From 9592be12666880a24b4c4dcff5cbb196c9700e06 Mon Sep 17 00:00:00 2001
From: Em Prestige <emiel200@gmail.com>
Date: Thu, 21 Mar 2024 14:57:22 +0000
Subject: [PATCH] update exclusion criteria for cohort flow chart

---
 analysis/cohort_criteria.R                | 28 +++++++++--------------
 analysis/dataset_definition_flow_chart.py | 13 +++++------
 2 files changed, 17 insertions(+), 24 deletions(-)

diff --git a/analysis/cohort_criteria.R b/analysis/cohort_criteria.R
index 3957f68..6bb073d 100644
--- a/analysis/cohort_criteria.R
+++ b/analysis/cohort_criteria.R
@@ -56,27 +56,21 @@ not_age_count <- if (cohort == "infants" | cohort == "infants_subgroup") {
   registered_count - age_count
 }
 
-if (cohort == "older_adults") {
-  included_count <- sum(patients_df$registered & patients_df$is_female_or_male 
-                        & patients_df$is_appropriate_age & patients_df$has_imd 
-                        & !patients_df$care_home, na.rm = TRUE)
-  excluded_count <- sum(!patients_df$is_female_or_male |!patients_df$has_imd 
-                        | patients_df$care_home, na.rm = TRUE) - not_age_count
-} else if (cohort == "infants" | cohort == "infants_subgroup") {
-  included_count <- sum(patients_df$is_female_or_male 
+if (cohort == "infants" | cohort == "infants_subgroup") {
+  included_count <- sum(!patients_df$severe_immunodeficiency
                         & patients_df$is_appropriate_age & patients_df$has_imd 
-                        & !patients_df$risk_group_infants 
-                        & !patients_df$severe_immunodeficiency, na.rm = TRUE)
-  excluded_count <- sum(!patients_df$is_female_or_male |!patients_df$has_imd 
-                        | patients_df$risk_group_infants 
-                        | patients_df$severe_immunodeficiency, na.rm = TRUE) - 
+                        & patients_df$is_female_or_male & !patients_df$care_home
+                        & !patients_df$risk_group_infants, na.rm = TRUE)
+  excluded_count <- sum(!patients_df$is_female_or_male|!patients_df$has_imd 
+                        |patients_df$risk_group_infants|patients_df$care_home
+                        |patients_df$severe_immunodeficiency, na.rm = TRUE) - 
                     not_age_count
 } else {
   included_count <- sum(patients_df$registered & patients_df$is_female_or_male 
-                        & patients_df$is_appropriate_age & patients_df$has_imd,
-                        na.rm = TRUE)
-  excluded_count <- sum(!patients_df$is_female_or_male 
-                        |!patients_df$has_imd, na.rm = TRUE) - not_age_count
+                        & patients_df$is_appropriate_age & patients_df$has_imd 
+                        & !patients_df$care_home, na.rm = TRUE)
+  excluded_count <- sum(!patients_df$is_female_or_male|!patients_df$has_imd 
+                        |patients_df$care_home, na.rm = TRUE) - not_age_count
 }
 
 ## create output directories ----
diff --git a/analysis/dataset_definition_flow_chart.py b/analysis/dataset_definition_flow_chart.py
index 457c5da..73aa45b 100644
--- a/analysis/dataset_definition_flow_chart.py
+++ b/analysis/dataset_definition_flow_chart.py
@@ -193,13 +193,12 @@ def first_infection_event(codelist, where = True):
   )
 
 #care home resident
-if cohort == "older_adults" :
-  care_home_tpp = (
-    addresses.for_patient_on(index_date)
-    .care_home_is_potential_match.when_null_then(False)
-  )
-  care_home_code = (has_prior_event(codelists.carehome_codelist))
-  dataset.care_home = care_home_tpp | care_home_code
+care_home_tpp = (
+  addresses.for_patient_on(index_date)
+  .care_home_is_potential_match.when_null_then(False)
+)
+care_home_code = (has_prior_event(codelists.carehome_codelist))
+dataset.care_home = care_home_tpp | care_home_code
 
 #define population
 dataset.define_population(practice_registrations.exists_for_patient())