Understanding local COVID-19 data

This webpage was made for a February 22 presentation to the Round Hill Neighborhood Circle of Northampton Neighbors. A recording of the presentation is at the bottom of this page.

How widespread is COVID-19 in my community?

It’s not as easy to get an answer to that question as we might like, and the answers often don’t include much guidance on what the data mean for us as community members, or important caveats about their limitations.

I’ll talk about where to find and how to interpret the data that are currently available, and also about how our understanding of the local situation could be increased by access to better publicly available data.

County and city new case data

The MassDPH releases a count of new cases for the county every day, and for the cities and towns every week. “Cases” are individuals with a positive PCR test. The actual number of people infected is probably about 4 times the new case number. The number of currently active cases is estimated by the MassDPH as the number of new cases for the last 21 days (UMass Amherst uses the last 10 calendar days).

The daily county count is updated at 5 pm each day on the dashboard (select COVID-19 cases, scroll down and select Hampshire). The MassDPH also releases downloadable current and historical raw data. The best presentations of the data over time and across counties and states that I have seen are on the New York Times website:

Track Coronavirus Cases in Places Important to You

Hampshire County, Massachusetts Covid Case and Risk Tracker

These include interactive graphs that show daily counts, as well rolling daily averages calculated over a 7-day or 14-day window.

The risk tracker also provides guidance for individuals relative to the current risk level “developed with public health experts at Johns Hopkins Bloomberg School of Public Health and Resolve to Save Lives, an initiative of Vital Strategies.” The risk level is calculated using a per capita rate:

A county is at an  extremely high risk level if it reported more than 640 cases per 100,000 people during the past two weeks. This is equivalent to a daily rate of 46 cases per 100,000 people. (Very high > 160 per 2 weeks or 11 per day, High > 40/3, Medium > 10/1, Low < 10/1).

Here is some of the guidance for Very High Risk, the category that we are currently in, even if you take UMass numbers out, or look only at Northampton, as we will see in a minute. (See this Atlantic article on discrepancies between state restrictions and advice for individuals).


The CDC also now provides county data, and transmission rate levels based on the last 7 days (so multiply by 2 to compare with the NY Times/Johns Hopkins levels).

From new CDC guidance released Feb. 12 2021:

The town and city data are released every Thursday at 5 pm. in the Weekly Public Health report (also available in the Dashboard):

MassDPH Weekly Public Health Report Feb. 11.

The total case count is current to the end of the day Tuesday before the report is released, while the “last 14 days” ends on the Saturday before. So you can get a more recent week’s worth of data by subtracting the previous week’s total case count from the current one, which gives us 29 cases for the week ending Feb. 16. This site provides the result of that calculation over time.

Northampton data from the Feb. 18 public health report.

The Northampton per 100K rate is about 3.4 times the week’s total (29*3.4=98.6), and the daily average per 100K is about half (29*0.49=14.2). This is at the top of the CDC’s orange “Substantial Transmission” category, and in the “Very High Risk”category of the NYTimes/Hopkins metric.

We can also get some more information about Northampton’s cases from the city’s webpage (though it’s hard to get the number of new cases from it):

Northampton cumulative case numbers from “Residential clusters” refers to cases in long-term health care facilities.

Higher ed testing and new case data

Testing at higher ed institutions is included in the local counts; cases are assigned to places of residence. This can skew both the new case data, and the percent positivity. A dramatic example of a new case skew comes from the effect of the recent outbreak at UMass Amherst on the Hampshire County numbers. We don’t know exactly how many of the UMass new cases are Hampshire residents, and the MassDPH dates and those from the UMass dashboard don’t align perfectly, but this graph provides an estimate of the effect, and what the rest of the county looks like (see further this page).

Data sources: Data downloaded from the MassDPH website and the UMass Amherst COVID-19 dashboard Feb. 21, 2021

In the week ending Feb. 20, Hampshire County had 420 new cases, and UMass Amherst had 221 in the week ending Feb. 18 (there is about a two-day discrepancy between the dates assigned to cases). Hampshire County’s per capita rate is 261.1 cases per week per 100K (37.3 per day). If we subtract the UMass numbers, we get 199 cases, and 136.5 per week per 100K (19.5 per day). On the CDC metric, both 261.1 and 136.5 are in in the red “high transmission” category.

The two week totals are 892 and 583, which gives Hampshire a two week rate of 554.6 per 100K (39.6 per day), and Hampshire minus UMass 211.9 (15.1). On the NYTimes/Hopkins metric, these are both in the red “Very High Risk” category.

Higher ed testing and test positivity

The MassDPH also releases a two-week total for municipalities and the county of the number of tests, the number of positive results for those tests, and the resulting percent positivity. This figure plays a problematically central role in the state’s color coding classification of towns and cities.

From Pater, Stein and Voss (2021) How might higher-ed COVID-19 asymptomatic testing influence testing rates and percent positivity in Hampshire County?
From the Feb. 18 MassDPH Public Health Report

The Public Health Report gives no guidance on how these categories should be interpreted, but red is usually called “high risk”, and yellow “moderate” (I have also seen “caution” for yellow, which raises the question of what green is supposed to mean). The CDC’s yellow “moderate” requires < 7 new cases per day, and NYTimes/Hopkins yellow “medium” requires < 3. There is no upper bound on the new case rate for a community classified as yellow in Massachusetts.

For a community the size of Northampton, the positivity rate must be 5% or greater for it to be classified as red. This is problematic because percent positivity is not a good measure of the incidence of the disease. Positivity is useful as a way of measuring whether enough testing is being done; a high positivity rate can mean that only (highly) symptomatic people are being tested. Including it as a supplementary measure, as in the CDC and NYTimes/Hopkins metrics, makes good sense, since it counteracts a low new case rate arising from too little testing. But it’s hard to see why one would require high test positivity for the “high risk” classification. It would seem to mean: “As long as there is enough testing, there is only a moderate risk of transmission, no matter how high the new case rate is.”

Test positivity is a particularly problematic measure in Hampshire County, where we have so much repeated asymptomatic surveillance testing at the Five Colleges. By the count in our paper, the number of tests at the Five Colleges was 90% of the number in Hampshire County. Here is an update to one of the tables in our paper, which looks at the effect of subtracting Smith College numbers from Northampton’s.

Data sources: MassDPH Public Health reports, and data supplied by the Smith College COVID-19 Testing & Contact Tracing Team. See further Pater, Stein and Voss (2021).

Because Northampton’s positivity rate as reported by the state has never exceeded 5%, Northampton has never been classified as “red”, although the new case numbers have been high enough for a red classification by the CDC or NYTimes/Hopkins metrics since November. Besides sending a seemingly false signal that high levels are “moderate”, this classification makes no distinction between the very different levels just before and just after the Thanksgiving-to-New Year’s holidays.

MassDPH Public Health Report data and classifications for Northampton, graph from The dashed line is percent positivity.

As a Gazette article points out, Amherst with its 637 cases in two weeks (113 per day per 100K!) was also classified yellow in the Feb. 18th report because of its 2.19% positivity rate.

New variant data

From the Feb. 18 Public Health Report

The same statement appeared in the previous report, with just this difference:

From the Feb. 11 Public Health Report

The City of Northampton website has no information about the new variants, but we know thanks to a Feb. 1 Boston Globe article, and a recent follow-up in the Gazette, that there was a case of the B.1.17 variant in Hampshire County. The CDC database cited in the Feb. 18 Public Health Report reported 44 cases for Massachusetts as of Feb. 18; there seems to be no publicly available currently updated county-level data.

Hampshire Daily Gazette web edition, published Feb. 15 2021

It’s not clear why “medical privacy reasons” stopped the DPH from releasing more details. Compare the Jan. 17 press release on the first detected case (in Boston). We now only have aggregated data on travel status:

Hampshire Daily Gazette web edition, published Feb. 15 2021

We also know thanks to that Gazette article that there is relatively little surveillance testing for new variants in Massachusetts.

Hampshire Daily Gazette web edition, published Feb. 15 2021

With a daily new case rate of 3248 on Feb. 1, 100 per week means about 0.4% of cases were being sampled. With a daily rate of 1818 on Feb. 19, about 0.8% are being sampled.

Given the current data, the B.1.1.7 variant may well be fairly widespread here. But we are being given little or no guidance about what we should be doing as individuals in light of that, and policy making does not seem to be taking it into account (see this critical take on state-level policy and the new variants). It is shocking that there was no local press release from the MassDPH or one of our local health boards when the Hampshire County case was discovered, and that we still know so little about it.

Local data in Kingston Ontario
“This is an extra new concern, and we have an actual plan to deal with it”. Dr. Kieran Moore, Medical Officer of Health, Kingston, Frontenac, Lennox & Addington (KFL&A) Public Health.

In Ontario, all of the positive tests with an N501Y mutation are sequenced, plus 5% of other tests.

Testing for new variants in Ontario as of Feb. 3.

Update Feb. 23: A second new variant case has been detected in KFL&A. The reporting on it is incredibly thorough, and gives lots of details about how these cases, as well as surveillance and quarantine for international travel are being handled in Ontario.

The KFL&A Region has a slightly higher population than Hampshire County. They have had only one death from COVID-19, we are currently averaging one a day.

The current Kingston data is also presented in easier to read tables by a local web-based newspaper (note especially the detailed case information at the end). The week ending Feb. 19 had 17 new cases, which with a population of 204,116 makes 8.3 per 100K (1.2 per day), which puts it in the CDC’s blue Low Transmission category (see this webpage on what Kingston’s “Green/Prevent” categorization by Ontario entails).

The KFL&A health board provides an example of how outstanding public health work can help curtail the spread of COVID-19, and of how information and guidance can be effectively shared with the community (its success has been discussed in the national media: see this article from July 24 2020, this one from Jan. 19 2021, and this one from Feb. 21 that is paywalled, here is an excerpt).

Looking ahead

As we hopefully move into a time where COVID-19 is less prevalent in our area, we might also hope that our health departments will improve the quality of local data so that local decision making, both by officials and individuals, is on firmer empirical ground. Perhaps the new Joint Committee on COVID-19 and Emergency Preparedness, chaired by our own Senator Comerford, will help create some motion in this direction.

A wish list:

  • Systematic public presentation of details about cases and their contacts, anonymized or aggregated in whatever way needed to protect privacy
  • Separation of higher ed data from other local testing data
  • An improved state-level risk metric (maybe adoption of the new CDC one?), and guidance about what the risk levels mean for the public
  • Better surveillance testing for new variants (and the disease in general), with guidance about how the results should guide decision making

Massachusetts residents may ultimately be better served by county rather than town/city health departments. Two advantages:

  • Resource sharing
  • Avoidance of data bottlenecks at the state level (MA has more boards of health than any other state in the country, which may well explain the lags in transmission of town/city data)