In mrc-ide/esft: WHO ESFT Calculations
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
library(esft)
Here we will go into greater details on the user input defaults, the parameters, and the capacity calculations.
User Input
user <- user_input()
# we set the country to be afghanistan
country <- "AFG"
User input contains four parameters:
- forecast_period - which is the length of time we are forecasting for, in weeks. The default is 12 weeks.
- delivery_leadtime - this is the lead time between forecast and impacted delivery. This only impacts calculations of the amounts in the first week in the ESFT sheet, and is not used in our calculations. Default is 0.
- week1 - this is one of the dates included in week 1 of our forecast. The default is "2022-01-01". I guess this should not be named like this. In the end everything is later than this - because this ends up being one of the dates in week 1 (because it's subset by week_end > date) and then patients needs that to get the correct patient count, but the correct patients weekly doesnt show up until the second week of that one
- scenario - default is medium transmission, with an R(0) number or R(eff) number of 0.94. This signifies no change, or keeping the status quo of transmission. Other options include "Low", with an R(0) or R(eff) of 0.47, which simulates a 50 percent decrease in transmission, or "High", with an R(0) or R(eff) of 1.41, which simulates a 50 percent increase in transmission.
Parameter Setting
# This is the standard list of all parameters, many from inputs tab of excel sheet
params <- get_parameters()
Case Severity Distributions:
- mild_i_proportion - proportion of cases that are mild, considered to be isolating to reduce onward transmission; default = 0.4
- mod_i_proportion - proportion of cases that are moderate, considered to be isolating to minimize onward transmission; default = 0.4
- sev_i_proportion - proportion of cases that are severe, depending on severe bed availability, are admitted to hospital and require oxygen; default = 0.15
- crit_i_proportion - proportion of cases that are critical, depending on critical bed availability, admitted to hospital and require ventilation; default = 0.05
Weeks of Stay in Isolation (WHO recommendation based on incubation period of COVID-19 and case management guidelines):
- stay_mild - average length of stay (in weeks) of mild cases in isolation; default = 2
- stay_mod - average length of stay (in weeks) of moderate cases in isolation; default = 2
Weeks of Stay in Hospital (based on studies of case severity):
- stay_sev - average length of stay (in weeks) of severe cases in hospital; default = 1
- stay_crit - average length of stay (in weeks) of critical cases in hospital; default = 2
Infection Fatality Rates (note: could be updated based on newer research):
- ifr_sev - infection fatality rate of severe cases, based on WHO China Joint Mission Report; default = 0.134
- ifr_crit - infection fatality rate of critical cases, based on Imperial College Report nr. 9; default = 0.5
Miscellaneous Parameters:
- perc_hcws_not_covid - assumption of percentage of HCWs not working on COVID; default = 0.4
- n_hosp_beds_per_care_unit - assumption, used to estimate triple packaging boxes (might not be necessary); default = 40
Healthcare Workers per Bed or Patient:
- ambulancews_per_bed - ambulance personnel ratio assumes 1 ambulance per 100 bed hospital with 2 operators (paramedic + driver) at all times (3x8 hour shifts) so 6/100 beds; default = 0.06
- bioengs_per_bed - biomedical engineer ratio assumes 2 biomedical engineers (on 8-hour shifts) per 100 bed hospital; default = 0.02
- n_inf_caregivers_hosp - reference assumption of zero is based on current guidance that no family members or other caretakers should be in hospitals; default = 0
- n_inf_caregivers_isol - reference is based on management of home care guidance, with estimates of 1 caregiver per patient for the duration of the roughly 2-week isolation - his calculation estimates the quantity of PPE required (e.g., masks and gloves) for the patient and caregiver; default = 1
Percent of Critical Patients that are Mechanically Ventilated:
- perc_crit_inv_mv - WHO recommendation; default = 0.667
- perc_crit_noninv_mv - WHO recommendation; default = 0.333
Oxygen Flow by Case Severity, in LPM:
- o2_flow_sev - severe patients require 5-15 LPM, so median taken; default = 10
- o2_flow_crit_inv_mv - critical patients with invasive mechanical ventilation; default = 30
- o2_flow_crit_noninv_mv - critical patients with non-invasive mechanical ventilation; default = 30
Testing Strategy
# Testing strategy parameters - population level testing parameters (i.e. percent
# tested)
test_strat <- set_testing_strategy()
Testing Strategies (there are two choices, input as string values):
- all - all cases that present, regardless of severity, using standard number of negatives per positive test
- targeted - restricted testing of mild/moderate presenting cases may be employed if limited tests available. all required severe and critical cases will still be tested. if selected, only the percent of suspected/mild/moderate cases input here will be tested (typically high-risk patients)
Testing Parameters (the rest of the testing strategy parameters, some of which depend on the testing strategy chosen):
- perc_tested_sev_crit - percent of severe or critical cases tested, always 100\%; default = 1
- num_neg_per_pos_test - estimated average number of negative tests per positive test; default = 10
- tests_per_hcw_per_week - tests per HCW or staff member per week. Includes tests for inpatient HCW, screening/triage HCW, ambulance personnel, cleaners, lab techs, and biomedical engineers. input can also be 0 (for no tests) or a decimal (e.g. 0.5, to represent tests every other week); default = 1
- testing_contacts - is testing done for contacts of positive cases; default = TRUE
- avg_contacts_pos_case - average number of contacts per positive case. Suggested options are 5 (High/Strong, e.g. stay at home regulations), 10 (Medium/Weak social distancing, e.g. travel restrictions), 15 (Low/No social distancing, e.g. advice only); default = 10
- perc_contacts_tested - percent of contacts of a positive case who get tested; default = 0.6
Diagnostic Parameters
# Diagnostic parameters - individual level testing parameters (i.e. number of tests
# per case per diagnosis, etc.)
test_params <- get_diagnostic_parameters()
Total tests and percent antigen:
- total_tests_mild_mod - total tests, used for diagnosis only; default = 1
- total_tests_sev_crit - total tests, used for diagnosis and release; default = 2
- perc_antigen_tests - percent testing done in hospital via antigen testing, max 80 percent; default = 20 percent
- tests_diagnosis_mild_mod - the number of tests for diagnosis for mild or moderate cases; default = 1
- tests_diagnosis_sev_crit - the number of tests for diagnosis for severe or critical cases; default = 1
- tests_release_mild_mod - the number of tests for release for mild or moderate cases; default = 0
- tests_release_sev_crit - the number of tests for release for severe or critical cases; default = 1
Lab Parameters
# Diagnostic lab parameters - number of lab staff per lab, safety boxes per unit,
# etc.
lab_params <- get_lab_parameters()
Requirements per Lab:
- lab_staff_per_lab - multiplier that helps estimate lab staff equipment requirements; default = 3
- hygienists_per_lab - multiplier that helps estimate lab hygienist/cleaner equipment requirements; default = 3
- safety_boxes_per_unit_week - WHO recommendation for safe sharp disposal; default = 8
- triple_packaging_per_unit - WHO recommendation for sample transport; default = 4
- perc_wastage_manual_test_kits - percentage wastage, only of manual test kits; default = 10\%
- num_tests_manual_test_kits - number of tests in an RT-PCR manual test kit; default = 100
Capacity Mapping
capacity <- get_country_capacity(iso3c = country)
Country-Specific Capacity Estimates:
- country - country name, English
- iso3c - country code, iso3c format
- population - population estimate from the latest available data from the United Nations Development Programme, reported in 2020
- yoy_growth - average year over year growth of country population, estimated from the change in growth in population between UNDP reported numbers in 2015 and 2020
- income_group - World Bank income group
- n_hcws - number of estimated HCWs. Combines latest available estimates of nurses (
Nursing personnel) and doctors (Medical doctors) from WHO GHO with the estimates YoY growth from UNDP to yield estimates of nurses and doctors, the sum of which are the number of healthcare workers. If these are unavailable, uses a combination of income group averages and yoy growth.
- n_labs - number of lab staff, estimated using the same method as the number of HCWs (codes:
Medical and Pathology Laboratory scientists and Medical and Pathology Laboratory Technicians).
- n_hosp_beds - estimated beds per country, calculated using similar procedure as above but with World Bank estimates of hospital beds per 1,000 people.
- perc_beds_crit_covid - percent beds for ICU patients per country, estimated from World Bank/Imperial College Report
- perc_beds_not_covid - percent beds not allocated to COVID, default = 0.4
- perc_beds_sev_covid - percent beds allocated to severe COVID- cases, $1 - (\text{perc_beds_crit_covid} + \text{perc_beds_not_covid})$
- beds_covid - number of hospital beds allocated to COVID $(\text{n_hosp_beds}* (1-\text{perc_beds_not_covid}))$
- severe_beds_covid - number of severe beds allocated to COVID, $\text{n_hosp_beds}*\text{perc_beds_sev_covid}$
- crit_beds_covid - number of critical beds allocated to COVID, $\text{n_hosp_beds}*\text{perc_beds_crit_covid}$
Testing Capacity
country_test_capacity <- get_country_test_capacity(iso3c = country)
This uses estimated provided by the WHO Operations, Supply and Logistics Team that are reviewed by diagnostics technical experts at the WHO to provide estimates of counts of various diagnostic processing machines in country.
Types of machines in output:
- roche_6800 - high throughput conventional platform
- roche_8800 - high throughput conventional platform
- abbott_m2000 - high throughput conventional platform
- hologic_panther - high throughput conventional platform
- hologic_panther_fusion - high throughput conventional platform
- genexpert - near-patient PCR machine modules (not platforms: here, one machine contains several modules)
- manual - manual real-time PCR platform
Diagnostic Capacity
The next function, called calc_diagnostic_capacity, takes these machine counts as inputs, in addition to data on their processing capability, hours per shift, and number of shifts per day to estimate the total test processing capability, what we term the diagnostic capacity.
data(throughput, package = "esft")
data(hours_per_shift, package = "esft")
diagnostic_capacity <- calc_diagnostic_capacity(
country_diagnostic_capacity = country_test_capacity,
throughput,
hours_per_shift = hours_per_shift,
shifts_per_day = 1
)
Diagnostic (Total Test Processing) Capacity:
- shifts_day - these are the shifts per day specified in function call
- platform_key - snake case version of the testing platform name
- platform - platform name in plain english
- type - type of diagnostics machine
- days_week - days it operates per week
- covid_capacity - capacity of test processing function dedicated to COVID-19
- country_name - country for which this capacity is calculated
- region - WHO region of that country
- income_group - WB income group of that country
- country_code - iso3c code of that country
- modules_activated - number of total modules (a machine can have several modules) in active use in that country
- hours - hours per shift
- throughput_per_day - number of tests that can be processed by that machine per day (with day being specified by combining hours per shift, and shifts per day and cross referencing that with the throughput input data
- total_test_capacity - this is the total number of tests that the different categories of machines can process in a week, depending on the days per week and the throughput per day ($Capacity_{test-total} = Modules_{activated}Days_{week}Throughput_{day}$).
- covid_test_capacity - this is the total test capacity multiplied by the covid capacity - or the number of tests that can be run for COVID-19 on the different machines every week ($Capacity_{test-COVID} = Capacity_{test-total}*p_{covid-capacity}$)
Total Labs
# Calculates the total labs available for COVID-19 given the diagnostic capacity
# (number of modules and machines) available
t_labs <- total_labs(diagnostic_capacity)
This calculates the number of labs conducting COVID-19 testing, the reference value assumes 1 lab for every 3 high-throughput, manual or other platforms and 1 lab for every 4 GeneXpert modules.
$$
\begin{align}
Modules_{sum} = Modules_{roche-6800} + Modules_{roche-8800} + Modules_{abbott-m2000} +\
Modules_{hologic-panther}+Modules_{hologic-panther-fusion}+Modules_{manual}\
\
Labs_{total} = \frac{Modules_{sum}}{3} + \frac{Modules_{genexpert}}{4}
\end{align}
$$
Max Tests
# Uses the country specific diagnostic capacity estimates to calculate the max
# number of tests per day. This is then used to help cap tests by total testing
# capacity in total_tests.
max_tests <- max_tests_per_day(diagnostic_capacity)
This calculates the total number of tests per week given covid test capacity, and divides by 7 to find the max number of tests per day (I am not certain that 7 is the correct way to calculate max tests - the machines are presumed to run for a pre-set number of days per week, so I am not sure why this would change here).
mrc-ide/esft documentation built on July 31, 2023, 2:30 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
knitr::opts_chunk$set( collapse = TRUE, comment = "#>" )
library(esft)
Here we will go into greater details on the user input defaults, the parameters, and the capacity calculations.
User Input
user <- user_input() # we set the country to be afghanistan country <- "AFG"
User input contains four parameters:
- forecast_period - which is the length of time we are forecasting for, in weeks. The default is 12 weeks.
- delivery_leadtime - this is the lead time between forecast and impacted delivery. This only impacts calculations of the amounts in the first week in the ESFT sheet, and is not used in our calculations. Default is 0.
- week1 - this is one of the dates included in week 1 of our forecast. The default is "2022-01-01". I guess this should not be named like this. In the end everything is later than this - because this ends up being one of the dates in week 1 (because it's subset by week_end > date) and then patients needs that to get the correct patient count, but the correct patients weekly doesnt show up until the second week of that one
- scenario - default is medium transmission, with an R(0) number or R(eff) number of 0.94. This signifies no change, or keeping the status quo of transmission. Other options include "Low", with an R(0) or R(eff) of 0.47, which simulates a 50 percent decrease in transmission, or "High", with an R(0) or R(eff) of 1.41, which simulates a 50 percent increase in transmission.
Parameter Setting
# This is the standard list of all parameters, many from inputs tab of excel sheet params <- get_parameters()
Case Severity Distributions:
- mild_i_proportion - proportion of cases that are mild, considered to be isolating to reduce onward transmission; default = 0.4
- mod_i_proportion - proportion of cases that are moderate, considered to be isolating to minimize onward transmission; default = 0.4
- sev_i_proportion - proportion of cases that are severe, depending on severe bed availability, are admitted to hospital and require oxygen; default = 0.15
- crit_i_proportion - proportion of cases that are critical, depending on critical bed availability, admitted to hospital and require ventilation; default = 0.05
Weeks of Stay in Isolation (WHO recommendation based on incubation period of COVID-19 and case management guidelines):
- stay_mild - average length of stay (in weeks) of mild cases in isolation; default = 2
- stay_mod - average length of stay (in weeks) of moderate cases in isolation; default = 2
Weeks of Stay in Hospital (based on studies of case severity):
- stay_sev - average length of stay (in weeks) of severe cases in hospital; default = 1
- stay_crit - average length of stay (in weeks) of critical cases in hospital; default = 2
Infection Fatality Rates (note: could be updated based on newer research):
- ifr_sev - infection fatality rate of severe cases, based on WHO China Joint Mission Report; default = 0.134
- ifr_crit - infection fatality rate of critical cases, based on Imperial College Report nr. 9; default = 0.5
Miscellaneous Parameters:
- perc_hcws_not_covid - assumption of percentage of HCWs not working on COVID; default = 0.4
- n_hosp_beds_per_care_unit - assumption, used to estimate triple packaging boxes (might not be necessary); default = 40
Healthcare Workers per Bed or Patient:
- ambulancews_per_bed - ambulance personnel ratio assumes 1 ambulance per 100 bed hospital with 2 operators (paramedic + driver) at all times (3x8 hour shifts) so 6/100 beds; default = 0.06
- bioengs_per_bed - biomedical engineer ratio assumes 2 biomedical engineers (on 8-hour shifts) per 100 bed hospital; default = 0.02
- n_inf_caregivers_hosp - reference assumption of zero is based on current guidance that no family members or other caretakers should be in hospitals; default = 0
- n_inf_caregivers_isol - reference is based on management of home care guidance, with estimates of 1 caregiver per patient for the duration of the roughly 2-week isolation - his calculation estimates the quantity of PPE required (e.g., masks and gloves) for the patient and caregiver; default = 1
Percent of Critical Patients that are Mechanically Ventilated:
- perc_crit_inv_mv - WHO recommendation; default = 0.667
- perc_crit_noninv_mv - WHO recommendation; default = 0.333
Oxygen Flow by Case Severity, in LPM:
- o2_flow_sev - severe patients require 5-15 LPM, so median taken; default = 10
- o2_flow_crit_inv_mv - critical patients with invasive mechanical ventilation; default = 30
- o2_flow_crit_noninv_mv - critical patients with non-invasive mechanical ventilation; default = 30
Testing Strategy
# Testing strategy parameters - population level testing parameters (i.e. percent # tested) test_strat <- set_testing_strategy()
Testing Strategies (there are two choices, input as string values):
- all - all cases that present, regardless of severity, using standard number of negatives per positive test
- targeted - restricted testing of mild/moderate presenting cases may be employed if limited tests available. all required severe and critical cases will still be tested. if selected, only the percent of suspected/mild/moderate cases input here will be tested (typically high-risk patients)
Testing Parameters (the rest of the testing strategy parameters, some of which depend on the testing strategy chosen):
- perc_tested_sev_crit - percent of severe or critical cases tested, always 100\%; default = 1
- num_neg_per_pos_test - estimated average number of negative tests per positive test; default = 10
- tests_per_hcw_per_week - tests per HCW or staff member per week. Includes tests for inpatient HCW, screening/triage HCW, ambulance personnel, cleaners, lab techs, and biomedical engineers. input can also be 0 (for no tests) or a decimal (e.g. 0.5, to represent tests every other week); default = 1
- testing_contacts - is testing done for contacts of positive cases; default = TRUE
- avg_contacts_pos_case - average number of contacts per positive case. Suggested options are 5 (High/Strong, e.g. stay at home regulations), 10 (Medium/Weak social distancing, e.g. travel restrictions), 15 (Low/No social distancing, e.g. advice only); default = 10
- perc_contacts_tested - percent of contacts of a positive case who get tested; default = 0.6
Diagnostic Parameters
# Diagnostic parameters - individual level testing parameters (i.e. number of tests # per case per diagnosis, etc.) test_params <- get_diagnostic_parameters()
Total tests and percent antigen:
- total_tests_mild_mod - total tests, used for diagnosis only; default = 1
- total_tests_sev_crit - total tests, used for diagnosis and release; default = 2
- perc_antigen_tests - percent testing done in hospital via antigen testing, max 80 percent; default = 20 percent
- tests_diagnosis_mild_mod - the number of tests for diagnosis for mild or moderate cases; default = 1
- tests_diagnosis_sev_crit - the number of tests for diagnosis for severe or critical cases; default = 1
- tests_release_mild_mod - the number of tests for release for mild or moderate cases; default = 0
- tests_release_sev_crit - the number of tests for release for severe or critical cases; default = 1
Lab Parameters
# Diagnostic lab parameters - number of lab staff per lab, safety boxes per unit, # etc. lab_params <- get_lab_parameters()
Requirements per Lab:
- lab_staff_per_lab - multiplier that helps estimate lab staff equipment requirements; default = 3
- hygienists_per_lab - multiplier that helps estimate lab hygienist/cleaner equipment requirements; default = 3
- safety_boxes_per_unit_week - WHO recommendation for safe sharp disposal; default = 8
- triple_packaging_per_unit - WHO recommendation for sample transport; default = 4
- perc_wastage_manual_test_kits - percentage wastage, only of manual test kits; default = 10\%
- num_tests_manual_test_kits - number of tests in an RT-PCR manual test kit; default = 100
Capacity Mapping
capacity <- get_country_capacity(iso3c = country)
Country-Specific Capacity Estimates:
- country - country name, English
- iso3c - country code, iso3c format
- population - population estimate from the latest available data from the United Nations Development Programme, reported in 2020
- yoy_growth - average year over year growth of country population, estimated from the change in growth in population between UNDP reported numbers in 2015 and 2020
- income_group - World Bank income group
- n_hcws - number of estimated HCWs. Combines latest available estimates of nurses (
Nursing personnel) and doctors (Medical doctors) from WHO GHO with the estimates YoY growth from UNDP to yield estimates of nurses and doctors, the sum of which are the number of healthcare workers. If these are unavailable, uses a combination of income group averages and yoy growth. - n_labs - number of lab staff, estimated using the same method as the number of HCWs (codes:
Medical and Pathology Laboratory scientistsandMedical and Pathology Laboratory Technicians). - n_hosp_beds - estimated beds per country, calculated using similar procedure as above but with World Bank estimates of hospital beds per 1,000 people.
- perc_beds_crit_covid - percent beds for ICU patients per country, estimated from World Bank/Imperial College Report
- perc_beds_not_covid - percent beds not allocated to COVID, default = 0.4
- perc_beds_sev_covid - percent beds allocated to severe COVID- cases, $1 - (\text{perc_beds_crit_covid} + \text{perc_beds_not_covid})$
- beds_covid - number of hospital beds allocated to COVID $(\text{n_hosp_beds}* (1-\text{perc_beds_not_covid}))$
- severe_beds_covid - number of severe beds allocated to COVID, $\text{n_hosp_beds}*\text{perc_beds_sev_covid}$
- crit_beds_covid - number of critical beds allocated to COVID, $\text{n_hosp_beds}*\text{perc_beds_crit_covid}$
Testing Capacity
country_test_capacity <- get_country_test_capacity(iso3c = country)
This uses estimated provided by the WHO Operations, Supply and Logistics Team that are reviewed by diagnostics technical experts at the WHO to provide estimates of counts of various diagnostic processing machines in country.
Types of machines in output:
- roche_6800 - high throughput conventional platform
- roche_8800 - high throughput conventional platform
- abbott_m2000 - high throughput conventional platform
- hologic_panther - high throughput conventional platform
- hologic_panther_fusion - high throughput conventional platform
- genexpert - near-patient PCR machine modules (not platforms: here, one machine contains several modules)
- manual - manual real-time PCR platform
Diagnostic Capacity
The next function, called calc_diagnostic_capacity, takes these machine counts as inputs, in addition to data on their processing capability, hours per shift, and number of shifts per day to estimate the total test processing capability, what we term the diagnostic capacity.
data(throughput, package = "esft") data(hours_per_shift, package = "esft") diagnostic_capacity <- calc_diagnostic_capacity( country_diagnostic_capacity = country_test_capacity, throughput, hours_per_shift = hours_per_shift, shifts_per_day = 1 )
Diagnostic (Total Test Processing) Capacity:
- shifts_day - these are the shifts per day specified in function call
- platform_key - snake case version of the testing platform name
- platform - platform name in plain english
- type - type of diagnostics machine
- days_week - days it operates per week
- covid_capacity - capacity of test processing function dedicated to COVID-19
- country_name - country for which this capacity is calculated
- region - WHO region of that country
- income_group - WB income group of that country
- country_code - iso3c code of that country
- modules_activated - number of total modules (a machine can have several modules) in active use in that country
- hours - hours per shift
- throughput_per_day - number of tests that can be processed by that machine per day (with day being specified by combining hours per shift, and shifts per day and cross referencing that with the throughput input data
- total_test_capacity - this is the total number of tests that the different categories of machines can process in a week, depending on the days per week and the throughput per day ($Capacity_{test-total} = Modules_{activated}Days_{week}Throughput_{day}$).
- covid_test_capacity - this is the total test capacity multiplied by the covid capacity - or the number of tests that can be run for COVID-19 on the different machines every week ($Capacity_{test-COVID} = Capacity_{test-total}*p_{covid-capacity}$)
Total Labs
# Calculates the total labs available for COVID-19 given the diagnostic capacity # (number of modules and machines) available t_labs <- total_labs(diagnostic_capacity)
This calculates the number of labs conducting COVID-19 testing, the reference value assumes 1 lab for every 3 high-throughput, manual or other platforms and 1 lab for every 4 GeneXpert modules.
$$ \begin{align} Modules_{sum} = Modules_{roche-6800} + Modules_{roche-8800} + Modules_{abbott-m2000} +\ Modules_{hologic-panther}+Modules_{hologic-panther-fusion}+Modules_{manual}\ \ Labs_{total} = \frac{Modules_{sum}}{3} + \frac{Modules_{genexpert}}{4} \end{align} $$
Max Tests
# Uses the country specific diagnostic capacity estimates to calculate the max # number of tests per day. This is then used to help cap tests by total testing # capacity in total_tests. max_tests <- max_tests_per_day(diagnostic_capacity)
This calculates the total number of tests per week given covid test capacity, and divides by 7 to find the max number of tests per day (I am not certain that 7 is the correct way to calculate max tests - the machines are presumed to run for a pre-set number of days per week, so I am not sure why this would change here).
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.