NORTHAMPTON — A particular graph caught Susan Voss’ eye when she looked at the state’s weekly COVID-19 data report for data to inform her family and her position as a School Committee member.
“All fall I really was concerned about this graph,” the Northampton resident said. “It showed the differences in the percent positivity if you remove higher education versus keeping higher education. It’s a really big difference.”
On Jan. 19, for example, the percent positivity rate — or the number of positive tests per total tests done — for the state over seven days was 5.9%, according to state data. But when higher education test results — which only 0.4% were positive — were taken out, the state’s percent positivity rate jumped to 7.3%. College and universities are doing regular testing of students and staff. Smith College and University of Massachusetts Amherst, for example, require on-campus students and staff to be tested twice a week, according to their websites.
The state started releasing testing data from higher education institutions in September, but it is aggregated across all the institutions and is not broken down by college or university. Its reports don’t include the percentage of tests that are positive in particular counties, towns or cities without higher education tests taken into account.
This fall, one area lawmaker, state Rep. Mindy Domb, D-Amherst, created her own data page to track cases of COVID-19 at colleges across the state.
Voss, a Smith College professor of engineering, was teaching remotely this fall, but she needed to be on campus for certain engineering equipment and to get strong internet, and she was tested twice a week. She wondered how those tests affect the percent positivity rate at the county and city level. “I know this has an effect, but how big?” she asked.
So, Voss teamed up with Joe Pater, a Northampton resident and professor of linguistics at the University of Massachusetts Amherst, and Michael Stein, a city resident with a Ph.D. in political science. They used the data that is available to calculate estimations about the effect of higher education testing on the percentage of tests in Hampshire County and Northampton that are positive. They detailed their findings in a paper published on Pater’s website and shared them with local officials. For towns of more than 10,000 people, the percent positivity rate is used to help determine which risk category the state Department of Public Health will label a city or town each week.
The Five Colleges publish COVID-19 testing data on their websites, so the authors used that information combined with what the state Department of Public Health publishes. Between mid-August and mid-December, they estimated 90% of tests in Hampshire County were done at one of the Five Colleges.
That percentage is the ratio of the number of tests done at the Five Colleges in the period divided by the total tests of county residents in the period. A more accurate estimation, the paper explains, would look at the number of tests done at the Five Colleges of Hampshire County residents, as some people at the colleges may live outside the county, and take into account anyone who lives in the county and is part of college testing elsewhere, but the state doesn’t report that information. Regardless of where the COVID-19 test is done, in the state data it’s assigned to wherever the person’s residence is, a spokesperson for the state Department of Public Health confirmed.
Still, the estimate of 90% “is large enough that it is almost certain that higher-ed testing plays a large role in Hampshire County COVID-19 testing data,” the paper reads.
The authors also make estimations about percent positivity in Hampshire County and Northampton using the data the state and colleges publish.
Between mid-August and mid-January, the authors calculated using the state data that the percent positivity for the county was 1.17%, while the figure for tests done at the Five Colleges during that period was 0.24%. Taking out the higher education tests for the county, they estimate that the positive test rate for that time period would be 7.48%.
That calculation assumes everyone tested at the Five Colleges lives in Hampshire County. But even if 25% of the people tested at the Five Colleges live outside of the county, the figure without higher education tests included would be 2.93% — more than doubled — the authors estimated.
Northampton numbersThe authors did a similar calculation for Northampton, subtracting out tests from Smith College, and looking at what the percent positivity would be each week. Depending on the week, the percent positivity increased by a factor of 1.17 to at most a factor of 3.84.
Voss emphasized that higher education testing is not a bad thing. “I actually think it is really good, and it’s too bad we’re not doing more of it across other domains,” she said.
The authors are pushing for more data. “We urge the local and state health authorities to provide local data that separates out the results of higher-ed testing and any other repeated testing performed on special populations (and perhaps also provide a population-weighted average),” the report reads.
City public health director Merridith O’Leary said she didn’t know what the percent positivity rate for the city was without higher education tests taken into account. “Right now there’s not that many students and faculty at Smith, so I don’t think it has that large of an impact,” she said in early January.
During the school’s fall semester, it had about 100 students on campus, according to a Smith College spokesperson.
“I think like at the state level where they do percent positivity with higher education and without higher education, they should also do on the local level,” O’Leary said of the state’s COVID data reports. “Again, it doesn’t have a big impact here in Northampton,” she said. But if 800 students come back to Smith this winter, “those numbers are huge. It definitely, definitely, has an impact in Amherst with three colleges there and the amount of testing.”
Pater said he’s hoping for more transparency and data. “I think it’s really important. I don’t think either of those goals are being met as well as they could be.” He later added, “This is something that I’m hoping our state and local officials will start to talk more about.”