN-India site (Figure 4A). Sampling sites were chosen based on security, availability of electricity, and willingness of the participant to host the monitor. Sites were selected to be in central areas relative to clusters of participants. Ambient samples followed a similar procedure of packing and shipping from the field site to the field laboratory to the central laboratory. To prevent cables from being plugged into the incorrect connector, connect with the unique pin layout as given in Supplementary figure 9. The results of ambient monitoring (PM2.5) measured between 2018 to 2020 is shown in Figure 4B. Also, the results of ambient monitoring using this method elsewhere is already published29.
Biomonitoring
Figure 5A displays the urine aliquoting procedure. The samples from the field site were stored in vaccine cooler bags and transported to the field laboratory, where they were aliquoted and stored in a deep freezer (-20 °C). Figure 5B summarizes the sample collection, transport, and field storage COC.
Figure 5C shows the DBSs; 5Ci shows valid spots before drying, and 5Cii shows valid spots after drying. Table 2 summarizes the pattern of valid DBS collection in follow-up visits among the HAPIN participants (M, OAW, C). The success rate of collecting valid DBSs from mothers for three visits is 100% (BL), 93% (P1), and 83% (P2). Similarly, for OAW, the success of DBS collection was consistent (100%-72%) for the first three (BL-P2) visits, but reduced (45%-35%) from B1 to B4 during the pandemic and during hurricane Gaja (2018). The success of DBS collection in children was 72.09% at birth (B0), 64% at B1, 62% at B2, and 45% at B4.
Figure 5D emphasizes that the cold chain with dry ice maintains sample integrity. Every month, biological samples were packed with dry ice in a separate thermal insulation box and shipped with temperature and relative humidity (RH) loggers. Correlation analysis of urinary specific gravity measured between the field laboratory and central laboratory showed good agreement, as shown in Figure 5E. Our results of biomonitoring method cross-validation in urine samples of polycyclic aromatic hydrocarbon metabolites shows the quality assurance (QA)/quality control (QC) of sample integrity21.
All sampling and CRF data were securely uploaded from SRIHER to Emory University's server. Data transfer occurred daily, thus reducing the likelihood of data loss. The list of CRFs used for data collection are given in Supplementary Table 3. The data collection flow from the field site to the Emory server is given in Supplementary Figure 10.
Figure 1: Personal and microenvironmental monitoring. (A) i) Pregnant mother wearing the vest with air sampling instruments (ECM, CO logger, and TLL); ii) Metallic stand with the vest; iii) Child vest with TLM units. (B) i) ECM cleaning; ii) ECM calibration. (C) Kitchen area monitoring with the ECM, CO logger, and TLL. (D) Outdoor area monitoring with the ECM, CO logger and TLL. (E) i) Damaged pre-weighed filters; ii) Damaged sampled filters. Please click here to view a larger version of this figure.
Figure 2: Chain-of-custody of filters. (A) Sampled ECM with filters wrapped up in aluminium foil. (B) Transport of sampled filters from participant households to the field lab in vaccine cooler bags containing gel packs. (C) Sample filters stored in a deep freezer (-20 °C) in the field laboratory. Please click here to view a larger version of this figure.
Figure 3: Stove use monitoring-temperature Loggers. (A) Calibration of Geocene Dot temperature loggers. (B) i) Valid pattern of peak for stove use monitoring; ii) Probe problem; iii) Technical error; iv) Thermocouple problem. (C) Stove use monitoring-Temperature Loggers. Please click here to view a larger version of this figure.
Figure 4: Ambient monitoring. (A) Ambient PM2.5 instrument installed at the field site. (B) Time series of ambient PM2.5 level measurements (2018-2020). Please click here to view a larger version of this figure.
Figure 5. Biomonitoring-sample collection, processing, and quality control. (A) Urine aliquoting. (B) Chain-of-custody for sample collection, storage, and transport. (C) Dried blood spot: i) before drying; ii) after drying. (D) Cold chain-of-custody of sample shipment. (E) QA/QC of sample integrity-data of urinary specific gravity measured in the field site and central laboratory. Please click here to view a larger version of this figure.
Table 1: Valid sampling threshold criteria of PM2.5 and CO monitors. Note that due to sensitivity of the accelerometer, values outside of expected ranges were flagged but not excluded from analyses. Please click here to download this Table.
Table 2: Summary of valid DBS collection among the study participants. *During a hurricane (cyclone) there was a drop in DBS collection. †During COVID19 lockdown there was a drop in DBS collection. ‡During COVID19 lockdown there was a drop and 2021 data is not included in DBS collection. Abbreviations: M = pregnant mother; OAW = other adult woman; C = child. Please click here to download this Table.
Supplementary Table 1: Guidelines for the installation of samplers for microenvironmental monitoring.
Supplementary Table 2: Difference between the blank slot and sample slot for loading the filter. *The diffuser can be replaced only if there is any visible damage or if used for ~750-1,000 filters.
Supplementary Table 3: List of CRFs in respect to exposure and biomarker sampling. The CRFs are available in the RedCap database, stored at Emory University, and are maintained with the data-sharing agreement between all the collaborators, which can be provided to the readers upon request.
Supplementary Figure 1: Data entry sheet for filter weighing. Please click here to download this File.
Supplementary Figure 2. Instruments and steps involved in microenvironment and personal air sampling. Please click here to download this File.
Supplementary Figure 3: Cartridges and filter slots. A: Blank slot (Position 1); B: Blank bottom cartridge containing both the blank diffuser and the blank filter in the cartridge; C: Top piece of the blank cartridge; D: Sample slot (Position 2); E: Bottom sample cartridge with sample diffuser; F: Top piece of the sample cartridge. Please click here to download this File.
Supplementary Figure 4: Instruments and steps involved in stove use monitoring. Please click here to download this File.
Supplementary Figure 5: Dots installed in different cookstoves. Please click here to download this File.
Supplementary Figure 6: Instruments and steps involved in ambient air monitoring. Please click here to download this File.
Supplementary Figure 7: Instruments and steps involved in biological sampling. Please click here to download this File.
Supplementary Figure 8: Summary-of-summary of carbon monoxide (CO) data logger. Please click here to download this File.
Supplementary Figure 9: E-sampler connector layout. Please click here to download this File.
Supplementary Figure 10: Data collection and processing. Please click here to download this File.
Supplementary Figure 11: Usage of vaccine bag. Please click here to download this File.
Supplementary Figure 12: Difference between pelican and vaccine bag. Temperature performance of two cooler bags (pelican vs. vaccine) are tested using a CO data logger monitor for 48 h in the laboratory at an average room temperature of 28.3 ± 0.6 °C and RH of 49.2% ± 3.6%. A urine sample (~60 mL) with an initial temperature of 36.4 °C was placed in two bags and kept undisturbed for 48 h in a storage room. Please click here to download this File.
We demonstrated and visually represented standard procedures to collect population-level data on personal exposures to household air pollution in the multi-country HAPIN trial19, 24. The field-based environmental and biomarker sampling methods described here are appropriate and feasible, particularly in vulnerable populations in resource-limited settings where PM2.5 exposures are several orders of magnitude higher than the WHO Air Quality Guideline (AQG) values (annual mean of 5 µg/m3 and 24 h mean of 15 µg/m3)37,38.
The instruments used in this study have been deployed in other studies with different sampling durations. In Jack et al., CO was measured every 6 weeks using a lightweight CO logger monitor, and PM2.5 measurement (microPEM) was co-located for 72 h on nearly half of pregnant participants9. Another study used ambient-adjusted personal exposures assessed with the Ultrasonic Personal Aerosol Sampler (UPAS) to address uncertainty in estimates of global impacts of HAP on cardiac health using a quantitative assessment of PM2.5 exposure14.
The procedures followed during the HAPIN trial and described here can serve as guidelines for personal/microenvironment HAP monitoring in other settings24,26. The field team first evaluated possible locations for placing the monitors in the kitchen, when sleeping, and in outdoor household locations. When there were no ideal conditions for placing the monitors (1.5 m above ground, 1 m away from the stove and any doors and windows), suitable places adjacent to the ideal sample location were chosen35. This occurred relatively infrequently-in <2% of all collected samples. Temperature loggers used as stove use monitors had waterproof covers to protect from spillage during kitchen activities, like cooking and heating water. However, temperature loggers that were placed on outdoor stoves were damaged during the monsoon and flooding (Figure 3C).
The compliance of wearing the vest was observed by field surveyors on the day of removing the monitors (Day 2). Compliance assessed via sensors was shown to be, at times, incorrect; in some instances, participants would be wearing the monitor, but would be sitting still and thus would be flagged as non-compliant. This recognition of sensor-based misclassification was only possible due to observant fieldworkers. As an additional check, our CRFs contained participant-reported compliance.
The handling of filters during conditioning, weighing, sampling (pre- and post-), transportation, and storage at the field and central laboratory are critical activities within any data collection phase. After 24 h of sampling, the personal monitor was completely covered in aluminium foil and placed in a biospecimen bag to transport in a cold chain and dust free environment. The current study has demonstrated the procedures for preserving the filters from the participants' home to the field laboratory to the central laboratory via chain-of-custody.
Few deviations from required temperature conditions during storage and transport are predicted during biological sample collection and transport from the field to the laboratory for analysis, which might lead to erroneous results. The cooler bag used in the multi-country installations were too expensive in India. Under various national programs in India, the vaccine cooler bag has been used widely used for vaccine transport. These vaccine bags were obtained locally at a reasonable cost, almost 30-fold lower than the cooler bag (Supplementary Figure 11). Before making a bulk purchase, the shipment temperature in these vaccine cooler boxes was compared with the cooler bag to ensure sample integrity (Supplementary Figure 12). In resource-constrained settings, collecting biological samples and maintaining their integrity is difficult. Transporting the samples in locally available vaccine bags from the participant's home to the field laboratory resolved this issue.
DBS collection is also known as spots of capillary blood obtained from a finger, heel, or ear lobe39. Using a DBS card for collecting blood samples is relatively painless and non-invasive, and can be collected in the participant's home through non-clinical but trained health personnel. The blood collected on the filter paper is easily dried and stored. A drop of whole blood occupies approximately 50 µL in a disc with a 12.7 mm diameter23. The ring finger is usually the preferred site for adults, and it is a common procedure in therapeutic monitoring. Although the steps involved in DBS collection for adult biomarker screening were visualized in earlier studies, the tasks and microsteps involved in resource-constrained settings have not been captured40,41. This study is among the first, to our knowledge, to capture DBSs from M, OAW, and C,) of the same household42. In rural settings, it is challenging, although the procedure is minimally invasive42. Frequent training for field surveyors about collecting valid DBSs, and technical clarifications on selecting the non-dominant hand, relaxing and massaging the arm, and selecting the ring or middle finger played a significant role in collecting valid DBSs33.
Similarly, for neonates, the capillary sampling through heel prick was performed for babies weighing from ~3 to 10 kg (birth to 6 months), and finger prick in follow-up (over 6 months) for babies weighing >10 kg. Following WHO guidelines, the choice and position (puncture at a 90° angle parallel to the heel) of the lancet for pricking has played a significant role in obtaining enough blood flow, successful collection of DBSs, and a slightly shorter estimated depth33,34. The length of the blade in a lancet varies by manufacturer (i.e., from 0.85-2 mm for neonates). In premature babies, lancets for heel prick (0.85 mm x 1.75 mm depth) and lancets for finger-prick (1 mm x 2.5 mm depth) were utilized with babies between 6 months and 8 years.
After heel prick, the blood wass drawn using PTS capillary tubes (Ref# 2866) to collect the sample without clots in the blood spots and to avoid stamping on the protein saver card. Based on our preliminary experiments, it is believed that placing the capillary tube downward takes up the blood promptly without any hindrances due to uniform surface tension.
Following the successful collection of valid DBSs from HAPIN participants in both the study sites, the sample collected in the protein saver card was dried overnight at room temperature (25 °C) in the field laboratory, and it was ensured that the protein saver card was free of any insects and houseflies by a covering insect net. After drying (brown color, Figure 5Cii), the DBS card was stored at -20 °C.
During the DBS collection at participants' homes, the dropped blood was within 12.7 mm spot, but after drying overnight at room temperature, the two individual spots became merged, in the NP site. The observed difference in the NP site might be due to the higher relative humidity, where the collected valid DBSs became invalid as the two individual dry blood spots became fused. Following the harmonized DBS collection procedure, the results of clinical biomarkers (oxidative stress, inflammation, endothelial dysfunction, lung insult) have been cross-validated on blinded samples in the LEADER laboratory at Emory University, and were found to be in good agreement (data not shown).
Biospecimen collection requires firm adherence to safety protocols. During the pandemic period (March 24, 2019 to June 2019), additional safety protocols were executed, following the advisories of the local government. The study staff was instructed to wear personal protective equipment (PPE) such as gloves, face masks, goggles, and aprons while traveling and in participant houses. Lab coats were mandatory while working in field offices, and field offices were equipped with biosafety cabinets for handling biospecimens. Training was provided to all staff on using and identifying damaged PPEs. The used PPEs were collected in separate disposal bags and handed over at the collaborating health centers for safe disposal to the common biomedical waste management facility authorized by the State Pollution Control Board.
Capturing high-resolution videos of field data collection, especially in challenging rural settings, will aid in bridging the training gaps in air pollution monitoring and field data collection. Overall, at every stage of project execution, the quality and reliability of data collection was ensured. Periodical trainings and retrainings of field staff built their capacity and confidence and avoided the expensive loss of sample integrity. The methods used are transferable and will assist other researchers in adopting environmental monitoring and biospecimen collection procedures in LMICs using cost-effective strategies.
The gaps and challenges faced throughout the HAPIN study, particularly in resource-limited rural areas, are also reported. We note that extensive preliminary work and training, reported elsewhere in publications detailing HAPIN's formative work, were critical for sorting out issues with the protocol, like the design of vests for sampling and the mechanisms for safe transport of both air pollution and biological specimens. Furthermore, during this period, many 'growing pains' were overcome, including handling of the very small 15 mm ECM gravimetric filters, techniques for placement of stove use monitors, etc.
Special attention was made while shipping the sampled filters, instruments, and biospecimens from households to the field laboratory. All air sampling equipment, accessories, and samples were tracked through inventory management at the central and field laboratories. It has allowed timely maintenance, repair, replacement, and evaluation of project supplies to provide uninterrupted data collection.
The data collection methods demonstrated here have proved reliable and consistent during the entire year-long study period. The use and adoption of affordable and smart technologies may point toward a future paradigm for randomized control trials (RCTs) and exposure-response studies, ensuring acceptable data collection to yield reliable results. Such endeavors are not without their challenges; however, as shown here, diligence and review of established protocols can ensure that field teams are able to adapt to changing circumstances, both expected (differences in household configurations, for example) and unexpected (COVID-19, hurricanes). For HAPIN, this began with hands-on training in the laboratory and the field provided by exposure and biomarker core experts prior to the trial. In addition, refresher training was given once every 6 months at various tiers throughout the study period. The periodic coaching increased the team's ability to efficiently sample and handle instruments, filters, and biospecimens. The visualization and field sampling procedures will be a valuable educational tool for researchers conducting similar large-scale epidemiological studies in India or LMICs.
The authors have nothing to disclose.
The investigators would like to thank the members of the advisory committee – Patrick Brysse, Donna Spiegelman, and Joel Kaufman – for their valuable insight and guidance throughout the implementation of the trial. We also wish to acknowledge all research staff and study participants for their dedication to and participation in this important trial.
This study was funded by the U.S. National Institutes of Health (cooperative agreement 1UM1HL134590) in collaboration with the Bill & Melinda Gates Foundation (OPP1131279). A multidisciplinary, independent Data and Safety Monitoring Board (DSMB) appointed by the National Heart, Lung, and Blood Institute (NHLBI) monitors the quality of the data and protects the safety of patients enrolled in the HAPIN trial. NHLBI DSMB: Nancy R. Cook, Stephen Hecht, Catherine Karr (Chair), Joseph Millum, Nalini Sathiakumar, Paul K. Whelton, Gail Weinmann and Thomas Croxton (Executive Secretaries). Program Coordination: Gail Rodgers, Bill & Melinda Gates Foundation; Claudia L. Thompson, National Institute of Environmental Health Science; Mark J. Parascandola, National Cancer Institute; Marion Koso-Thomas, Eunice Kennedy Shriver National Institute of Child Health and Human Development; Joshua P. Rosenthal, Fogarty International Center; Conception R. Nierras, NIH Office of Strategic Coordination Common Fund; Katherine Kavounis, Dong- Yun Kim, Antonello Punturieri, and Barry S. Schmetter, NHLBI.
HAPIN Investigators: Vanessa Burrowes, Alejandra Bussalleu, Devan Campbell, Eduardo Canuz, Adly Castañaza, Howard Chang, Yunyun Chen, Marilú Chiang, Rachel Craik, Mary Crocker, Victor Davila-Roman, Lisa de las Fuentes, Oscar De León, Ephrem Dusabimana, Lisa Elon, Juan Gabriel Espinoza, Irma Sayury Pineda Fuentes, Dina Goodman, Meghan Hardison, Stella Hartinger, Phabiola M Herrera, Shakir Hossen, Penelope Howards, Lindsay Jaacks, Shirin Jabbarzadeh, Abigail Jones, Katherine Kearns, Jacob Kremer, Margaret A Laws, Pattie Lenzen, Jiawen Liao, Fiona Majorin, McCollum, John McCracken, Julia N McPeek, Rachel Meyers, Erick Mollinedo, Lawrence Moulton, Luke Naeher, Abidan Nambajimana, Florien Ndagijimana, Azhar Nizam, Jean de Dieu Ntivuguruzwa, Aris Papageorghiou, Usha Ramakrishnan, Davis Reardon, Barry Ryan, Sudhakar Saidam, Priya Kumar, Meenakshi Sundaram, Om Prashanth, Jeremy A Sarnat, Suzanne Simkovich, Sheela S Sinharoy, Damien Swearing, Ashley Toenjes, Jean Damascene Uwizeyimana, Viviane Valdes, Kayla Valentine, Amit Verma, Lance Waller, Megan Warnock, Wenlu Ye.
BD adult lancet | BD Biosciences | 366594 | DBS collection from finger |
BD Quikheek infant safety lancet | BD Biosciences | 368100 & 368101 | Heel prick DBS collection |
Beacon | Roximity | O/EM | Time and location monitor [TLM] (Personal monitor) |
Beacon Logger | Berkley Air Monitoring group | xxxx | Time and location logger [TLL] (Indirect measurement) |
Cr![]() |
Peli Biothermal USA | Cooler bag | |
Enhanced Children MicroPEM (ECM) | RTI International, Durham, NC, US | xxxx | Personal monitor of PM2.5 |
E-sampler | Met One Instruments | 9800 | Indirect measurement of ambient PM2.5 |
Geocene | Geocene Inc., Vallejo,CA | xxxx | for stove use monitoring |
Humidity indicating card | DESSICARE, INC. | 04BV14C10 | Sample integrity indicator |
Lascar | Lascar Electronics | EL-USB-300 | Carbon monoxide (CO) data logger |
PTS collect capillary tubes- 40 µL | PTS collect | 2866 | To collect heel prick DBS from children |
Sartorius | Sartorius Lab Instruments, GmbH & Co, Germany | MSA6-6S-000-DF | Microbalance (Weighing filters) |
SootScanTM | Magee Scientific Co, Berkeley, USA | OT21 | Black carbon measurement |
Vaccine Bag | Apex International, India | AIVC-46 | Vaccine Bag |
Whatman 903 Protein Saver card | GE Healthcare Life Sciences | 10534612 | Collection of capillary blood samples (Dried Blood Spot) |