February’s journal club was led by Jenny Bratburd (Graduate Student in Microbiology). Journal club review was written by Amanda Hurley (Postdoc in Plant Pathology).
Recently, CaSP discussed a local case on how policy-makers are expected to make decisions when the data is lacking or imperfect. In February’s journal club, we met to discuss a potentially more frustrating situation: what happens when policy-makers make decisions despite sound science?
In the wake of Flint, Michigan’s lead-poisoning crisis in 2014, the city was tasked with replacing the lead pipes with copper ones and there were several issues. Not all pipes contained lead and, due to shoddy record-keeping, the map of lead pipes in the city was woefully inaccurate. Randomly digging up pipes and checking for lead would have been expensive; each dig would cost between $2,500-5,000 and, with over 40,000 households, the conservative price tag was one hundred million dollars.
To prevent wasting money digging up pipes that were safe, city officials teamed up with academics to build a mathematical model that would predict high-risk homes. The model included variables such as age, value, and location of the home. If the home had a high probability of being hazardous, the model recommended pipe replacement. In addition, the model could evolve. By digging up pipes and determining the validity of the prediction, the success or failure of the prediction was fed back into the model so that it could adapt. This evolutionary process is known as “machine learning.”
The model worked splendidly, predicting the presence of lead pipes with 70% accuracy, but the support didn’t last. Early use of the model in 2017 helped the city replace lead pipes for 6,228 homes. Flint even signed a $5 million contract with a large engineering firm to bolster replacement efforts. However, as the project changed hands, the model was abandoned. While many factors surely influenced this decision, Mayor Karen Weaver played a key role. Caving to political pressure from fearful residents, Weaver “demanded” the firm inspect homes in wards that were unlikely to have lead pipes taking resources away from wards with predicted high concentrations of lead pipes. In addition, she supported the policy of inspecting every house on a given block to appease the concerned neighbors of high-priority homes. By mid-December 2018, out of 10,531 homes inspected only a mere 15% had lead pipes, leading to the criticism that the politically driven inspections were ineffective and more costly than the science-based model.
Was Mayor Weaver wrong for wanting each of her constituents to feel safe? Possibly yes, if it meant knowingly endangering the subset of constituents actually living with lead in their water. So how does an elected official embody the will of the people without totally succumbing to it?
CaSP discussed how the constituents themselves could have been the key to remedy the situation. A community supportive of the academic model wouldn’t have put pressure on city officials to make questionable decisions about ditching the model and inspecting wards equally instead of inspecting wards predicted to have high amounts of lead pipes. The model needed a messenger to explain the benefits and garner such support and save the city piles of cash. Humble and reassuring scientists would have been the ideal choice, and authors of the model did try to reach out to the engineering firm. While it makes sense to target the people actually doing the digging, the real boss of the firm was the city of Flint. And the real boss of the city, was the frightened public. In short, the scientists should have done public outreach about their model.
‘How could we have done this better?’ we asked ourselves. We considered many ways to explain the importance of the model: it’s fiscally responsible, it’s very accurate, it’ll help the ones who need it most. But in the traumatic experience of lead-poisoned water, the people of Flint were understandably mistrustful. It’s hard to predict what message will resonate. Drawing on our skills from the 2019 American Association for the Advancement of Science meeting, we realized that a one-way lecture from scientists wouldn’t have been the solution. Only listening to constituents would provide the pathway to change. Listening builds trust, making room for conversations that can actually be useful.
While the city believed they were making the right decisions, the state did not. Flint officials were unable to explain why the homes in 2018 were targeted, especially since the model in 2017 worked so well. Thus, the state has cut off funding reimbursement for contractors, placing the lead pipe replacement program on hold. The unfortunate truth is that the folks who are hurt by both the random digging and the State’s withholding of funds are those with lead pipes. Thankfully all parties came to an agreement earlier this month (Feb 2019) to reinstate the successful model and replace the pipes in high priority properties first. While this is definitely a win, I worry that the root of public mistrust has not been addressed. Unless we (scientists) can communicate the value of the tools we provide, the policy is vulnerable to the completely justified concerns of the public.