Abstract

Mortality data on deaths due to pneumonia & influenza through the week ending 2020-03-28

Background Tracking the severity and public health impact of emerging diseases, such as the novel coronavirus diseases COVID-19, is a critical need. These efforts are often hampered by testing issues and reporting lags for key epidemiological data. Evaluating unexplained increases in deaths attributed to non-specific causes, such as pneumonia, can provide a more complete picture of the burden caused by COVID-19.

Methods We evaluated increases in the occurrence of deaths due to P&I above a seasonal baseline across the United States and compared the rate of excess deaths to reported rates of death due to COVID-19, as well as state-level coronavirus testing rates and excess rates of influenza-like illness.

Results There were notable increases in the rate of death due to pneumonia and influenza. In a number of states, these increases pre-dated the increase in testing rates and went uncounted. There was substantial variability between states in the discrepancy between reported rates of death due to COVID-19 and the estimated burden of excess deaths due to P&I. In some states, the burden of reported deaths was 10-fold lower than the burden of excess deaths due to P&I.

Conclusions Given the lack of consistent virological testing, tracking spikes in deaths due to non-specific causes provides the most complete estimate of the burden of COVID-19. This approach provides a framework for tracking the progression of the epidemic and its public health impact.

Figure 1: Observed weekly death rate vs seasonal baseline (+/-95% Prediction Interval)

The black line shows the observed proportion of deaths that were due to Pneumonia & Influenza (P&I) per week. The red line and shaded area represent the 95% Prediction Interval. The latest P&I data is for the week ending 2020-03-28.

Figure 2: Observed deaths/expected deaths by state

Excess Deaths due to pneumonia and influenza

These plots show the Observed/Expected number of deaths due to pneumonia and influenza in each week for the 2019-20 year (red) compared to previous years (gray). Values close to 1 indicate that the values for that week are close to what would be expected based on the time of year and influenza activity.

Figure 3: Reported number of COVID-19 deaths,compared with the excess deaths due to pneumonia and influenza in each week, by state.

The blue line shows the number of excess P&I cases, the red solid line shows the reported number of COVID-19 deaths for the same week (as compiled by the Johns Hopkins CSEE), and the dotted red line shows the reported COVID-19 deaths for weeks in which the CDC data were not yet available. Excess P&I estimates from recent weeks are adjusted for incomplete records using the “percent_complete” variable.

Table 1: Excess P&I deaths and reported deaths due to COVID-19

Observed and Excess deaths due to pneumonia & influenza, and COVID-19, from February 9, 2020 through Mar 28, 2020
State Total P&I deaths Excess P&I deaths Reported COVID-19 deaths Excess P&I deaths/1,000,000 people
NY 2654 620(526, 714) 728 31.9 (27.1, 36.7)
WA 736 187(139, 235) 207 24.6 (18.2, 30.9)
NJ 860 206(154, 258) 140 23.2 (17.4, 29)
IL 1266 185(117, 253) 47 14.6 (9.2, 20)
PA 1237 182(115, 249) 34 14.2 (9, 19.4)
CA 3428 399(286, 512) 101 10.1 (7.2, 13)
ME,MA,NH,RI,VT 1258 114(44, 183) 59 10.1 (3.9, 16.3)
IA,KS,MO,NE 1498 124(47, 200) 20 8.7 (3.3, 14.2)
CO,MT,ND,SD,UT,WY 953 105(45, 165) 36 8.6 (3.7, 13.5)
FL 2039 180(93, 267) 54 8.4 (4.3, 12.4)
IN,MN,WI 1795 150(65, 235) 49 8.2 (3.6, 12.9)
AK,ID,OR 466 51(10, 93) 19 7.6 (1.4, 13.8)
AR,LA,NM,OK 1402 95(21, 169) 158 6.9 (1.5, 12.3)
GA 677 68(18, 118) 69 6.4 (1.7, 11.2)
TX 2105 176(86, 266) 27 6.1 (3, 9.2)
AZ,HI,NV 994 62(0, 125) 25 5.3 (0, 10.6)
DE,MD,VA 1188 67(-3, 136) 25 4.3 (-0.2, 8.7)
OH 973 44(-19, 106) 25 3.7 (-1.6, 9.1)
MI 979 37(-27, 101) 92 3.7 (-2.7, 10.1)
AL,KY,MS,SC,TN 2661 47(-58, 152) 43 1.9 (-2.4, 6.2)

Figure 4: Compare Excess P&I mortality vs Excess ILI

Here we compare the observed vs expected number of deaths due to pneumonia and influenza in each week compare 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)

Table S1: Excess deaths by week across the US, compared with provisional death counts from NCHS

Note excess P&I deaths and covidtracking.com numbers do not include Connecticut, North Carolina, and West Virginia
Table S1: Comparison of data sources: Observed and Excess deaths due to pneumonia & influenza, and COVID-19, from February 9, 2020 through Mar 28, 2020
Week ending Total P&I deaths Excess P&I deaths NCHS Reported COVID-19 Deaths, with pneumonia code NCHS Reported COVID-19 Deaths All reported COVID-19 deaths (covidtracking.com)
2020-02-15 3971 -71(-202, 60) 0 0 0
2020-02-22 3867 -132(-262, -2) 0 0 0
2020-02-29 3998 67(-62, 196) 3 5 1
2020-03-07 4092 270(143, 397) 11 19 22
2020-03-14 4046 455(332, 578) 22 44 36
2020-03-21 4317 887(767, 1007) 200 435 229
2020-03-28 4877 1622(1505, 1739) 928 2034 1670 
#> [1] "3098(2766, 3430)"
#> [1] 1958
#> [1] 2537

Figure S1: Excess all-cause deaths

Now instead of modeling the proportion of deaths that were due to pneumonia and influenza, we will model all-caused deaths, using the proportion under-reporting as the denominator

The black line shows the observed number of all deaths per week, regardless of cause. The red line and shaded area represent the 95% Prediction Interval. The latest data is for the week ending 2020-03-28. Note that these are adjusted for percent completeness of the data. There are clear jumps in all-cause mortality in NY and NJ, other states less clear, probably due to reporting delays

Figure S2: Excess all-cause deaths for NY state (incl NYC) vs Excess deaths due to pneumonia and influenza

Fold difference between excess all-cause mortality and excess P&I mortality in NJ and NY. Shows that excess P&I accounts for ~50-60% of the excess deaths

#> [1] 0.5667752
#> [1] 0.352065

Figure S3: NYC only data P&I vs all-cause excess deaths

Ratio of all-cause vs P&I deaths in NYC. In NYC, only 25% of excess deaths were coded as P&I

#> [1] 0.237500 0.283148
#> [1] 1409

Table S2: Proportion of COVID-19 deaths with a pneumonia code, by state, through most recent date

Note these values will be greater than those in other tables, which are 2 weeks behind

State NCHS Reported COVID-19 Deaths NCHS Reported COVID-19 Deaths, with pneumonia code Proportion with pneumonia
25 Total US 6930 3161 0.46
26 Alabama 17 6 0.35
27 Alaska 1 1 1
28 Arizona 26 13 0.5
29 Arkansas 10 2 0.2
30 California 187 102 0.55
31 Colorado 107 60 0.56
33 Delaware 2 2 1
34 District of Columbia 5 5 1
35 Florida 157 93 0.59
36 Georgia 74 42 0.57
37 Hawaii 1 0 0
38 Idaho 13 4 0.31
39 Illinois 77 47 0.61
40 Indiana 43 19 0.44
41 Iowa 12 2 0.17
42 Kansas 9 3 0.33
43 Kentucky 6 3 0.5
44 Louisiana 154 64 0.42
45 Maine 10 5 0.5
46 Maryland 75 42 0.56
47 Massachusetts 172 85 0.49
48 Michigan 309 142 0.46
49 Minnesota 17 6 0.35
50 Mississippi 34 17 0.5
51 Missouri 22 9 0.41
52 Montana 3 1 0.33
53 Nebraska 3 2 0.67
54 Nevada 27 21 0.78
55 New Hampshire 15 8 0.53
56 New Jersey 593 303 0.51
57 New Mexico 1 0 0
58 New York 1130 650 0.58
59 New York City 2862 1065 0.37
61 North Dakota 1 0 0
62 Ohio 8 4 0.5
63 Oklahoma 17 5 0.29
64 Oregon 21 13 0.62
65 Pennsylvania 220 98 0.45
66 Rhode Island 1 1 1
67 South Carolina 36 13 0.36
68 South Dakota 4 1 0.25
69 Tennessee 25 11 0.44
70 Texas 47 18 0.38
71 Utah 8 3 0.38
72 Vermont 17 6 0.35
73 Virginia 39 15 0.38
74 Washington 278 142 0.51
76 Wisconsin 34 7 0.21
78 Puerto Rico 18 13 0.72

Table S3: Do not adjust for flu

Observed and Excess deaths due to pneumonia & influenza, and COVID-19, from February 9, 2020 through Mar 28, 2020
State Excess P&I deaths (unadjusted) Excess P&I deaths (adjusted) Reported COVID-19 deaths
NY 620(526, 714) 620(526, 714) 728
CA 393(280, 506) 399(287, 512) 101
NJ 206(154, 258) 206(154, 258) 140
WA 187(139, 235) 187(139, 235) 207
IL 185(117, 253) 185(117, 253) 47
FL 180(93, 267) 180(93, 267) 54
TX 159(71, 247) 176(88, 264) 27
PA 156(96, 216) 182(122, 242) 34
IN,MN,WI 146(62, 230) 150(66, 234) 49
ME,MA,NH,RI,VT 110(40, 180) 114(44, 183) 59
IA,KS,MO,NE 110(35, 185) 124(48, 199) 20
CO,MT,ND,SD,UT,WY 101(41, 161) 105(46, 165) 36
AR,LA,NM,OK 74(3, 145) 95(24, 166) 158
DE,MD,VA 65(-4, 134) 67(-2, 136) 25
AZ,HI,NV 61(-1, 123) 62(0, 125) 25
GA 47(-2, 96) 68(20, 117) 69
AK,ID,OR 45(5, 85) 51(11, 92) 19
AL,KY,MS,SC,TN 38(-65, 141) 47(-57, 150) 43
MI 37(-27, 101) 37(-27, 101) 92
OH 34(-23, 91) 44(-14, 101) 25

Table S4: do not adjust for delayed reporting

Excess deaths due to pneumonia & influenza from February 9, 2020 through Mar 28, 2020 with or without adjustment for delayed reporting
state Excess P&I (unadjusted) Excess P&I (adjusted for reporting delay)
NY 514(420, 608) 620(526, 714)
CA 370(257, 482) 399(287, 512)
NJ 206(154, 258) 206(154, 258)
WA 156(108, 204) 187(139, 235)
IL 165(97, 233) 185(117, 253)
PA 75(15, 135) 182(122, 242)
FL 179(92, 266) 180(93, 267)
TX 88(0, 177) 176(88, 264)
IN,MN,WI 101(17, 185) 150(66, 234)
IA,KS,MO,NE 62(-13, 137) 124(48, 199)
ME,MA,NH,RI,VT 61(-9, 130) 114(44, 183)
CO,MT,ND,SD,UT,WY 53(-7, 113) 105(46, 165)
AR,LA,NM,OK 67(-5, 138) 95(24, 166)
GA 40(-9, 88) 68(20, 117)
DE,MD,VA 39(-30, 108) 67(-2, 136)
AZ,HI,NV 46(-16, 109) 62(0, 125)
AK,ID,OR 17(-23, 58) 51(11, 92)
AL,KY,MS,SC,TN 42(-61, 145) 47(-57, 150)
OH 13(-44, 70) 44(-14, 101)
MI 46(-18, 110) 37(-27, 101)