John Locke Update / Research Brief

The Fog of COVID-19 Data: An Introduction

posted on in COVID-19 Series
Featured Image

For weeks now, Americans have lived in a suspended state as they grapple with the response to COVID-19, a disease caused by a “novel” coronavirus that appears to have originated in China. As more is learned about the novel coronavirus, claims that appeared to be accurate seem to be invalidated daily.

While the Cooper administration has proclaimed that data and “science” drive their decisions, the nature and use of the data are sometimes unclear. In a previous research update, the John Locke Foundation assessed the use of models in supporting policy decisions. The Cooper administration has since dropped its reliance on modeling. Here we discuss the question of deaths attributable to COVID-19 versus the flu or other related illnesses.

COVID-19 and the Flu

We are familiar with many types of viruses. Influenza viruses (A and B) cause the flu, while rhinoviruses and some coronaviruses can cause the common cold. Viruses are known to mutate in nature. Flu vaccines must be prepared months before flu season and are developed based on educated guesses of the flu strains that will be prevalent at the start of the flu season. Some guesses are better than others.

All viruses can be serious or even lethal. When a patient succumbs to a viral disease, the actual cause of death is not necessarily the virus itself.  Respiratory diseases often bring complications, such as pneumonia, that can be the actual cause of death. As a result, the reporting of deaths from viruses has been imprecise.  Long before the current COVID-19 pandemic, doctors have debated how to distinguish between flu and pneumonia when identifying the cause of death.

The distinctions are still unclear.  For example, the number of flu deaths in the United States and in North Carolina over the past few years varies significantly. According to data compiled by the Center for Disease Control (CDC):

 Symptomatic IllnessesSymptomatic IllnessesMedical VisitsMedical VisitsHospitalizationsHospitalizationsDeathsDeaths
SeasonEstimate95% U IEstimate95% U IEstimate95% U IEstimate95% U I
2010-201121000000(20,000,000 – 25,000,000)10000000(9,300,000 – 12,000,000)290000(270,000 – 350,000)37000(32,000 – 51,000)
2011-20129300000(8,700,000 – 12,000,000)4300000(4,000,000 – 5,600,000)140000(130,000 – 190,000)12000(11,000 – 23,000)
2012-201334000000(32,000,000 – 38,000,000)16000000(15,000,000 – 18,000,000)570000(530,000 – 680,000)43000(37,000 – 57,000)
2013-201430000000(28,000,000 – 33,000,000)13000000(12,000,000 – 15,000,000)350000(320,000 – 390,000)38000(33,000 – 50,000)
2014-201530000000(29,000,000 – 33,000,000)14000000(13,000,000 – 16,000,000)590000(540,000 – 680,000)51000(44,000 – 64,000)
2015-201624000000(20,000,000 – 33,000,000)11000000(9,000,000 – 15,000,000)280000(220,000 – 480,000)23000(17,000 – 35,000)
2016-201729000000(25,000,000 – 45,000,000)14000000(11,000,000 – 23,000,000)500000(380,000 – 860,000)38000(29,000 – 61,000)
2017-2018*45000000(39,000,000 – 58,000,000)21000000(18,000,000 – 27,000,000)810000(620,000 – 1,400,000)61000(46,000 – 95,000)
2018-2019*35520883(31,323,881, 44,995,691)16520350(14,322,767, 21,203,231)490561(387,283, 766,472)34157(26,339, 52,664)

Annual deaths have ranged from 20,000 to 60,000 per year, symptomatic illnesses were about 30 million per year, and hospitalizations fluctuated from a low of 200,000 to a high of 800,000. The 2017-2018 flu season was particularly deadly. The CDC first reported almost 80,000 deaths during that season. This number was later revised downwards to 60,000, which illustrates the uncertainty in the process. North Carolina averages approximately 2,000 flu/pneumonia deaths per year.

Again, these numbers lump flu and pneumonia together as the causes of death. The NC Department of Health and Human Services (DHHS) reports flu deaths alone, which are those deaths confirmed by a flu test and are typically one-tenth of the CDC reported flu/pneumonia deaths.   NC DHHS also reports a combined number.  Pneumonia appears to be a far greater cause of death than the flu.

 

Geographical AreaNumber of Deaths 2015Death Rate* 2015*Number of Deaths 2011-2015Death Rate* 2011-2015*Age-Adjusted Death Rate* 2011-2015*
North Carolina211321942719.217.8
Alamance3119.6*15119.515.5
Alexander821.4*4122.1*18.8*
Alleghany327.7*1120.2*11.8*
Anson415.5*2519.1*16.6*
Ashe1244.4*4936.1*22.7*
Avery739.6*4045.2*31.5*
Beaufort612.6*3816*11.4*
Bertie524.8*1918.5*13.3*
Bladen617.5*3419.6*15.7*
Brunswick1512.2*9416.211.6
Buncombe6324.927622.315.8
Burke2932.6*12928.821.8
Cabarrus552822724.125.4
Caldwell2935.7*12430.325.2
Camden329.1*1325.6*26.5*
Carteret1318.9*7321.414.8
Caswell939.2*3328.5*21.2*
Catawba3824.5*19625.421.9
Chatham2231*7020.812
Cherokee622.1*3525.8*14.8*
Chowan427.8*1621.8*12.9*
Clay328*1731.9*19.3*
Cleveland3132*16233.328.2
Columbus1221.2*6823.819.6
Craven1312.6*991915.9
Cumberland6520.126816.520.3
Currituck2287.1*11291.596.3
Dare1233.6*9956.848.6
Davidson4627.9*19924.320.9
Davie2047.9*6129.420.3
Duplin1220.3*5719.216.8
Durham3812.6*18312.714.2
Edgecombe814.8*6423.218.5
Forsyth9726.339922.119.7
Franklin1523.5*6922.220.9
Gaston7434.733531.929.2
Gates217.5*1423.9*18*
Graham446.4*1432.2*20.3*
Granville610.2*4214.5*13.3*
Greene29.5*1413.1*12.5*
Guilford9818.942516.815.5
Halifax1426.7*6725.118.5
Harnett107.8*7411.914.4
Haywood2643.4*9231.119.6
Henderson4136.4*16830.616.8
Hertford28.3*2318.9*13.8*
Hoke59.5*2710.6*17.1*
Hyde00*27*5.5*
Iredell4023.5*18222.121.6
Jackson512.1*3718.1*16.5*
Johnston2614*1161315.1
Jones330*1427.5*17.1*
Lee1728.5*6120.517.9
Lenoir1424.1*7124.218.5
Lincoln1518.5*7318.317.6
McDowell817.8*6127.120.8
Macon1440.9*4124.2*14.9*
Madison733.1*3432.3*24*
Martin938.5*2521.1*14.3*
Mecklenburg1551559912.115.2
Mitchell532.8*2431.3*20*
Montgomery1036.3*4029*23.4*
Moore2728.6*1012212.7
Nash3840.5*13528.524.5
New Hanover3315*1391311.2
Northampton524.5*2321.9*13.5*
Onslow2211.8*778.413.9
Orange96.4*669.511.6
Pamlico431.3*1116.9*12.6*
Pasquotank820.1*4522.5*19.4*
Pender1322.6*4817.4*14.9*
Perquimans752.1*2435.6*23.2*
Person1333.1*5829.523.2
Pitt169.1*748.510.1
Polk1049.1*3635.4*15.4*
Randolph2920.3*15521.819.1
Richmond1022*3414.8*12.7*
Robeson2115.6*8913.214
Rockingham3538.1*19542.331
Rowan574127239.332.4
Rutherford1522.6*7422.117
Sampson1320.4*5617.515
Scotland822.5*3016.7*13.8*
Stanly3049.4*8628.422.9
Stokes1838.8*7130.423.3
Surry3244*11130.322
Swain962.4*2433.9*26.6*
Transylvania1133.1*4326.1*12.9*
Tyrrell00*314.4*9.9*
Union3113.9*13812.917.9
Vance1942.6*7633.928.3
Wake878.5389810.7
Warren839.7*3433.2*20.7*
Washington324.2*1726.8*16.7*
Watauga1018.9*3814.5*14.3*
Wayne1915.3*8513.712.8
Wilkes2739.4*15143.832
Wilson1923.3*9924.320
Yadkin1334.6*6534.226.2
Yancey1056.9*2932.9*20.5*

*Death rates with a small number (<50) of deaths in the numerator should be interpreted with caution.

Similarly, NC is currently reporting more than 600 COVID-19 deaths. However, NC DHHS uses a different policy where COVID-19 is concerned. Unlike flu deaths, deaths coded as COVID-19 are deaths that may or may not be confirmed by a test.  Even with this more relaxed reporting requirement, the number is still lower than the pneumonia deaths in a typical year.

Was the cause of death the flu, pneumonia, and COVID-19?

The potential mix-up between the three causes of death has caused some researchers to take a different approach. First, this approach simply looks at previous year averages for all deaths in the nation and each state from all causes. Based on these statistics, the expected average (to a 95% confidence interval) can be calculated. This means that we would expect this average to be exceeded only 5% of the time, which might be considered unusual. In other words, if the actual deaths exceed this average, some new or unusual cause is to blame. The analysis of New York is illustrative.

Here both the 2017-2018 flu season and the recent COVID deaths are clearly implicated as unusual (excess) deaths. This approach is attractive because no judgment calls are made.

Here are data for Tennessee:

Again, the 2017-2018 flu season is implicated as being unusual. The more recent COVID-19 season is not as clear, however.

Data for North Carolina are here:

There are marked differences between New York, Tennessee, and North Carolina.  Interestingly,  North Carolina and Connecticut are the only two states in the nation that have not reported data since April 1. The NC Senate recently cited this reporting gap as the reason why it championed legislation that compelled NC DHHS to provide certain data, including COVID-19 deaths. The more expansive definition used by the agency for COVID-19 deaths (compared with flu deaths) means that NC DHHS may overcount coronavirus deaths. The “excess deaths” method described above would be helpful in this regard.

Similar graphs for North Carolina counties would allow elected officials to determine if coronavirus response restrictions should be more localized. Governor Cooper proclaimed that COVID-19 “does not respect county borders.”  Yet, substantial differences currently exist between different counties, that is, COVID-19 appears to be much more prevalent in some counties than others.

Going forward

Each week, the John Locke Foundation will use the Coronavirus (COVID-19) Global Data Tracker and other sources to provide an analysis of the latest COVID-19 data available.  We will include commentary on data collection, reporting policies and practices, health care outcomes, coronavirus mitigation measures, and related matters.  We will also offer policy recommendations that seek to slow the spread of COVID-19 without sacrificing vital economic activity.

Dr. Donald van der Vaart is a Senior Fellow for the John Locke Foundation. Dr. van der Vaart earned a B.S. in Chemistry from UNC Chapel Hill, an M.S. in Chemical Engineering from N. C. State University, a Ph.D. in Chemical… ...

Donate Today

About John Locke Foundation

We are North Carolina’s Most Trusted and Influential Source of Common Sense. The John Locke Foundation was created in 1990 as an independent, nonprofit think tank that would work “for truth, for freedom, and for the future of North Carolina.” The Foundation is named for John Locke (1632-1704), an English philosopher whose writings inspired Thomas Jefferson and the other Founders.

The John Locke Foundation is a 501(c)(3) research institute and is funded solely from voluntary contributions from individuals, corporations, and charitable foundations.