Log In
NACCHO Applications
My Information
My Subscriptions
My Transactions
Model Practice Options:   Print Practice   Provide Feedback   Overall Feedback
Please press CTRL+P to print this page

2013 Model Practices (Public)

Application Name: 2013 Model Practices (Public) : Cuyahoga County Board of Health : Using Predictive Models for Rapid Assessments of Recreational Water Quality
Applicant Name: Mr. Barry J. Grisez
Application Title:
Using Predictive Models for Rapid Assessments of Recreational Water Quality
Please enter email addresses you would like your confirmation to be sent to.
bgrisez@ccbh.net; jlis@ccbh.net
Practice Title
Using Predictive Models for Rapid Assessments of Recreational Water Quality
Submitting LHD/Agency/Organization
Cuyahoga County Board of Health
Head of LHD/Agency/Organization
Terrance M. Allan, R.S., MPH - Commissioner
Street Address
5550 Venture Drive
Practice Contact Person
Barry Grisez, R.S.
Program Manager

Email Address

Submitting LHD/Agency/Organization Web Address (if applicable)

Provide a brief summary of the practice in this section. This overview will be used to introduce the model or promising practice in the Model Practices Database. Although this section is not judged, the judges use it to get an overall idea about your practice. You must include answers to the following questions in your response:

• Size of population in your health department’s jurisdiction
• Who is your target population/audience, what is the size of your intended population/audience for this practice and what percent of your target population did you reach?
•Provide the demographics of your target population (i.e. age, gender, race/ethnicity, socio-economic status) 
• Describe the nature and gravity of the public health issue addressed
• List the goal’s and objective(s) of the practice and clearly link them to the problem or issue the practice is addressing.
• Describe the potential public health impact of the practice, and the likely effectiveness of the practice being implemented as intended, and the ease of adoption of the practice by other LHDs.

In your description, also address the following
• When (month and year) the practice was implemented.
• Briefly describe how the practice was implemented, what were major activities, and any start-up and in-kind costs and funding services.
• Outcomes of practice (list process milestones and intended/actual outcomes and impacts.
• Were all of the objectives met? 
• What specific factors led to the success of this practice?

The Cuyahoga County Board of Health (CCBH) serves a population of 854,859, based on the 2010 Census. Services are provided to all communities in Cuyahoga County, excluding the cities of Cleveland and Shaker Heights, which have their own city health department. Other pertinent Census statistics are below: • The County is 64% white, 30% black/African-American, 3% Asian, and 3% other. Approximately 5% of the population is Hispanic. • Median age: 40.2 years. • Mean household income: $61,369.00, median: $43,603.00. Approximately 16% of individuals claimed income below the poverty level. • Target audience: Cuyahoga and surrounding counties, approximately 2.1 million The CCBH Bathing Beach Program was established in 1993 and is designed to protect the public from potential health risks associated with swimming in recreational waters. Water quality monitoring and reporting however, continues to be an obstacle for developing a successful public notification strategy. Water quality advisories in Ohio are based on standards for concentrations of Escherichia coli (E. coli) which can change drastically during the 18-24 hour delay that exists between sample collection and laboratory analysis. Beach-goers could potentially be exposed to contaminated water or beaches may unnecessarily get posted with a “water quality advisory” during this time, as day-old results are being reported on. In response to this issue, the United States Geological Survey (USGS) partnered with the CCBH to develop a program to estimate beach water quality. The USGS performed statistical analysis on environmental variables known to impact water quality (wave height, rainfall, etc.). The result: a model that could be used to provide near-real time water quality assessments. The output: the probability that the standard for E. coli (235 colony-forming units per 100 mL) was exceeded. Using a specific combination of variables, a threshold probability was determined to distinguish “good” from “poor” water quality. This threshold, expressed as a percentage, was derived from data collected during 2000-2005 and was the basis for issuing “water quality advisories”. In 2006, the model, known as the Nowcast System, went live and became the primary source for public notification of water quality at Huntington Beach in Bay Village, Ohio. Water quality information was now available within an hour, eliminating the 18-24 hour lag period. This rapid availability of water quality information was made available to the public via a dedicated website, www.ohionowcast.info, by 9:30 A.M. It was important to have this information available as early as possible to aid in the public’s decision making about whether or not to go to the beach based on the model prediction. The goal: to utilize near-real-time data to prevent exposures to waterborne contaminants through rapid public notification. Since 2006, monitoring at Huntington involves interns assisting with sample collection and public notification. Social media accounts were created recently via Twitter (@ccbh_net, 140 followers) and Facebook (www.facebook.com/ccbh.net, 250 contacts). Data is also posted on the Ohio Department of Health’s (ODH) BeachGuard website (www.odh.ohio.gov). For quality assurance purposes, the USGS visits the beach annually to verify that sample collection, beach conditions assessments, turbidity readings, and data entry were consistent with protocol. On average, the model has achieved an accuracy rate of 85%. In 2012, the CCBH began using the USEPA’s Virtual Beach software program, under the direction of the USGS. Similar to those run manually in earlier years, the CCBH created models for Huntington Beach based on

Overflow: Please finish the response to the question above by using this text area.  Please be mindful of the word limits.

the most significant variables. A successful beach program, particularly one involving predictive models, is dependent on two factors - appropriate beaches and funding. A useful model requires data on water quality and environmental variables that are statistically related to E. coli concentrations and that can be easily obtained. Beaches that are too “clean” or “dirty” cannot be modeled accurately because significant variables cannot be identified using statistical analysis. Financial support is a necessity. The USGS has been funded through various grants, the USGS cooperative water program, and currently the Great Lakes Restoration Initiative. The USEPA is self-funded; resources allotted to bathing beaches are accounted for federally. The CCBH covers the remainder of the cost through grants, permit fees, and general revenue funds. Program costs include personnel, travel, and equipment. Sufficient funding, strong partnerships with state and federal agencies, and a beach conducive to modeling are all critical factors this practice’s success. In an effort to increase awareness with beach-goers, the interns were also responsible for administering surveys at Huntington and other local beaches. Questions asked are designed to gauge their understanding of water contamination, water-borne illnesses, and predictive modeling. Attached to this application is a statistical analysis of survey responses.
You may provide no more than two supplement materials to support your application. These may include but are not limited to graphs, images, photos, newspaper articles, etc.
Describe the public health issue that this practice addresses. (350 word limit)
Rapid assessment of beach water quality as a public health practice is a need that has been recognized throughout the research and regulatory communities. While other methods of rapid water sample analysis are being tested, predictive models have been implemented for the past few years and are the basis for all public notification activities at several beaches in the Greater Cleveland region. The current laboratory analysis procedure, referred to as “standard culture method” or “membrane filtration”, requires an 18-24 hour delay between sample collection and reporting of results. Beach-goers could potentially be exposed to pathogenic bacteria during this lag period. Conversely, water quality advisories could be posted unnecessarily for an extended period of time affecting the overall usage of the beach. In 2012, for example, water quality information would have been properly conveyed only 69% of the time if this practice was still employed at Huntington Beach. Inaccurate information would have been passed onto the public roughly one-third of the time, translating into approximately thirty-three days of the summer. This means of risk management is dependent on laboratory analysis of numerous samples throughout the season, and determining the accuracy of this method is highly subjective as the number of samples collected varies from year to year. As a result of predictive modeling, public notification activities can be expedited and beach-goers made aware of water quality within an hour of sample collection. More importantly, beach-goers can take comfort in the fact that the information they are receiving is the most current information available. Referring back to the 2012 season, the model achieved an accuracy rate of 87.9%. The USEPA has recognized the benefit of predictive models and amended the National Beach Guidance Required Performance Criteria for Grants to include Use of Predictive Tools in Beach Monitoring Programs (Criterion 4.5).
What process was used to determine the relevancy of the public health issue to the community? (350 word limit)
Two factors drive the development of a beach monitoring program and the implementation of predictive models; the popularity of beaches and old infrastructure. According to the USEPA, an estimated one-third of all Americans visit coastal areas annually, making 910 million trips and spending approximately $44 billion. The USEPA also reported that over half the U.S. population lives in coastal watershed counties, and roughly one-half of the nation's gross domestic product is generated in those counties and adjacent ocean waters. A Michigan Sea Grant analysis of annual estimates provided by the Bureau of Labor Statistics shows approximately 1.5 million jobs, and $62 billion in wages, directly connected to the Great Lakes for 2009. Of that, 217,635 jobs fall into the tourism and recreation category. In 2012, the average number of individuals visiting Huntington Beach on any given weekday was about 1,000, with about 3,000 people per day on the weekends. These statistics provide overwhelming evidence that coastal regions are invaluable resources across the United States. To help ensure the long-term vitality of these areas, Congress passed the Beaches Environmental Assessment and Coastal Health (BEACH) Act in 2000; establishing uniform criteria for testing, monitoring, and notifying public users of possible coastal recreation water problems. The passage of the BEACH Act allowed for CCBH to enhance beach monitoring activities. Point-source (storm sewer outfalls) and non-point source (runoff) pollution could now be identified, quantified, and the data used to help guide decision-making at the local, state, and federal levels. Revenue obtained through the USEPA provided the means necessary to begin the collection of water quality data and research predictive modeling. In 2003, the USEPA and Center for Disease Control (CDC) began the National Epidemiological and Environmental Assessment of Recreational Water (NEEAR) Study at Great Lakes and ocean beaches, including Huntington Beach. Analysis of data collected through this study revealed an association between recreational water exposure and the incidence of illness. The incidence of gastrointestinal illness ranged from 12-14%, as compared to 10% among non-swimmers. An epidemiological link between poor water quality and human illness was established as a result of the NEEAR study.
How does the practice address the issue? (350 word limit)
In 2007, the Natural Resource Defense Council's (NRDC) annual survey of water quality at U.S. beaches reported that over 70% of the more than 20,000 beach advisory postings resulted from monitoring that detected pathogen-indicator levels exceeding beach water quality standards (NRDC, 2008). Furthermore, in 2007, Great Lakes beaches in Illinois, Ohio, Wisconsin, and Indiana ranked highest in percent of samples exceeding national standards. Persistent fecal pollution of recreational waters is an established public health hazard, as pollution introduced through animal or human waste can contain pathogenic bacteria, viruses, and protozoa. In particular, bacterial contamination is responsible for over 14% of the impaired waterways recognized by the USEPA (USEPA, 2004) and is the most common reason for an “impaired” classification of Ohio’s waterways. The USEPA estimates that over 5,000 impairments nationwide are attributed to microbial pathogens (approximately 24% of all impairments) reported under the national 303(d) Total Maximum Daily Load program. Approximately 40% of all beach impairments resulted from pollution associated with storm sewer overflow and overland runoff, while over 50% of the closing and advisory days were attributed to unknown bacteria sources (USEPA, 2003; NRDC, 2006). Sources of pathogens commonly implicated in beach related illnesses are associated with runoff from impervious surfaces, groundwater drainage, releases of untreated sewage, faulty sewage pipes, or from the direct release of fecal material (NRDC, 2008; Mallin, et al., 2000). The implementation of a predictive model allows for near-real time assessment of water quality. Since predictions are made on an everyday basis, updated water quality information is available daily throughout the summer. Public notification of water quality is posted, on average, by 9:30 A.M. each morning via the Nowcast website and social media. The 18-24 hour time period between sample collection and reporting is no longer an impediment to timely and accurate water quality information. Bacterial contamination at Huntington Beach is primarily attributed to non-point source pollution impacts after periods of heavy rain. By using a predictive model, the number of beach-goers exposed to contaminated water and, by default, possibly contracting an infection caused by a pathogenic microorganism can be greatly reduced.
Does this practice address any of the CDC Winnable Battles? If yes, select from the following
Does this practice address any of the CDC Winnable Battles? If yes, select from the following
Nutrition, Phyiscal Activity and Obesity
Please list any evidence based strategies used in developing this practice. (Provide links or other materials for support)
USEPA National Beach Guidance Required Performance Criteria for Grants: Criterion 4.5 - Use of Predictive Tools in Beach Monitoring Programs, http://water.epa.gov/grants_funding/beachgrants/chapter4.cfm#section5 USEPA and Center for Disease Control (CDC) National Epidemiological and Environmental Assessment of Recreational Water (NEEAR) Study at Great Lakes and ocean beaches, http://www.epa.gov/nheerl/neear/ 2007 Natural Resource Defense Council (NRDC) annual survey of water quality at U.S. beaches, http://www.nrdc.org/water/oceans/ttw/ The USGS posted several published documents on the Nowcast System/predictive modeling: http://www.ohionowcast.info/nowcast_technical.asp
Is the practice new to the field of public health? If so, answer the following questions.

What process was used to determine that the practice is new to the field of public health? Please provide any supporting evidence you may have, e.g. literature review.

The idea of beach water quality monitoring is still a relatively new concept when compared to more traditional environmental health programs. The fact that it wasn’t until 2000 that the USEPA provided uniformity to beach monitoring programs, with financial assistance, is a strong indication that programs created prior to that date were very localized and subject to fiscal constraint. The CCBH program was enacted in 1993, and sample analysis was conducted using the only available method; membrane filtration/standard culture method. In 1998, the USGS began collecting water quality data and exploring the use of multiple linear regression statistical analysis. For the Huntington Beach model, water quality data was collected for 5 years prior to the implementation of the Nowcast System. The model was validated during the 2005 season and went live in May of 2006. Up to this point, there were only rudimentary rainfall-based alerts where water quality advisories were posted when the amount of rain exceeded a pre-determined threshold. According to the USGS, other models are in various stages of development; however there are only three used for decision-making. All three are in the Great Lakes region, one being the Nowcast System developed for Huntington Beach.
How does this practice differ from other approaches used to address the public health issue?
Over the past several years, attempts have been made to improve sample analysis techniques to keep pace with the demand for faster turn-around times. Known as rapid assessment methods, the goal was to provide an E.coli concentration within 4-6 hours of sample collection. The two most common techniques are quantitative polymerase chain reaction (qPCR) and immunomagnetic separation/adenosine triphosphate (IMS/ATP). As opposed to predictive modeling that uses a series of variables to predict water quality, qPCR and IMS/ATP provide same-day laboratory analysis of E. coli. However, there are several drawbacks to both methods. The start-up costs for qPCR are prohibitive for many local health departments; the equipment is expensive, staff must be hired and trained, and there must be sufficient field personnel to collect samples and deliver to the lab by mid-morning. The equipment for IMS/ATP is not as costly, but this method requires more hands-on activity. Since the equipment is not portable, all samples need to be taken to a central location for analysis. The Northeast Ohio Regional Sewer District (NEORSD) completed a project in 2011 comparing rapid assessment methods with membrane filtration and predictive models. From their research, it was concluded that the qPCR method is very reliable from an accuracy standpoint. No correlation could be made for IMS/ATP. Both of these methods are still in the research phase. At this point in time, predictive models are an economical way to obtain real-time, accurate water quality information using a field-validated methodology.
Is the practice a creative use of an existing tool or practice? If so, answer the following questions.
What process was used to determine that the practice is a creative use of an existing tool or practice?  Please provide any supporting evidence you may have, for example, literature review.

What tool or practice (e.g., APC development tool, The Guide to Community Preventive Services, HP 2020, MAPP, PACE EH, etc.); did you use in a creative way to create your practice?  (if applicable) (300 word limit total)
a. Is it in NACCHO’s Toolbox; (if not, have you uploaded it in the Toolbox)?
b. If you used a tool or practice to implement your practice, how was your approach to implementing the tool unique and innovative for your target area/population?

How does this practice differ from other approaches used to address the public health issue? 

If this practice is similar to an existing model practice in NACCHO’s Model Practices Database (www.naccho.org/topics/modelpractices/database), how does your practice differ? (if, applicable)
Who were the primary stakeholders in the practice?
There are several entities that are critical to the success of the CCBH beach monitoring program, particularly where predictive models are implemented. The USGS, USEPA, Ohio Department of Health (ODH), Northeast Ohio Regional Sewer District, and Cleveland Metroparks all maintain a vested interest in the CCBH beach monitoring program. However, the beach-going public in the Greater Cleveland region is the most important stakeholder in the practice. Without their interest and enthusiasm for accessible bathing beaches, resource allocation to monitoring programs and predictive model development could not be justified.
What is the LHD's role in this practice?
Until the 2012 beach season, the primary role for CCBH was to implement the predictive models created by the USGS; specifically, collect beach water samples and notify the public. A DOS-based software program would be uploaded to the laptop computer(s) used for beach monitoring. As samples were collected, the turbidity levels would be measured along with the other parameters and the water quality prediction would be made. The CCBH staff would then notify the public by updating the Nowcast, ODH BeachGuard, and CCBH websites as well as the social media outlets. Signage at the beach would then be changed (if necessary) to indicate the water quality was predicted to be “poor” or “good”. The Nowcast website would be updated later in the day as the previous days’ sample results were received from the lab. The CCBH was also the point of contact for media inquiries on beach water quality.
What is the role of stakeholders/partners in the planning and implementation of the practice?
All of the stakeholders mentioned in the previous section play a role in planning and/or implementation of predictive models. The USGS has been the primary source of data analysis and model development since the inception of this practice. The USGS staff would review the sample results from the previous beach season, identify those variables significantly impacting water quality, and produce a model that would be maintained by CCBH. The USGS would also visit the beach annually to conduct the quality control assessments. The ODH (through the USEPA BEACH Act Funding) provides a significant amount of financial assistance which allows this practice to continue and refine model development. At the end of each season, the CCBH furnishes a final report to the ODH summarizing all sampling activities. This report is then forwarded to the USEPA. All beach sampling information is available to the general public, media outlets, and any other interested parties. The NEORSD provides all laboratory analysis for the CCBH. Furthermore, the NEORSD researches new techniques in laboratory analysis of water quality (i.e.: rapid assessment methods). Huntington Beach is located on a parcel of land maintained by the Cleveland Metroparks. A strong partnership has been fostered over the past several years with the Metroparks and assistance is provided whenever requested. The professional relationship established between the Metroparks and the CCBH has resulted in concerted efforts beyond the beach monitoring program. In addition to notifying the general public of the daily water quality conditions, the CCBH conducts beach user surveys in an attempt to raise their awareness of this issue. General demographic information is collected, and participants are asked various questions pertaining to water quality, advisories, and symptoms associated with bacterial or viral infections. This information is summarized as part of the final report drafted for the ODH. Also, because of the location of Huntington Beach, the Cleveland Metroparks also receives the survey results. The unique function of each stakeholder is a necessary component to the success of this practice.

What does the LHD do to foster collaboration with community shareholders?
Describe the relationship(s) and how it furthers the practice's goals.

Maintaining strong working relationships with community stakeholders is primarily driven by programmatic responsibilities. The CCBH and USGS are in continuous contact throughout the year as predictive models are created, implemented, and then evaluated for accuracy. Reporting requirements serve as a means of communication between the Ohio Department of Health and the USEPA. Sample results and beach conditions are recorded on a daily basis via the ODH BeachGuard website. The CCBH attends an annual pre-season meeting with other stakeholders hosted by the ODH. There, emerging issues impacting Lake Erie are discussed as well as general information and the expanding use of modeling. Beach managers, including Metroparks representatives, are furnished with an annual report which is tailored to their specific beach. The report includes data on sample results, accuracy of that year’s predictive model, “new” concerns adversely affecting Lake Erie (i.e.: Harmful Algal Blooms), and information other issues related to beaches and water quality. Any concerns expressed by beach managers or beach-goers are addressed on an individual basis, separate from the normal daily activities performed as part of the bathing beach program. To help further public awareness of water quality and illness prevention, the CCBH interns survey beach-goers throughout the summer at Huntington and other public beaches in the area. Survey questions are designed to gauge the general public’s knowledge on water contamination, illness symptoms, predictive modeling, and notification procedures carried out by the CCBH. Constant contact with stakeholders has been a cornerstone to the success of predictive modeling and the overall CCBH beach monitoring program.
Describe lessons learned and barriers to developing collaborations.
The implementation of the Nowcast System has fortunately gone smoothly due to partnerships already being in place with agencies such as the USGS and the ODH. Prior to launching the Nowcast System in 2006, the CCBH and the USGS met with the ODH to discuss communication of data to the public. For example, if the actual water sample result exceeded the standard, which would typically result in a water quality advisory being issued, we needed to determine how to handle communicating that the model prediction, made the previous day, may have conflicted with the actual result. It was agreed that the most effective means of communication was to base each day's water quality upon the specific prediction. If the public was interested in the actual lab results, they could view it on the Nowcast website. Although this was not a barrier, it was necessary to have all pertinent agencies on the same page with the communication of data. No barriers have been encountered with collaborations to date. Effective communication with partners and the public is essential to the success of this practice.

Evaluation assesses the value of the practice and the potential worth it has to other LHDs and the populations they serve. It is also an effective means to assess the credibility of the practice. Evaluation helps public health practice maintain standards and improves practice.

Two types of evaluation are process and outcome. Process evaluation assesses the effectiveness of the steps taken to achieve the desired practice outcomes. Outcome evaluation summarizes the results of the practice efforts. Results may be long-term, such as an improvement in health status, or short-term, such as an improvement in knowledge/awareness, a policy change, an increase in numbers reached, etc. Results may be quantitative (empirical data such as percentages or numerical counts) and/or qualitative (e.g., focus group results, in-depth interviews, or anecdotal evidence).

List up to three primary objectives for the practice. For each objective, provide the following information:  (750 word limit per objective)

• Performance measures used to evaluate the practice: List the performance measures used in your evaluation. Depending on the type of evaluation conducted, these might be measures of processes (e.g., number of meetings held, number of partners contacted), program outputs (e.g., number of clients served, number of informational flyers distributed), or program outcomes (e.g., policy change, change in knowledge or attitude, change in a health indicator)
• Data: List secondary and primary data sources used for the evaluation.  Describe what primary data, if any were collected for each performance measure, who collected them, and how.
• Evaluation results: Summarize what the LHD learned from the process and/or outcome evaluation. To what extent did the LHD successfully implement the activities that supported that objective? To what extent was the objective achieved?
• Feedback:  List who received the evaluation results, what lessons were learned, and what modifications, if any, were made to the practice as a result of the data findings.


• Performance measures used to evaluate the practice: List the performance measures used in your evaluation. Depending on the type of evaluation conducted, these might be measures of processes (e.g., number of meetings held, number of partners contacted), program outputs (e.g., number of clients served, number of informational flyers distributed), or program outcomes (e.g., policy change, change in knowledge or attitude, change in a health indicator)
• Data: List secondary and primary data sources used for the evaluation.  Describe what primary data, if any were collected for each performance measure, who collected them, and how.
• Evaluation results: Summarize what the LHD learned from the process and/or outcome evaluation. To what extent did the LHD successfully implement the activities that supported that objective? To what extent was the objective achieved?
• Feedback:  List who received the evaluation results, what lessons were learned, and what modifications, if any, were made to the practice as a result of the data findings.

Objective 1:

Objective: Implementation of predictive models as the primary means of water quality notification at Huntington Beach by May of 2006 Performance Measures: There were three phases to model development; data collection, statistical analysis, and validation. From 2000-2005, the CCBH staff visited Huntington Beach daily Monday –Thursday for sample collection and documentation of conditions. 532 samples were analyzed along with information on beach conditions during this time period. After statistical analysis of the data, the USGS determined that modeling was feasible and proceeded to identify variables significantly impacting water quality. In May 2005, testing began for the model, comparing predictions to the results obtained through laboratory analysis (using standard culture method/membrane filtration). 58 laboratory-analyzed sample results were available to compare with the Nowcast predictions. The modeling technique was deemed effective and implemented for the entire 2006 beach season as the sole source of public notification. Data: From 2000-2011, the CCBH and USGS created a recordkeeping form specific to Huntington Beach. Personnel collecting the sample(s) are responsible for documenting conditions including wave height, rainfall, turbidity, wind speed, wind direction, water and air temperatures, wildlife (geese, gulls, dogs, etc.), the number of people at the beach, refuse in the area, and other conditions impacting water quality (i.e.: algae). Currently, the CCBH uses the USEPA “Routine On Site Sanitary Survey” form at the time of sample collection. This form contains water quality variables used in model development, and has been modified slightly to include those not listed (such as lake level). For the majority of the summer, a college intern handles sample collection, completion of paperwork, and data entry. The intern also manages public notification activities; updating the Nowcast and ODH BeachGuard websites, and sending an e-mail message to the CCBH Communications Officer for Facebook and Twitter postings. Full-time staff fills in as necessary. The program manager oversees the interns and ensures that the sample results are reported once laboratory analysis has been completed. This individual was also responsible for communication between the USGS and CCBH as well as other stakeholders. The collection of primary data (described above) is used cumulatively to develop predictive models from one year to the next. Also, secondary data in the form of census and Bureau of Labor Statistics information is used to justify

Overflow (Objective 1): Please finish the response to the question above by using this text area.  Please be mindful of the word limits.

continued investment in the practice. Evaluations: As a result of predictive modeling, water quality information is available almost immediately; the 18-24 hour timeframe is now reduced to about half an hour. Signage at the beach, websites, and recently-added social media outlets are used to update the beach-going public daily. People can now be informed of contamination prior to entering the water and potentially being exposed to pathogenic microorganisms. Conversely, beaches where the water quality is determined to be “good” will not need to be posted with advisory signs for an extended period of time. Referring back to the 2012 season, the model was 87.9% accurate. This was approximately 20% better than the traditional method of public notification where there would have been 9 days of swimmers exposed to contaminated water and 9 days of advisories posted unnecessarily. By using the predictive model for notification, there was only 4 days where swimmers were exposed to contaminated water and 3 days of unnecessary postings. Keep in mind that statistical analysis of water quality variables will never result in a model that is 100% accurate. Feedback: Predictive model development is a continual work in progress, refined by the statistical analysis of cumulative data obtained throughout each summer. The USGS plays a critical role in this evaluation process, and routine data sharing allows for constant model validation. The ODH and USEPA are made aware of this process through written reports submitted at the end of the season; one section is dedicated only to modeling-related activities. As of May 2006, the beach-going public is notified of water quality information on an everyday basis through beach signage and website postings, and now social media outlets. The beach user surveys administered by the interns are analyzed statistically to uncover trends (or shortcomings) in public knowledge of water quality. This information can be used to gauge water quality awareness and illness prevention within the region as well as potentially identifying previously unused means of communicating information to the beach-going public. Cleveland Metroparks staff involved with oversight of Huntington Beach receives an annual summary of activities including the effectiveness of the predictive model(s). The feedback provided by these various partners is essential to maintaining an effective beach monitoring program.

Objective 2:

Overflow (Objective 2): Please finish the response to the question above by using this text area.  Please be mindful of the word limits.

Objective 3:

Overflow (Objective 3): Please finish the response to the question above by using this text area.  Please be mindful of the word limits.

What are the specific tasks taken that achieve each goal and objective of the practice?
The overarching goal, rapid assessments of recreational water quality, was achieved through the completion of a series of tasks as determined by the USGS and the CCBH. Data collection was the first phase of the process; from 2000-2005, full-time staff and summer interns collected water samples and conducted beach assessments. Sample analysis was provided by the Cuyahoga County Sanitary Engineers laboratory. The completion of this first objective provided a substantial amount of data that was needed to move on to phase two. The second task was to analyze the data accumulated during the first five years. The USGS performed the statistical analysis using the multiple linear regression technique. Their findings indicated that a combination of turbidity, wave height, rainfall (from Hopkins International Airport and the USGS-developed Radar Rain), and day of the year were considered predictors for bacterial concentrations. A model was developed, and a threshold of 30% was established to determine “good” versus “poor” water quality. As opposed to predicting the specific E. coli concentration, the threshold expressed as a percentage is an indication of water quality, where exceeding that level results in a significant risk of exposure to contaminated water. Thresholds are designed to reduce the number of false positive and false negative predictions to an acceptable level. Although undesirable, these scenarios are also unavoidable. Once created, the model was tested during the summer of 2005 for validation. The USGS and CCBH were satisfied with the results obtained during 2005, and progress was made toward completing the objective of model implementation. For the entire summer of 2006, the model was the basis for public notification of water quality information. Either full-time CCBH staff or a summer intern collected the water samples, obtained turbidity levels, calculated the probability (percentage) of water quality exceeding the threshold, and ensured that the signage at the beach reflected the model. The Nowcast website was also updated as soon as predictions were made. This process continued for the entire summer, and subsequent years until 2012 when the CCBH was responsible for both the development and implementation of predictive models for Huntington Beach.
What was the timeframe for carrying out these tasks?
In general, the only significant dates when working with predictive models are the beginning of the season and the last day of the season. If a model is to be used as the basis for public notification, it must be ready to go by the first day of the sampling season; normally on or around Memorial Day. Note that, as stated in previous sections, water quality predictions should be made and posted by 9:00-9:30 A.M. each morning to ensure that beach-goers benefit from this practice. Water samples and data collection will continue on a regular schedule throughout the summer until the last day of the season; Labor Day. Data analysis and model enhancement can take place anytime between the end of one season and the beginning of the next.
Please provide a succinct outline of some basic steps taken in implementing your practice.
From 2000-2005, the CCBH collected beach water samples and documented weather conditions, physical characteristics, and any potential activities that may impact water quality. This information was provided to the USGS, where statistical analysis of these variables was conducted to determine significance. In 2005, the USGS compiled enough data to create a predictive model for Huntington Beach and test it for accuracy. At this point, public notification was still based on laboratory analysis of sample results since the model had not yet been validated. The 2005 beach season confirmed that the methodology used to create the model was accurate. Beginning in 2006, all public notification was based on model predictions only and all notification-related activity was completed prior to leaving the beach. At the conclusion of each season, the USGS would re-evaluate water quality parameters in order to fine tune the model for the upcoming season. Often, the model would be slightly modified mid-season to account for changes observed in previous years; thus “subseason 1” and “subseason 2” were used to distinguish two models used during the same season. In 2012, the USGS relegated the responsibilities to local health departments and provided technical support for data analysis and model development. Using Virtual Beach software, the CCBH Researcher used water quality data from 2005-2011 to create a predictive model that would be implemented for the 2012 season. After regular discussion with the USGS, the model was refined and implemented on May 21, 2012. Water quality data continued to be collected throughout the 2012 season using the routine sanitary survey forms explained in previous sections. The data obtained during the 2012 beach season will be statistically analyzed to further enhance the Huntington Beach model for 2013.

What were some lessons learned as a part of your program's implementation process?

The implementation of the models required training of CCBH staff by the USGS on the intricacies of the software program. A greater understanding of model development was gained prior to the 2012 season when the CCBH was responsible for not only implementing the models, but also developing them. The biggest lesson learned was the knowledge and benefits gained through its success. The model has proved that its use can reduce water quality advisories, and reduce unnecessary public health exposures to waterborne pathogens. It was also learned that its proven accuracy can lead to reduced program expenses. As society is becoming more reliant on the use of technology, it is important that programs evolve and use technology as a means of better reaching the target audience. The implementation of the predictive model has been a perfect example of how technology can be used for communication to the public. Reviews of Nowcast website statistics allowed us to determine what times of day the website was being visited the most, as well as what days of the week it was being visited the most. Statistics also showed peaks in usage, such as during heat waves. Over the years, the CCBH has also experienced a decline in phone calls to the office about Huntington Beach water quality, which may be attributed to the public relying on the website as their information source. The success of the Nowcast System has brought national attention to our beach program. CCBH was the first local health department to implement a predictive beach model in Ohio, which was also highly recognized by the USEPA. The CCBH was asked to participate in the development of a DVD, sponsored by the USEPA, about predictive modeling. The DVD, Beach Models: Predicting Water Quality, was made available to other health departments and numerous stakeholders at USEPA sponsored national conferences. The uniqueness of the Nowcast System, especially upon its initial implementation, also led to an article in the Wall Street Journal, an interview by NBC News, in addition to local media attention.
Provide a breakdown of the overall cost of implementation, including start-up and in-kind costs and funding services.
The services provided by the USGS and the software program developed by USEPA are budgeted for by their respective agencies. The costs for these goods and services are not required to be covered by the CCBH. The following is an approximated breakdown of costs associated with model development and implementation at Huntington Beach in 2012, including all field activities and laboratory analysis of water samples: CCBH Supervisor (wages/benefits and mileage for: model development, field activities, program oversight) - $3700.00 CCBH Beach Program Manager (wages/benefits and mileage for: model development, field activity, program oversight) - $6400.00 CCBH Researcher (wages/benefits and mileage for: model development) - $7200.00 Intern (wages and mileage for field activities and data entry) - $3000.00 Administrative Support - $70.00 Laboratory Fees ($20 per sample, 58 samples for 2012, 2 sampling locations @ beach) - $2320.00 Revenue from Metroparks Beach Permit Fee - $1150.00 Revenue from ODH Beach Monitoring Grant (USEPA Beach Act Funds)- $17,355.00 Total Cost for CCBH to Maintain Predictive Modeling Activities - $3885.00 Keep in mind that these figures are an approximation for the resources allocated to Huntington Beach for the 2012 season. A laptop computer and turbidimeter did not need to be purchased. Funding from the ODH varies from one year to the next, and the number of laboratory-analyzed samples fluctuates per season depending on the level of financial assistance from ODH. This budget reflects costs associated with a well-established modeling program. Newer programs or those in the data collection phase may have increased expenditures for field activities, laboratory sample analysis, and model development. Salary/benefits and mileage will vary per local health department.
Is there sufficient stakeholder commitment to sustain the practice?  Describe how this commitment is ensured.
Based on the economic importance of bathing beaches in the region, there is strong demand to continue to provide the beach-going public with near-real time water quality information. The current funding levels allow for an almost self-sustaining program, however financial support varies from year-to-year. The CCBH is committed to maintaining this practice at Huntington Beach, and possibly expanding to other public beaches in the region within the next few years. Annually, resources are allocated to the bathing beaches program from the organization’s general revenue fund. If financial assistance at the federal/state level declines or is eliminated, the CCBH will explore funding opportunities with the Cleveland Metroparks and local foundations by way of grant applications. A substantial amount of federal, state, and local resources have been invested in this practice. As a result, predictive modeling has become a cost-effective means of accurately evaluating water quality. Regression should not be an option, and the CCBH is obligated to maintain a standard of care to which the public has become accustomed.
Describe plans to sustain the practice over time and leverage resources.
One important benefit of predictive modeling is this; the more refined the model, the less laboratory analysis is required for validation. Data has been collected from Huntington Beach since 2000. In 2012, sampling was conducted only four days per week, however predictions were made every day throughout the summer. As budget concerns surface, predictive modeling can play a crucial role in bridging the gap between addressing financial shortcomings and maintaining public safety. Substantial cost savings can be realized in personnel, travel, and laboratory expenditures when models can be used in place of the traditional method of water quality analysis/public notification. An indirect benefit of models, as opposed to the rapid assessment methods IMS-ATP and qPCR, is that there are no substantial up-front costs for equipment and personnel training. A laptop computer, wireless internet access, and a turbidimeter are the only tools needed to upgrade to a model-based beach monitoring program. The USGS and the CCBH are in the process of attaining the next benchmark in predictive modeling; improving Nowcast predictions using locally-collected data. These include measuring photosynthetically-active radiation (PAR) using instrumentation installed at Huntington and measuring turbidity and wave height at a buoy installed near the swimming area at another beach in Cuyahoga County – Edgewater Beach. All other variables (rainfall, lake level, wind speed, etc.) can be obtained via the Internet. If successful, water quality predictions can be made from any location at any time. The initial investment in the buoy will be offset by the savings in personnel, travel, and lab fees. This form of modeling is still being researched, and will only be pursued if acceptable accuracy rates can be obtained.
Practice Category Choice 1:
Water Quality
Practice Category Choice 2:
Practice Category Choice 3:

Please Describe:

Check all that apply.
Colleague in my health department

Other (please specify):

Are you a previous applicant?