Dan Weinberger,1 Jenny Chen,7 Ted Cohen,1 Forrest W. Crawford,2 Farzad Mostashari,3 Don Olson,4 Virginia E Pitzer,1 Nicholas G Reich,5 Marcus Russi,1 Lone Simonsen,6 Annie Watkins,1 Cecile Viboud7

1Department of Epidemiology of Microbial Diseases and the Public Health Modeling Unit, Yale School of Public Health, New Haven, CT 2Department of Biostatistics and the Public Health Modeling Unit, Yale School of Public Health, New Haven, CT; Yale Departments of Ecology and Evolutionary Biology, Statistics & Data Science, Yale School of Management 3Aledade, Inc 4Department of Health and Mental Hygiene, New York City, NY 5Department of Biostatistics and Epidemiology, School of Public Health and Health Sciences, University of Massachusetts, Amherst, MA 6Department of Science and Environment, Roskilde University, Denmark 7Division of International Epidemiology and Population Studies, Fogarty International Center, National Institutes of Health, Bethesda, MD

Mortality data on deaths due to pneumonia & influenza through the week ending 2020-05-30

Overview

Background Efforts to track the severity and public health impact of the novel coronavirus, COVID-19, in the US have been hampered by testing issues, reporting lags, and inconsistency between states. Evaluating unexplained increases in deaths attributed to broad outcomes, such as pneumonia and influenza or due to all causes, can provide a more complete and consistent picture of the burden caused by COVID-19.

Methods We evaluated increases in the occurrence of deaths due to any cause above a seasonal baseline (adjusted for influenza activity) or deaths due to coronaivirus/P&I (P&I&C) for March-early-May across the United States. These estimates are compared with reported deaths due to COVID-19 and with testing data. Detailed methods can be found here

Conclusions Excess deaths provide an estimate of full COVID-19 burden and indicate that official tallies likely undercount deaths due to the virus.

Estimate for national excess deaths:

National all-cause excess

Excess deaths for the United States from March 1, 2020 through May 30, 2020. The observed number of deaths is indicated by the black solid line, and the expected number of deaths, adjusting for seasonality, influenza epidemics, and reporting delays, is indicated by the black dashed line. The area between these two lines represents the total number of excess deaths. The blue area represents deaths recorded as due to COVID-19, the red shaded area presents additional pneumonia and influenza excess deaths not coded as due to COVID-19, and the green shaded area represents deaths that were not attributed to COVID-19 or pneumonia or influenza.

Excess deaths in New York City and the rest of New York

Excess deaths for New York state (excluding New York City) and for New York City from March 1, 2020 through May 30, 2020. The observed number of deaths is indicated by the black solid line, and the expected number of deaths, adjusting for seasonality, influenza epidemics, and reporting delays, is indicated by the black dashed line. The area between these two lines represents the total number of excess deaths. The blue area represents deaths recorded as due to COVID-19, the red shaded area presents additional pneumonia and influenza excess deaths not coded as due to COVID-19, and the green shaded area represents deaths that were not attributed to COVID-19 or pneumonia or influenza.

Excess deaths in four high-incidence states

Excess deaths for four states that had large numbers of reported COVID-19 deaths from March 1, 2020 through May 30, 2020. The observed number of deaths is indicated by the black solid line, and the expected number of deaths, adjusting for seasonality, influenza epidemics, and reporting delays, is indicated by the black dashed line. Area between these two lines represents the total number of excess deaths. The blue area represents deaths recorded as due to COVID-19, the red shaded area presents additional pneumonia and influenza excess deaths not coded as due to COVID-19, and the green shaded area represents deaths that were not attributed to COVID-19 or pneumonia or influenza.

Excess deaths in select additional states

Excess deaths for additional states from March 1, 2020 through May 30, 2020. The observed number of deaths is indicated by the black solid line, and the expected number of deaths, adjusting for seasonality, influenza epidemics, and reporting delays, is indicated by the black dashed line. Area between these two lines represents the total number of excess deaths. The blue area represents deaths recorded as due to COVID-19, the red shaded area presents additional pneumonia and influenza excess deaths not coded as due to COVID-19, and the green shaded area represents deaths that were not attributed to COVID-19 or pneumonia or influenza.

Summary Table: Excess deaths and reported deaths due to COVID-19

Observed and Excess deaths due to COVID-19, pneumonia/influenza/Covid-19, and all-causes COVID-19, from March 1, 2020 through May 30, 2020
State Observed Deaths Expected deaths Excess all-cause deaths COVID-19 deaths (U07.1) Excess Pneumonia/Influenza/Covid-19 deaths Excess all-cause deaths/100,000
US 780975 658700 122300 (116800,127000) 95235 94130 (92350,95690) 37.3 (35.6,38.7)
NYC 38170 13000 25100 (24800,25400) 18603 17500 (17280,17670) 299.1 (295.3,302.7)
NJ 34013 17800 16200 (15800,16500) 11952 12060 (11910,12180) 182.3 (178.1,186.2)
NY 36871 24600 12300 (11900,12700) 9773 9220 (9040,9380) 111.4 (107.4,114.9)
MA 22146 14700 7400 (7100,7700) 6627 6400 (6280,6510) 107.6 (103,112.2)
DC 1950 1400 500 (400,600) 400 390 (350,420) 73.5 (58.9,86.6)
LA 13256 10200 3000 (2700,3300) 2310 2340 (2270,2400) 64.7 (59.1,70.3)
MD 16092 12300 3700 (3500,4000) 2651 2650 (2560,2740) 62 (57.2,66.8)
MI 30044 23900 6100 (5700,6600) 4839 4740 (4620,4870) 61.4 (56.7,65.6)
IL 33296 25800 7500 (7100,7900) 4911 4920 (4800,5050) 59.2 (55.7,62.3)
RI 2881 2300 500 (400,700) 586 550 (510,580) 51.7 (39.3,62.9)
DE 2537 2100 400 (300,600) 365 370 (330,400) 45.3 (32.6,57.5)
PA 37258 31800 5400 (4900,5900) 5840 5610 (5480,5730) 42.4 (38.5,45.9)
MS 8787 7600 1100 (900,1400) 677 710 (640,780) 38.3 (30.6,45.9)
CO 11804 9900 1900 (1600,2100) 1354 1330 (1260,1410) 32.4 (27.6,37)
IN 18336 16200 2100 (1800,2500) 2012 1990 (1900,2080) 31.9 (26.7,36.7)
VT 1580 1400 200 (100,300) 48 10 (-20,40) 29.8 (13.6,44.9)
SC 13601 12200 1400 (1100,1700) 528 480 (400,550) 27.9 (21.8,33.6)
VA 19219 16900 2300 (1900,2600) 1471 1400 (1310,1480) 26.7 (22.8,30.6)
AZ 16929 15200 1800 (1400,2100) 846 930 (840,1010) 24 (19.6,28.5)
GA 22105 19900 2200 (1800,2600) 1686 1800 (1710,1880) 21 (17.2,24.4)
NH 3383 3100 300 (100,400) 244 240 (200,280) 19.9 (7.9,30)
AL 13436 12500 900 (600,1200) 672 790 (720,860) 18 (12,24.1)
WA 15367 14000 1300 (1000,1700) 1021 1010 (930,1090) 17.5 (13.4,21.7)
CA 72407 65600 6800 (6100,7500) 4046 4130 (3940,4330) 17.2 (15.5,19)
FL 56462 52900 3500 (2900,4100) 2457 2940 (2780,3080) 16.4 (13.4,19.3)
MN 12112 11200 900 (600,1200) 1020 980 (910,1060) 15.4 (10.6,20.6)
OH 30429 28800 1600 (1200,2000) 1843 1890 (1780,1980) 13.8 (9.9,17.5)
TX 51994 48400 3600 (3000,4200) 1622 2100 (1960,2260) 12.4 (10.4,14.5)
MO 16043 15300 700 (400,1000) 718 800 (720,880) 12.1 (6.6,17.1)
TN 19080 18300 800 (400,1100) 348 380 (280,480) 11.5 (5.6,16.8)
WI 14191 13500 600 (300,900) 607 580 (510,660) 11 (5.6,16.2)
OK 9493 9100 400 (100,600) 328 320 (250,400) 9.1 (2.8,15.3)
NV 6735 6500 200 (0,400) 389 400 (350,460) 6.8 (-0.5,13.8)
NE 4241 4100 100 (0,300) 153 130 (80,180) 6.4 (-2.1,15)
OR 9030 8800 300 (0,500) 183 150 (90,200) 6.1 (0.1,11.5)
NM 4715 4600 100 (-100,300) 303 290 (240,340) 5.5 (-3.2,14.4)
IA 7801 7700 100 (-100,400) 529 470 (400,530) 4.4 (-2.4,11.5)
UT 5044 4900 100 (-100,300) 109 100 (50,140) 4.2 (-2,9.8)
AR 7905 7800 100 (-200,300) 123 170 (110,220) 2.9 (-5,10.4)
MT 2465 2400 0 (-100,100) 21 0 (-30,30) 2.2 (-10.1,13.5)
ID 3656 3600 0 (-100,200) 77 70 (40,110) 1.7 (-7,10.9)
KS 6640 6600 0 (-200,200) 222 190 (140,240) 1.5 (-5.5,8.2)
KY 11382 11300 100 (-200,300) 413 410 (320,480) 1.5 (-4.9,7.6)
ME 3749 3800 0 (-200,100) 86 0 (-60,50) -1.6 (-12.9,10.5)
HI 2821 2900 -100 (-200,100) 15 0 (-30,40) -5.2 (-15.5,4.4)
SD 1961 2000 -100 (-200,0) 52 60 (20,90) -7.7 (-21.3,4.7)
AK 932 1000 -100 (-200,0) 9 -10 (-30,0) -10.8 (-23,-0.4)
WV 4446 5100 -600 (-800,-500) 94 70 (20,120) -36.2 (-46.7,-26)
WY 830 1200 -300 (-400,-300) 16 20 (0,40) -58.4 (-76.2,-43.5)
ND 1350 1900 -500 (-700,-400) 36 70 (50,100) -71.9 (-87.4,-55.9)

Observed weekly death rate due to any cause vs seasonal baseline (+/-95% Prediction Interval)

The black line shows the observed proportion of deaths that were due to any cause per week. The red line and shaded area represent the 95% Prediction Interval. The latest data are for the week ending 2020-05-30.

Time series of observed and expected prior to March 2020

This is to demonstrate the fit of the model in the pre-pandemic period. Note: the axes do not extend to 0 because these plots are meant to highlight the fit of the model

Observed weekly death rate due to pneumonia/influenza/covid-19 vs seasonal baseline for pneumonia&influenza (+/-95% Prediction Interval)

The black line shows the observed proportion of deaths that were due to Pneumonia & Influenza & Covid-19 (P&I&C) per week. The red line and shaded area represent the 95% Prediction Interval. The latest P&I&C data is for the week ending 2020-05-30.

Reported number of COVID-19 deaths,compared with the excess deaths due to all causes in each week, by state.

Trends in excess mortality due to all causes (red solid line +/- 95% prediction intervals) or reported deaths due to COVID-19 (blue dotted line) for March 1, 2020 through May 9, 2020. The thick dashed gray line shows the volume of tests performed per 1000 people in that week.

Same thing, for additional states

NCHS vs COVIDtracking.com data

Compare Excess all-cause mortality vs Excess ILI

Here we compare the observed vs expected number of deaths due to all causes in each week compared to the observed vs expected number of outpatient visits for influenza-like illness (ILI) in each week. we would expect ILI (blue line) to increase earlier than deaths (red line)

Evaluation of reporting delays with provisional data

Comparison of excess deaths week 10->Two weeks ago, compared with earlier time period pre-covid

There were 1.56810^{5} more deaths reported in provisional statistics for 2020 than 2019 since March 1. Adjusting for variations in deaths between the two years, there were 1.4810^{5} more deaths in 2020.

Sensitivity analysis: do not Adjust for flu

Comparison of excess deaths due to any cause, as estimated from a regression that adjusts for influenza with a regression that does not, with or without adjustments for reporting delays
State Excess deaths, adjusted for influenza Excess deaths, unadjusted for influenza
1 AK -100 (-200,0) -70 (-160,0)
2 AL 900 (600,1200) 710 (420,1000)
3 AR 100 (-200,300) 80 (-170,300)
4 AZ 1800 (1400,2100) 1250 (900,1570)
5 CA 6800 (6100,7500) 6190 (5500,6850)
6 CO 1900 (1600,2100) 1730 (1450,1980)
7 DC 500 (400,600) 490 (400,590)
8 DE 400 (300,600) 440 (320,550)
9 FL 3500 (2900,4100) 2770 (2140,3370)
10 GA 2200 (1800,2600) 1960 (1580,2310)
11 HI -100 (-200,100) -80 (-230,50)
12 IA 100 (-100,400) 150 (-90,380)
13 ID 0 (-100,200) 10 (-160,160)
14 IL 7500 (7100,7900) 7370 (6910,7760)
15 IN 2100 (1800,2500) 2120 (1780,2440)
16 KS 0 (-200,200) 10 (-200,220)
17 KY 100 (-200,300) -40 (-330,220)
18 LA 3000 (2700,3300) 2680 (2410,2940)
19 MA 7400 (7100,7700) 7160 (6830,7490)
20 MD 3700 (3500,4000) 3680 (3380,3960)
21 ME 0 (-200,100) -70 (-230,90)
22 MI 6100 (5700,6600) 5950 (5540,6330)
23 MN 900 (600,1200) 790 (500,1050)
24 MO 700 (400,1000) 660 (320,960)
25 MS 1100 (900,1400) 1120 (880,1340)
26 MT 0 (-100,100) 10 (-120,130)
27 ND -500 (-700,-400) -560 (-680,-450)
28 NE 100 (0,300) 130 (-40,290)
29 NH 300 (100,400) 270 (100,400)
30 NJ 16200 (15800,16500) 15950 (15600,16300)
31 NM 100 (-100,300) 50 (-140,220)
32 NV 200 (0,400) 200 (-20,410)
33 NY 12300 (11900,12700) 11990 (11580,12400)
34 NYC 25100 (24800,25400) 25000 (24700,25280)
35 OH 1600 (1200,2000) 1520 (1060,1940)
36 OK 400 (100,600) 370 (110,600)
37 OR 300 (0,500) 240 (-10,480)
38 PA 5400 (4900,5900) 5470 (4980,5910)
39 RI 500 (400,700) 550 (410,660)
40 SC 1400 (1100,1700) 1330 (1020,1600)
41 SD -100 (-200,0) -80 (-200,30)
42 TN 800 (400,1100) 440 (90,790)
43 TX 3600 (3000,4200) 2170 (1590,2750)
44 UT 100 (-100,300) 100 (-90,280)
45 VA 2300 (1900,2600) 2070 (1740,2410)
46 VT 200 (100,300) 190 (80,280)
47 WA 1300 (1000,1700) 1160 (860,1470)
48 WI 600 (300,900) 490 (170,790)
49 WV -600 (-800,-500) -640 (-830,-470)
50 WY -300 (-400,-300) -330 (-410,-250)

Empirical baseline

#> [1] 109490.2
#> [1] 122335.5