Every state counts COVID deaths differently. A few report something as a COVID death if the dead person was COVID positive and COVID might have contributed to the death. So, they end up over-counting. Far more states err on the side of exclusion, by declining to report a COVID death unless it's very clear the death would not otherwise have occurred. So, they end up under-counting.
Because of that, I've instead relied on excess death figures. Before COVID, the CDC was very good at predicting how many people would die in a particular state over a particular time simply by projecting forward past mortality rates, so comparing their pre-pandemic projected deaths with the actual death counts, you can get a pretty good feel for how many COVID has killed in each state.
What you can also do with that data is figure out how much a particular state has been "overcounting" or "undercounting," simply by comparing the number of excess deaths in the state to the number of COVID deaths they've counted.
So, for instance, if the CDC predicted 5,000 people would die in a state during a given year, based on pre-pandemic mortality trends, and 6,000 actually died, then that's 1,000 excess deaths. If the state reported 500 COVID deaths, that suggests they probably were undercounting by half, and their "COVID Multiplier" would be 2 (1,000/500). If, on the other hand, they reported 2,000 COVID deaths, that suggests they were probably overcounting and their multiplier would be 0.5 (1,000/2,000).
Here's the multiplier for each state, for the twelve months ending the last week of April:
Vermont 3.30
Oregon 1.79
Alaska 1.63
Utah 1.50
West Virginia 1.49
Arizona 1.47
Washington 1.46
Delaware 1.41
New Mexico 1.39
Virginia 1.38
California 1.38
Maine 1.35
Montana 1.35
Colorado 1.34
Georgia 1.33
North Carolina 1.32
Arkansas 1.30
Louisiana 1.26
Mississippi 1.26
Alabama 1.26
Nevada 1.25
Idaho 1.24
Florida 1.24
Connecticut 1.24
Tennessee 1.21
Texas 1.21
Michigan 1.20
Illinois 1.20
United States 1.19
Kentucky 1.17
Wyoming 1.13
Oklahoma 1.13
South Carolina 1.09
Wisconsin 1.09
Kansas 1.08
Minnesota 1.03
Missouri 1.01
Ohio 0.97
New York 0.96
Pennsylvania 0.96
Iowa 0.94
Indiana 0.92
New Hampshire 0.86
Hawaii 0.79
Nebraska 0.79
Maryland 0.72
South Dakota 0.70
New Jersey 0.60
North Dakota 0.58
Rhode Island 0.54
Massachusetts 0.38
Note, that doesn't mean that Vermont had the most COVID deaths or Massachusetts the least. It's just that Vermont undercounted the most and Massachusetts overcounted the most. You would need to apply the multiplier to their claimed COVID death count to get the "adjusted COVID death count."
So, using that data, if Arizona announced that 100 people in their state died of COVID in a given week, the real number would probably be closer to 147. If Maryland announced 100 died of COVID in that week, the real number would likely be closer to 72. They could be reporting the same thing, and yet COVID could actually be killing twice as many people in Arizona, simply because AZ tends to decide gray-area calls against calling something a COVID death and Maryland tends to go the opposite direction and err on the side of including it.
On average, as you can see, the national trend has been to underreport, with a 1.19 multiplier. So, when the CDC says 1,000 Americans died of COVID in a week, it's probably more like 1,190.
Because of that, I've instead relied on excess death figures. Before COVID, the CDC was very good at predicting how many people would die in a particular state over a particular time simply by projecting forward past mortality rates, so comparing their pre-pandemic projected deaths with the actual death counts, you can get a pretty good feel for how many COVID has killed in each state.
What you can also do with that data is figure out how much a particular state has been "overcounting" or "undercounting," simply by comparing the number of excess deaths in the state to the number of COVID deaths they've counted.
So, for instance, if the CDC predicted 5,000 people would die in a state during a given year, based on pre-pandemic mortality trends, and 6,000 actually died, then that's 1,000 excess deaths. If the state reported 500 COVID deaths, that suggests they probably were undercounting by half, and their "COVID Multiplier" would be 2 (1,000/500). If, on the other hand, they reported 2,000 COVID deaths, that suggests they were probably overcounting and their multiplier would be 0.5 (1,000/2,000).
Here's the multiplier for each state, for the twelve months ending the last week of April:
Vermont 3.30
Oregon 1.79
Alaska 1.63
Utah 1.50
West Virginia 1.49
Arizona 1.47
Washington 1.46
Delaware 1.41
New Mexico 1.39
Virginia 1.38
California 1.38
Maine 1.35
Montana 1.35
Colorado 1.34
Georgia 1.33
North Carolina 1.32
Arkansas 1.30
Louisiana 1.26
Mississippi 1.26
Alabama 1.26
Nevada 1.25
Idaho 1.24
Florida 1.24
Connecticut 1.24
Tennessee 1.21
Texas 1.21
Michigan 1.20
Illinois 1.20
United States 1.19
Kentucky 1.17
Wyoming 1.13
Oklahoma 1.13
South Carolina 1.09
Wisconsin 1.09
Kansas 1.08
Minnesota 1.03
Missouri 1.01
Ohio 0.97
New York 0.96
Pennsylvania 0.96
Iowa 0.94
Indiana 0.92
New Hampshire 0.86
Hawaii 0.79
Nebraska 0.79
Maryland 0.72
South Dakota 0.70
New Jersey 0.60
North Dakota 0.58
Rhode Island 0.54
Massachusetts 0.38
Note, that doesn't mean that Vermont had the most COVID deaths or Massachusetts the least. It's just that Vermont undercounted the most and Massachusetts overcounted the most. You would need to apply the multiplier to their claimed COVID death count to get the "adjusted COVID death count."
So, using that data, if Arizona announced that 100 people in their state died of COVID in a given week, the real number would probably be closer to 147. If Maryland announced 100 died of COVID in that week, the real number would likely be closer to 72. They could be reporting the same thing, and yet COVID could actually be killing twice as many people in Arizona, simply because AZ tends to decide gray-area calls against calling something a COVID death and Maryland tends to go the opposite direction and err on the side of including it.
On average, as you can see, the national trend has been to underreport, with a 1.19 multiplier. So, when the CDC says 1,000 Americans died of COVID in a week, it's probably more like 1,190.