Checking in on Pittsburgh’s Lead Problem

Raanan Gurewitsch
9 min readAug 13, 2020

The PWSA just announced their lowest lead levels in twenty years. Great! What does this mean? Is it true? What’s next? In this article I try to provide some context by answering some important questions. Let’s start from the beginning.

There’s lead in the water?

Yes. In 2015, the Pittsburgh Water and Sewer Authority (PWSA) informed its 80,000+ customers that they detected ‘Elevated Lead Levels’ in Pittsburgh’s tap water. They had just joined ranks of Flint, MI and hundreds of small and large water systems in the US that have exceeded the Environmental Protection Agency’s (EPA) Action Level of 15 parts per billion of lead in their regular compliance sampling. However, it’s important to know that the Action Level is not a health based standard, and no amount of lead in water is safe. The main source of lead water contamination is from lead service lines (LSLs), which are underground pipes that connect a house to the water system. There are other sources of lead as well, like indoor plumbing.

Sources of Lead in Water

Why are there lead pipes in the water system?

Whose brilliant idea was it to make water pipes out of a toxic chemical? Well, the Romans, actually. The word ‘plumbing’ literally comes from the Latin ‘plumbum’ for lead, or “liquid silver” (this why lead is ‘Pb’ on the periodic table). The widely-used metal was first linked to health problems in the early 20th century. However, the American lead industry did all they could to silence the scientists and promote the use of leaded gasoline, paint and plumbing all over the country. This campaign was so successful that lead pipes were actually mandated in residential building codes in cities throughout Pennsylvania in the 1930’s, until the Safe Drinking Water Act ultimately banned lead plumbing in 1986. During that time period, over 150,000 houses were built in Pittsburgh.

What’s wrong with lead?

Lead is a neurotoxin. When it enters the blood stream, your body confuses it for other good metals, like iron or calcium. When your brain takes in the lead instead of a different kind of metal that it actually needs, it can cause a number of serious health problems. Lead poisoning has been linked to memory loss, hearing loss, behavioral disorders, IQ deficiencies, learning disabilities, etc. The most vulnerable are infants, young children and pregnant women, particularly in low-income, minority communities.

Health Effects of Lead Poisoning

Is there lead contamination where I live?

This can be a difficult question to answer. But, if you live in Pittsburgh, there’s a decent chance. The PWSA has been monitoring the water lead levels in homes throughout the city in three ways:

  1. EPA Lead and Copper Rule Testing

This is a small sample of about 150 homes where the PWSA claims to know there is a LSL. They submit the results of these tests to the EPA every quarter, hoping that the highest detected lead levels are below the Action Level. This is the data, unfortunately, that determines whether a water system is in compliance with the Safe Drinking Water Act’s Lead and Copper Rule (LCR).

The problem here is that such a small sample size cannot possible be representative of the whole city. In addition, previous research has pointed to glaring issues related to tap water sampling protocols as well as sampling bias in Flint, MI. Studies like these call into question the reliability of the LCR’s monitoring provisions, suggesting more (and different) testing is necessary. Nonetheless, the PWSA continues to publicly justify claims about local water quality using charts like this, from July 2019. Check out that convenient downward trend line!

PWSA Press Release July 2019 — “Orthophosphate Continues to Reduce Lead Levels”

2. Voluntary Residential Sampling

This is a much larger data set, consisting of test results from PWSA customers throughout the city between 2016 and 2020. While the individual results are no more reliable than those in the EPA compliance data, this particular data set is clearly much more representative of the whole city. Using an interactive map, we can explore the lead levels detected at blocks (not individual houses, for privacy reasons) throughout the city. To account for some of the inherent variability in the data, I computed the median lead levels at each block in the data set observed over the entire 2016–2020 time period.

Taking a look at the results, we can see that lead levels are not consistent across the city, nor are they noticeably higher or lower in specific neighborhoods, with a few exceptions. However, it remains clear that lead has been detected in tap water samples taken all over the city. You can request a tap water test from the PWSA here.

Screenshot of my interactive map. Green and yellow bubbles represent low lead levels (below 5 parts per billion) and orange and red bubbles represent higher lead levels (above 10 parts per billion). Larger bubbles represent where more tests have taken place.

3. Orthophosphate Monitoring Program

Next there’s the orthophosphate monitoring program. When I reached out to the PWSA with my concerns about their July 2019 press release, they shared with me a detailed memo about the monitoring program. This is a comprehensive, state-approved study meant to examine whether the new treatment method is working by evaluating key water quality parameters in a controlled environment. The program is run by the consulting firm Mott MacDonald and includes weekly sampling from six water distribution monitoring stations, placed throughout the city. As far as I know, there’s no other program like this. It could actually provide some very valuable insights into how lead gets into residential tap water, including whether the orthophosphate treatment method is effectively reducing lead levels.

From the memo: Lead Levels Measured at PRS Monitoring Stations before and after the switch to orthophosphate treatment

So, lead levels are going down?

Yeah, maybe! And I hope they are. But this is an even more difficult question to answer. The charts from the orthophosphate monitoring program are very promising, but these results are not from actual homes. Maybe that is why the PWSA/Mott Mac have not yet made public any data from the monitoring program, and possibly why the PWSA continues to use convenient trends in the compliance data to justify their claims that lead levels are going down. Anyway, I tried to take a crack at answering the question myself using the data that is currently available.

Using the same voluntary tap testing data, I mapped the observed change in lead levels following the addition of orthophosphate. Because 82.5% of blocks in the sample lack sufficient data to measure any change since April 2019, it shows the changes at the neighborhood level. While the map shows lead levels have gone down in many neighborhoods, it is clear that 1) lead levels have increased in neighborhoods like Oakland, Hill District, Bloomfield. Homewood North, Troy Hill and Glen Hazel and 2) more testing is surely needed to get a better sense of the situation in actual households.

It would seem based on my analysis of the voluntary testing data that not all neighborhoods have benefited equally from the orthophosphate treatment. In these areas, specifically, it would make the most sense not only to monitor lead levels more closely, but also to prioritize lead service line replacements.

Neighborhood-level map of observed changes in water lead levels after April 2019, when the orthophosphate treatment was introduced into the Pittsburgh water system. Areas in grey lack sufficient data.

What about lead service line replacements?

The ultimate solution to the lead problem is to remove lead from the water system. All of it. But how much lead is in the system? That’s a good question (you’re learning fast!). The PWSA has been locating and replacing LSLs throughout the city for a few years now, and they claim to be making exceptional progress. To their credit, they have. But the rosy picture that they paint with metrics like total miles of LSLs replaced, doesn’t address the follow-up question, how many miles lay ahead?

Screenshot from PWSA Home Page

Studies done at the University of Pittsburgh have estimated the number of LSLs in the PWSA service area to be in the ballpark of 39,000, but note that a lack of reliable records on a significant portion of the city’s housing stock makes it very difficult to predict the actual number. But how do you replace all the lead pipes if you don’t know where all of them are?

To figure this out, the PWSA has been going house to house since 2017 conducting inspections and excavations to verify the material of public and private side service lines. The total cost of the program is expected to amount to $96 million by next year. Notably absent from this 2017 estimate, however, is the cost of LSL replacements in 2020 and 2021. The PWSA, required by law to replace 7% of remaining LSLs per year, expects to wrap up the program in 2026.

Cost Breakdown from PWSA Lead Program Summary (2017)

As you can see, these curb box inspections and excavations are very expensive. And the last time I looked at the numbers, they were only successfully identifying the public-private-side material about 29% of the time. Below is a chart from my undergraduate thesis, which looked at machine learning as a potential remedy for this issue. The data is not up to date, as the last time the PWSA shared their LSL inventory with me was the summer of 2018.

The pie chart breaks down the results of roughly 5,300 curb box inspections performed as of July 2018. For the most part, the results are some variety of ‘unknown,’ either because they can’t find the service line, can’t determine the material at a particular address. Consisting of these results combined with 44,000 digitized historical records, the PWSA’s data indicated (in 2018) lead at 57% of addresses where the material was known.

Breakdown of Curb Box Inspection Results (as of July 2018)

This is where I think Pittsburgh can learn some lessons from Flint, MI. In the beginning of the city’s response to the lead crisis, expensive excavations continued to find homes without lead pipes, wasting scarce resources needed to fix the city’s water crisis. The city began to use an algorithm developed at the University of Michigan to guide its exploration of the city’s housing stock by identifying the houses with the highest risk of having a LSL.

With the help of the AI-driven map, the success rates for Flint’s excavations shot up to ~80%, saving the city tens of millions of dollars. When the city stopped using the algorithm, costs rose by millions and success rates fell dramatically. Now, the city is required by the state of Michigan to use the algorithm’s recommendations as it replaces its remaining LSLs. The same research team that built the algorithm now run a social venture called BlueConduit, which is using the AI to help cities in the United States and Canada find and remove LSLs. Here’s a video from NOVA on their story:

So, now what?

Based on all the research I’ve done, it seems to me like the best thing for Pittsburgh to do at this point is:

  1. Do more tap water testing
  2. Make data more accessible and understandable to the general public
  3. Prioritize lead service line replacements in neighborhoods where lead levels are increasing or remain above zero.

If you’re a Pittsburgh resident, you can:

  1. See if there have been high lead levels in your area
  2. Use a water filter that is certified to remove lead
  3. Test your water
  4. Learn more about lead in water


I am currently a consultant for BlueConduit. BlueConduit is not responsible for any of the language, research or opinions stated in this article.

Thank you for reading!