diff --git a/vignettes/internal-impact.Rmd b/vignettes/internal-impact.Rmd
index f38abfd..122483d 100644
--- a/vignettes/internal-impact.Rmd
+++ b/vignettes/internal-impact.Rmd
@@ -1,7 +1,7 @@
---
title: "Using vimpact for estimating vaccine impact - internal"
-author: "Rob Ashton"
-date: "2022-02-02"
+author: "Rob Ashton, Xiang Li, Katy Gaythorpe"
+date: "`r Sys.Date()`"
output: rmarkdown::html_vignette
vignette: >
%\VignetteIndexEntry{Using vimpact for estimating vaccine impact - internal}
diff --git a/vignettes/internal-impact.html b/vignettes/internal-impact.html
new file mode 100644
index 0000000..56ffa5f
--- /dev/null
+++ b/vignettes/internal-impact.html
@@ -0,0 +1,628 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Using vimpact for estimating vaccine impact - internal
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+Using vimpact for estimating vaccine impact
+- internal
+Rob Ashton, Xiang Li, Katy Gaythorpe
+2025-02-26
+
+
+
+
+This vignette describes how to use vimpact to calculate impact as a
+member of VIMC. This requires a connection to the montagu database so
+can only be used internally. Note that this is all in development and
+the interface is likely to change.
+N.B. For VIMC Rubella models, disease burden is measured in terms of
+CRS cases affecting newborns, linked to their pregnant mothers. This is
+different from many other diseases (e.g., HepB, measles or YF in this
+vignette), where the burden is directly attributed to the infected
+individual. Consequently, when you see vimpact outputs for Rubella,
+please refer to the following interpretations.
+a.) Age-specific calendar impact
+Definition: The disease burden averted for newborns in a given
+calendar year, attributed to mothers in specific age groups during that
+year. Interpretation: How many CRS related burden are prevented among
+newborns, broken down by the age of the mother in the year of
+pregnancy.
+b.) Calendar impact
+Definition: The total disease burden averted for newborns in a given
+calendar year (regardless of the mother’s age). Interpretation: How many
+CRS related burden are prevented among newborns overall in that
+year.
+c.) Cohort impact
+Definition: The disease burden averted for newborns whose mothers
+were born in a specific birth year. Interpretation: How many CRS related
+burden are prevented among newborns, categorized by the mother’s year of
+birth.
+d.) Impact by year of vaccination
+Definition: The disease burden averted that is linked to immunization
+programs carried out in a particular year. Interpretation: How many CRS
+related burden are prevented among newborns, attributed to vaccinations
+given in that specific year.
+
+
Impact by calendar year & impact by birth year
+
+
Function interface
+
impact <- vimpact::calculate_impact(
+ con, method = "calendar_year", touchstone = "201910gavi-5",
+ modelling_group = "CDA-Razavi", disease = "HepB",
+ focal_scenario_type = "default", focal_vaccine_delivery = list(
+ list(vaccine = "HepB_BD", activity_type = "routine"),
+ list(vaccine = "HepB", activity_type = "routine")
+ ),
+ baseline_scenario_type = "novac",
+ burden_outcomes = c("hepb_deaths_acute", "hepb_deaths_dec_cirrh",
+ "hepb_deaths_hcc"))
+str(impact)
+#> tibble [9,898 × 4] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:9898] "AFG" "AFG" "AFG" "AFG" ...
+#> $ burden_outcome: chr [1:9898] "deaths" "deaths" "deaths" "deaths" ...
+#> $ year : int [1:9898] 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 ...
+#> $ impact : num [1:9898] 0 0 0 0 0 0 0 53 81 97 ...
+
Or for impact by birth year, use method = "birth_year".
+This will get burden_estimate_set ids for the baseline and
+focal scenarios for this particular touchstone, modelling group,
+disease. Then uses those to pull the burden_estimate data
+for specified burden_outcomes. Optionally filtering on
+country via countries arg, year via
+vaccination_years and under 5 age groups if
+is_under5 = TRUE. We then use raw impact from
+burden_estimate table and call relevant public facing
+impact method. For method = "calendar_year"
+impact_by_calendar_year. For
+method = "birth_year"
+impact_by_birth_year.
+
+
+
Recipe interface
+
Define a recipe either as a csv or using
+recipe_template
+
recipe <- data.frame(
+ touchstone = "201910gavi-5",
+ modelling_group = "CDA-Razavi", disease = "HepB",
+ focal = "default:HepB_BD-routine;HepB-routine",
+ baseline = "novac",
+ burden_outcome = "hepb_deaths_acute,hepb_deaths_dec_cirrh,hepb_deaths_hcc;hepb_cases_acute_severe,hepb_cases_dec_cirrh,hepb_cases_hcc;dalys")
+t <- tempfile(fileext = ".csv")
+write.csv(recipe, t, row.names = FALSE)
+
This is a set of properties defining what data we want to extract
+from the db e.g. touchstone, modelling_group,
+disease, focal and baseline scenarios and
+burden_outcome. It captures the same info as the args to
+calculate_impact above. Then use the recipe to define meta
+data frame
+
meta <- vimpact:::get_meta_from_recipe(default_recipe = FALSE, recipe = t, con = con)
+
And use this to calculate impact
+
old_impact <- vimpact:::get_raw_impact_details(con, meta, "deaths")
+str(old_impact)
+#> 'data.frame': 9898 obs. of 7 variables:
+#> $ country : int 104 104 104 104 104 104 104 104 104 104 ...
+#> $ time : int 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 ...
+#> $ baseline_value: num 5636 5765 5901 6034 6172 ...
+#> $ focal_value : num 5636 5765 5901 6015 6075 ...
+#> $ value : num 0 0 0 19 97 180 252 312 344 381 ...
+#> $ index : int 1 1 1 1 1 1 1 1 1 1 ...
+#> $ burden_outcome: chr "deaths" "deaths" "deaths" "deaths" ...
+
+
+
Impact by year of vaccination
+
+
Function interface
+
Very similar to examples for calendar year and birth year, to
+calculate impact by year of vaccination stratified by activity type
+run
+
impact <- vimpact::calculate_impact(
+ con, method = "yov_activity_type", touchstone = "201910gavi-5",
+ modelling_group = "CDA-Razavi", disease = "HepB",
+ focal_scenario_type = "default", focal_vaccine_delivery = list(
+ list(vaccine = "HepB_BD", activity_type = "routine"),
+ list(vaccine = "HepB", activity_type = "routine")
+ ),
+ baseline_scenario_type = "novac",
+ burden_outcomes = "dalys")
+str(impact)
+#> tibble [4,279 × 8] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:4279] "AFG" "AFG" "AFG" "AFG" ...
+#> $ vaccine : chr [1:4279] "HepB" "HepB" "HepB" "HepB" ...
+#> $ activity_type : chr [1:4279] "routine" "routine" "routine" "routine" ...
+#> $ year : int [1:4279] 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 ...
+#> $ burden_outcome: chr [1:4279] "dalys" "dalys" "dalys" "dalys" ...
+#> $ impact : num [1:4279] 103859 106575 105852 111953 115165 ...
+#> $ impact_ratio : num [1:4279] 0.154 0.154 0.154 0.154 0.154 ...
+#> $ fvps : num [1:4279] 676396 694088 689380 729111 750028 ...
+
or use method = "yov_birth_cohort for impact by year of
+vaccination stratified by birth cohort. This uses the same logic as
+other impact methods but before calling the public facing impact
+function it will also extract FVP data from the data base for this
+touchstone and vaccine delivery. It then calls either
+impact_by_year_of_vaccination_activity_type or
+impact_by_year_of_vaccination_birth_cohort to get
+impact.
+
+
+
Recipe interface
+
We can re-use the recipe from above and get meta table for this
+method
+
meta <- vimpact:::get_meta_from_recipe(default_recipe = FALSE, recipe = t,
+ method = "method2a", con = con)
+
Then use this to get raw impact
+
old_raw_impact <- vimpact:::get_raw_impact_details(con, meta, "dalys")
+
Extract the fvp data and map output to required interface
+
fvps <- vimpact::extract_vaccination_history(con, "201910gavi-5", year_min = 2000,
+ year_max = 2030,
+ disease_to_extract = "HepB")
+#> User defined touchstone version is used.
+#> Converting input coverage data......
+#> Extracted interpolated population.
+#> Extracted raw coverage data...
+#> Transformed coverage data.
+fvps$fvps <- fvps$fvps_adjusted
+fvps$country <- fvps$country_nid
+
Then can use meta, raw_impact and
+fvps to calculate impact by year of vaccination
+
old_impact <- vimpact:::impact_by_year_of_vaccination(
+ meta, old_raw_impact, fvps, vaccination_years = 2000:2030)
+str(old_impact)
+#> 'data.frame': 4964 obs. of 27 variables:
+#> $ country : int 4 4 4 4 4 4 4 4 4 4 ...
+#> $ vaccine : chr "HepB_BD" "HepB" "HepB_BD" "HepB_BD" ...
+#> $ activity_type : chr "routine" "routine" "routine" "routine" ...
+#> $ scenario_description: chr "best-estimates" "best-estimates" "best-estimates" "best-estimates" ...
+#> $ coverage_set : int 1750 1751 1750 1750 1751 1751 1751 1751 1751 1751 ...
+#> $ delivery_id : int 487 3093 1169 1175 3376 4029 4272 3375 4564 3096 ...
+#> $ disease : chr "HepB" "HepB" "HepB" "HepB" ...
+#> $ scenario_type : chr "default" "default" "default" "default" ...
+#> $ gavi_support_level : chr "with" "with" "with" "with" ...
+#> $ year : int 2016 2017 2014 2025 2030 2015 2019 2027 2029 2020 ...
+#> $ gavi_support : logi FALSE FALSE FALSE FALSE FALSE FALSE ...
+#> $ gender : chr "Both" "Both" "Both" "Both" ...
+#> $ age : num 0 0 0 0 0 0 0 0 0 0 ...
+#> $ target_source : num 1118424 1131834 1103729 1180725 1188620 ...
+#> $ coverage_source : num 0.18 0.66 0.04 0.473 0.749 ...
+#> $ cohort_size : num 1118424 1131834 1103729 1180725 1188620 ...
+#> $ delivery_population : num 1118424 1131834 1103729 1180725 1188620 ...
+#> $ fvps_source : num 201316 747010 44149 558483 890514 ...
+#> $ fvps_adjusted : num 201316 747010 44149 558483 890514 ...
+#> $ coverage_adjusted : num 0.18 0.66 0.04 0.473 0.749 ...
+#> $ country_nid : int 4 4 4 4 4 4 4 4 4 4 ...
+#> $ fvps : num 201316 747010 44149 558483 890514 ...
+#> $ time : num 2016 2017 2014 2025 2030 ...
+#> $ burden_outcome : chr "dalys" "dalys" "dalys" "dalys" ...
+#> $ impact_ratio : num 0.154 0.154 0.154 0.154 0.154 ...
+#> $ impact : num 30912 114701 6779 85754 136736 ...
+#> $ index : int 1 1 1 1 1 1 1 1 1 1 ...
+
+
+
Comparison
+
calculate_impact has a very similar interface as a
+single row in an impact recipe. The output is slightly different format
+but output from previous method can be transformed into same format.
+
country <- dplyr::tbl(con, "country")
+old_impact <- old_impact %>%
+ dplyr::left_join(country, by = c("country" = "nid"), copy = TRUE) %>%
+ dplyr::select(country = id, vaccine, activity_type, year = time,
+ burden_outcome, impact) %>%
+ dplyr::filter(!is.na(impact)) %>%
+ dplyr::arrange(activity_type, country, year, vaccine)
+str(old_impact)
+#> 'data.frame': 4279 obs. of 6 variables:
+#> $ country : chr "AFG" "AFG" "AFG" "AFG" ...
+#> $ vaccine : chr "HepB" "HepB" "HepB" "HepB" ...
+#> $ activity_type : chr "routine" "routine" "routine" "routine" ...
+#> $ year : num 2007 2008 2009 2010 2011 ...
+#> $ burden_outcome: chr "dalys" "dalys" "dalys" "dalys" ...
+#> $ impact : num 103859 106575 105852 111953 115165 ...
+
and we can see from a plot that the output is very similar
+
+
HepB
+
htmltools::tags$iframe(
+ src = "impact_comparisons_hepb.html",
+ width = "100%",
+ height = "400",
+ scrolling = "no",
+ seamless = "seamless",
+ frameBorder = "0",
+ `data-external` = "1"
+)
+
+
+
+
Measles
+
htmltools::tags$iframe(
+ src = "impact_comparisons_measles.html",
+ width = "100%",
+ height = "400",
+ scrolling = "no",
+ seamless = "seamless",
+ frameBorder = "0",
+ `data-external` = "1"
+)
+
+
+
+
Comparisons of HepB, Measles & YF
+
This shows bars of mean difference between old impact & new
+impact method for each country for HepB, Measles and YF. Tooltips show
+the number of observations and the min and max order of the values.
+i.e. an impact of 23432.342 has order 5.
+
htmltools::tags$iframe(
+ src = "impact_comparisons.html",
+ width = "100%",
+ height = "400",
+ scrolling = "no",
+ seamless = "seamless",
+ frameBorder = "0",
+ `data-external` = "1"
+)
+
+
There are small differences in impact because of differences in
+precision. calculate_impact tries to do large parts of the
+aggregation on the database where burden_estimate values
+are stored as Postgres real type which has 4 bytes of
+storage size and 6 decimal digits of precision. Whereas if we pull this
+before aggregation the values are stored in R as doubles which are much
+higher precision leading to small differences when aggregated.
+
calculate_impact at the moment won’t be able to generate
+impact for you for more than 1 set of scenario comparisons. Next steps
+will be to add a wrapper which can take some data like the impact recipe
+and call calculate_imapct for multiple scenarios.
+
+
Using vimpact for estimating vaccine
+impact
+Rob Ashton, Xiang Li, Katy Gaythorpe
+2025-02-26
+
+
+
+This vignette describes how to use impact calculation methods used in
+VIMC as an external user.
+N.B. For VIMC Rubella models, disease burden is measured in terms of
+CRS cases affecting newborns, linked to their pregnant mothers. This is
+different from many other diseases (e.g., HepB, measles or YF in this
+vignette), where the burden is directly attributed to the infected
+individual. Consequently, when you see vimpact outputs for Rubella,
+please refer to the following interpretations.
+a.) Age-specific calendar impact
+Definition: The disease burden averted for newborns in a given
+calendar year, attributed to mothers in specific age groups during that
+year. Interpretation: How many CRS related burden are prevented among
+newborns, broken down by the age of the mother in the year of
+pregnancy.
+b.) Calendar impact
+Definition: The total disease burden averted for newborns in a given
+calendar year (regardless of the mother’s age). Interpretation: How many
+CRS related burden are prevented among newborns overall in that
+year.
+c.) Cohort impact
+Definition: The disease burden averted for newborns whose mothers
+were born in a specific birth year. Interpretation: How many CRS related
+burden are prevented among newborns, categorized by the mother’s year of
+birth.
+d.) Impact by year of vaccination
+Definition: The disease burden averted that is linked to immunization
+programs carried out in a particular year. Interpretation: How many CRS
+related burden are prevented among newborns, attributed to vaccinations
+given in that specific year.
+
+
Data
+
The simplest possible example is a static model with constant routine
+immunisation and a one-off campaign. Burden estimates from a theoretical
+country are found in 3 csv titled example_novax_burden.csv,
+example_routine_burden.csv and
+example_routine_campaign_burden.csv. These include
+estimates of burden in years 1900 - 2100 and display the format used by
+the functions.
+
example_novax <- read.csv(system.file("extdata/example_novax_burden.csv", package = "vimpact"))
+example_routine <- read.csv(system.file("extdata/example_routine_burden.csv", package = "vimpact"))
+example_routine_campaign <- read.csv(system.file("extdata/example_routine_campaign_burden.csv", package = "vimpact"))
+
+str(example_novax)
+#> 'data.frame': 10201 obs. of 6 variables:
+#> $ country : chr "ABC" "ABC" "ABC" "ABC" ...
+#> $ year : int 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ...
+#> $ age : int 0 1 2 3 4 5 6 7 8 9 ...
+#> $ value : int 16 15 14 13 13 12 12 11 11 10 ...
+#> $ activity_type : chr "novax" "novax" "novax" "novax" ...
+#> $ burden_outcome: chr "deaths" "deaths" "deaths" "deaths" ...
+str(example_routine)
+#> 'data.frame': 10201 obs. of 6 variables:
+#> $ country : chr "ABC" "ABC" "ABC" "ABC" ...
+#> $ year : int 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ...
+#> $ age : int 0 1 2 3 4 5 6 7 8 9 ...
+#> $ value : int 16 15 14 13 13 12 12 11 11 10 ...
+#> $ activity_type : chr "routine" "routine" "routine" "routine" ...
+#> $ burden_outcome: chr "deaths" "deaths" "deaths" "deaths" ...
+str(example_routine_campaign)
+#> 'data.frame': 10201 obs. of 6 variables:
+#> $ country : chr "ABC" "ABC" "ABC" "ABC" ...
+#> $ year : int 2000 2000 2000 2000 2000 2000 2000 2000 2000 2000 ...
+#> $ age : int 0 1 2 3 4 5 6 7 8 9 ...
+#> $ value : int 16 15 14 13 13 12 12 11 11 10 ...
+#> $ activity_type : chr "campaign" "campaign" "campaign" "campaign" ...
+#> $ burden_outcome: chr "deaths" "deaths" "deaths" "deaths" ...
+
+
Impact by calendar year (cross-sectional impact)
+
Impact by calendar year calculates impact accrued over all ages for a
+specific year. This calculates the difference in disease burden between
+baseline and focal scenarios for a given year.
+
To use this method you need to provide a data frame of baseline
+impact and a data frame of focal impact for the vaccine of interest for
+a given age, year and country.
+
These data frames must contain columns for country, burden_outcome
+(e.g. cases, deaths, DALYs), year, age and value. It will sum the values
+over all ages for each year and return the impact as the difference
+between baseline and focal scenarios for each group.
+
If the baseline and focal impact data contains groups which do not
+overlap then this will return impact only for groups contained in both
+datasets.
+
calendar_year <- vimpact::impact_by_calendar_year(baseline_burden = example_novax, focal_burden = example_routine_campaign)
+str(calendar_year)
+#> tibble [101 × 4] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:101] "ABC" "ABC" "ABC" "ABC" ...
+#> $ burden_outcome: chr [1:101] "deaths" "deaths" "deaths" "deaths" ...
+#> $ year : int [1:101] 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 ...
+#> $ impact : int [1:101] 0 0 0 0 0 3 11 18 27 35 ...
+
+
Impact by birth year (lifetime impact)
+
When calculating impact by birth year we account for the long-term
+impact of vaccination by finding the sum of the difference in burden
+between focal and baseline scenarios over the lifetime of a particular
+cohort. This can be written as \(D(c, k) =
+\sum_{y-a=k}D_{b-f}(a, c, y)\) where \(y \in Y_{m}\) and \(a \in A_{m}\) where \(D\) is the difference in burden, \(b\) is baseline, \(f\) is focal, \(a\) is age, \(c\) is country, \(y\) is year, \(k\) is birth cohort, \(Y_{m}\) are the years modelled and \(A_{m}\) are the ages modelled. This
+requires the same data as impact_by_calendar_year. You need
+to provide data frames containing baseline and focal impact for the
+vaccine of interest for a given age, year and country.
+
These data frames must contain columns for country, burden_outcome
+(e.g. cases, deaths, DALYs), year, age and value. It will sum the impact
+grouped by country, burden_outcome, and birth year. Where birth year is
+year - age. Then calculates the impact for these
+groups.
+
If the baseline and focal impact data contains groups which do not
+overlap then this will return impact only for groups contained in both
+datasets.
+
birth_year <- vimpact::impact_by_birth_year(baseline_burden = example_novax, focal_burden = example_routine_campaign)
+str(birth_year)
+#> tibble [201 × 4] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:201] "ABC" "ABC" "ABC" "ABC" ...
+#> $ burden_outcome: chr [1:201] "deaths" "deaths" "deaths" "deaths" ...
+#> $ birth_cohort : int [1:201] 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 ...
+#> $ impact : int [1:201] 0 0 0 0 0 0 0 0 0 0 ...
+
birth_year$year <- birth_year$birth_cohort
+birth_year$impact_birth_year <- birth_year$impact
+calendar_year$impact_calendar_year <- calendar_year$impact
+plot_data <- merge(birth_year, calendar_year, by = c("country", "burden_outcome", "year"))
+plot_data$year <- as.Date(paste(plot_data$year, "01", "01", sep = "-"))
+ggplot2::ggplot(plot_data, ggplot2::aes(x = year)) +
+ ggplot2::geom_line(ggplot2::aes(y = impact_birth_year, colour = "birth year")) +
+ ggplot2::geom_line(ggplot2::aes(y = impact_calendar_year, colour = "calendar year")) +
+ ggplot2::expand_limits(y = 0) +
+ ggplot2::ylab("Impact") +
+ ggplot2::xlab("Year")
+
+
+
Impact by year of vaccination
+
The final view of impact is by attributing it to the year in which
+vaccination took place. Impact by year of vaccination methods are vital
+for determining the long-term impact of of vaccination due to activities
+carried out in a particular year. We obtain the impact ratio as the
+impact attributable per fully vaccinated person (FVP) calculated as the
+coverage × cohort size. This ratio can be stratified by different
+characteristics, such as birth cohort in order to catch temporal changes
+in transmission or healthcare or by activity type to capture the
+differing effects of routine and campaign vaccination. The impact ratio
+allows effects due to a particular year’s worth of vaccination to be
+attributed to that year.
+
+
Impact by year of vaccination: impact ratio stratified by activity
+type
+
To calculate impact by year of vaccination with impact ratio
+stratified by activity type use function
+impact_by_year_of_vaccination_activity_type. You need to
+provide 3 data frames. Like the previous two impact methods this needs a
+data frame of baseline impact and focal impact. These must contain the
+same columns as in the previous two methods with the addition of a
+column for the activity_type of the vaccination - either
+routine or campaign.
+
You also need to provide a data frame of FVP (fully vaccinated
+person) data which contains country, year, activity type, age and fvp.
+And a set of vaccination_years which is a range of years
+for which you want calculate impact for. This will be used to filter the
+baseline impact, focal impact and FVPs.
+
This method produces impact values for a particular disease
+stratified by country, activity type, year and burden outcome.
+
example_fvps <- read.csv(system.file("extdata/example_fvps.csv", package = "vimpact"))
+
+routine <- vimpact::impact_by_year_of_vaccination_activity_type(baseline_burden = example_novax, focal_burden = example_routine, fvps = example_fvps[example_fvps$activity_type == "routine",], 2000:2030)
+
+campaign <- vimpact::impact_by_year_of_vaccination_activity_type(baseline_burden = example_routine, focal_burden = example_routine_campaign, fvps = example_fvps[example_fvps$activity_type == "campaign",], 2000:2030)
+
+activity_type <- rbind(routine, campaign)
+str(activity_type)
+#> tibble [35 × 8] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:35] "ABC" "ABC" "ABC" "ABC" ...
+#> $ vaccine : chr [1:35] "YF" "YF" "YF" "YF" ...
+#> $ activity_type : chr [1:35] "routine" "routine" "routine" "routine" ...
+#> $ year : int [1:35] 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 ...
+#> $ burden_outcome: chr [1:35] "deaths" "deaths" "deaths" "deaths" ...
+#> $ impact : num [1:35] 184 323 404 478 479 ...
+#> $ impact_ratio : num [1:35] 0.168 0.168 0.168 0.168 0.168 ...
+#> $ fvps : int [1:35] 1090 1919 2397 2838 2842 4008 2971 2310 1494 2996 ...
+
+
+
Impact by year of vaccination: impact ratio stratified by birth
+cohort
+
Impact by year of vaccination can be stratified by birth cohort to
+catch temporal changes in transmission or healthcare. This method is
+invariant to vaccination activity type. However, vaccine effect is
+assumed to vary over time through birth cohorts.
+
This method requires basline impact, focal impact and FVPs
+disaggregated by country, burden outcome, year and age. The data doesn’t
+have to be disaggregated by activity_type. This returns
+impact for a particular vaccine by country, year and burden outcome.
+
cohort_type <- vimpact::impact_by_year_of_vaccination_birth_cohort(baseline_burden = example_novax, focal_burden = example_routine_campaign, fvps = example_fvps, 2000:2030)
+str(cohort_type)
+#> tibble [35 × 7] (S3: tbl_df/tbl/data.frame)
+#> $ country : chr [1:35] "ABC" "ABC" "ABC" "ABC" ...
+#> $ year : int [1:35] 2013 2018 2019 2020 2021 2022 2023 2024 2004 2005 ...
+#> $ burden_outcome: chr [1:35] "deaths" "deaths" "deaths" "deaths" ...
+#> $ vaccine : chr [1:35] "YF" "YF" "YF" "YF" ...
+#> $ activity_type : chr [1:35] "campaign" "campaign" "campaign" "campaign" ...
+#> $ impact : num [1:35] 619 868 1314 1348 1891 ...
+#> $ fvps : int [1:35] 9987 12378 18313 18310 25002 24999 24997 26681 1090 1919 ...
+
activity_sum <- stats::aggregate(impact ~ country + year + burden_outcome,
+ activity_type, sum)
+cohort_sum <- stats::aggregate(impact ~ country + year + burden_outcome,
+ cohort_type, sum)
+activity_sum$impact_activity_strat <- activity_sum$impact
+cohort_sum$impact_cohort_strat <- cohort_sum$impact
+plot_data <- merge(activity_sum, cohort_sum, by = c("country", "burden_outcome", "year"))
+plot_data$year <- as.Date(paste(plot_data$year, "01", "01", sep = "-"))
+ggplot2::ggplot(plot_data, ggplot2::aes(x = year)) +
+ ggplot2::geom_line(ggplot2::aes(y = impact_cohort_strat, colour = "cohort")) +
+ ggplot2::geom_line(ggplot2::aes(y = impact_activity_strat, colour = "activity")) +
+ ggplot2::expand_limits(y = 0) +
+ ggplot2::ylab("Impact") +
+ ggplot2::xlab("Year")
+
+
+
+
Glossary
+
All definitions and terminology can be found in the associated
+manuscript available at: https://www.medrxiv.org/content/10.1101/2021.01.08.21249378v1.full-text
+; we include a summary below.
+
FVP - Fully vaccinated person
+
Focal scenario - The vaccination scenario to be assessed in
+comparison to a baseline scenario
+
Baseline scenario - The vaccination scenario to be compared against,
+often the situation without any vaccination (no-vac)
+
Impact- The effect of vaccination often measure in deaths or cases
+averted and calculated as the difference in burden between a focal
+scenario and a baseline scenario
+
Impact ratios - The impact per fully vaccinated person
+