| Title: | AvonCap Study Analysis |
|---|---|
| Description: | A WIP set of functions allowing data load, wrangling of the AvonCap data set. |
| Authors: | Rob Challen [aut, cre] |
| Maintainer: | Rob Challen <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.0.0.9028 |
| Built: | 2024-11-06 05:43:50 UTC |
| Source: | https://github.com/bristol-vaccine-centre/avoncap |
Clear data from the passthrough cache for complex or long running operations
.cache_clear( .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = ".*", interactive = TRUE ).cache_clear( .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = ".*", interactive = TRUE )
.cache |
the location of the cache as a directory. May get its value from options("ggrrr.cache.dir") or the default value of rappdirs::user_cache_dir("ggrrr") |
.prefix |
a regular expression matching the prefix of the cached item, so that do selective clean up operations. defaults to everything. |
interactive |
suppress |
nothing. called for side effects
Staleness is determined by the number of days from 2am on the current day in the current time-zone. A item cached for only one day becomes stale at 2am the day after it is cached. The time is configurable and option(cache.time_day_starts = 0) would be midnight. Automated analysis using caches and updated data should ensure that analysis does not cross this time point otherwise it may end up using old data.
.cache_delete_stale( .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = ".*", .stale = Inf ).cache_delete_stale( .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = ".*", .stale = Inf )
.cache |
the location of the cache as a directory. May get its value from options("cache.dir") or the default value of rappdirs::user_cache_dir("ggrrr") |
.prefix |
a name of the operation so that you can namespace the cached files and do selective clean up operations on them |
.stale |
the length of time in days to keep cached data before considering it as stale. |
nothing. called for side effects.
This function copies a remote file to a local cache once and makes sure it is reused.
.cache_download( url, ..., .nocache = getOption("cache.disable", default = FALSE), .cache = getOption("cache.download", rappdirs::user_cache_dir(utils::packageName())), .stale = Inf, .extn = NULL ).cache_download( url, ..., .nocache = getOption("cache.disable", default = FALSE), .cache = getOption("cache.download", rappdirs::user_cache_dir(utils::packageName())), .stale = Inf, .extn = NULL )
url |
the url to download |
... |
ignored |
.nocache |
if set to TRUE all caching is disabled |
.cache |
the location of the downloaded files |
.stale |
how long to leave this file before replacing it. |
.extn |
the file name extension |
the path to the downloaded file
executes expr and saves the output as an RDS file indexed by has of code in expr and the hash of input variables (which should contain any variable inputs)
.cached( .expr, ..., .nocache = getOption("cache.disable", default = FALSE), .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = "cached", .stale = Inf ).cached( .expr, ..., .nocache = getOption("cache.disable", default = FALSE), .cache = getOption("cache.dir", rappdirs::user_cache_dir(utils::packageName())), .prefix = "cached", .stale = Inf )
.expr |
the code the output of which requires caching. Other than a return value this should not create side effects or change global variables. |
... |
inputs that the code in expr depends on and changes in which require the code re-running, Could be Sys.Date() |
.nocache |
an option to defeat the caching which can be set globally as options("cache.disable"=TRUE) |
.cache |
the location of the cache as a directory. May get its value from options("cache.dir") or the default value of rappdirs::user_cache_dir("ggrrr") |
.prefix |
a name of the operation so that you can namespace the cached files and do selective clean up operations on them |
.stale |
the length of time in days to keep cached data before considering it as stale. can also be set by options("cache.stale") |
the output of .expr which will usually be a value
Extracting metadata from the filename where present - particularly hospital, and year number
all_files()all_files()
a dataframe containing filename, path, date, hospital, and study_year fields
AvonCap data has lots of columns which are named in a difficult to remember fashion, composed of data items that have enumerated values with no semantics. This makes displaying them difficult and any filtering done on the raw data inscrutable. Depending on the source of the data some different columns may be present due to differences in the NHS and UoB data sets. The redcap database has some options that may be checklists and some that are radio buttons, both of these end up with mysterious names in the data.
augment_data(x, ...)augment_data(x, ...)
x |
|
... |
Arguments passed on to
|
This function maps the data into a tidy dataframe with consistently named columns, and named factors where appropriate. If not present in the data the ethnicity
files Most of the sanitisation code is held in the
zzz-avoncap-mappings.R file.
a tracked dataframe with
This sequences, catches errors and allows parameters to be passed by name
augment_generic(df, ...)augment_generic(df, ...)
df |
a data frame |
... |
unnamed parameters are a list of functions, named parameters are passed to those functions (if they match formal arguments). |
the altered df
fn1 = function(df,v) {df %>% dplyr::filter(cut=="Fair") %>% dplyr::mutate(x_col = color)} fn2 = function(df,v) {df %>% dplyr::filter(color==v$color$J)} df = ggplot2::diamonds %>% augment_generic(fn1, fn2)fn1 = function(df,v) {df %>% dplyr::filter(cut=="Fair") %>% dplyr::mutate(x_col = color)} fn2 = function(df,v) {df %>% dplyr::filter(color==v$color$J)} df = ggplot2::diamonds %>% augment_generic(fn1, fn2)
This function plots a stacked bar of proportions for an input set of data
binomial_proportion_points(data, mapping, ..., width = 0.8, size = 0.5)binomial_proportion_points(data, mapping, ..., width = 0.8, size = 0.5)
data |
the data |
mapping |
a aes mapping with at least |
... |
passed to |
width |
width of position dodge |
size |
the bar size |
a ggplot
Deals with some annoying issues classifying integer data sets, such as ages, into groups. where you want to specify just the change over points as integers and clearly label the resulting ordered factor.
cut_integer( x, cut_points, glue = "{label}", lower_limit = -Inf, upper_limit = Inf, ... )cut_integer( x, cut_points, glue = "{label}", lower_limit = -Inf, upper_limit = Inf, ... )
x |
a vector of integer valued numbers, e.g. ages, counts |
cut_points |
a vector of integer valued cut points which define the lower boundaries of conditions |
glue |
a glue spec that may be used to generate a label. It can use |
lower_limit |
the minimum value we should include (this is inclusive for the bottom category) (default -Inf) |
upper_limit |
the maximum value we should include (this is also inclusive for the top category) (default Inf) |
... |
not used |
an ordered factor of the integer
cut_integer(stats::rbinom(20,20,0.5), c(5,10,15)) cut_integer(floor(stats::runif(100,-10,10)), cut_points = c(2,3,4,6), lower_limit=0, upper_limit=10)cut_integer(stats::rbinom(20,20,0.5), c(5,10,15)) cut_integer(floor(stats::runif(100,-10,10)), cut_points = c(2,3,4,6), lower_limit=0, upper_limit=10)
default column naming mappings
default_column_names(...)default_column_names(...)
... |
additional named items to add |
a set of mappings
The denominator is a time varying quantity
data(denom_by_age_by_day)data(denom_by_age_by_day)
A dataframe containing the following columns:
method (character) - estimation method. The default is "Campling 2019"
age (character) - the age category
date (date) - the date for which this estimate is valid
population (integer) - the esimtate of the population size for that age group on that day
No default value.
32592 rows and 4 columns
This also will calculate a time interval between admissions. There is also a repeat admission instrument that this does not use.
derive_admission_episode(df, v)derive_admission_episode(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
The 3 category classifications
derive_aLRTD_categories(df, v, ...)derive_aLRTD_categories(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
aetiological:
Confirmed SARS-CoV-2 - implies Infective
No evidence SARS-CoV-2 - implies Infective but not confirmed as SARS-CoV-2
Non-infective - presumed non infective
clinical presentation:
Pneumonia - implies Infective
NP-LRTI - implies Infective
No evidence LRTI (include CRDE and HF)
Some cases do not get a clinical presentation in this. Typically they are people who have an infective cause, but LRTI and pneumonia have been excluded. These could be URTI and or incidental COVID cases.
a dataframe
Create a flag for patients who have been given antivirals
derive_antiviral_status(df, v)derive_antiviral_status(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
Names are normalised by removing commonly mixed up components and
derive_catchment_status(df, v)derive_catchment_status(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
Vaccination is deemed to have had effect if given > 14 days before admission for 1st dose or >7 days before admission for subsequent doses. This does not account for previous infection which is not in the data set.
derive_completed_vaccination_status(df, v, ...)derive_completed_vaccination_status(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
Typically used in regression models with non-linear effects over splines
derive_continuous_categories(df, v, ...)derive_continuous_categories(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Age category - UK demographic data ends at 85, and 65 key cut off in 5 year bands, so 10 year bands age categories end at 85 (N.b.) there is a more principled reason here. Boundaries fall approx 0.1, 0.2, 0.4, 0.6, 0.8 quantiles. Could merge first two groups but outcomes are usually different. Covid vaccination cohorts were in 5 year age groups, but vaccination prioirity was in these groups approximately.
Age of eligibility for vaccines: 65+ Age of pneumovax eligibility
CCI - 4 bands as defined in original Charleson paper: ** https://pubmed.ncbi.nlm.nih.gov/3558716/ ** in https://link.springer.com/article/10.1007/s10654-021-00802-z there is rationale given for not using the charleson score as a continuous value.
Alternate CCI - 0,1,2,3+ is also used as a grouping in the original charleson paper
Rockwood score - Completely independent versus dependent frailty levels.
CURB65 categorisation - As per derivation study (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1746657/): 0-1 consider home treatment; 2 consider admit as inpatient; 3-5 admit, consider ICU.
a dataframe
This should be consistent with AvonCAP age / CURB cateories.
derive_continuous_categories_pneumo(df, v, ...)derive_continuous_categories_pneumo(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
SARS-CoV-2 PCR positive only lab confirmed diagnosis.
derive_covid_status(df, v, ...)derive_covid_status(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
admission.covid_pcr_result:
based on fields: c19_adm_swab and covid_19_diagnosis
Patient reported, clinical diagnoses are assumed PCR negative (although possible in some cases they may not have been done).
Lateral flows done in hospital are counted as PCR negative.
negative admission swabs are counted as negative
NA signifies test not done.
admission.is_covid:
Binary confirmed or no-evidence.
PCR results count as confirmed,
Lateral flow results count as confirmed,
anything else is no evidence (includes negatives and test not done)
a dataframe
Pneumonia if one of:
Standard of care diagnosis of CAP (radiologically or clinically)
Empyema or abscess
Admission chest X-ray shows pneumonia
derive_diagnosis_categories(df, v)derive_diagnosis_categories(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
NP-LRTI if:
Not pneumonia and Standard of care LTRI diagnosis
Exacerbation of CRDE:
Standard of care exacerbation COPD
Standard of care exacerbation Non-COPD
(N.B. may be pneumonia or NP-LRTI)
Heart failure:
Standard of care congestive heart failure.
a dataframe
This does not account for previous infection which is not in the data set.
derive_effective_vaccination_status(df, v, ...)derive_effective_vaccination_status(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
This relies on date period during which we are very confidence that the only variants circulating are of a given type. These are quite conservative estimates based on the frequency of sequenced cases in the bristol area (according to the Sanger centre and to cases identified in the hospital testing)
derive_genomic_variant(df, v, ...)derive_genomic_variant(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Pre-alpha before 05 Dec 2020
Alpha between 13 Feb 2021 and 15 May 2021
Delta between 01 Jun 2021 and 07 Nov 2021
Omicron from 07 Feb 2022 to present
a dataframe
Identify patients from the GP surgeries in linked primary care study
derive_gp_linkage(df, v)derive_gp_linkage(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
Elevated troponin : > 18: 18ng/L is simply the 99th percentile value Beckman assay we use as quoted by the IFCC. We elected to not use sex-specific 99th percentile values although they are also quoted here and you could incorporate into your analysis. I am sure you are aware of the 4th Universal definition of MI that requires a rise or fall above the 99th percentile etc.
derive_haematology_categories(df, v, ...)derive_haematology_categories(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
These outcomes were tested in the Delta vs Omicron severity paper and sensitivity analysis. These are only defined for COVID cases.
derive_hospital_burden_outcomes(df, v, ...)derive_hospital_burden_outcomes(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
O2 requirement within 7 days (various cut-offs)
Any respiratory support in 7 days (various cut-offs)
LOS > X days in first 7 days (various cut-offs)
a dataframe
Infective admissions are defined as any of:
pneumonias
NP-LRTI
laboratory confirmed COVID diagnosis
admission swab COVID positive
derive_infective_classification(df, v)derive_infective_classification(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
Infective admissions are excluded if:
Standard of care states non-infectious process
SOC non-LRTI (and none of the other categories above)
Any unknowns are defined as non-Infective
a dataframe
Pneumococcal invasive status and binary test category
derive_invasive_status(df, ...)derive_invasive_status(df, ...)
df |
the dataframe. |
... |
ignored |
a dataframe
Only relevant to SARS-CoV-2 PCR positive patient. Timing of positive test compared to admission: This relies on knowing dates and hence only works on the identifiable data sets,
derive_nosocomial_covid_status(df, v, ...)derive_nosocomial_covid_status(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Logic is:
Community if PCR result predates admission
Probably commuinity if PCR result within 7 days of admission
Probably nosocomial if 7-28 days after admission
Otherwise is it undefined.
a dataframe
Hospital acquired COVID is recorded explicitly in 2 places for some patients. A large difference between admission date and enrollment date (<21 days) is suggestive in other cases. The data is probably only collected in COVID cases so shoudl be treated with caution.
derive_nosocomial_status(df, v)derive_nosocomial_status(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
Date columns
derive_pandemic_timings(date_col, prefix)derive_pandemic_timings(date_col, prefix)
date_col |
the date column |
prefix |
a prefix for the columns to be added |
a derive_... style function to augment a data set containing date_col
with a set of columns describing the timing.
The patient identifier is derived from the record number or the first record number (ensuring it matches) an entry in the record number. This deals with multiple admissions in the data set. In the patient identifiable NHS data this is the NHS number.
derive_patient_identifier(df, v)derive_patient_identifier(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
a dataframe
A range of useful serotype groups is defined in the list uad_groups. The
default_pcv_map gives a set of mappings to group headings that gives the
overall serotype distribution by vaccine.
derive_pcv_groupings( df, ..., pcv_map = uad_pcv_map, not_matched = "Other", col_name = "pneumo.pcv_group" )derive_pcv_groupings( df, ..., pcv_map = uad_pcv_map, not_matched = "Other", col_name = "pneumo.pcv_group" )
df |
the normalised urine antigen data |
... |
ignored |
pcv_map |
a 2 column data frame mapping |
not_matched |
what to call the column of non-matched serotypes? Default is
|
col_name |
the target column name for the pcv grouping (defaults
to |
The logic employed in combining elements is:
any(result == "Unknown") ~ "Unknown"
any(result == "Positive") ~ "Positive"
all(result == "Negative") ~ "Negative"
TRUE ~ "Other"
an augmented data frame with an additional column defined by col_name
For the longitudinal oneumocococcal data, a range of useful serotype groups
is defined in the list avoncap::serotype_data. The avoncap::serotype_pcv_map gives a set of
mappings to (multiple) group headings that gives the overall serotype distribution by
vaccine.
derive_phe_pcv_group(df, v, ...)derive_phe_pcv_group(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
A list of presentations based on site which
LRTI
Meningitis
Effusion/Empyema
Septic arthritis
URTI
Other
derive_pneumo_clinical_syndrome(df, v, ...)derive_pneumo_clinical_syndrome(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
Needed for:
derive_simpler_comorbidities
derive_pneumococcal_high_risk
derive_pneumococcal_risk_category
derive_pneumo_polyfill(df, ...)derive_pneumo_polyfill(df, ...)
df |
the dataframe. |
... |
ignored |
a dataframe
The panels are UAD1 for PCV13 serotypes, UAD2 for PPV23 serotypes.
derive_pneumo_uad_panel(df, ...)derive_pneumo_uad_panel(df, ...)
df |
a pneumo serotype dataframe |
... |
ignored |
a dataframe with additional columns pneumo.uad1_panel_result,
pneumo.uad2_panel_result, pneumo.non_uad_panel_result,
pneumo.serotype_summary_result
logic is defined in derive_pcv_groupings().
derive_pneumo_uad_status(df, ...)derive_pneumo_uad_status(df, ...)
df |
a pneumo serotype dataframe |
... |
ignored |
a dataframe with additional columns pneumo.uad1_panel_result,
pneumo.uad2_panel_result, pneumo.non_uad_panel_result,
pneumo.serotype_summary_result
The 4 category disjoint classification.
derive_pneumococcal_categories(df, v, ...)derive_pneumococcal_categories(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
pneumo.presentation_class:
CAP+/RAD+ - radiologically proved pneumonia
CAP+/RAD- - pneumonia without x-ray confirmation
NP-LRTI - non-pneumonic lower respiratory tract infection
No evidence LRTI - believed to be non-infective at admission, this last group is usually discarded from analysis, however it only really describes people without a clinical diagnosis of LRTI on admission. There could still be undiagnosed infection there, and some of these patients have COVID (possibly without lower respiratory symptoms?).
a dataframe
High pneumococcal risk defined if any of the following:
over 65 years old
other pneumococcal risks
comorbid copd
interstitial lung disease
cystic fibrosis
hypertension
CCF
ischaemic heart disease
chronic kidney disease
chronic liver disease
diabetes
asthmatic with immunodeficiency
on immunosupression
derive_pneumococcal_high_risk(df, v, ...)derive_pneumococcal_high_risk(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
Original algorithm from B1851202 SAP defines a 3 class risk group:
derive_pneumococcal_risk_category(df, v, ...)derive_pneumococcal_risk_category(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
High-risk (immunocompromised)
Asplenia - not supported
Cancer/Malignancy, Hematologic - OK
Cancer/Malignancy, Solid Tumor - OK
Chronic Kidney Disease - OK
Human Immunodeficiency Virus (HIV) – AIDS - OK
Human Immunodeficiency Virus (HIV) – No AIDS - OK
Immunodeficiency - OK
Immunosuppressant Drug Therapy - OK
Organ Transplantation - OK
Multiple Myeloma - not supported
At Risk (immunocompetent)
Asthma - OK
Alcoholism - OK
Celiac Disease - not supported
Chronic Liver Disease without Hepatic Failure - OK
Chronic Liver Disease with Hepatic Failure - OK
Chronic Obstructive Pulmonary Disease - OK
Cochlear Implant - not supported
Congestive Heart Failure - OK
Coronary Artery Disease (CAD) - OK
Chronic Neurologic Diseases - OK
Coagulation factor replacement therapy - not supported
CSF Leak - not supported
Diabetes Treated with Medication - OK
Down syndrome - OK
Institutionalized in nursing home or LTC facility (Nursing home or long-term care facility for those with disability or dependency on subject characteristics/risk determinants eCRF page) - OK
Occupational risk with exposure to metal fumes - OK
Other Chronic Heart Disease - OK
Other Chronic Lung Disease - OK
Other pneumococcal disease risk factors - OK
Previous Invasive Pneumococcal Disease - not supported
Tobacco smoking (Tobacco/E-Cigarettes) - OK
Anything else is low risk
a dataframe
Some basic context to allow comparison to ED data.
derive_polyfill_central(df, v, ...)derive_polyfill_central(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
All of the patients admitted
a dataframe
The ED data has some different fields from the main avoncap data.
derive_polyfill_ed(df, v, ...)derive_polyfill_ed(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
It is missing an admission cxr summary field needed to calculate pneumonia
It has a fixed admission route of "A&E" (i.e. ED to non UK people)
None of the patients admitted
Hospital admission length of stay is zero
a dataframe
Pneumonia if one of:
Initial diagnosis of CAP (supported by initial radiology or clinically)
Empyema or abscess
derive_presumed_diagnosis_categories(df, v)derive_presumed_diagnosis_categories(df, v)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
Presumed clinical presentation:
Pneumonia - implies Infective
NP-LRTI - implies Infective
No evidence LRTI (include CRDE and HF)
a dataframe
uses inbuilt imd_to_townsend map. This implements a cut down version of the QCovid2 score depending on what data is available.
derive_qcovid(df, v = avoncap_df %>% get_value_sets())derive_qcovid(df, v = avoncap_df %>% get_value_sets())
df |
a normalised avoncap data source |
v |
a value set |
the same dataframe with additional columns,
qcovid2.log_hazard, covid2.hazard_ratio: a log hazard rate for the QCOVID2 score where missing data is substituted with the reference value for the QCOVID2 population.
qcovid2.log_comorbid_hazard, qcovid2.comorbid_hazard_ratio: a log hazard rate for the comorbid conditions and not including age and BMI.
Split a continuous variable into quintiles
derive_quintile_category(col, labels = c("1-short", "2", "3", "4", "5-long"))derive_quintile_category(col, labels = c("1-short", "2", "3", "4", "5-long"))
col |
the continuous data column that is to be categorised by quintile. |
labels |
the category labels |
a derive_... style function that augments a data set with col xxx
with col xxx_quintile containing the quintiles
Confirmed death within 30 days (subject to potential censoring)
Confirmed death within 1 year (subject to potential censoring). The date of censoring depends on when the mortality data was updated. Currently this is 04 Oct 2024
Confirmed death (any length follow up)
Any ICU admission
derive_severe_disease_outcomes(df, v, ...)derive_severe_disease_outcomes(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
described in aLRTD paper. These outcomes are
a dataframe
and generate some summary values
derive_simpler_comorbidities(df, v, ...)derive_simpler_comorbidities(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
simple DM without insulin dependence
Solid / Haematological / Any cancer present binary indicators
any chronic resp dx: i.e. any of asthma, bronchiectasis, chronic pleural disease, COPD, interstitial lung dx, cyctic fibrosis, other chronic resp dx
any chronic heart disease: pulmonary htn, CCF, IHD, previous MI, congential heart dx, hypertension, AF, other arrythmia, other heart dx, other other heart dx
Stroke or TIA binary
Any immune compromise binary (immunodeficient or on immune suppressants)
a dataframe
Expects as days since admission:
survival.length_of_stay - length of stay until discharge or death (NA if still in hosptial),
survival.uncensored_time_to_death - time until death (NA if alive at last obs),
survival.last_observed_event - last time patient observed alive.
derive_survival_censoring(df, v, ...)derive_survival_censoring(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Calculates
a 30 day survival duration and censoring status for survfit
a 1 year survival duration and censoring status for survfit
Hospital length of stay and censoring status for survfit
Categorical length of stay and 30 day survival 0-3, 4-6, 7-13, 14-29, gte 30
Survival data will be of the form:
survival.30_day_death_xxx, survival.1_yr_death_xxx, survival.30_day_discharge
xxx_time: for this is the follow up time to event in days (max 30 or 365).
xxx_event: The event type indicator
0 = alive at event (censored),
1 = dead.
or for length of stay:
0 = still inpatient / died (censored),
1 = discharged from hospital
A survival model will be of the form:
survival::Surv(time = xxx_time, event=xxx_event) ~ ...
a dataframe
Fixes a data issue with length of stay and survival duration being filled in across 2 columns. and missing last observation dates so that we can calculate survival censoring consistently in other data sets.
derive_survival_times_avoncap(df, v, ...)derive_survival_times_avoncap(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Calculates:
A consistent length of stay - shortest of length of stay and 30 day and 1 yr survival duration
A consistent uncensored time to death - shortest of 30 day and 1 yr survival duration
A consistent time to last observation
a dataframe
Fixes a data issue with length of stay and survival duration being filled in across 2 columns. and missing last observation dates so that we can calculate survival censoring consistently in other data sets.
derive_survival_times_pneumo(df, v, ...)derive_survival_times_pneumo(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
Calculates:
A consistent length of stay - shortest of length of stay and 30 day and 1 yr survival duration
A consistent uncensored time to death - shortest of 30 day and 1 yr survival duration
A consistent time to last observation
a dataframe
Derived data function template
derive_template(df, v, ...)derive_template(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
If symptom duration is not given it is assumed to be zero.
derive_vaccination_timings(df, v, ...)derive_vaccination_timings(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
There are too many potential combinations with 4th, 5th and sixth dose to make this useful.
derive_vaccine_combinations(df, v, ...)derive_vaccine_combinations(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
a dataframe
Scores 0-3 are for community cases.
derive_WHO_outcome_score(df, v, ...)derive_WHO_outcome_score(df, v, ...)
df |
the dataframe. |
v |
the value set. usually precomputed by the augment framework the value
set can be explicitly supplied with |
... |
ignored |
We generally can't tell the difference between 7 and 8.
4: Hospitalised; no oxygen therapy
5: Hospitalised; oxygen by mask or nasal prongs
6: Hospitalised; oxygen by NIV or high flow
7: Intubation and mechanical ventilation, pO2/FiO2 >= 150 or SpO2/FiO2 >= 200
8: Mechanical ventilation pO2/FIO2 <150 (SpO2/FiO2 <200) or vasopressors
9: Mechanical ventilation pO2/FiO2 <150 and vasopressors, dialysis, or ECMO
10: Dead
a dataframe
When a data set is normalised or augmented the original column names are stored as metadata. This helps us determine how a particular item was created. In future this will be useful for documentation.
extract_dependencies(data, col, original = TRUE)extract_dependencies(data, col, original = TRUE)
data |
the dataframe |
col |
the column as a symbol |
original |
map the names to the original column names from the data. If this is false the function returns a list of current normalised column names. |
a named list of dependencies and original column names for a given column
Get the transformed columns from original field names
find_new_field_names(normalised, fields)find_new_field_names(normalised, fields)
normalised |
the transformed data set. |
fields |
a vector of field names |
a named list mapping original to new columns
The list of validation, normalisation and augmentation frameworks. There should be one validation per data set. The may be mulitple normalisations and augmentations depending on the aspect of the data we are extracting (e.g. re-nesting flattened data.)
This function examines a dataframe and returns a list of the columns with sub-lists as all the options for factors. This provides programmatic access (and automcomplete) to the values available in a dataframe, and throws and early error if we try and access data by a variable that does not exist.
get_value_sets(df)get_value_sets(df)
df |
a dataframe to examine |
a list of lists with the column name and the factor levels as list, as a checked list.
The denominator relates only to patients coming from these GP surgeries
data(icb_surgeries)data(icb_surgeries)
A dataframe containing the following columns:
code - an official ODS code for the GP surgery
name - the official surgery name.
82 rows and 2 columns
A high level mapping from IMD to Townsend score This is inaccurate as townsend score
A data frame with 10 rows and 2 columns:
The IMD
the average townsend score for this IMD
...
Locate the input directory
input(...)input(...)
... |
the sub paths within the input directory |
a path to the input directory and sub paths if provided
# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })
A list of key dates:
mortality_updated - the last time the NHS mortality data was extracted and added to AvonCAP
min_alpha - earliest observation of the alpha variant
max_wuhan - last observation of the wuhan variant
min_delta - earliest observation of the delta variant
max_alpha - last observation of the alpha variant
min_omicron - earliest observation of the omicron variant
max_delta - last observation of the delta variant
The default catchment population for AvonCAP is limited to the Bristol, North Somerset and South Gloucestershire Integrated Care Board (BNSSG ICB). This list is the list of GP surgeries considered part of the denominator.
code - the NHS ODS organisational code of the practice.
name - the official name of the practice
Faceted Kaplan-Meier plot
km_plot( df, coxmodel, facet = NULL, ..., maxtime = NULL, ylab = if (!invert) "surviving (%)" else "affected (%)", xlab = "time (days)", facetlab = NULL, ylim = (if (invert) c(0, NA) else c(NA, 100)), n_breaks = 5, heights = c(10, 1), invert = FALSE, show_label = FALSE, show_legend = TRUE )km_plot( df, coxmodel, facet = NULL, ..., maxtime = NULL, ylab = if (!invert) "surviving (%)" else "affected (%)", xlab = "time (days)", facetlab = NULL, ylim = (if (invert) c(0, NA) else c(NA, 100)), n_breaks = 5, heights = c(10, 1), invert = FALSE, show_label = FALSE, show_legend = TRUE )
df |
the data |
coxmodel |
the cox model output of survival::coxph from the data |
facet |
the division to highlight in the KM strata. Defaults to first term on the lhs of the cox model formula |
... |
Arguments passed on to
|
maxtime |
the longest x value to plot (optional) |
ylab |
the y axis label |
xlab |
the x axis label |
facetlab |
a label to add as a facet title |
ylim |
the range to show on the KM plot |
n_breaks |
number of x axis breaks to display this also determines the timing and number of "at risk" counts to display. |
heights |
the relative height between the KM plot and the "at risk" table |
invert |
reverse survival statistics to count number of affected |
show_label |
show the label on the at risk table ( which is somewhat redundant as items are coloured) |
show_legend |
show the legend for the strata. (This is sometimes redundant if the at risk table is labelled) |
a ggplot patchwork.
cox = survival::coxph(survival::Surv(time, status) ~ trt + celltype + karno + diagtime + age + prior , data = survival::veteran) km_plot(survival::veteran, cox) km_plot(survival::veteran, cox, facet = 1) km_plot(survival::veteran, cox, "celltype", show_label=TRUE) & ggplot2::theme(legend.position="bottom") km_plot(survival::veteran, cox, "trt", show_label=TRUE) & ggplot2::theme(legend.position="bottom")cox = survival::coxph(survival::Surv(time, status) ~ trt + celltype + karno + diagtime + age + prior , data = survival::veteran) km_plot(survival::veteran, cox) km_plot(survival::veteran, cox, facet = 1) km_plot(survival::veteran, cox, "celltype", show_label=TRUE) & ggplot2::theme(legend.position="bottom") km_plot(survival::veteran, cox, "trt", show_label=TRUE) & ggplot2::theme(legend.position="bottom")
Loads the AvonCap data from a set of csv files, which may optionally be
qualified by site ('BRI' or 'NBT') and database year ('y1', 'y2', 'y3')
as part of the file name. This selects the most recent files earlier than the
reproduce_at date and detects whether they are in a set of files.
load_data( type, subtype = NULL, reproduce_at = as.Date(getOption("reproduce.at", default = Sys.Date())), merge = TRUE, ... )load_data( type, subtype = NULL, reproduce_at = as.Date(getOption("reproduce.at", default = Sys.Date())), merge = TRUE, ... )
type |
the file category see |
subtype |
the subtype from |
reproduce_at |
|
merge |
|
... |
|
The files are loaded as csv as checked that files have (A) the same columns, (B) the same type (or are empty) (C) have any major parse issues. It then merges the files into a single dataframe, if possible, otherwise it will return the individually loaded files as a list of dataframes.
either a list of dataframes or a single merged dataframe
try(load_data("nhs-extract","deltave"))try(load_data("nhs-extract","deltave"))
record_number -> admin.record_number (name)
what_was_the_first_surveil -> admin.first_record_number (name)
ac_study_number -> admin.consented_record_number (study_id)
nhs_number -> admin.patient_identifier (ppi)
duplicate -> admin.duplicate (yesno)
enrollment_date -> admin.enrollment_date (date)
admission_type -> admission.admission_route (list)
study_year -> admin.study_year (name)
file -> admin.data_file (name)
week_number -> admin.week_number (name)
c19_diagnosis -> diagnosis.standard_of_care_COVID_diagnosis (list)
clinical_radio_diagnosis -> diagnosis.clinical_or_radiological_LRTI_or_pneumonia (yesno)
c19_adm_swab -> diagnosis.admission_swab (list)
c19_test_type -> diagnosis.test_type (list)
qualifying_symptoms_signs -> diagnosis.qualifying_symptoms_signs (name)
cc_critieria -> diagnosis.meets_case_control_criteria (yesno)
cc_pos_date -> diagnosis.first_COVID_positive_swab_date (date)
gender -> demog.gender (list)
age_at_admission -> demog.age (double)
age_march -> demog.age_in_march_2021 (double)
imd -> demog.imd_decile (name)
gp_practice -> admin.gp_practice_old (name)
gp_practice_drop_down -> admin.gp_practice (list)
smoking -> demog.smoker (list)
ethnicity2 -> demog.ethnicity (list)
care_home -> demog.care_home_resident (yesno)
hapcovid_screening -> admission.non_lrtd_hospital_acquired_covid (yesno)
hospital_covid -> admission.hospital_acquired_covid (yesno)
drugs -> demog.no_drug_abuse, demog.alcohol_abuse, demog.ivdu_abuse, demog.marijuana_abuse, demog.other_inhaled_drug_abuse (checkboxes)
vaping -> demog.vaping (list)
alc_units -> demog.units_of_alcohol (name)
np_swab -> admin.np_swab_taken_1 (list)
adm_np_type -> admin.np_swab_site_1 (list)
np_date -> admin.np_swab_date_1 (date)
days_adm_npswab -> admin.np_swab_day_since_admission (double)
np_swab_2 -> admin.np_swab_taken_2 (list)
adm_np_type_2 -> admin.np_swab_site_2 (list)
np_date_2 -> admin.np_swab_date_2 (date)
np_swab_3 -> admin.np_swab_taken_3 (list)
adm_np_type_3 -> admin.np_swab_site_3 (list)
np_date_3 -> admin.np_swab_date_3 (date)
saliva -> admin.saliva_sample_taken (list)
saliva_date -> admin.saliva_sample_date (date)
days_adm_saliva -> admin.saliva_sample_day_since_admission (double)
sputum -> admin.sputum_sample_taken (list)
sputum_date -> admin.sputum_sample_date (date)
days_adm_sputum -> admin.sputum_sample_day_since_admission (double)
pt_ad_ur -> admin.urine_sample_needed (yesno)
adm_ur_taken -> admin.urine_sample_taken (list)
nourine_reason -> admin.urine_sample_failure_reason (list)
adm_np_type_2 -> admin.urine_sample_site (list)
adm_ur_date -> admin.urine_sample_date (date)
days_adm_urine -> admin.urine_sample_day_since_admission (double)
adm_serum_tak -> admin.serum_sample_taken (list)
adm_seru_date -> admin.serum_sample_date (date)
days_adm_serum -> admin.serum_sample_day_since_admission (double)
contraindication -> vaccination.covid_vaccine_contraindicated (yesno)
covid19_vax -> vaccination.covid_vaccination (list)
covidvax_date -> vaccination.first_dose_date (date)
covidvax_dose_2 -> vaccination.second_dose_date (date)
covidvax_dose_3 -> vaccination.third_dose_date (date)
covidvax_dose_4 -> vaccination.fourth_dose_date (date)
covidvax_dose_5 -> vaccination.fifth_dose_date (date)
covidvax_dose_6 -> vaccination.sixth_dose_date (date)
brand_of_covid19_vaccinati -> vaccination.first_dose_brand (list)
covid19vax_brand_2 -> vaccination.second_dose_brand (list)
covid19vax_brand_3 -> vaccination.third_dose_brand (list)
covid19vax_brand_4 -> vaccination.fourth_dose_brand (list)
covid19vax_brand_5 -> vaccination.fifth_dose_brand (list)
covid19vax_brand_6 -> vaccination.sixth_dose_brand (list)
c19vaxd1_adm -> admission.time_since_first_vaccine_dose (name)
c19vaxd2_adm -> admission.time_since_second_vaccine_dose (name)
c19vaxd3_adm -> admission.time_since_third_vaccine_dose (name)
c19vaxd4_adm -> admission.time_since_fourth_vaccine_dose (name)
c19vax5_adm -> admission.time_since_fifth_vaccine_dose (name)
c19vax6_adm -> admission.time_since_sixth_vaccine_dose (name)
flu_date -> vaccination.last_flu_dose_date (date)
fluvax_adm_d1 -> admission.time_since_last_flu_vaccine_dose (name)
ppv23_date -> vaccination.last_pneumococcal_dose_date (date)
ppv23vax_adm_d -> admission.time_since_last_pneumococcal_vaccine_dose (name)
c19_variant -> genomic.variant (variant)
year -> admission.year (double)
study_week -> admission.study_week (double)
admission_date -> admission.date (date)
hospital -> admin.hospital, toupper (text_to_factor)
adm_diagnosis -> admission.presumed_CAP_radiologically_confirmed, admission.presumed_CAP_clinically_confirmed, admission.presumed_CAP_no_radiology, admission.presumed_LRTI, admission.presumed_Empyema_or_abscess, admission.presumed_exacerbation_COPD, admission.presumed_exacerbation_non_COPD, admission.presumed_congestive_heart_failure, admission.presumed_non_infectious_process, admission.presumed_non_LRTI (checkboxes)
ics -> admission.on_inhaled_corticosteroids (yesno)
immsup -> admission.on_immunosuppression (yesno)
psi_class -> admission.pneumonia_severity_index_class (list)
crb_test_mai -> admission.curb_65_severity_score (list)
news_2_total -> admission.news2_score (name)
pulse_ox -> admission.oximetry (name)
rr -> admission.respiratory_rate (name)
fio2 -> admission.max_oxygen (name)
systolic_bp -> admission.systolic_bp (name)
diastolic_bp -> admission.diastolic_bp (name)
hr -> admission.heart_rate (name)
temperature -> admission.temperature (list)
symptom_days_preadmit -> admission.duration_symptoms (double)
previous_infection -> admission.previous_covid_infection (yesno_unknown)
previousinfection_date -> admission.previous_covid_infection_date (date)
c19d_preadm -> admission.time_since_covid_diagnosis (name)
rockwood -> admission.rockwood_score (name)
cci_total_score -> admission.charlson_comorbidity_index (name)
height -> admission.height (name)
weight -> admission.weight (name)
bmi -> admission.BMI (double)
first_radio -> admission.cxr_normal, admission.cxr_pneumonia, admission.cxr_heart_failure, admission.cxr_pleural_effusion, admission.cxr_covid_changes, admission.cxr_other (checkboxes)
c19_peep -> day_7.max_peep (name)
c19_hospadm -> day_7.length_of_stay (list)
c17_high -> day_7.max_care_level (list)
c19icuon -> day_7.still_on_icu (yesno)
c19_icudays -> day_7.icu_length_of_stay (list)
c19_vent -> day_7.max_ventilation_level (list)
c19_ox -> day_7.max_o2_level (list)
c19_ionotropes -> day_7.ionotropes_needed (yesno)
c19_complication -> day_7.PE, day_7.DVT, day_7.ARF, day_7.NSTEMI, day_7.STEMI, day_7.cardiac_failure, day_7.new_AF, day_7.new_other_arrythmia, day_7.inpatient_fall, day_7.other_complication, day_7.no_complication (checkboxes)
c19_death7d -> day_7.death (yesno)
c19_meds -> treatment.dexamethasone, treatment.remdesevir, treatment.tocilizumab, treatment.sarilumab, treatment.in_drug_trial, treatment.no_drug_treatment, treatment.sotrovimab (checkboxes)
hospital_length_of_stay -> outcome.length_of_stay, floor (integer)
survival_days -> outcome.survival_duration, round (integer)
ip_death -> outcome.inpatient_death (yesno)
days_in_icu -> outcome.icu_duration (double)
did_the_patient_have_respi -> outcome.respiratory_support_needed (yesno)
number_of_days_of_ventilat -> outcome.ventilator_duration (double)
ett_days -> outcome.endotracheal_tube_duration (double)
renal_replacement_therapy -> outcome.renal_support_duration (double)
complications -> outcome.acute_renal_failure, outcome.liver_dysfunction, outcome.hospital_acquired_infection, outcome.acute_respiratory_distress_syndrome, outcome.NSTEMI, outcome.STEMI, outcome.new_AF, outcome.new_other_arrhthmia, outcome.stroke, outcome.DVT, outcome.PE, outcome.heart_failure, outcome.fall_in_hospital, outcome.reduced_mobility, outcome.increasing_care_requirement, outcome.no_complications (checkboxes)
ventilatory_support -> outcome.highest_level_ventilatory_support (list)
did_the_patient_receive_ec -> outcome.received_ecmo (yesno)
inotropic_support_required -> outcome.received_ionotropes (yesno_unknown)
lrtd_30d_outcome -> outcome.functional_status (list)
survive_1yr -> outcome.one_year_survival (yesno)
survival_1yr_days -> outcome.one_year_survival_duration (integer)
yr_survival_complete -> outcome.one_year_survival_complete (list)
fever2 -> symptom.abnormal_temperature (yesno)
pleurtic_cp -> symptom.pleuritic_chest_pain (yesno)
cough2 -> symptom.cough (yesno)
sput_prod -> symptom.productive_sputum (yesno)
dyspnoea -> symptom.dyspnoea (yesno)
tachypnoea2 -> symptom.tachypnoea (yesno)
confusion -> symptom.confusion (yesno)
anosmia -> symptom.anosmia (yesno_unknown)
ageusia -> symptom.ageusia (yesno_unknown)
dysgeusia -> symptom.dysguesia (yesno_unknown)
fever -> symptom.fever (yesno_unknown)
hypothermia -> symptom.hypothermia (yesno_unknown)
chills -> symptom.chills (yesno_unknown)
headache -> symptom.headache (yesno_unknown)
malaise -> symptom.malaise (yesno_unknown)
wheeze -> symptom.wheeze (yesno_unknown)
myalgia -> symptom.myalgia (yesno_unknown)
worse_confusion -> symptom.worsening_confusion (yesno_unknown)
general_det -> symptom.general_deterioration (yesno_unknown)
ox_on_admission -> symptom.oxygen_required_on_admission (yesno_unknown)
resp_disease -> comorbid.no_resp_dx, comorbid.copd, comorbid.asthma, comorbid.resp_other (checkboxes)
other_respiratory_disease -> comorbid.bronchiectasis, comorbid.interstitial_lung_dx, comorbid.cystic_fibrosis, comorbid.pulmonary_hypertension, comorbid.chronic_pleural_dx, comorbid.other_chronic_resp_dx (checkboxes)
chd -> comorbid.no_heart_dx, comorbid.ccf, comorbid.ihd, comorbid.hypertension, comorbid.other_heart_dx (checkboxes)
mi -> comorbid.previous_mi (yesno)
other_chd -> comorbid.congenital_heart_dx, comorbid.af, comorbid.other_arrythmia, comorbid.pacemaker, comorbid.valvular_heart_dx, comorbid.other_other_heart_dx (checkboxes)
diabetes -> comorbid.diabetes (list)
dm_meds -> comorbid.diabetes_medications (list)
neurological_disease -> comorbid.neuro_other, comorbid.cva, comorbid.tia, comorbid.hemiplegia, comorbid.paraplegia, comorbid.no_neuro_dx (checkboxes)
dementia -> comorbid.no_dementia, comorbid.dementia, comorbid.cognitive_impairment (checkboxes)
cancer -> comorbid.solid_cancer (list)
haem_malig -> comorbid.no_haemotological_cancer, comorbid.leukaemia, comorbid.lymphoma (checkboxes)
ckd -> comorbid.ckd (list)
liver_disease -> comorbid.liver_disease (list)
gastric_ulcers -> comorbid.gastric_ulcers (yesno)
pvd -> comorbid.periph_vasc_dx (yesno)
ctd -> comorbid.connective_tissue_dx (yesno)
immunodeficiency -> comorbid.immunodeficiency (yesno)
other_pn_disease -> comorbid.other_pneumococcal_risks (yesno)
transplant -> comorbid.transplant_recipient (yesno)
pregnancy -> comorbid.pregnancy (list)
hiv -> comorbid.no_HIV, comorbid.HIV, comorbid.AIDS (checkboxes)
final_soc_lrtd_diagnosis -> diagnosis.SOC_CAP_radiologically_confirmed, diagnosis.SOC_CAP_clinically_confirmed, diagnosis.SOC_CAP_no_radiology, diagnosis.SOC_LRTI, diagnosis.SOC_Empyema_or_abscess, diagnosis.SOC_exacerbation_COPD, diagnosis.SOC_exacerbation_non_COPD, diagnosis.SOC_congestive_heart_failure, diagnosis.SOC_non_infectious_process, diagnosis.SOC_non_LRTI (checkboxes)
covid_19_diagnosis -> diagnosis.covid_19_diagnosis (list)
ppv23 -> vaccination.pneumovax (list)
flu_vaccine -> vaccination.influenza_vaccination (list)
abx_14d_prior -> admission.pre_admission_antibiotics_given (yesno_unknown)
antibiotic_used -> admission.pre_admission_antibiotic (checkboxes_to_nested_list)
antiplatelets -> admission.antiplatelet_therapy (list)
anticoagulants -> admission.anticoagulant_therapy (list)
statins -> admission.cholesterol_lowering_therapy (list)
hypertensives -> admission.antihypertensive_therapy (list)
antiviral_14d_prior -> admission.pre_admission_antiviral (checkboxes_to_nested_list)
map_avoncap_central()map_avoncap_central()
a list
consented -> admin.consented (list)
ppc -> admin.pp_consented (list)
withdrawal -> admin.withdrawal (yesno)
consent_urine -> admin.consent_for_urine (yesno)
consent_blood -> admin.consent_for_blood (yesno)
consent_resp_samples1 -> admin.consent_for_respiratory_samples (yesno)
map_avoncap_consent()map_avoncap_consent()
a list
All the ED data is also mapped using the map_avoncap_central() list
as it si quite similar
map_avoncap_ed()map_avoncap_ed()
ed_hours -> outcome.emergency_dept_length_of_stay (name)
ed_reattendance -> admin.ed_episodes_in_last_30_days (name)
hosp_adm_30d -> outcome.admitted_within_30_days (yesno)
hosp_adm_7d -> outcome.admitted_within_7_days (yesno)
home_d_1 -> outcome.days_since_last_ed_episode (name)
radiology_result_1___2 -> radio.consistent_with_pneumonia_1 (yesno)
radiology_result_2___2 -> radio.consistent_with_pneumonia_2 (yesno)
a list
consented -> admin.consented (list)
ppc -> admin.pp_consented (list)
withdrawal -> admin.withdrawal (yesno)
consent_urine -> admin.consent_for_urine (yesno)
consent_blood -> admin.consent_for_blood (yesno)
consent_resp_samples1 -> admin.consent_for_respiratory_samples (yesno)
map_avoncap_ed_consent()map_avoncap_ed_consent()
a list
record_number -> admin.record_number (name)
ac_study_number -> admin.consented_record_number (study_id)
ph_7_35 -> haem.blood_gas_ph (double)
glucose -> haem.glucose (double)
albumin -> haem.albumin (double)
wcc -> haem.white_cell_count (double)
eos -> haem.eosinophils (double)
hb -> haem.haemoglobin (double)
haematocrit -> haem.haemotocrit (double)
pmn -> haem.neutrophils (double)
lymphocytes -> haem.lymphocytes (double)
crp -> haem.crp (double)
na_result -> haem.sodium (double)
ur_result -> haem.urea (double)
egfr -> haem.egfr (double)
sars_cov2_antigen -> haem.sars_cov2_antigen (trunc_double)
ferritin -> haem.ferritin (double)
troponin -> haem.troponin (double)
nt_probnp -> haem.pro_bnp (double)
d_dimer -> haem.d_dimer (double)
patient_blood_group -> haem.blood_group (list)
map_avoncap_haem()map_avoncap_haem()
a list
microtest_done -> micro.test_performed (yesno)
microtest_date -> micro.test_date (date)
microday -> micro.test_days_from_admission (pos_integer)
micro_test -> micro.test_type (list)
micro_isolates -> micro.pathogen_detected (yesno_unknown)
isolate_identified -> micro.pathogen, .micro_isolate_list (checkboxes_to_nested_list)
pn_result -> micro.pneumo_serotype_status (list)
pn_st -> micro.pneumo_serotype (pneumo_serotype)
micro_lab -> micro.sent_to_central_lab (yesno_unknown)
pen_susceptibility -> micro.penicillin_susceptibility (checkboxes_to_list)
septrin_susceptibility -> micro.septrin_susceptibility (checkboxes_to_list)
doxy_susceptibility -> micro.doxycycline_susceptibility (checkboxes_to_list)
levoflox_suscept -> micro.levofloxacin_susceptibility (checkboxes_to_list)
cef_susceptibility -> micro.ceftriaxone_susceptibility (checkboxes_to_list)
pn_uat_result -> micro.pneumo_binax_now (list)
lg_uat_result -> micro.pneumo_legionella_uat (list)
micro_final_report -> micro.is_final_report (yesno)
map_avoncap_micro(instrument)map_avoncap_micro(instrument)
instrument |
the numeric instrument number |
a list
participant_number -> admin.record_number (name)
hospital -> admin.hospital (list)
nhs_number -> admin.patient_identifier (ppi)
age_at_admission -> demog.age (double)
sex -> demog.gender (list)
test_date -> pneumo.test_date (date)
test -> pneumo.test_type (list)
serotype -> pneumo.phe_serotype (pneumo_serotype)
smoker -> demog.smoker (list)
resp_disease -> comorbid.no_resp_dx, comorbid.copd, comorbid.asthma, comorbid.bronchiectasis, comorbid.pulmonary_fibrosis, comorbid.resp_other (checkboxes)
chd -> comorbid.no_heart_dx, comorbid.ccf, comorbid.ihd, comorbid.hypertension, comorbid.af, comorbid.other_heart_dx (checkboxes)
mi -> comorbid.previous_mi (yesno)
ckd -> comorbid.ckd (list)
liver_disease -> comorbid.liver_disease (list)
diabetes -> comorbid.diabetes (list)
dm_meds -> comorbid.diabetes_medications (list)
dementia -> comorbid.no_dementia, comorbid.dementia, comorbid.cognitive_impairment (checkboxes)
neurological_disease -> comorbid.neuro_other, comorbid.cva, comorbid.tia, comorbid.hemiplegia, comorbid.paraplegia, comorbid.no_neuro_dx (checkboxes)
gastric_ulcers -> comorbid.gastric_ulcers (yesno)
dysphagia -> comorbid.dysphagia (yesno)
pvd -> comorbid.periph_vasc_dx (yesno)
ctd -> comorbid.connective_tissue_dx (yesno)
immunodeficiency -> comorbid.immunodeficiency (yesno)
other_pn_disease -> comorbid.other_pneumococcal_risks (yesno)
hiv -> comorbid.no_HIV, comorbid.HIV, comorbid.AIDS (checkboxes)
cancer -> comorbid.solid_cancer (list)
haem_malig -> comorbid.no_haemotological_cancer, comorbid.leukaemia, comorbid.lymphoma (checkboxes)
recent_chemo -> comorbid.recent_chemotherapy (yesno)
recent_radiotherapy -> comorbid.recent_radiotherapy (yesno)
transplant -> comorbid.transplant_recipient (yesno)
pregnancy -> comorbid.pregnancy (list)
drugs -> demog.no_drug_abuse, demog.alcohol_abuse, demog.ivdu_abuse, demog.marijuana_abuse, demog.other_inhaled_drug_abuse (checkboxes)
immsup -> admission.on_immunosuppression (yesno)
weight_problem -> comorbid.bmi_status (list)
concomittant_flu -> comorbid.influenza_infection (yesno)
hcv -> comorbid.hepatitis_c (yesno)
ppv23 -> vaccination.ppv23_vaccination (list)
flu_vaccine -> vaccination.flu (list)
cci_total_score -> admission.charlson_comorbidity_index (name)
los_days -> outcome.length_of_stay (double)
amts -> admission.triage_score (list)
resp_rate -> admission.respiratory_rate (double)
sats_ra -> admission.saturations_on_room_air (double)
systolic_bp -> admission.systolic_bp (double)
diastolic_bp -> admission.diastolic_bp (double)
crb65_score -> admission.crb_65_severity_score (list)
curb65_score -> admission.curb_65_severity_score (list)
antibiotic_route -> outcome.antibiotic_route (list)
antibiotic_days -> outcome.antibiotic_duration (double)
infection_site -> admission.infection_site (list)
deranged_lfts -> outcome.abnormal_lft (yesno)
aki -> outcome.acute_kidney_injury (yesno)
pleural_effusion -> outcome.pleural_effusion (yesno)
empyema -> outcome.empyema (yesno)
discharge_destination -> outcome.discharge_to (list)
icu -> outcome.admitted_icu (yesno)
niv -> outcome.non_invasive_ventilation (yesno)
intubation -> outcome.intubation (yesno)
recurrent_pneumonia -> outcome.recurrent_pneumonia (yesno)
ecmo -> outcome.received_ecmo (yesno)
inotropes -> outcome.received_ionotropes (yesno)
trachy -> outcome.tracheostomy (yesno)
inpatient_death -> outcome.inpatient_death (yesno)
death_30days -> outcome.death_within_30_days (yesno)
death_1year -> outcome.death_within_1_year (yesno)
survival_days -> outcome.survival_duration (name)
albumin -> haem.albumin (double)
wcc -> haem.white_cell_count (double)
hb -> haem.haemoglobin (double)
pmn -> haem.neutrophils (double)
lymphocytes -> haem.lymphocytes (double)
crp -> haem.crp (double)
na_result -> haem.sodium (double)
ur_result -> haem.urea (double)
egfr -> haem.egfr (double)
creatinine -> haem.creatinine (double)
cxr_sides -> radio.cxr_infection (list)
cxr_lobes -> radio.cxr_lobar_changes (list)
death_5year -> outcome.death_within_5_years (yesno)
survival_days_2 -> outcome.5_yr_survival_duration (name)
imd_decile -> demog.imd_decile (name)
map_avoncap_pneumococcal()map_avoncap_pneumococcal()
a list
radio_exam -> radio.test_performed (yesno)
radiology_date -> radio.test_date (date)
radiodays -> radio.test_days_from_admission (pos_integer)
radio_test -> radio.test_type (list)
radiology_result -> radio.alrtd_finding (checkboxes_to_nested_list)
radiology_other_result -> radio.non_alrtd_finding (checkboxes_to_nested_list)
map_avoncap_radio(instrument)map_avoncap_radio(instrument)
instrument |
the numeric instrument number |
a list
viral_testing_performed -> virol.test_performed (yesno)
virology_date_of_asst -> virol.test_date (date)
viroldays -> virol.test_days_from_admission (pos_integer)
specimen_type -> virol.test_type (list)
virus_isolated -> virol.pathogen_detected (yesno)
test_type -> virol.test_type (list)
virus_pathogen -> virol.pathogen, .virol_isolate_list (checkboxes_to_nested_list)
virol_patient_lab -> virol.test_provenance (list)
map_avoncap_virol(instrument)map_avoncap_virol(instrument)
instrument |
the numeric instrument number |
a list
RESULT -> pneumo.urine_antigen_result, .x (text)
EVENT_DATE -> pneumo.test_date (date)
ANALYSIS -> pneumo.urine_antigen_test (name)
SUBJECT -> admin.consented_record_number (study_id)
BARCODE -> pneumo.urine_antigen_sample_id (name)
map_urine_antigens()map_urine_antigens()
a list
RESULT -> pneumo.binax_result, .x (text)
EVENT_DATE -> pneumo.test_date (date)
SUBJECT -> admin.consented_record_number (study_id)
BARCODE -> pneumo.urine_antigen_sample_id (name)
RESULT -> pneumo.binax_result, .x (text)
EVENT_DATE -> pneumo.test_date (date)
SUBJECT -> admin.consented_record_number (study_id)
BARCODE -> pneumo.urine_antigen_sample_id (name)
map_urine_binax() map_urine_binax()map_urine_binax() map_urine_binax()
a list
a list
find most recent files of a specific type
most_recent_files( type = "", subtype = NULL, reproduce_at = as.Date(getOption("reproduce.at", default = Sys.Date())) )most_recent_files( type = "", subtype = NULL, reproduce_at = as.Date(getOption("reproduce.at", default = Sys.Date())) )
type |
see valid_inputs() for current list of supported types in input directory |
subtype |
see valid_inputs() for list of supported filenames |
reproduce_at |
after this date new files are ignored. This enforces a specific version of the data. |
a list of the file paths to the most up to date files of the given type relevant to each site and study year
# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })
AvonCap data has lots of columns which are named in a difficult to remember fashion, composed of data items that have enumerated values with no semantics. This makes displaying them difficult and any filtering done on the raw data inscrutable. Depending on the source of the data some different columns may be present due to differences in the NHS and UoB data sets. The redcap database has some options that may be checklists and some that are radio buttons, both of these end up with mysterious names in the data.
normalise_data(rawData, instrument = NULL, ...)normalise_data(rawData, instrument = NULL, ...)
rawData |
|
instrument |
the numeric instrument number if applicable |
... |
Arguments passed on to
|
This function maps the data into a tidy dataframe with consistently named columns, and named factors where appropriate. The mapping is defined in data.
files Most of the sanitisation code is held in the
normalise-xxx.R file. but these in turn may depend on the mapping-xxx.R
files
a tracked dataframe with n
Get the mapping of transformed columns back to original
original_field_names(data, inverse = TRUE)original_field_names(data, inverse = TRUE)
data |
the transformed data set. |
inverse |
give the data as a old -> new mapping for finding normalised names of original columns. if false gives it as new->old for finding original names of normalised columns |
a named list mapping original to new columns
A somewhat complete list of pneumococcal serotypes as seen in Bristol
Get a label for a column
readable_label(columnVar, colNames = default_column_names())readable_label(columnVar, colNames = default_column_names())
columnVar |
the column name as a string |
colNames |
bespoke column names mapping (see |
a mapped column name
Get a readable label for the AvonCap data as a named list (for ggplot)
readable_label_mapping(x, ...) ## S3 method for class 'data.frame' readable_label_mapping(x, colNames = default_column_names(...), ...) ## S3 method for class 'list' readable_label_mapping(x, colNames = default_column_names(...), ...) ## S3 method for class 'character' readable_label_mapping(x, colNames = default_column_names(...), ...) ## Default S3 method: readable_label_mapping(x, colNames = default_column_names(...), ...)readable_label_mapping(x, ...) ## S3 method for class 'data.frame' readable_label_mapping(x, colNames = default_column_names(...), ...) ## S3 method for class 'list' readable_label_mapping(x, colNames = default_column_names(...), ...) ## S3 method for class 'character' readable_label_mapping(x, colNames = default_column_names(...), ...) ## Default S3 method: readable_label_mapping(x, colNames = default_column_names(...), ...)
x |
either the column names as strings, or a dataframe |
... |
ignored |
colNames |
a mapping to convert a column name (as a string) to a readable label |
a named list of the labels for the columns
readable_label_mapping(data.frame): for data frames
readable_label_mapping(list): for lists
readable_label_mapping(character): for character vectors
readable_label_mapping(default): defaults
Relevel serotype data into an factor based on PCV group status and serotype name.
relevel_serotypes(serotypes, ..., exprs)relevel_serotypes(serotypes, ..., exprs)
serotypes |
a vector of serotypes as a factor or character. |
... |
an unwrapped version of the |
exprs |
a list of formulae with a predicate on the LHS and a PCV group name on
the RHS. which are interpreted as the parameters for a |
x = rlang::exprs( PCV7 ~ "PCV7", PCV15 ~ "PCV15-7", TRUE ~ "Non-PCV15 serotype" ) relevel_serotypes(avoncap::phe_serotypes, exprs=x) relevel_serotypes(avoncap::phe_serotypes) relevel_serotypes(avoncap::phe_serotypes, PCV24Affinivax ~ "Affinivax", TRUE ~ "Non-affinivax" )x = rlang::exprs( PCV7 ~ "PCV7", PCV15 ~ "PCV15-7", TRUE ~ "Non-PCV15 serotype" ) relevel_serotypes(avoncap::phe_serotypes, exprs=x) relevel_serotypes(avoncap::phe_serotypes) relevel_serotypes(avoncap::phe_serotypes, PCV24Affinivax ~ "Affinivax", TRUE ~ "Non-affinivax" )
Write file source information out to a text files
save_data_source_info(..., .file)save_data_source_info(..., .file)
... |
A list of data frames loaded with the |
.file |
the output file location |
the file name written (invisibly)
The scale groups colours by PCV group, but it is important to have the source
data using the same levels as this scale otherwise the colour legend will be
ordered in a different sequence. This can be achieved using relevel_serotypes,
scale_fill_serotype( ..., palette_fn = scales::brewer_pal(palette = "Dark2"), undefined = "#606060", exprs = rlang::exprs() )scale_fill_serotype( ..., palette_fn = scales::brewer_pal(palette = "Dark2"), undefined = "#606060", exprs = rlang::exprs() )
... |
Arguments passed on to
|
palette_fn |
a function that returns a set of colours for a number of
levels. Such functions can be obtained from things like |
undefined |
the colour for the last group which is assumed to be the
|
exprs |
a list of formulae with a predicate on the LHS and a PCV group name on
the RHS. which are interpreted as the parameters for a |
A ggplot2 scale
A list of pneumococcal serotype / UAD cross mappings
Also performs some structure checks and makes sure that the README files are in place.
set_input(path)set_input(path)
path |
the path to the input directory |
the full path to the directory
Spline term marginal effects plot
spline_term_plot( coxmodel, var_name, xlab = var_name, max_y = NULL, n_breaks = 7 )spline_term_plot( coxmodel, var_name, xlab = var_name, max_y = NULL, n_breaks = 7 )
coxmodel |
an output of a coxph model |
var_name |
a variable that is involved in a spline term |
xlab |
x axis label |
max_y |
maximium hazard ratio to display on y axis. Inferred from the central estimates if missing, which will most likely cut off confidence intervals |
n_breaks |
The number of divisions on the y axis |
a ggplot
This function plots a stacked bar of proportions for an input set of data
stacked_barplot(data, mapping, ...)stacked_barplot(data, mapping, ...)
data |
the data |
mapping |
a aes mapping with at least |
... |
passed to |
a ggplot
stacked_barplot( ggplot2::diamonds, ggplot2::aes(x=cut, fill=clarity, group=color) )+ ggplot2::facet_wrap(dplyr::vars(color))stacked_barplot( ggplot2::diamonds, ggplot2::aes(x=cut, fill=clarity, group=color) )+ ggplot2::facet_wrap(dplyr::vars(color))
This is poorly named as only give the start date is the input is an integer
start_date_of_week(study_week)start_date_of_week(study_week)
study_week |
does accept decimals and returns the nearest whole date to the value |
a vector of sudy_week numbers
Convert a date to a study week
study_week(dates)study_week(dates)
dates |
a list of date objects |
an integer number of weeks since 2019-12-30
Upset plot with counts stratified by a categorical column
upset_plot(df, boolean_cols, categorical_col, lbl_size = 5)upset_plot(df, boolean_cols, categorical_col, lbl_size = 5)
df |
the data |
boolean_cols |
a tidyselect specification selecting the columns to be used as binary one-hot encoded classes |
categorical_col |
a column containing a disjoint category as a factor |
lbl_size |
font sise of the label |
a ggplot
load_data(...)
A valid set of types of file that can be loaded by load_data(...)
valid_inputs()valid_inputs()
a dataframe of type, subtype
# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })# devtools::load_all() try({ avoncap::set_input("~/Data/avoncap") avoncap::input("nhs-extract") avoncap::all_files() # exact match on filename column of all_data() avoncap::most_recent_files("AvonCAPLRTDCentralDa") # or matches by lower case startWith on directory avoncap::most_recent_files("nhs-extract","deltave") avoncap::most_recent_files("metadata") avoncap::valid_inputs() })
Runs a set of QA checks. This function dispatches the call
in a data set specific function using the type and subtype of the
data set. The checks are in source files named validate-xxx.R depending
on the data source.
validate_data(rawData, ...)validate_data(rawData, ...)
rawData |
|
... |
not used / passed to the validation function specific to the type of data. |
the same input with a new data_quality_failures attribute containing
issues.
Write out data quality issues
write_issues(df, file)write_issues(df, file)
df |
the raw data frame |
file |
the output data quality file |
the list of failures as a dataframe
table
Wrapper around table
xglimpse(data, ...)xglimpse(data, ...)
data |
a dataframe |
... |
columns or named expressions to cross-tabulate |
the cross-tabulation