Momtezuma Tuatara
12-01-09, 07:37 PM
Sent: Monday, July 19, 1999 5:32 PM
To: nipinfo@cdc.gov (nipinfo@cdc.gov)
Subject: CDC Measles Statistics
(To Whom It May Concern :The calculations in this mail seem to be correct. Could someone please explain the discrepancy?)
According to the CDC - www.cdc.gov/nip/vacsafe/va (http://www.cdc.gov/nip/vacsafe/va)
"MEASLES - Before measles immunizations were available, nearly everyone in the U.S. got measles. There were approximately 3 to 4 million measles cases each year. An average of 450 measles-associated deaths were reported each year between 1953 and 1963.
In industrialized countries, up to 20% of persons with measles are hospitalized, and 7% to 9% suffer from complications such as pneumonia, diarrhea, or ear infections.
Although less common, some persons with measles develop encephalitis, resulting in brain damage. It is estimated that as many as 1 of every 1,000 persons with measles will die."
Now, if there were approximately 3 to 4 million measles cases each year and an average of 450 measles-associated deaths were reported each year between 1953 and 1963... that means, if my maths is correct, approx 0.01285% of people infected with measles died, that is, one in 7777 cases... so how do they arrive at "1 of every 1000 persons with measles will die"?
If there is an average of 3.5 million cases of measles per year.. and 1 in 1000 ie: 0.1% of people are expected to die.. then the CDC predict 3500 to die each year... There is a BIG difference between past experience ie. 450 people dying every year and a predicted 3500...
Answer:
Thank you for your question concerning measles death rates. I can try to provide an explanation of the seemingly contradictory data you ask about.
Let's start with the raw numbers. Before 1963, approximately 500,000 cases and 500 deaths were REPORTED annually (with epidemic cycles every 2-3 years). The ratio of REPORTED deaths to REPORTED cases was approximately 1:1000.
You can do the calculations yourself. Here are the data for reported> measles in the U.S. from 1953 to 1963 (source: Summary of Reportable> Diseases, United States, Morbidity and Mortality Weekly Report, various issues):
Year # cases # deaths rate
1953 449146 462 (1.0 per 1000)
1954 682720 518 (0.8 per 1000)
1955 555156 345 (0.6 per 1000)
1956 611936 530 (0.9 per 1000)
1957 486799 389 (0.8 per 1000)
1958 763094 552 (0.7 per 1000)
1959 406162 385 (0.9 per 1000)
1960 441703 380 (0.9 per 1000)
1961 423919 434 (1.0 per 1000)
1962 481530 408 (0.8 per 1000)
1963 385156 364 (0.9 per 1000)
The average annual rate of reported deaths to reported cases during this time period (1953-1963) was 0.85 deaths per 1000 cases.
It is assumed that cases were severely under-reported - the ACTUAL number of cases (reported and non-reported) in the pre-vaccine era is estimated to have been 3-4 million annually, significantly higher than the reported number of 500,000 cases.
The ACTUAL number of deaths is also estimated to have been higher than the reported 500. Exactly how much higher is an complicated issue. We generally assume that surveillance of measles-related severe events, such as deaths, is likely to be better reported than sicknesses that don't end in death. However, it is possible that many deaths due to measles were not reported as such. They may have been reported as death from pneumonia, which is often a complication of measles. If we assume that the ratio of deaths due to measles among unreported cases was the same as the ratio among reported cases, then your figure of 3500 might be an accurate guess at the ACTUAL number of deaths in these years. For this reason, we can use only REPORTED data (cases and deaths) to estimate mortality rates (in this case, about 1 reported death per 1000 reported cases).
An additional complicating factor is that death from measles is quite age-specific. We know that infants and children less than 5 years of age, and adults 20 and over, are at much higher risk of death than school-age children. So the number of deaths is a function of the age group that is most affected. During the prevaccination years, measles was mostly a disease of school-aged children, which would tend to lower the number of total deaths, and the death rate. Most adults were immune from having survived measles as a child. During 1989-1991, on the other hand, measles mostly affected preschool-aged children, and the death rate increased considerably, as you can see below.
Here are the actual data on cases and deaths from measles in the United States for 1990 through 1998 (source: Summary of Reportable Diseases,
United States, Morbidity and Mortality Weekly Report, various issues):
Year # cases # deaths rate
1990 27786 64 (2.3 per 1000)
1991 9643 27 (2.8 per 1000)
1992 2237 4 (1.8 per 1000)
1993 312 0 (0.0 per 1000)
1994 963 0 (0.0 per 1000)
1995 309 2 (6.5 per 1000)
1996 508 1 (2.0 per 1000)
1997 138 0 (0.0 per 1000)
1998 89 0 (0.0 per 1000)
The average annual rate of reported deaths to reported cases during this time period (1990-1998) was 1.7 deaths per 1000 cases.
In recent years, surveillance has been much better, and it is believed that REPORTED numbers of cases and deaths more accurately reflect ACTUAL numbers of cases and deaths.
In any case, you can see that the rate (at least among reported cases) has remained more or less the same. 1 to 2 reported deaths occur per 1000 reported cases. It is statistically reasonable to assume that this is also the rate among unreported cases, depending on the age group affected.
On this basis, we predict that about 1 person will die from measles, ON AVERAGE, for every 1000 persons with the disease.
Now, in addition to the numbers, there are a lot of other things to consider in determining number of cases and numbers of deaths and rates.
First, old data are likely to be way off. Reporting systems were simply not very good. The numbers given are the absolute minimums, and are probably very, very low.
Second, when denominators (# of cases) are higher, rates of death tend to be lower then when denominators (# of cases) become very small - this is because a single death in a population of 500 cases, for example, gives a very, very high death rate of 2 per 1000). As the size of the denominator decreases, it is very hard to find a rate with any confidence - it is hard to ensure that the rate obtained is applicable to the general population.
A single death in 1998 would have made the death rate 10%!
Third, the absolute numbers (because of the fact that reported data is likely to be inaccurate and also because the denominators are diminishing (over time) are generally meaningless. What is meaningful, in contrast, is the general trend. In this case, the trend that we see is that the numbers of cases and deaths due to measles have steadily and significantly decreased over time to almost nothing.
I hope this information is helpful to you.
William Atkinson, MD
Medical Epidemiologist
National Immunization Program
Centers for Disease Control and Prevention
To: nipinfo@cdc.gov (nipinfo@cdc.gov)
Subject: CDC Measles Statistics
(To Whom It May Concern :The calculations in this mail seem to be correct. Could someone please explain the discrepancy?)
According to the CDC - www.cdc.gov/nip/vacsafe/va (http://www.cdc.gov/nip/vacsafe/va)
"MEASLES - Before measles immunizations were available, nearly everyone in the U.S. got measles. There were approximately 3 to 4 million measles cases each year. An average of 450 measles-associated deaths were reported each year between 1953 and 1963.
In industrialized countries, up to 20% of persons with measles are hospitalized, and 7% to 9% suffer from complications such as pneumonia, diarrhea, or ear infections.
Although less common, some persons with measles develop encephalitis, resulting in brain damage. It is estimated that as many as 1 of every 1,000 persons with measles will die."
Now, if there were approximately 3 to 4 million measles cases each year and an average of 450 measles-associated deaths were reported each year between 1953 and 1963... that means, if my maths is correct, approx 0.01285% of people infected with measles died, that is, one in 7777 cases... so how do they arrive at "1 of every 1000 persons with measles will die"?
If there is an average of 3.5 million cases of measles per year.. and 1 in 1000 ie: 0.1% of people are expected to die.. then the CDC predict 3500 to die each year... There is a BIG difference between past experience ie. 450 people dying every year and a predicted 3500...
Answer:
Thank you for your question concerning measles death rates. I can try to provide an explanation of the seemingly contradictory data you ask about.
Let's start with the raw numbers. Before 1963, approximately 500,000 cases and 500 deaths were REPORTED annually (with epidemic cycles every 2-3 years). The ratio of REPORTED deaths to REPORTED cases was approximately 1:1000.
You can do the calculations yourself. Here are the data for reported> measles in the U.S. from 1953 to 1963 (source: Summary of Reportable> Diseases, United States, Morbidity and Mortality Weekly Report, various issues):
Year # cases # deaths rate
1953 449146 462 (1.0 per 1000)
1954 682720 518 (0.8 per 1000)
1955 555156 345 (0.6 per 1000)
1956 611936 530 (0.9 per 1000)
1957 486799 389 (0.8 per 1000)
1958 763094 552 (0.7 per 1000)
1959 406162 385 (0.9 per 1000)
1960 441703 380 (0.9 per 1000)
1961 423919 434 (1.0 per 1000)
1962 481530 408 (0.8 per 1000)
1963 385156 364 (0.9 per 1000)
The average annual rate of reported deaths to reported cases during this time period (1953-1963) was 0.85 deaths per 1000 cases.
It is assumed that cases were severely under-reported - the ACTUAL number of cases (reported and non-reported) in the pre-vaccine era is estimated to have been 3-4 million annually, significantly higher than the reported number of 500,000 cases.
The ACTUAL number of deaths is also estimated to have been higher than the reported 500. Exactly how much higher is an complicated issue. We generally assume that surveillance of measles-related severe events, such as deaths, is likely to be better reported than sicknesses that don't end in death. However, it is possible that many deaths due to measles were not reported as such. They may have been reported as death from pneumonia, which is often a complication of measles. If we assume that the ratio of deaths due to measles among unreported cases was the same as the ratio among reported cases, then your figure of 3500 might be an accurate guess at the ACTUAL number of deaths in these years. For this reason, we can use only REPORTED data (cases and deaths) to estimate mortality rates (in this case, about 1 reported death per 1000 reported cases).
An additional complicating factor is that death from measles is quite age-specific. We know that infants and children less than 5 years of age, and adults 20 and over, are at much higher risk of death than school-age children. So the number of deaths is a function of the age group that is most affected. During the prevaccination years, measles was mostly a disease of school-aged children, which would tend to lower the number of total deaths, and the death rate. Most adults were immune from having survived measles as a child. During 1989-1991, on the other hand, measles mostly affected preschool-aged children, and the death rate increased considerably, as you can see below.
Here are the actual data on cases and deaths from measles in the United States for 1990 through 1998 (source: Summary of Reportable Diseases,
United States, Morbidity and Mortality Weekly Report, various issues):
Year # cases # deaths rate
1990 27786 64 (2.3 per 1000)
1991 9643 27 (2.8 per 1000)
1992 2237 4 (1.8 per 1000)
1993 312 0 (0.0 per 1000)
1994 963 0 (0.0 per 1000)
1995 309 2 (6.5 per 1000)
1996 508 1 (2.0 per 1000)
1997 138 0 (0.0 per 1000)
1998 89 0 (0.0 per 1000)
The average annual rate of reported deaths to reported cases during this time period (1990-1998) was 1.7 deaths per 1000 cases.
In recent years, surveillance has been much better, and it is believed that REPORTED numbers of cases and deaths more accurately reflect ACTUAL numbers of cases and deaths.
In any case, you can see that the rate (at least among reported cases) has remained more or less the same. 1 to 2 reported deaths occur per 1000 reported cases. It is statistically reasonable to assume that this is also the rate among unreported cases, depending on the age group affected.
On this basis, we predict that about 1 person will die from measles, ON AVERAGE, for every 1000 persons with the disease.
Now, in addition to the numbers, there are a lot of other things to consider in determining number of cases and numbers of deaths and rates.
First, old data are likely to be way off. Reporting systems were simply not very good. The numbers given are the absolute minimums, and are probably very, very low.
Second, when denominators (# of cases) are higher, rates of death tend to be lower then when denominators (# of cases) become very small - this is because a single death in a population of 500 cases, for example, gives a very, very high death rate of 2 per 1000). As the size of the denominator decreases, it is very hard to find a rate with any confidence - it is hard to ensure that the rate obtained is applicable to the general population.
A single death in 1998 would have made the death rate 10%!
Third, the absolute numbers (because of the fact that reported data is likely to be inaccurate and also because the denominators are diminishing (over time) are generally meaningless. What is meaningful, in contrast, is the general trend. In this case, the trend that we see is that the numbers of cases and deaths due to measles have steadily and significantly decreased over time to almost nothing.
I hope this information is helpful to you.
William Atkinson, MD
Medical Epidemiologist
National Immunization Program
Centers for Disease Control and Prevention