Is the Juice Worth the Squeeze? Collecting Advisories for the National Drinking Water Tool

The Bottom Line

Our National Drinking Water Explorer tool will make drinking water information available to a wide range of users. As part of this work we  collected and standardized state reported boil water, do not drink, and do not use advisories, and are reaching out to states for permission to use this data in our tool. Often written as BWA or BWNs, these notices reflect communications to drinking water users to not drink or to boil their water before use. These advisories can be scheduled or unscheduled, and can affect fractions of a system, the entire system, or just select groups (sensitive populations, infants, etc.). These advisories typically last for a few hours or days, but our data shows some systems have persistent notices lasting years. There is also no requirement or standard for reporting advisories beyond affected customers – states may report a wealth of information about the advisory and the date it was lifted, and some may only provide a link to the news report. We did the “squeezing” of collecting and standardizing these advisories to bring customer-oriented experiences into the data landscape and alongside reported compliance data in our tool. 

A Case Study of the Problem

Drinking water advisories are the primary way customers learn about acute issues with their drinking water supply. While these advisories are communicated to customers within 24 hours under the Public Notification Rule, the state is not required to collect and share this information with the general public. For example, during my freshman year of college, the community water system serving my university had a major water main break that resulted in the loss of nearly 1.5 million gallons of water. This is one of the most common causes of advisories, but there can be different practices between utilities for reporting advisories and even between events for the same utility. In this instance, I was notified of a  “do not drink” order via text from my university while living on campus. However, a much larger break occurred a year later while I was off campus, and I didn’t find out about it until recently. Currently, the only way to find out about these advisories is from reading the news or incident reports.

Like many other states, North Carolina doesn’t maintain a public-facing list of drinking water advisories. As a result, these customer experiences may be evident to the utility serving them but are practically invisible to the broader drinking water sector. Drinking water regulators, researchers, and interested parties typically tap into data from the Safe Drinking Water Information System (SDWIS) to understand a system’ s public health protection performance. SDWIS has not reported any significant violations of the Safe Drinking Water Act for my college drinking water system over the past 10 years. While according to contaminant testing data in SDWIS my college water system is high performing, that doesn’t tell the full story. Without standardized advisory data, we are lacking a critical piece of information to  help us understand utilities that may be struggling to provide consistent service to their customers. The good news is 13 states report their advisories in a publicly accessible format, which is the first step in making this information more accessible to others in the drinking water sector. 

Our Gold Stars

There were three states with excellent practices for reporting their advisories:

  • Oregon: maintains a long historic record of advisories in a simple table that was  very straightforward to collect using web scraping techniques. 

  • West Virginia: maintains a record of advisories over the past year, has a data download option, and uses an application programming interface (API, allows software to communicate with each other) which makes it easy to access this data using programming languages.

  • Massachusetts: logs historic data in a simple table, which makes it easy to collect using web drivers (a method to navigate and scrape data from webpages using code). Our only suggestion would be to report the water system ID in the table, rather than hiding it in a pop-up. 

Here are common traits shared among states whose data were easy to access:

  • Options to download data! This could either be a simple data download option, or an API to give users easier access to advisories. 

  • Simple tables are better! Although they might not look as fancy, simple tables were easy to read using code. There was one state whose website was so interactive, we could not build a method that would reliably collect their advisories.

  • Providing Water system IDs as they appear in SDWIS makes it easy to relate water system information to other datasets. While providing the water system name is a great first step, it is difficult to match names to water system IDs.

  • Provide a historic record of advisories. For states that only reported active advisories, we had to add additional infrastructure to build a historic record over time. Providing a full record would make data more useful for researchers. 

  • Consistent data reporting removed the headache of re-standardizing dates of various formats, or reconstructing water system IDs. 

  • Policies to facilitate data sharing like reporting requirements and interoperable data systems would ease the burden on utilities to share data with the state.

What We Learned Along the Way

States are not required to collate or report boil water advisories on public websites, and previous research suggests some do not have plans to track this information at all. As a result, there is not a national data standard for reporting, and we built custom code to collect data for each state. Of the 13 states that provided accessible information on drinking water advisories, 8 noted when the advisory was lifted, and the rest only reported active advisories. While most states reported the water system name associated with the advisory, only 9 consistently provided the water system ID from SDWIS – a primary field for relating water systems to other drinking water databases. We are not suggesting another reporting requirement, but having a set of “best practices” for sharing advisories outside of affected areas (similar to the EPA’s Public Notification templates) would make this information more accessible for researchers. 

State Reported Boil Water Advisories - Feature Coverage

In addition to feature coverage,  we found states with hundreds of advisories and some with only ~50 over a similar time period. Does this actually reflect fewer advisories, or do the states simply have different reporting requirements? It’s difficult to tell with the data we have available to us, but these caveats and considerations will be clear in our National Drinking Water Explorer tool to avoid apples-to-oranges comparisons. 

While Consumer Confidence reports provide customers an annual overview of their water system’s public health performance, drinking water advisories are the primary way customers receive communication about acute or immediate risks to water safety. We’ve collected a first pass of drinking water advisories and are in the process of reaching out to states for permission to use this data in our tool and alongside appropriate citations, caveats, and disclaimers. Our National Drinking Water Explorer tool will show what's possible and encourage better data sharing practices to bring these lived experiences into the drinking water conversation.