Measles: GBD 2017

Disease Description

Measles is a highly contagious, serious disease caused by the measles virus (Measles morbillivirus). Symptoms usually develop 10-12 days after exposure to the virus, and last 7-10 days. Symptoms include fever, cough, runny nose, conjunctivitis, characteristic white spots inside the cheek (called Koplik’s spots), and a red, flat, blotchy skin rash that develops on average 14 days after exposure to the virus (range, 7-21 days) and lasts 5-6 days. Recovery from measles confers lifelong immunity. [WHO-Measles], [CDC-Measles], [Wikipedia-Measles], [GBD-2017-YLD-Capstone-Appendix-1-Measles]

Most measles-related deaths are caused by complications associated with the disease. The most serious complications include blindness, encephalitis, severe diarrhea, ear infections, and pneumonia. Serious complications are more common in children under the age of 5 or adults over the age of 30, especially those with vitamin A deficiency or those whose immune systems have been weakened by HIV/AIDS or other diseases [WHO-Measles].

Measles is spread by coughing and sneezing, close personal contact, or direct contact with infected nasal or throat secretions. The virus remains active and contagious in the air or on infected surfaces for up to 2 hours. It can be transmitted by an infected person from 4 days prior to the onset of the rash to 4 days after the rash erupts [WHO-Measles], [CDC-Measles], [Wikipedia-Measles].

Before the introduction of a measles vaccine in 1963 and widespread vaccination, major epidemics occurred approximately every 2–3 years, and measles caused an estimated 2.6 million deaths each year. Despite the availability of a vaccine, approximately 110,000 people died from measles in 2017, mostly children under the age of 5 years. However, due to accelerated immunization activities, global measles deaths have decreased 80% during the period 2000–2017, from an estimated 545,000 to 110,000, and measles vaccination prevented an estimated 21.1 million deaths during 2000–2017 [WHO-Measles].

The ICD 10 codes for measles are B05-B05.9, Z24.4, and ICD 9 codes are 055-055.9, 484.0, V04.2, V73.2 [GBD-2017-YLD-Capstone-Appendix-1-Measles].

Todo

Add data about global vaccine coverage and efficacy. Perhaps start with these references:

Also perhaps note this recent New York Times article (Oct 31, 2019):

Measles Makes Your Immune System’s Memory Forget Defenses Against Other Illnesses: New research shows the virus can have devastating effects on the immune system that persist much longer than the illness itself.

Modeling Measles in GBD 2017

Todo

Add relevant detail about measles modeling process from [GBD-2017-YLD-Capstone-Appendix-1-Measles] and from the CoD Capstone Appendix. Note that each country’s vaccine coverage went into the estimation of measles incidence rates, which are then multiplied by an average disease duration of 10 days to compute prevalence.

Describe enough of the data sources and modeling process to verify that even though measles can lead to diarrhea or other causes that we include in our Vivarium models, we won’t be double counting mortality and morbidity from these causes. For example, a death caused by diarrheal dehydration due to measles should be counted in the GBD as a death due to measles, not as a death due to diarrheal diseases.

The relationship with vitamin A deficiency may also be important for our models.

Make sure to check on measles sequelae as well. Our models so far have not paid much attention to the nonfatal side, but it looks like some of the complications can persist well after someone recovers from measles, so maybe that’s important to think about.

GBD Hierarchy

Hierarchy Diagram:

Measles GBD hierarchy diagram

Cause Model Diagram

Simple SIR Measles cause model diagram

Model Assumptions and Limitations

This model is designed to be used for estimating DALYs due to measles that are averted from a country-level intervention (e.g. food fortification or supplementation given to a percentage of the population) that can reduce measles incidence as a downstream effect.

In particular, there are various uses for which this model is not suitable. For example:

1. The simple measles model described here does not explicitly incorporate vaccine coverage or efficacy, hence cannot be used to model the impact of a vaccination campaign.

2. This model uses country-level data, and cannot be used to model local measles outbreaks due to lack of vaccination in small communities.

Some of the assumptions made in this model are:

1. There is no data available for population in recovered state in GBD. So, we considered all the population who do not have measles as susceptible This includes both susceptible and recovered population. To compensate this, the incidence rate among susceptible and recovered population is used for transition.

2. There is no data avaialable for remission rate in GBD. So a constant remission rate is calculated from average case duration assumption of 10 days [GBD-2017-YLD-Capstone-Appendix-1-Measles].

Restrictions

Restriction Type

Value

Notes

Male only

False

Female only

False

YLL only

False

YLD only

False

YLL age group start

Post Neonatal

GBD age group id is 4

YLL age group end

50 to 54 years

GBD age group id is 15

YLD age group start

Post Neonatal

GBD age group id is 4

YLD age group end

50 to 54 years

GBD age group id is 15

Todo

Describe more limitations and assumptions of the model as appropriate. For example,

  • There are 2 ways people can be in the “recovered” state - either they get measles and then recover, or they get vaccinated and move directly into the “recovered” state without ever having the disease. We should look into measles vaccination rates in the countries we’re interested in (Nigeria, India, Ethiopia) and compare this to the number of people who actually get measles. If the number of vaccinated people is much higher than the number who get the disease, then our assumption will have a smaller effect, because the few people who enter the recovered state in our model will be be a small proportion of the total number of people in the recovered state, and the GBD incidence rate is already accounting for people who are “recovered” by vaccination.

  • We should also look at the case fatality rate / excess mortality rate for measles, as this will also have an impact on the effect of this assumption, as well as on our assumption of a constant remission rate.

  • There are ways we could try to estimate the people who are in the recovered state, but it is probably not worth the effort and added complexity for this model, particularly because we are not explicitly modeling vaccinations.

  • For our assumption of a constant remission rate (below), we should think about what the actual hazard function for remission should look like (we should be able to get some idea about this from the disease description), and estimate how replacing it with a constant rate will affect our results.

  • Also include about GBD’s assumption of 50% of measles cases as moderate and other 50% as severe.

Data Description

Definitions

State

State Name

Definition

S

Susceptible

Susceptible to measles

I

Infected

Infected with measles

R

Recovered

Recovered from measles
States Data

State

Measure

Value

Notes

S

prevalence

1-prevalence_c341

S

excess mortality rate

0

S

disabilty weights

0

I

prevalence

prevalence_c341

I

excess mortality rate

\(\frac{\text{deaths\_c341}}{\text{population} \times \text{prevalence\_c341}}\)

I

disability weights

disability_weight_s117 \(\times\) prevalence_s117+ disability_weight_s118 \(\times\) prevalence_s118

GBD assumes 50% of measles cases as severe and other 50% as moderate [GBD-2017-YLD-Capstone-Appendix-1-Measles].

R

prevalence

0

Clearly room for improvement. There is no data available for the number of recovered people in GBD. So we included recovered among susceptible during initialization and calculating incident rate. This is done to simplify the model as the focus is on LSFF but not on measles.

R

excess mortality rate

0

R

disabilty weights

0

ALL

cause specific mortality rate

\(\frac{\text{deaths\_c341}}{\text{population}}\)
Transition Data

Transition

Source

Sink

Value

Notes

i

S

I

\(\frac{\text{incidence\_rate\_c341}}{\text{1 - prevalence\_c341}}\)

Incidence rate in total population is divided by 1-prevalence_c341 to get incidence rate among the recovered and susceptible population.

r

I

R

remission_rate_c341 \(= \frac{\text{365 person-days}}{\text{10 person-days} \times \text{1 year}}\) \(= \frac{\text{36.5}}{\text{year}}\)

GBD assumes average case duration as 10 days [GBD-2017-YLD-Capstone-Appendix-1-Measles]. So constant remission rate is approximated to this calculation.
Data Sources

Measure

Sources

Description

Notes

prevalence_c341

como

Prevalence of cause measles

deaths_c341

codcorrect

Deaths from measles

population

demography

Mid-year population for given country

incidence_rate_c341

como

Incidence rate for measles

remission_rate_c341

YLD appendix

Remission rate for measles

GBD assumes average case duration as 10 days [GBD-2017-YLD-Capstone-Appendix-1-Measles]. So constant remission rate is calculated from this assumption.

disability_weight_s{sid}

YLD appendix

Disability weights associated with each sequelae

prevalence_s{sid}

como

Prevalence of each sequelae

Validation Criteria

Todo

Describe tests for model validation.

References

[WHO-Measles] (1,2,3,4)

Measles Fact Sheet. World Health Organization, 9 May 2019. Retrieved 13 Nov 2019. https://www.who.int/news-room/fact-sheets/detail/measles

[CDC-Measles] (1,2)

Chapter 13: Measles. Epidemiology and Prevention of Vaccine-Preventable Diseases (The Pink Book, 13th Edition). Centers for Disease Control and Prevention, 2015. Retrieved 13 Nov 2019. https://www.cdc.gov/vaccines/pubs/pinkbook/meas.html

[Wikipedia-Measles] (1,2)

Measles. From Wikipedia, the Free Encyclopedia. Retrieved 13 Nov 2019. https://en.wikipedia.org/wiki/Measles