#covid19 in #southcentralpa
As a public service to the local 10-county region.
This page is updated regularly (usually around Noon-1pm, sometimes later depending on when the State releases the data and my calendar).
Wed 25-Nov: Top lines: Daily New Cases still up around a 1000 for our 10-county area. Current hotspots are Mifflin and Juniata counties (see the "flatten the curve spaghetti chart"). Fatalities are starting to shift back from the General Public to the Nursing Homes (LTCF). Both are now (14-day trailing average) at around 6.5 fatalities a day - or 13 total. Contrast this to the combined rate of 3 about a month ago.
Tue 24-Nov: Refreshing a data table today that I used to report on regularly a while ago. I think it is telling. You decide.
Mon 23-Nov: As we approach Thanksgiving, with so very much for which to be grateful, we can add a couple small signals in the data which may or may not offer a sigh of relief (depends on whether or not they persist). A sharp drop in the daily new case counts, is awesome. And the death rate dropped from mid-twenties to below 10. Let's keep it going!
Sat 21-Nov: OK, 2nd day in a row of declining new cases. This is an encouraging signal. Let's try and keep it up. Please note the "spaghetti chart" (which is no longer flat) that Mifflin and Juniata counties are climbing appreciably faster than the other eight counties. However, fourth day in a row of 20+ deaths in the region (27, 25, 27, 20). No Nursing Home vs Gen Public breakout today because the State DoH did not publish Nursing Home data.
Fri 20-Nov: Another 27 deaths reported today for yesterday (26 of the 27 in the general public) and another 988 new cases (97% in the general public). Percentage positivity is up around 25%. The 14-moving average for fatalities is rising sharply, as is the deaths per 100k of population - for the first time - in the general public. These are the facts.
Thu 19-Nov: Yest, 27 deaths reported in our region from Tue. Today, the state reports 25 for yest. Deaths typically lag onset from three to four weeks. That many weeks ago, our daily new infections was around 300 to 400/day. We are now at three to four times that rate. So, in three to four more weeks (following Thanksgiving and approaching Christmas) if the ratios hold - our region will be approaching 100 deaths/day. Are we prepared?
Wed 18-Nov: Neighbors, another new record high today crossing the 1000 line for new cases. 90% of those are in the general public and 59% of the 27 deaths were also in the general public. The 14-moving average of fatalities now stands at around 5.5 and 3 for the Gen Public and LTCF populations respectively; and rising sharply. All charts updated below.
Tue 17-Nov: Neighbors, new record high for the region of 964 new cases (54 probables, the rest confirmed PCR positive) (945 - or 98% - of those are in the general public). The 14-day moving average for the region is now over 4 times the peak during the red phase. The 5 deaths yesterday comprised 2 in nursing homes and 3 in the general public. This is a reflection of the larger trend, where the deaths are moving from the LTCF to the general public. Note:
Mon 16-Nov: Neighbors, mixed signals in the data today. You'll note that the daily new cases has dropped and hopefully that is the start of a trend back downward. However, the fatalities chart continues upward - for both the general public and the nursing homes. A friend of mine told me yesterday (lives in a LTCF) that they have seen 80 new cases in just the last week or two, and five deaths... in just one facility. This, once again, amplifies how important it is to protect our LTCF - and to separate these two populations in data and in policy. So, I went back to an old set of charts I hadn't updated in a while. I personally find these charts very curious. In particular, Cumberland and Lebanon counties. You decide (click image to enlarge):
Fri 13-Nov: Another day up the curve. Third day in a row of record new cases. 14-day rolling average now over 500 (more than 300% of the peak during the red phase in April). Then there is the "bump up" in fatalities in the nursing homes. The general public is still weathering better than the nursing homes, but fatalities there are also rising.
Thu 12-Nov: The regional daily new case doubling-rate is now down to 4 days which is commensurate with what we were seeing in the original runup back in April. And the fatality curves are now elbowing up sharply as well. This is not an emotional "sky is falling" bulletin, these are just a statement of facts.
Tue 10-Nov: Four of the last five days our ten-county region has scored in the high 500's of new cases. Probables account for less than 30 of those on any given day. Confirmed new cases are definitely on the rise. Deaths are also starting to rise. These are the facts.
Mon 9-Nov: Fellow citizens, just data and analysis; not politics (which is why this box is purple). I'm not sure how anyone can look at these charts and not be alarmed. Daily new cases in our region are now well over double the peak of the red phase. And the 14-day rolling average is now 416 which is 267% of the peak 156 during the red phase. Furthermore, since I (alone apparently) have been breaking out the general public and nursing home (LTCF) populations - we can see that the vast majority of these new cases are in the large and uncontainted general public; which is unlike the nursing home population which are contained and thus can "limit" the spread. I guess it is a mercy that it isn't in the vulnerable population of the nursing homes as the virus is much more devastating there. The deaths per 100k in that population is still around 7400 but you can see a very slow flattening of that curve. Conversely, the deaths per 100k in the general public while only 20, that curve is now bending in the wrong direction. So, fellow citizens... PLEASE mask up, wash your hands and keep your distance. Our scientists are making great progress on therapeutics and prevention (vaccine) but we still have many months ahead of us before we can relax. Please be responsible.
Fri 6-Nov: Citizens, not much good to report today. We've hit new record levels of new cases across the board. Our fatalities are starting to track upward now also. Percentage positivity in testing is also rising. All metrics are up, period.
Wed 4-Nov: A couple things to share today.
Mon 2-Nov: We may be at the end of the road here. Last week, the State DoH completely revamped their website(s) concerning Covid-19 and have not posted raw overall data since the 30th, nor nursing home data since the 26th of Oct. Without access to the data, curious on the eve of the election, I am blind and cannot offer any additional insights. All charts and tables are as up to date as of the last available data on either 26th or 30th.
Wed 28-Oct: Curious how our region compares to state as a whole...
Tue 27-Oct: New page format today. All "regular" charts now at the top; some 'one-time' charts still interspersed below. Today's summary: New cases: up. Fatalities: up. Percentage positivity: up. Most of this is in the general population, not the nursing homes... which is what is different than back in April. Apparently, that population is now "under control" though we are indeed still losing too many people there too. One other note about the data. I've commented in the past that it is unclear what testing is included in the new daily positives. Does it include Antibody, Antigen or Rapid (notoriously high false positive rate) tests? Or just PCR? So I called the State DoH directly and asked them. The person I spoke with said she did not know, but would pass my query onto someone who could answer it and get back to me. I'm still waiting...
Mon 26-Oct: No question, the numbers are climbing. Daily new cases: up. Deaths: up. Percentage positive in testing: up. If we don't get this under control - we're headed for another lockdown like Northern Ireland, the Netherlands, Israel and more. We can do this! Mask up!
Wed 21-Oct: Key charts updated today. If the State DoH Covid-19 data is to be believed, the above two charts should make you concerned. Please be responsible and vigilant.
Mon 19-Oct: Key charts updated today. The new cases chart looks a wee bit less alarming today. However, the percentage positivity chart looks more alarming. That said, there is a suspicious single day spike in testing volume - a literal 500% increase in a single day. This seems unlikely, so I speculate that a bunch of tests were "held up" in the lab or in reporting or something. Why? No idea. The "good news" is that the spike in testing volume correlates to the single day "420 new cases alarm" a few days ago. So that appears to be more of an administrative reporting snafu than a literal spike in new cases. But who knows?!
Fri 16-Oct: Key charts updated today. While the new case count dropped significantly to 267 from the prior day's 420, 267 is still one of the peak days since April. The 14-day moving average is now at 211. But the question is: are these truly active cases? 34 of these are so-called "probables". How many are antibody positives (not active cases)? Will the State DoH separate the numbers? If not, why not? Inclusion is a clear "artificial inflation" of the reality and not science. What is the agenda here? Does the State DoH and the Governor's Office not understand that by "spinning the numbers" the only thing they succeed in accomplishing is sowing distrust and cynicism by the population - the consequence of which is that everyone ends up not trusting the numbers and then deciding to do whatever they want anyway. How is this helpful?
Thu 15-Oct: Key charts updated today. Neighbors, yesterday saw a new record high of new cases in our region - of 420. That is DOUBLE the highest new case peak during the red phase. Of those 420, 18 of them are "probables" which are thus "likely but not confirmed by actual science. Of the remaining 402 "new cases" - I wonder how many are antibody positives, not PCR positives? Is this an actual spike? Or is it politics? What if the antibody positive number accounts for 200 of the remaining 402? Are we sounding the alarm "not factually"? Does anyone out there have a data source other than the State DoH that reports PCR Positive, separate from Antibody Positive separate from "Probable"? I would surely appreciate that data.
Wed 14-Oct: Key charts updated today. Again, 14-day moving averages are rising rapidly - particularly in the general public.
Tue 13-Oct: Key charts updated today. Note - 14-day moving average of new cases in the general public remains on the rise, sharply. Deaths per 100k of population - in the nursing homes is still climbing - but improving - still at over 7100/100k of pop. Deaths in the general population, has risen to 17. However, given the sharp rise in the last seven days (with the highest peak ever for a single day 356 new cases reported on Sat 10-Oct, and four peaks in the last seven days, the death rate will rise in a couple weeks... Please remain vigilant.
Tue 6-Oct: Key charts updated today. Note - 14-day moving average of new cases in the general public is definitely on the rise, and deaths per 100k of population - in the general public - has risen to 16. While still relatively low (particularly compared to Nursing Homes), these trends are concerning.
Thu 1-Oct: Colleagues and Neighbors, some new analysis today. Continuing to think about the relationship between Long Term Care Facilities and the General Public, but also about the Employee who bridges the two populations. Also thinking about the relationship between the fatality rates in the two populations - as they relate to the new cases. Here is a familiar chart showing the new cases; split by LTCF v Public. As is clear, the LTCF has flattened if not trending down (whew!), but the Public is rising. How does this relate to the fatality rates in each population? The Fatalities chart shows the 14 day moving average of fatalities in each population. As you can see, both are 'unstable' patterns; no clear trend up or down. Conclusion? Fatalities are not yet 'under control' in either population and seemingly unrelated to infection trends. Additionally, I charted the new case rates in the LTCF alone - and plotted those along with fatalities just in that population: What I was seeking to see here were a couple things. First, is there a 'pattern' to the Employee incidents and the Resident incidents? In other words: which way is the virus moving; from Emp to Res or Res to Emp? When we see a large green bar - is there a spike a few days later in the orange, or vice-versa? My two examples, Dauphin and Lancaster counties, don't appear to shown any such pattern. Also, I was expecting to see a spike in the red line somewhere around two weeks after a spike in an orange bar - assuming it takes an elderly and comorbid Resident two weeks to progress from onset to fatality. I can see no such pattern. I can speculate lots of reasons for this, but I'll resist that urge. A good time series analysis could potentially show this - but the data is too spotty and unreliable for that. So, bottom line: I cannot draw any reliable correlations relative to Emp vs Res or Onset vs Fatality. That said, it is still clear that the virus is still infecting an ever increasing number of the Public (assuming the data is all PCR when it is most likely PCR and Antibody thus making the new infection rate truly unclear) and the fatality rate still unpredictable in relation to infection rate.
Mon 21-Sep: The usual refreshes. Some progress on new cases, the 14-day moving average in the general population has come back down. Whew! But if you look at the new cases split by Nursing Homes vs General Public, you can see that from about mid-July onward, there is a fairly linear trajectory upward. Similarly, when looking at the deaths per 100k of population, the general public is rising on a fairly linear trajectory. Now, I hasten to add that the deaths per 100k in the general population has just crossed the 15 mark. Contrast that number with the 6741 per 100k in the nursing home population. This means the mortality rate for general public: 0.015% vs 6.74% in the nursing homes. This make death from Covid-19 450 times more likely if you live in a nursing home, or 450 times LESS likely if you don't. The data is clear: in the general public Covid is about 7-8x more deadly than the seasonal flu (2/100k), but, 13x less deadly than good old fashioned heart disease (197/100k). (source: CDC) So, just keeping it all in relative perspective.
Mon 14-Sep: All charts and tables refreshed today. Again, new cases are up and the trailing 14-day average is now ABOVE the peak during the red phase. I'm surprised that the State DoH and the Press are not all over that. ;-) Second, the percentage positive numbers are also rather dramatically jagging up and down - but generally upwards. Both of these are concerning trends. That said, the fatalities continue to remain low (relatively speaking to the red phase). Is this a consequence of better medical treatment protocols? Or could the virus be weakening? Or could it be that the rise in new cases is a false signal because it includes antibody test results (which measure past cases) not current active virus cases? Who knows! More transparency in the numbers, please.
Tue 8-Sep: All charts and tables refreshed today. Fellow citizens, the sky is not falling - but, a couple concerning signals in the data. First, be careful and responsible with the lives of others please... The daily 14-day average for new cases for the general public is rising and we haven't even begun measuring the spread of Covid from this past Labor Day weekend. In fact, if you look at the above chart carefully, you will see two concerning things: the 14-day trailing average is now equal to the peak in during the red phase and the general daily pattern is also not unlike the red phase. Second, the percentage positive cases is also rising sharply. It is never clear what the State DoH is reporting in their testing number and positive numbers (PCR vs Antibody) - so this could be a false signal if it represents a rise in Antibody test percentages (which reports cases that happened a month or more ago so do not represent current active virus cases).
Mon 31-Aug: All charts and tables refreshed today.
Mon 24-Aug: All charts and tables refreshed today. Regarding testing and positivity rates: both the state and the region have DECLINING testing volumes and positive test rates continue to be highly volatile. However, the linear regressions for the Region and the State are relatively flat at about 6% and 5% respectively. Finally, the fatality rates continue to flatten with the General Public around 13 deaths per 100,000 and the Nursing Homes at 6300 per 100,000.
Mon 17-Aug: All charts and tables refreshed today. Note: the percentage positivity in testing for our local region continues to be higher than the overall rate for the commonwealth, however, our fatality rate continues to be low(er). Cannot offer any data-driven explanations for this "contrapositive trend".
Mon 10-Aug: All charts refreshed today. A few notes: the fatality curve in the nursing homes does indeed appear to be flattening. Something appears to be happening there that is helping. Kudos and thanks to all the extraordinary workers on those communities that are saving lives every day. Also, there does appear to be a slight uptick in infection rates in our local region. Not alarming, but a signal in the data regardless. However, and this I must once again powerfully draw attention to - the infection rate and mortality rate in the general public is still very, very low. See this chart and note the orange (public) versus blue (nursing) lines. In Cumberland County in particular, the fatality rate in the general public is a literal ZERO and the CUMULATIVE infection rate is still only 0.3%. It is still appropriate to take all reasonable precautions when leaving our homes, but is the economic damage wrought by public policy consistent with this DATA?
Mon 3-Aug: All charts refreshed today, one new analysis. The good news is that the cumulative mortality rate is leveling off. It appears that our region is "flattening the death curve."
Mon 27-Jul: All charts refreshed today. Good news! The fatalities in the Nursing Home population is flattening. These two charts show that the two populations are "decoupling" as the nursing home population new case rate is finally trending downward in a consistent pattern. However, the new case rate in the general public is trending the other way. And the Fatalities chart further amplifies the differences in mortality rate between nursing homes and the general public. Again, suggesting that these two populations are distinctly different.
Mon 20-Jul: All charts refreshed today. According to the State DoH, the 10-county region has not suffered any new fatalities in the last four days. According to the State DoH, in those same four days, there have been 7 new fatalities in our Long Term Care Facilities. Just reporting the facts.
Mon 14-Jul: All charts refreshed today. Still trying to make sense of the data regarding Nursing Homes vs General Population. Today, an expanded analysis and a new analysis. The following chart is an expansion of the chart from 6-July that showed new daily cases just for nursing homes. Today, a similar chart (immediately below) shows the new daily cases for the general public.
Mon 06-Jul: Here is a two-week update with what data the State DoH has made available; a lot of gaps in the data. A few notes: The daily new cases in our region is jagged with a slight uptick. Also, again I draw your attention to the infection/fatality rate charts by county and you will note that the non-nursing home rate, even in the green phase, has remained a fraction of a percent... compared to percentages in the teens or twenties for the nursing homes. A dramatic difference. Meanwhile, you will note on the deaths per 100k charts, there is a clear flattening that has started - about a month after the edict went out to focus on the nursing homes. Finally, the chart that immediately follows shows daily new infections in the nursing homes; I think it speaks for itself.
Tue 23-Jun: Well, here we are again. Two weeks ago I gave up because the numbers coming out of DoH were SO erratic that they could not be considered credible. THEN, for several days there were no numbers. THEN, they were reporting numbers, but changed all kinds of delivery method and formatting. So, I grit my teeth and worked through the impediments to report for a few days. THEN, they ceased reporting numbers on weekends. NOW, they say they will only report nursing home numbers once a week on Tuesdays. Bottom line: they have manipulated the data flow so much over the life of this exercise, but in particular the last two weeks or so and now the flow is virtually dried up. Why is this clear pattern of poor data collection and decision-making plus obstructing public transparency/scrutiny being ignored by the press and the legislature? Cannot fathom an answer and I am beyond flummoxed by it all. I'm NOT a quitter, but my hands are tied. Nothing more I can do without the data.
Mon 22-Jun: State DoH posted data today so I have new charts for you. They have also now posted that they will only be updating the data Mon-Fri from now on.
Sun 21-Jun: The State DoH has not posted data yesterday (Sat 20-Jun) or today (Sun 21-Jun) - so I'm not ignoring this - I just have no data to illuminate. Perhaps tomorrow will be a better day for transparency.
Thu 18-Jun: Friends, many of you refer to what I'm doing here as "your data, Joe" - please - I implore you with love and civility - this is not my data. It is my analysis, yes, or perhaps even "illuminated insight," on the State DoH data, but it is not my data; that dubious distinction belongs with the State DoH. That needs to be made clear. That said, analysis/charts refreshed again today.
Tue 16-Jun: Having continued to track whatever numbers are made available - and continuing to think about this problem. I offer the following new "illuminated insight":
Both thumbnails are hyperlinks to the larger full-size images. PLEASE examine both. NOTES: Gaps in the charts are a consequence of the State DoH not reporting Nursing Home numbers on those dates. Also, the clear "aberrations" in the charts are clear indications of a lack of veracity in the data reported on those dates. Linear trend lines are ignoring the low-veracity data points.
The charts (and accompanying data table) show the Deaths per 100,000. The first chart shows what the numbers look like if you treat the two populations as if they are one. The other two chart break out the populations and thus you can see how dramatically different are those populations: Nursing Homes vs. General Population. Please pay close attention to the y-axis of each chart. You will see that the Nursing Home y-axis is over 400 times larger than that of the General Population, and the slope is still higher. In other words, you are more than 400 times more likely to die of Covid-19 if you reside in a nursing home, than you are if you live in the general population. In fact, it so eclipses the general population numbers that by NOT BREAKING THEM OUT, the general population is lead to believe that Covid-19 is much more deadly to THEM than it actually is and far LESS deadly to those in Nursing Homes than it actually is. Now, I hasten to add, and with emphasis, that I do not mean to say that Covid-19 is not deadly, or isn't more deadly than influenza or that the general population should cease and desist from all sensible PPE when out of their homes - but I AM saying that we should tell them the truth about their risks. Stop conflating the numbers and causing undue anxiety for no legitimate reason. The FACTS do not support it. According to the CDC: the flu is 2 per 100,000 per season. So, with benefit of lockdowns - Covid-19 is still at least 4 times more deadly than the flu. Also for comparison, the CDC shows the Cardio Vascular Disease kills at the rate of 176.1 per 100k per year.
Finally, again, employing some critical thinking here - I'd like to draw your attention to a very small signal in the Nursing Home chart. At the end of the chart, you will notice what could be a flattening trend in the death rate in those homes. This signal comes about 4 weeks after the Sec of Health decided to make nursing homes a priority and ramped up both testing and care. Is it a coincidence that 4 weeks later the death rate may be slowing? Hindsight is 20/20, true, so it would be easy to say now that the nursing homes should have been a priority sooner. On the other hand, if someone had actually been thinking about the data sooner, they would have seen the signals much earlier that it was better for everyone concerned (both demographic populations) to focus on the nursing homes rather than the general public. A point I personally made back on 4-May when I first introduced my nursing home vs general population analysis. I realize that statement may sound self-serving, but it isn't meant to. It IS meant to say that there were signals in the data back in mid April, had anyone bothered to look and THINK about it. How many needless deaths could have been avoided? How much needless economic hardship could have been avoided?
|County Infections Data||Infection/Fatality Percentages|
for Two Distinct Populations:
|Road to Reopening|
|Nursing Home Data Tables|
8-June: "This too shall pass" - and so it shall. In mid March I began tracking Covid-19 data and in late March I began posting my findings here, every day, for several reasons:
1) While the press was reporting raw numbers day to day (still do), they were not putting those numbers into context with anything as easy as a line chart that showed each day so you, the public, could easily and visually see if the situation was good, bad or indifferent. I am still mystified by this absense, which persists today. Furthermore, by only reporting the cumulative numbers day to day, without noting the daily increases so you can try to remember yesterday or the day before and mentally decide if the numbers are getting better or worse, this "constant climbing" of the totals fans fears. Fears borne not of reality - but absence of facts in context.
2) To encourage thoughtful discussion by the public (you) but also, in the futile attempt to engage a wider audience including press and policy makers. That has failed to happen. Numerous attempts by a number of you (and some by me too) to encourage Legislators, the Governor's office and the Sec for Health to examine these findings have failed to garner a conversation. In short, "crickets" is all I have heard from Capitol Hill. And for the record, from both political parties. This is not a time for partisanship.
3) This has also been an exercise in showing the power of "illuminated insight" - which is to say - in going deeper. It doesn't take a rocket (or data) scientist to go deeper on the data; just a desire and a little thought. Charts have power. They illuminate things.
Being frustrated by yet another day of "bizarre numbers" coming out of the DoH, I've decided that this is no longer worth the effort. The data is not credible and reporting of it is spotty and unreliable. The "viewership" is equally spotty. And policy has been completely unaffected by any insights shared herein over the last 73 days of sharing. So, this is, sadly, my last update to this page. However, the page will remain intact as a reference.
If you have any comments, criticisms, questions, ideas, anything that you'd like to share relative to this effort - please don't hesitate to email me directly (if you have my address) or click here to send me a private message via web-form. For those few of you who have been consistent and ardent supporters of this effort - I thank you.
Stay safe everyone! And remember, critical thinking is ALWAYS your friend. Go deeper!
2-Jun, 11:43pm: I'm taking the unprecendented step tonight of updating my post from earlier today because the State DOH released the LCTF (Long Term Care Facility) numbers late in the day. And the numbers (notice I didn't say "data") are simply not logical or credible. I am thus, hereby, and publicly, challenging the Veracity of the numbers.
Here is my rationale and I invite you to CAREFULLY review several "illuminations" herein. Please read and THINK carefully about these *facts* which are in contradiction.
If you examine the "Nursing Home Tables" below, you can see the numbers (all reported by the State DOH) as they existed on the dates shown. There are calculated numbers on those tables, but they are simple percentages or subtractions that anyone is free to verify. Each date shows wildly different "conclusions" based on the data. If the Five Vs of Data (Volume, Velocity, Value, Variety and Veracity) are inadequate, then so are the decisions you are making based on them.
This kind of wild fluctuation in the numbers suggests a problem in reporting. I'm saying plainly, these numbers are not credible as DATA. I am open to being wrong about my conclusions, if someone can offer a rational explanation for how a county can have more deaths in nursing homes alone - than they have in the entire county. There is only one logical explanation: the data collection and reporting is not consistent, reliable or "validated". If policy decisions are being made predicated on "unreliable data" then those policies should be reconsidered.