JOURNAL OF f i f t e e n d o l l a r s Environmental Health Published by the National Environmental Health Association www.neha.org Dedicated to the advancement of the environmental health professional Volume 85, No. 3 October 2022
October 2022 • Journal of Environmental Health 3 ADVANCEMENT OF THE SCIENCE Estimation of High Blood Lead Levels Among Children in Georgia: An Application of Bayesian Analysis ........................................................................................... 8 Exploring Foodborne Illness and Restaurant Cleanliness Reporting in Customer-Generated Online Reviews Using Business Analytics............................................... 16 Lead-Based Paint and Other In-Home Health Hazards in Las Vegas, Nevada: Findings of the Las Vegas Lead Hazard Control and Healthy Homes Program ....................................... 24 ADVANCEMENT OF THE PRACTICE Building Capacity: Capacity Building for Retail Food Regulatory Programs ............................... 32 Direct From CDC/Environmental Health Services: The Environmental Health Nexus: A Communication Hub ................................................................................................................ 34 Direct From ecoAmerica: The United Nations, Climate Change, Environmental Health, and You .......................................................................................................................... 36 NEW The Practitioner’s Tool Kit: An Introduction and Checking Field Thermometer Accuracy ............................................................................................................... 40 ADVANCEMENT OF THE PRACTITIONER EH Calendar .............................................................................................................................. 44 Resource Corner........................................................................................................................ 45 JEH Quiz #2............................................................................................................................... 46 YOUR ASSOCIATION President’s Message: Back in the Saddle ..................................................................................................6 NEHA 2023 AEC....................................................................................................................... 47 Special Listing ........................................................................................................................... 48 NEHA 2022 AEC Wrap-Up....................................................................................................... 50 NEHA News .............................................................................................................................. 62 DirecTalk: Seersucker Thursday ................................................................................................. 66 On June 28–July 1, 2022, the National Environmental Health Association held its 85th Annual Educational Conference (AEC) & Exhibition in Spokane, Washington, bringing together environmental health professionals from across the country and globe. The 2022 AEC marked the first time since 2019 that we were able to reconnect in person. The 2022 AEC was o¥ered as a hybrid event, providing education and interaction for in-person and virtual attendees. We feature a special wrap-up of the 2022 AEC in this issue, highlighting our featured speakers, educational sessions, social events, exhibition, and award and scholarship winners. See page 50. Cover image © iStockphoto: Flash vector A B O U T T H E C O V E R A D V E R T I S E R S I N D E X Accela ................................................................... 39 GOJO Industries................................................... 23 Hedgerow Software U.S., Inc. ............................... 31 HS GovTech (Formerly HealthSpace) .................. 68 Industrial Test Systems, Inc. ................................. 43 Inspect2GO Environmental Health Software ......... 2 NEHA-FDA Retail Flexible Funding Model Grant Program .......................................... 67 NSF International ................................................... 5 Ozark River Manufacturing Co. ........................... 15 Sweeps Software, Inc. ............................................. 7 JOURNAL OF Environmental Health Dedicated to the advancement of the environmental health professional Volume 85, No. 3 October 2022
4 Volume 85 • Number 3 Of f i c i a l Pub l i ca t i on Journal of Environmental Health (ISSN 0022-0892) Kristen Ruby-Cisneros, Managing Editor Ellen Kuwana, MS, Copy Editor Hughes design|communications, Design/Production Cognition Studio, Cover Artwork Soni Fink, Advertising For advertising call (303) 802-2139 Technical Editors William A. Adler, MPH, RS Retired (Minnesota Department of Health), Rochester, MN Gary Erbeck, MPH Retired (County of San Diego Department of Environmental Health), San Diego, CA Thomas H. Hatfield, DrPH, REHS, DAAS California State University, Northridge, CA Dhitinut Ratnapradipa, PhD, MCHES Creighton University, Omaha, NE Published monthly (except bimonthly in January/February and July/ August) by the National Environmental Health Association, 720 S. Colorado Blvd., Suite 105A, Denver, CO 80246-1910. Phone: (303) 7569090; Fax: (303) 691-9490; Internet: www.neha.org. E-mail: kruby@ neha.org. Volume 85, Number 3. Yearly subscription rates in U.S.: $150 (electronic), $160 (print), and $185 (electronic and print). Yearly international subscription rates: $150 (electronic), $200 (print), and $225 (electronic and print). Single copies: $15, if available. Reprint and advertising rates available at www.neha.org/JEH. CPM Sales Agreement Number 40045946. Claims must be filed within 30 days domestic, 90 days foreign, © Copyright 2022, National Environmental Health Association (no refunds). All rights reserved. Contents may be reproduced only with permission of the managing editor. Opinions and conclusions expressed in articles, columns, and other contributions are those of the authors only and do not reflect the policies or views of NEHA. NEHA and the Journal of Environmental Health are not liable or responsible for the accuracy of, or actions taken on the basis of, any information stated herein. NEHA and the Journal of Environmental Health reserve the right to reject any advertising copy. Advertisers and their agencies will assume liability for the content of all advertisements printed and also assume responsibility for any claims arising therefrom against the publisher. Full text of this journal is available from ProQuest Information and Learning, (800) 521-0600, ext. 3781; (734) 973-7007; or www.proquest. com. The Journal of Environmental Health is indexed by Current Awareness in Biological Sciences, EBSCO, and Applied Science & Technology Index. It is abstracted by Wilson Applied Science & Technology Abstracts and EMBASE/Excerpta Medica. All technical manuscripts submitted for publication are subject to peer review. Contact the managing editor for Instructions for Authors, or visit www.neha.org/JEH. To submit a manuscript, visit http://jeh.msubmit.net. Direct all questions to Kristen Ruby-Cisneros, managing editor, email@example.com. Periodicals postage paid at Denver, Colorado, and additional mailing offices. POSTMASTER: Send address changes to Journal of Environmental Health, 720 S. Colorado Blvd., Suite 105A, Denver, CO 80246-1910. Printed on recycled paper. in the next Journal of Environmental Health Bystander Chemical Exposures and Injuries Associated With Nearby Plastic Sewer Pipe Manufacture: Public Health Practice and Lessons A Rapid Screening Method for Detecting Hazardous Chemicals in Consumer Products, Food Contact Materials, and Thermal Paper Receipts Using ATR-FTIR Spectroscopy Risks and Understanding of Carbon Monoxide Poisoning in an Ice Fishing Community don’t miss NOW AVAILABLE: The updated REHS/RS Study Guide Fifth Edition! EDUCATION & TRAINING Recreated in a fresh visual layout to enhance the reading and studying experience Helps identify content areas of strength and areas where more studying is needed Incorporates insights of 29 subject matter experts Includes 15 chapters covering critical exam content areas Visit our Study References page for more information! NEHA.ORG/REHS-STUDY-REFERENCES
October 2022 • Journal of Environmental Health 5 IN PUBLIC HEALTH YOUR PARTNER FOOD SAFETY WASTEWATER POOLS & SPAS DRINKING WATER TRAINING SUSTAINABILITY On Farm Food Processing Distribution and Retail Food Equipment Dietary Supplements Organic Foods Performance and Safety Energy Efficiency Filtration and Recirculation Components HACCP Allergens Plan Review SQF, BRC, IFS Food Equipment Traceability and Recall Supply Chain Food Safety Life Cycle Analysis Green Building Products Environmental Declarations WaterSense® Energy Star Individual Onsite Wastewater Treatment Systems Advanced Treatment Systems Water Reuse Residential Point-of-Entry/ Point-of-Use Treatment Units Municipal Treatment Chemicals Distribution System Components Plumbing and Devices Visit www.nsf.org/regulatory to submit inquiries, request copies of NSF standards or join the regulatory mailing list. NSF International • 1-800-NSF-MARK • www.nsf.org/regulatory Standards • Audits • Testing • Certification Code Compliance • Webinars • Regulatory Support
6 Volume 85 • Number 3 YOUR ASSOCIATION D. Gary Brown, DrPH, CIH, RS, DAAS Back in the Saddle PRES IDENT ’ S MESSAGE As I write this next column I have just returned from beautiful Spokane, Washington, after attending the successful National Environmental Health Association (NEHA) 2022 Annual Educational Conference (AEC) & Exhibition with approximately 1,000 in-person and 400 virtual attendees. Thank you for making the 2022 AEC a success. Words cannot express how wonderful it was to see colleagues, friends, and members of my NEHA family. I have attended NEHA AECs since 2001 and have made numerous friends, many of whom have become part of my family. A shared passion to advance environmental health science—while helping people have clean air, food, and water, along with a safe place to live, work, and play— means we have an instant connection when meeting fellow professionals. Great minds think alike. I have been lucky enough to live in various parts of our beautiful country. People from NEHA—as well as Eastern Kentucky University, the Kentucky Environmental Health Association, and Jamaican Association of Public Health Inspectors—have all become a part of my family. I grew up as a Bualonian but am now an Alabamian, Kentuckian, Jamerican (i.e., Jamaican American), Manhattanite, and “NEHAian.” The NEHA AEC allows me the opportunity to reconnect with many members of my dierent families. Aristotle wrote, “Man is by nature a social animal.” We are inherently social creatures, something the COVID-19 pandemic brought to the forefront for many people. As we learned during the pandemic, people around the world experienced increased loneliness that can have implications for long-term mental and physical health, longevity, and well-being (Ernst et al., 2022) The 2022 AEC was a celebration of the return of being in person. Everyone who I spoke with at the conference was ecstatic about reconnecting in person. The internet, social media, Zoom, Teams, etc. are useful tools, but they cannot replicate the in-person experience. I met numerous new people in Spokane, making personal connections that would have been much harder electronically. I learned something from everyone I met at the 2022 AEC, including students and professionals at all ends of the career spectrum. Many attendees do not realize that there are a number of preconference oerings at the AECs, many of which are oered for free or at a minimal cost for NEHA members. Preconference oerings this year included: • review courses for the NEHA Certified Professional–Food Safety (CP-FS) and Registered Environmental Health Specialist/ Registered Sanitarian (REHS/RS) credential exams, • inspector training for body art facilities, • the Environmental Health and Land Reuse Certificate Program from NEHA and the Agency for Toxic Substances and Disease Registry, • a National Retail Food Regulatory Program Standards Self-Assessment and Verification Audit Workshop from NEHA and the Food and Drug Administration, • Climate for Health Ambassador Training, • and many others. The NEHA AEC is much more than a conference. It is the nexus for environmental health training, education, networking, and advancement. The NEHA AEC is the most comprehensive training and education investment you and your organization can make to achieve immediate and long-term benefits. Attendees at the NEHA AEC acquired practical and real-world information and expertise from like-minded professionals who share your passion for environmental health. Attendees leave the conference trained, motivated, inspired, and empowered to further advance themselves and their organizations. Attendees gain the skills, knowledge, and expertise needed to help solve daily and strategic challenges within their organizations, as A shared passion to advance environmental health science means we have an instant connection when meeting fellow professionals.
October 2022 • Journal of Environmental Health 7 well as improve bottom-line results. An added bonus is that attendees can earned continuing education contact hours to maintain their professional credentials. I thank the NEHA sta who ran our first hybrid AEC, which entailed running two conferences at the same time—in person and virtual. The NEHA sta worked 12 days straight before and during the conference from early in the morning to late in the evening. NEHA is lucky to have such dedicated, hardworking sta . The Washington State Environmental Health Association also deserves high praise for their role in helping to make this conference a success. Many attendees do not realize how valuable the AEC sponsors are. The generous contributions from the sponsors help reduce the cost of attendance, as well as provide their expertise, products, and services throughout the year. NEHA is incredibly lucky to have sponsors who are true partners and who help spread the vision and mission of NEHA. A special thank you goes out to Dr. Priscilla Oliver and Sandra Long, past presidents ofNEHA,whokepttheNEHAshipsteeredin the right directionand helpedour organization gain steam during the pandemic.Their stellar leadershipduring their presidencies allowed NEHA to increase revenue along with reserves. Theydid not have the opportunityto attend and lead an in-personAEC. Theyhave my gratitudeas well as that from the environmentalhealthprofessionandour members and staff . We want our members to have the best possible experience when attending the AEC. When NEHA considers a location for the AEC, numerous factors are taken into consideration including cost, desirability of the location, and the facilities. Numerous hours go into the groundwork before a location is selected. Once NEHA determines the location of the AEC, the challenging work really begins. For the 2023 AEC, New Orleans is much more than Bourbon Street and world-renowned food. New Orleans has something for everyone including one of the country ’s top-rated aquariums and zoos, historic homes, the National World War II Museum, the New Orleans Museum of Art, and architectural gems. All environmental health professionals have unique knowledge and experiences. I hope you took advantage of the 2023 AEC Call for Abstracts that was open in September to share your story and knowledge. I look forward to seeing you at the 2023 AEC in New Orleans, Louisiana, on July 31–August 3. Reference Ernst, M., Niederer, D., Werner, A.M., Czaja, S. J., Mikton, C., Ong, A.D., Rosen, T., Brähler, E., & Beutel, M.E. (2022). Loneliness before and during the COVID19 pandemic: A systematic review with meta-analysis. American Psychologist, 77(5), 660–677. https://doi.org/10.1037/ amp0001005 firstname.lastname@example.org Software Incorporated Environmental Health Software ® Call today! It was great seeing you at the NEHA 2022 AEC in Spokane! Contact Information (800) 327-9337 www.SweepsSoftware.com Info@SweepsSoftware.com SWEEPSsoftware is designed to effectively manage your resources, staff and programs. SWEEPS software is designed to exceed the requirements for available grant programs. “Make Your Data Work as Hard as You Do!”
8 Volume 85 • Number 3 A D VANC EME N T O F T H E SCIENCE Introduction Lead exposure can seriously a ect the health of children (World Health Organization, 2022). High levels of lead exposure can harm the brain and central nervous system of children. High levels of lead exposure can also cause coma, convulsions, and death in children. Children who survive severe lead poisoning can su er from mental deficiencies and behavioral disorders. Lead is known to a ect children’s brain development and can result in reduced IQ and behavioral changes such as short attention span and reduced educational attainment. Most importantly, these neurological and behavioral e ects of lead are irreversible (Centers for Disease Control and Prevention [CDC], 2022; Egan et al., 2021). Georgia Department of Public Health (n.d.) guidelines for blood lead screening recommend screening children who belong to high-risk groups such as families receiving Medicaid or Peach Care for Kids (i.e., health coverage for children in low-income families). The guidelines also recommend screening in 16 counties in which children are at greater risk for lead exposure. Following these guidelines, the resulting group of children to be tested for elevated blood lead levels (BLLs), however, is limited and some children with elevated BLLs might be missed. In 2012, the Centers for Disease Control and Prevention (CDC, 2021) defined a BLL of 5 µg/dL as a reference value for children <6 years. Note, this reference value was changed to a more stringent level of 3.5 µg/dL but at the time of our study the limit was 5 µg/dL. Bayesian analysis with limited beliefs about a parameter can be helpful in modeling the exposure of lead in children by suitably matching these beliefs with some known distribution. The primary objective of our study was to estimate and validate the observed number of children with BLLs of 5–9 µg/dL among children <6 years in di erent counties of Georgia, selected by region. This objective was important to investigate if screening of a limited group of children in Georgia resulted in underreporting of children with elevated BLLs. Although some studies have connected targeted screening and missed children with elevated BLLs (Roberts et al., 2017), no such research work has been found evaluating the impact of targeted screening on the rate of children <6 years with elevated BLLs in a region, especially in Georgia. Methods Data Collection We used data collected by the Healthy Homes and Lead Poisoning Prevention Program of the Georgia Department of Public Health for 2015. Child blood lead surveillance data was used, including the number of children <6 years who were tested and the number of children with BLLs of 5–9 µg/dL, by race and county. Estimates of children <6 years were available from the Georgia Governor’s O£ce of Planning and Budget (2016). Bayesian Model The variable z was used to represent the number of children <6 years with BLLs of 5–9 µg/dL in a county in Georgia. Because this event is rare, one can safely assume that z follows a statistical distribution known as Poisson distribution shown by: p(z/θ) = e–(m . θ)(m . θ)z/z! (1) Where θ is the rate of children with BLLs of 5–9 µg/dL (i.e., θ = children with BLLs of Abs t r ac t In Georgia, children in high-risk counties are at increased risk for lead exposure. Those children and others in high-risk groups, such as families receiving Medicaid and Peach Care for Kids (i.e., health coverage for children in low-income families), are screened for blood lead levels (BLLs). Such screening, however, might not include all children at high risk for having BLLs above the reference levels (≥5 µg/dL) in the state. In our study, Bayesian methods were used to estimate the predictive density of the number of children <6 years with BLLs of 5–9 µg/dL in a targeted county from each of five selected regions of Georgia. Furthermore, the estimated mean number of children with BLLs of 5–9 µg/dL in each targeted county, along with its 95% credible interval, were calculated. The model revealed likely underreporting of some children <6 years with BLLs of 5–9 µg/dL in counties of Georgia. Further investigation might help reduce underreporting and better protect children who are at risk for lead poisoning. Estimation of High Blood Lead Levels Among Children in Georgia: An Application of Bayesian Analysis Shailendra N. Banerjee, PhD National Center for Environmental Health, Centers for Disease Control and Prevention
October 2022 • Journal of Environmental Health 9 5–9 µg/dL/children tested for BLL); m is the number of children <6 years who were tested for BLL; m . θ is the number of children with BLLs of 5–9 µg/dL; and p(z/θ) is the probability that there are z number of children <6 years with BLLs of 5–9 µg/dL under the assumption that θ is the rate of children with BLLs of 5–9 µg/dL. Clearly, θ is unknown or a parameter, and under the Bayesian principle, one tries to estimate it based on a reasonable assumption of its statistical distribution, called “prior distribution” or simply “prior.” It is reasonable to assume that a parameter coming from a Poisson distribution should follow a statistical distribution called gamma distribution. Thus, this model assumes that θ follows a gamma (α, β) prior: p(θ) = e–(β θ) βα θα-1/Γ(α) (2) Where θ > 0, and α and β are its unknowns or parameters. Then, according to Bayesian rule, actual or simply put, posterior distribution, p(θ/z) of θ, will be given by p(θ/z) = p(z/θ) × p(θ)/p(z), which is the distribution of the observed number multiplied by the prior of its parameter divided by the constant p(z). That is: p(θ/z) = e–(m . θ)(m . θ)z × e–(β θ)βα θα-1/z! Γ(α) p(z) (2a) or, p(θ/z) = e– θ(β + m) (θ)z+α-1 × constant (3) Here, the right-hand side of Equation 2 and that of the posterior distribution in Equation 3 are similar, which indicates that the posterior is also a gamma (α1, β1) distribution with parameters α1 and β1 where: α1 = z + α and β1 = β + m (3a) This equation means that if one assumes that the prior information about parameter θ (the rate of children with BLLs of 5-9 µg/ dL) can be obtained from a small group of counties in Georgia, each of which is believed to have the same rate (θ) of 5–9 µg/dL BLLs among children <6 years, then applying Bayesian rule, the posterior for θ can be estimated from a gamma distribution as shown in Equation 3. Moreover, if one supposes zj is the number of children <6 years with BLLs of 5–9 µg/ dL among xj children from county j, then, assuming zj follows a Poisson distribution, one would have, as in Equation 1: p(zj/θ) = e –(xj θ)(x jθ) zj/z j! (4) Where θis the same as defined earlier. Thus, the likelihood function for n counties with the same parameter θ is given as follows: L(∑zj/θ) = e –(∑xj θ)∏(x jθ) zj/z 1! z2 ! ….. zn! (5) This equation is obtained by multiplying density functions like Equation 4 for n counties. Omitting the constant terms, one has: L(∑zj/θ) ∝ e –(∑xj θ)(θ)∑zj (6) Where ∝ indicates proportionality. If for all these n counties, one assumes that θ follows a noninformative prior 1/θ (i.e., p(θ) = 1/θ), then as was done in Equation 2a and from Equation 6, the posterior distribution of θis given by the following: p(θ/∑zj) ∝ e –(∑xjθ)(θ)∑zj . 1/θ (i.e., p(θ/∑zj) ∝ e –(∑xjθ)(θ)∑zj-1) (7) This is a gamma (α2, β2), where: α2 = ∑zj and β2 = ∑xj (8) Here, ∑zj is the shape parameter and ∑xj is the rate parameter of this gamma distribution, where zj is the number of children <6 years with BLLs of 5–9 µg/dL in county j and xj is the number of children tested for BLL in county j. The assumption is that the rate of children with BLLs of 5–9 µg/dL among children <6 years in these counties is similar to that in a targeted county where one wants to estimate that rate. One can then use known α and β from Equation 8 in Equations 2 and 3 to evaluate the prior and posterior distributions of the parameter θin the targeted county. According to the multiplication rule of probability, the joint distribution of data z and the parameter θare given by the following: p(z,θ) = p(θ) × p(z/θ), and also p(z,θ) = p(z) × p(θ/z) Thus, p(z) × p(θ/z) = p(θ) × p(z/θ), giving: p(z) = p(θ) × p(z/θ)/p(θ/z) (9) Here, p(θ) and p(θ/z) are the known prior and posterior distributions, respectively, of the parameter θ. Thus, p(θ) is a gamma density with the known shape and rate parameters from Equation 8. Similarly, p(θ/z) is a gamma density with known shape and rate parameters from Equations 8 and 3a. Assuming that p(z/θ) is the sampling distribution of data in the targeted county, one can estimate the predictive density p(z) of z in the targeted county from Equation 9 before any data are observed, where p(z/θ) is a Poisson density with known mean (mθ) as shown in Equation 1. If our model assumptions for sampling distribution of data and prior density are valid, one can check the validity of the observed values of the number of children <6 years with BLLs of 5–9 µg/dL in the targeted county. Detailed information about this Bayesian model can be found at www.neha.org/jeh/ supplemental. County and Region Selection The model was applied by dividing Georgia into five di¨erent regions: North, South, East, West, and Central. Then 11 neighboring counties were arbitrarily selected in each region, assuming similarity of BLL rates of 5–9 µg/dL among children ages <6 years in these counties. For each region, the county with the lowest observed proportion of children with BLLs 5–9 µg/dL was selected as the targeted county. The remaining 10 counties from each region provided data for estimation of parameters α and β for the prior distribution. The parameter θ, the rate of children with BLLs of 5–9 µg/dL in the targeted county, was estimated from the mean value α/β of the gamma distribution, as the predictive density (Equation 9) is valid for all θ. Data Analysis Data were analyzed using statistical software SAS (version 9.4) and R package. For each region, predictive density was calculated for the targeted county from Equation 9 for all children, and separately for White and non-White children. We assumed that the observed value for the number of children with BLLs of 5–9 µg/dL among children <6 years within the three largest predictive probabilities was compatible. Additionally, the mean number of children with BLLs of 5–9 µg/dL was estimated in the
10 Volume 85 • Number 3 A D VANC EME N T O F T H E SCIENCE Observed Blood Lead Levels (BLLs) for Children <6 Years From 11 Neighboring Counties in the North Region of Georgia, 2015 County a # of Children <6 Years # of Children <6 Years With BLLs of 5–9 µg/dL Total # of Children <6 Years Tested All White Non-White All White Non-White A 2,401 5 2 3 319 169 150 B 1,067 6 4 2 323 213 110 C 834 3 1 2 194 157 37 D 400 0 0 0 113 91 22 E 3,552 1 1 0 193 142 51 F 1,581 9 3 6 651 330 321 G 1,423 4 3 1 219 130 89 H 1,387 4 3 1 368 208 160 I 2,571 10 3 7 740 415 325 J 1,625 9 7 2 529 365 164 X b 743 0 0 0 246 148 98 a These 11 counties were chosen arbitrarily because they are contiguous. The assumption was that because they are contiguous, these counties will have similar BLL rates of 5–9 µg/dL among children <6 years. b X indicates the targeted county. A targeted county is one with the lowest observed proportion of tested children with BLLs of 5–9 µg/dL among children <6 years. TABLE 1 Observed Blood Lead Levels (BLLs) for Children <6 Years From 11 Neighboring Counties in the East Region of Georgia, 2015 County a # of Children <6 Years # of Children <6 Years With BLLs of 5–9 µg/dL Total # of Children <6 Years Tested All White Non-White All White Non-White A 400 4 3 1 39 13 26 B 13,956 49 10 39 1,817 303 1,514 C 1,595 8 1 7 393 104 289 D 829 2 0 2 205 57 148 E 535 2 0 2 62 16 46 F 1,467 11 1 10 177 47 130 G 985 3 2 1 162 38 124 H 494 3 2 1 123 50 73 I 4,196 20 9 11 1,203 255 948 J 1,132 16 5 11 722 228 494 X b 9,328 2 1 1 458 233 225 a These 11 counties were chosen arbitrarily because they are contiguous. The assumption was that because they are contiguous, these counties will have similar BLL rates of 5–9 µg/dL among children <6 years. b X indicates the targeted county. A targeted county is one with the lowest observed proportion of tested children with BLLs of 5–9 µg/dL among children <6 years. TABLE 2
October 2022 • Journal of Environmental Health 11 targeted county from Equation 9 by simultaneously simulating 1,000 values from each of the probability densities p(θ), p(θ/z), and p(z/θ). A 95% credible interval for the mean number of children with BLLs of 5–9 µg/dL was estimated from the simulated values. An observed number of children with BLLs of 5–9 µg/dL in the targeted county was considered an acceptable number if within the boundaries of the credible interval for that county. The estimated mean number of children <6 years with BLLs of 5–9 µg/dL in the targeted county was recommended as the true value if the observed value was outside the boundaries of the credible interval. Results Tables 1, 2, and 3 show the observed numbers of White, non-White, and total children who had their BLLs tested and those children with BLLs of 5–9 µg/dL in the North, East, and South regions of Georgia. The 11 counties chosen in each of the regions, including West and Central regions (not shown in the tables), were next to each other. For our study, it was assumed that the BLL rates among children <6 years could be similar in each county because of their proximity to each other. County X in the last row of each table represents the targeted county where the proportion of children <6 years with BLLs of 5–9 µg/dL was found to be lowest among the 11 counties and the value of county X was estimated by the model. Tables 1, 2, and 3 (representing North, East, and South regions of Georgia, respectively) have slightly dierent distributions of proportion of children with BLLs of 5–9 µg/ dL between White and non-White children. In the North region (Table 1), a smaller proportion of non-White children were tested for BLL in almost all the counties—and yet a higher percentage of them were found to have BLLs of 5–9 µg/dL. Thus, in county I in the North region, only 3 (0.07%) out of 415 White children tested had BLLs of 5–9 µg/dL, compared with 7 (2.15%) out of 325 nonWhite children tested. This finding is similar to that of county C in the North region: 1 (0.06%) out of 157 White children tested had BLLs of 5–9 µg/dL, compared with 2 (5.4%) out of 37 non-White children tested. In the East region (Table 2) and South region (Table 3), however, the situation was found to be completely the opposite. In both these regions, a smaller proportion of White children were tested, with a higher proportion of children with BLLs of 5–9 µg/dL in almost all the counties. Thus, in county A in the East region, 3 (23.08%) out of 13 White children had BLLs of 5–9 µg/dL, compared with 1 (3.84%) out of 26 non-White children. Similarly, in county A in the South region, 5 (8.77%) out of 57 White children tested had BLLs of 5–9 µg/dL, compared with 7 (1.00%) out of 70 non-White children tested. Tables 4, 5, and 6 show the predictive densities or estimated probabilities for 0–15 children <6 years with BLLs of 5–9 µg/dL in the targeted county for all, White, and nonWhite children, respectively. Each of these tables show probabilities for the five regions calculated based on Equation 9. According to Table 4, the estimated probabilities were found to be highest (0.190, 0.212, 0.181) at moderately three smaller numbers (2, 3, and 4, respectively) of all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the North region. This finding indicates that the number of all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the North region should be small, which is corroborated by its 95% credible interval [0.0, 9.3] shown in Table 7. Moreover, this Observed Blood Lead Levels (BLLs) for Children <6 Years From 11 Neighboring Counties in the South Region of Georgia, 2015 County a # of Children <6 Years # of Children <6 Years With BLLs of 5–9 µg/dL Total # of Children <6 Years Tested All White Non-White All White Non-White A 473 12 5 7 127 57 70 B 1,757 9 2 7 109 39 70 C 1,686 11 5 6 554 212 342 D 968 6 2 4 151 54 97 E 7,952 27 9 18 1,206 643 563 F 347 8 3 5 79 41 38 G 1,326 6 2 4 371 124 247 H 3,235 18 7 11 776 366 410 I 1,154 13 6 7 215 112 103 J 769 4 3 1 102 65 37 X b 2,910 15 4 11 990 519 471 a These 11 counties were chosen arbitrarily because they are contiguous. The assumption was that because they are contiguous, these counties will have similar BLL rates of 5–9 µg/dL among children <6 years. b X indicates the targeted county. A targeted county is one with the lowest observed proportion of tested children with BLLs of 5–9 µg/dL among children <6 years. TABLE 3
12 Volume 85 • Number 3 A D VANC EME N T O F T H E SCIENCE finding proves that the “0” observed number of all children with BLLs of 5–9 µg/dL in the targeted county (Table 1) is acceptable according to our model. The same findings holds true for the Central region, where the probabilities are highest (0.256, 0.270, 0.189) for a relatively smaller number (1, 2, and 3, respectively) of all children <6 years with BLLs of 5–9 µg/dL in the targeted county. The probabilities are, however, highest for a slightly larger number (9, 10, and 11) of all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the East region. For the South and West regions, the highest probabilities are not reached within a number of 15 for all children <6 years with BLLs of 5–9 µg/dL in the targeted county, indicating the number of children should be higher (Table 4). Clearly, an observed number of 14 for all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the West region (Table 7) is not acceptable because its 95% credible interval based on our model is [30.7, 65.3]. The same trend is observed for estimated probabilities for White and non-White children as shown in Tables 5 and 6. Table 7 shows the observed number of children <6 years with BLLs of 5–9 µg/dL in the targeted county, along with their estimated number and their 95% credible interval based on simulation. It is important to note from Table 7 that in only two regions—North and Central—the estimated numbers of children <6 years with BLLs of 5–9 µg/dL in the targeted county concurred with the observed values, which is true for all, White, and nonWhite children. Figure 1 shows the estimated probability distribution for all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the West and Central regions. The distribution in the West region, where the observed value of those children was not acceptable according to the model, is markedly dierent from the distribution in the Central region, where the model supported the observed value. The estimated probability is shown to be highest around 40 in the West region, indicating that the number of all children ages <6 years with BLLs of 5–9 µg/dL in the targeted county should be much higher than the observed value of 14, which is not acceptable. In the Central region, however, the estimated probability is shown to be Predictive Density for All Children <6 Years With Blood Lead Levels of 5–9 µg/dL in the Targeted County by Region in Georgia, 2015 # of Children Probability by Region North East South West Central 0 0.036 0 0 0 0.121 1 0.116 0 0 0 0.256 2 0.190 0.001 0 0 0.270 3 0.212 0.005 0 0 0.189 4 0.181 0.012 0 0 0.100 5 0.125 0.025 0 0 0.042 6 0.074 0.044 0 0 0.015 7 0.038 0.066 0 0 0.005 8 0.017 0.088 0 0 0.001 9 0.007 0.106 0 0 0 10 0.003 0.115 0 0 0 11 0.001 0.114 0 0 0 12 0 0.105 0 0 0 13 0 0.090 0 0 0 14 0 0.072 0.001 0 0 15 0 0.054 0.002 0 0 TABLE 4 Predictive Density for White Children <6 Years With Blood Lead Levels of 5–9 µg/dL in the Targeted County by Region in Georgia, 2015 # of Children Probability by Region North East South West Central 0 0.175 0.002 0 0 0.436 1 0.295 0.011 0 0.001 0.361 2 0.259 0.031 0 0.004 0.150 3 0.156 0.064 0.002 0.011 0.042 4 0.073 0.010 0.005 0.025 0.009 5 0.028 0.128 0.010 0.045 0.002 6 0.009 0.140 0.019 0.069 0 7 0.003 0.135 0.032 0.092 0 8 0.001 0.117 0.048 0.109 0 9 0 0.093 0.064 0.118 0 10 0 0.067 0.079 0.115 0 11 0 0.045 0.090 0.105 0 12 0 0.029 0.096 0.088 0 13 0 0.017 0.096 0.070 0 14 0 0.010 0.091 0.052 0 15 0 0.005 0.082 0.036 0 TABLE 5
October 2022 • Journal of Environmental Health 13 highest around 2 or 3, indicating that the number of all children <6 years with BLLs of 5–9 µg/dL is closer to the observed value of 1, which is acceptable. Discussion The estimated probabilities for all children <6 years with BLLs of 5–9 µg/dL in the targeted county in the Central region was highest for 1, 2, and 3 children (Table 4). The observed number of all children <6 years with BLLs of 5–9 µg/dL in the targeted county was 1 (Table 7). These results support the observed value. As further corroboration, the estimated number of all children with BLLs of 5–9 µg/dL in the targeted county in the Central region was found to be 2.1 through simulation. Its 95% credible interval was [0.0, 5.9] (Table 7), which included 1. Similar results were found for all, White, and non-White children for the North and Central regions. For the East region, however, the observed number of all children with BLLs of 5–9 µg/dL in the targeted county was 2 (Table 2) and the highest estimated probabilities were for 9, 10, and 11 children (Table 4). Similarly, the number of all children with BLLs of 5–9 µg/dL in the targeted county in the East region was estimated to be 11.9 by simulation and its 95% credible interval was [5.1, 20.2] (Table 7), which did not include 2. This finding shows discrepancies between the observed and estimated values of children with BLLs of 5–9 µg/dL in the targeted county. Similar results were found in the East region for White and non-White children. Discrepancies between observed and estimated numbers of children <6 years with BLLs of 5–9 µg/dL were also found for the targeted county in the South and West regions (Table 7). Our model shows the possibility of checking the validity of observed numbers of children with BLLs of 5–9 µg/dL and, if necessary, replacing those numbers with estimates that better reflect the actual probable numbers in the targeted counties. The model could reveal incorrect reporting of elevated BLLs in children <6 years, which might be the case if many of the targeted counties in dierent regions of a state show discrepancies between the observed and estimated numbers of children with BLLs of 5–9 µg/ dL. Therefore, this finding might also point to inadequacies in the screening process Predictive Density for Non-White Children <6 Years With Blood Lead Levels of 5–9 µg/dL in the Targeted County by Region in Georgia, 2015 # of Children Probability by Region North East South West Central 0 0.204 0.007 0 0 0.265 1 0.313 0.035 0 0 0.352 2 0.251 0.085 0 0 0.233 3 0.140 0.139 0 0 0.104 4 0.061 0.171 0 0 0.035 5 0.022 0.170 0.001 0 0.009 6 0.007 0.143 0.003 0 0.002 7 0.002 0.104 0.007 0 0 8 0 0.067 0.013 0 0 9 0 0.039 0.021 0 0 10 0 0.020 0.032 0 0 11 0 0.010 0.045 0 0 12 0 0.004 0.058 0 0 13 0 0.002 0.070 0 0 14 0 0.001 0.080 0.001 0 15 0 0 0.087 0.002 0 TABLE 6 Observed and Estimated Mean Number of Children <6 Years With Blood Lead Levels (BLLs) of 5–9 µg/dL and 95% Credible Interval in the Targeted County by Region in Georgia, 2015 Region Mean # of Children <6 Years With BLLs of 5–9 µg/dL All White Non-White North Observed 0 0 0 Estimated 3.8 2.0 1.9 95% credible interval [0, 9.3] [0, 5.9] [0, 5.6] East Observed 2 1 1 Estimated 11.9 8.4 5.3 95% credible interval [5.1, 20.2] [2.5, 17.1] [1.1, 11.1] South Observed 15 4 11 Estimated 34.6 16.2 17.9 95% credible interval [21.5, 50.8] [7.8, 28.7] [8.8, 30.0] West Observed 14 1 13 Estimated 46.0 11.8 35.0 95% credible interval [30.7, 65.3] [4.5, 22.4] [21.9, 51.6] Central Observed 1 0 1 Estimated 2.1 0.8 1.3 95% credible interval [0, 5.9] [0, 3.3] [0, 4.1] TABLE 7
14 Volume 85 • Number 3 used in the state, and thus lead to modifications to improve the process. Some studies have observed this inadequacy in the screening process of BLL surveillance data. Based on estimates of elevated BLL (≥10 µg/dL) data for children 1–5 years from 1999–2010 for 39 states (including Washington, DC) that were reported to CDC, Roberts et al. (2017) found that approximately 1.2 million children had elevated BLLs. Among these, 337,405 (approximately 28%) were not reported because of incomplete case ascertainment and far fewer cases were ascertained in the South and West regions. In Georgia, the case ascertainment ratio (i.e., the number reported/number of cases) was only 0.10. This finding points to undertesting of children with elevated BLL in many states, including Georgia. Similar results have been observed from other studies. According to data from the California Department of Health Care Services during 2009–2010 through 2017–2018, fewer than 27% of eligible children in California received all the required blood tests they should have, although many of these children lived in areas of the state with occurrences of elevated BLLs (Auditor of the State of California, 2020). Although these studies point to the inadequacy of the screening process for children, no study showed how inadequacy can aect actual BLLs among children <6 years. Our study fills the gap in that research and detects the discrepancy between estimated and observed numbers of children with higher (i.e., 5–9 µg/dL) BLLs—a discrepancy that resulted, most likely, from an undertesting of children with elevated BLLs. Most importantly, we find the corrected number of children with higher (i.e., 5–9 µg/dL) BLLs. Limitations Our study is subject to several limitations. For example, we assumed that the neighboring counties have similar BLL rates to what was found in the targeted county, which might not be true. If the neighboring counties do not have similar BLL rates, then the prior and posterior distributions of the parameter θ in the targeted county (Equation 9) will be distorted. The equation might still provide a reasonably reliable estimate, however, of the number of children with BLLs of 5–9 µg/dL in the targeted county, which is possible because prior p(θ) and posterior p(θ/z) occur in the numerator and denominator, respectively, of Equation 9 and might, to some extent, nullify each other’s distorting eect. If the risk factors for elevated BLLs in the targeted county, however, vastly dier from those in the neighboring counties, then this approach might not give a good estimate. We also assumed that the number of children with BLLs of 5–9 µg/dL followed a Poisson distribution and the BLL rate was distributed as gamma. The results might change if these model assumptions were modified. Conclusion We observed underreporting of children <6 years with BLLs of 5–9 µg/dL in some counties of Georgia. This finding is based on the application of a Bayesian model on county data. More research is needed to investigate BLLs among children to ensure they are adequately protected from lead poisoning. Our study has the appeal of being applied in any situation where surveillance data are collected to obtain vital information in institutions or communities, such as hospital-acquired infection in a specific hospital. For example, assuming that the rate of infection is similar to other hospitals in the vicinity, one can check the validity of the rates in this specific hospital and possibly correct it, if necessary, as we did in our study. Similar situations can arise in estimating heart transplant mortality in a hospital, or, as another example, estimating crime rate in a community from self-reported statistics. Our study, then, highlights a general approach to verify useful information and details an opportunity to estimate an actual value or index from observed data. Disclosure: The findings and conclusions in this article are those of the author and do not necessarily represent the o¤cial position of CDC. A D VANC EME N T O F T H E SCIENCE Plot for the Predictive Density of All Children <6 Years With Blood Lead Levels (BLLs) of 5–9 μg/dL in the Targeted Counties in the West and Central Regions of Georgia, 2015 Note. The observed value of children with BLLs of 5–9 μg/dL in the targeted county in the West region was 14 among 1,587 children tested. The observed value of children with BLLs of 5–9 μg/dL in the targeted county of the Central region was 1 among 170 children tested. 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 # of Children <6 Years With BLLs of 5–9 µg/dL 0 0.05 0.10 0.15 0.20 0.25 0.30 Probability Probability in the Central Region Probability in the West Region FIGURE 1
October 2022 • Journal of Environmental Health 15 Auditor of the State of California. (2020). Childhood lead levels: Millions of children in Medi-Cal have not received required testing for lead poisoning. https://www.auditor.ca.gov/pdfs/reports/2019-105. pdf Centers for Disease Control and Prevention. (2021). Blood lead reference value. https://www.cdc.gov/nceh/lead/data/blood-lead-refer ence-value.htm Centers for Disease Control and Prevention. (2022). Health e ects of lead exposure. https://www.cdc.gov/nceh/lead/prevention/healthe ects.htm Egan, K.B., Cornwell, C.R., Courtney, J.G., & Ettinger, A.S. (2021). Blood lead levels in U.S. children ages 1–11 years, 1976–2016. Environmental Health Perspectives, 129(3), Article 37003. https:// doi.org/10.1289/EHP7932 Georgia Department of Public Health. (n.d.). Healthy homes and lead poisoning prevention. https://dph.georgia.gov/environ mental-health/healthy-homes-and-lead-poisoning-prevention Governor’s Oce of Planning and Budget. (2016). Population estimates: County population by age, 2016. https://opb.georgia.gov/ census-data/population-estimates Roberts, E.M., Madrigal, D., Valle, J., King, G., & Kite, L. (2017). Assessing child lead poisoning case ascertainment in the U.S., 1999–2010. Pediatrics, 139(5), Article e20164266. https://doi. org/10.1542/peds.2016-4266 World Health Organization. (2022). Lead poisoning. https://www. who.int/news-room/fact-sheets/detail/lead-poisoning-and-health References Acknowledgements: The author is extremely grateful to Yu Sun, MPH, MD, an epidemiologist for the Healthy Homes and Lead Poisoning Prevention Program within the Environmental Health Section at the Georgia Department of Public Health, for her generous help in providing data for this research project. Corresponding Author: Shailendra N. Banerjee, Mathematical Statistician, Division of Environmental Health Science and Practice, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA 30341. Email: email@example.com. If Rex had washed his hands in our Titan PRO 1 portable sink, maybe — just maybe he wouldn’t be extinct. • Indoor & Outdoor • Self-Contained • On-Demand Hot Water • Out-of-the-Box Ready • NSF-Certified • Quick-Connect Tanks • Requires 110V 20A electric • Compact Design Dimensions: 25.75”W x 18.50”D x 53.75”H ©2022 Ozark River Manufacturing Free Catalog 1.866.663.1982 www.ozarkriver.com Find a Job Fill a Job Where the “best of the best” consult... NEHA ’ s Ca r e e r Cen t e r First job listing FREE for state, tribal, local, and territorial health departments with a NEHA member. For more information, please visit neha.org/careers.
16 Volume 85 • Number 3 A D VANC EME N T O F T H E SCIENCE Introduction The Centers for Disease Control and Prevention (CDC, 2020) estimates that 48 million people get sick from a foodborne illness (FBI) annually. Between 2009 and 2015, the Foodborne Disease Outbreak Surveillance System received reports of 5,760 outbreaks that caused 100,939 illnesses, 5,699 hospitalizations, and 145 deaths. Dewey-Mattia et al. (2018) included the specific location for food preparation for 5,022 outbreaks and showed that restaurants were the most common location (61%), followed by catering/banquet facilities (14%). FBI outbreaks are chronically underreported, however, because individuals and health professionals do not report a sizable number of cases to public health channels (CDC, 2018a). Therefore, there is a need for investigators to use novel and innovative methods to identify food safety issues and potential areas for improvement. While individuals who have a case of FBI might not report their cases to public health ocials, prior studies suggest that reviews— posted by restaurants patrons on online restaurant review forums such as Yelp.com—contain information related to FBI events (Nsoesie et al., 2014). Previous studies have used Yelp reviews as a tool to identify FBI outbreaks and have compared these reviews with health inspection scores (Harris et al., 2017; Park et al., 2016). No studies yet, however, have explored food safety or restaurant cleanliness issues in customer-generated reviews and examined how these issues aect customer satisfaction. Consequently, our primary research objectives were to 1) explore customer-generated reviews on an online review platform (i.e., Yelp) to identify FBI and restaurant cleanliness issues and 2) examine the relationship of FBI and restaurant cleanliness issues with customer satisfaction. For our study, we collected and analyzed a database containing 231,381 Yelp reviews of 954 restaurants in the Greater Houston area from 2005–2017. We selected Houston as the city for our research because it is one of the best U.S. food cities and has been recognized as a dynamic dining destination (Nelson, 2016). Research Background Food Safety Regulations The Food and Drug Administration (FDA, 2022) created the Food Code to serve as a model set of food safety regulations for U.S. states and municipalities to adopt for good food safety practices in food service establishments. The Food Code lists the following as the five major risk factors that cause the majority of FBI outbreaks: 1. improper holding temperatures, 2. inadequate cooking (e.g., undercooking raw shell eggs or chicken), 3. contaminated equipment, 4. food from unsafe sources, and 5. poor personal hygiene. The Food Code (U.S. Department of Health and Human Services, 2017) serves as a baseline set of regulations and individual jurisdictions can modify these regulations to fit the needs of their states. For example, in 2015 Texas legislators modified the Texas Food Establishment Rules (2021) and stipulated that all food service workers, regardless of their job descriptions, have to be food handler certified by September 2016. Companies such as ServSafe (National Restaurant Association Educational Foundation, 2019) that create food safety training programs design their curriculum to ensure all food handlers have a basic understanding of food safety as per the Food Code. Topics that are covered in these training programs include the following components: 1. basic food safety, 2. personal hygiene, Jack R. Hodges Minwoo Lee, PhD Agnes DeFranco, PhD Sujata A. Sirsat, PhD Conrad N. Hilton College of Global Hospitality Leadership, University of Houston Exploring Foodborne Illness and Restaurant Cleanliness Reporting in Customer-Generated Online Reviews Using Business Analytics Abs t r ac t Foodborne illness cases are chronically underreported, and it is crucial to investigate nontraditional strategies and approaches to identify food safety challenges that could lead to outbreaks. This point is especially important in the context of the food service industry because 61% of all foodborne illness outbreaks are attributed to restaurants. The overarching goal of our study was to data mine customer-generated restaurant reviews on an online review website and analyze the frequency at which restaurant patrons report specific terms related to foodborne illness and restaurant cleanliness. Our data analysis indicated statistically significant inverse correlations between the increased frequency of keywords in online reviews and customer satisfaction. The results from our study can be used to incentivize restaurateurs to implement enhanced food practices. Furthermore, the text mining methodology can be used in future studies to monitor food safety reporting in global markets.www.neha.org