04_metrics-components.Rmd

# Metrics Components

## Introduction {-}
- Last week: Methods and Calculations
- This week: Metrics Components
- Specifically: life expectancy, mortality rates, incidence and prevalence and disability weights

## Cause-specific or Population-wide {-}
- The metric components are the building blocks for analyses
- We need to tailor and select them to meet the needs of our analysis
- There are different variations of the metrics
  - Cause-specific => looking at specific disease and intervention
  - Population-wide => health policy changes
  - Many factors to include (age, sex, disease, location, point in time vs cumulative)

## Life tables and Life expectancy (Motivation) {-}
- We need life expectancy to calculate YLLs
- We use life tables to calculate life expectancy
- We use expected life remaining at age *x* (*ex*) rather than life expectancy from birth (*le*)

## Life tables (Example data) {-}
- Global Health Observatory Life Tables are included in the hmsidwR package

```{r, message = FALSE}
library(tidyverse)
library(hmsidwR)
library(knitr)
gho_2000 <- gho_lifetables %>%
  filter(sex == "both", year == 2000 ) %>% 
  select(-sex,-year)

gho_2000_wide <- gho_2000 %>% 
  pivot_wider(names_from = indicator, values_from = value) %>% 
  select(age,lx,nLx,Tx,ex) 

kable(gho_2000_wide)
```

## Life expectancy calculation {-}
- ex = Tx/lx
- Tx is the total number of person-years lived by the cohort from age x until all members of the cohort have died (look back at previous slide)
- That is sum(nLx) from the row for age x to the last row in the table
- For the exercise, Tx = 2,370,099 and lx = 89,867. So, ex = 26.4 years

## Mortality level and rates {-}
- Crude Mortality Rate = (Number of deaths/Total population)*100000
- Other types include: age-specific mortality rate, cause-specific mortality rate, infant mortality rate and maternal mortality rate
- We can use ranges of mortality rates to create simulations to estimate deaths

## Incidence and Prevalence {-}
- Incidence is the number of cases **developed** during a specific time period
- Prevalence is the number of cases **in the population** (new and existing) during a specific time period
- Prevalence is used in the calculation of DALYs to measure the current impact
- The first example calculates YLD using prevalence.
- The second example simulates incidence and prevalence using Poisson distributions to simulate the new cases and recovery

## Disability Weights and Severity Levels {-}
- Disability weights range from 0 to 1 based on severity 0 being no disability and 1 being equivalent to death
- There are different methods to compute or assign disability weights
- Global Burden of Disease study is one of the most thorough and extensive projects estimating disability weights
- These weights are available in hmsidwR
```{r, message = FALSE}
library(tidyverse)
library(hmsidwR)
library(knitr)
disability_weights <- hmsidwR::disweights

select_dws <- disability_weights %>%
  filter(year == 2019, sequela == 'Mild early syphilis infection') %>%
  select(sequela, dw, upper, lower) %>% 
  head()

kable(select_dws)
```

## Summary of DALYs' Components {-}
- YLLs Components: Age at death, Life expectancy, Standard life expectancy, Population size, Cause of death
- YLDs Components: Prevalence and Incidence, Disability weights, Age, Population size, and Duration of disability
- DALYs Components: Both YLLs and YLDs

## Meeting Videos {-}

### Cohort 1 {-}

`r knitr::include_url("https://www.youtube.com/embed/URL")`

<details>
<summary> Meeting chat log </summary>

```
LOG
```
</details>