Update phenotype_sensitivity.R

add totals
opensafely · Aug 30, 2024 · 2d91e0a · 2d91e0a
1 parent 8930504
commit 2d91e0a
Showing 1 changed file with 122 additions and 0 deletions.
diff --git a/analysis/sensitivity_analyses/phenotype_sensitivity.R b/analysis/sensitivity_analyses/phenotype_sensitivity.R
@@ -168,6 +168,125 @@ if (study_start_date >= covid_season_min) {
   )
 }
 
+#calculate totals for 'specific' outcomes
+if (study_start_date >= covid_season_min) {
+  specific_totals_mild_full <- df_input_specific %>%
+    mutate(
+      RSV_Mild_Total = sum(rsv_primary_inf),
+      Flu_Mild_Total = sum(flu_primary_inf),
+      COVID_Mild_Total = sum(covid_primary_inf)
+    )
+  specific_totals_severe_full <- df_input_specific %>%
+    mutate(
+      RSV_Severe_Total = sum(rsv_secondary_inf),
+      Flu_Severe_Total = sum(flu_secondary_inf),
+      COVID_Severe_Total = sum(covid_secondary_inf)
+    )
+} else {
+  specific_totals_mild_full <- df_input_specific %>%
+    mutate(
+      RSV_Mild_Total = sum(rsv_primary_inf),
+      Flu_Mild_Total = sum(flu_primary_inf)
+    )
+  specific_totals_severe_full <- df_input_specific %>%
+    mutate(
+      RSV_Severe_Total = sum(rsv_secondary_inf),
+      Flu_Severe_Total = sum(flu_secondary_inf)
+    )
+}
+
+#calculate totals for 'sensitive' outcomes
+if (study_start_date >= covid_season_min) {
+  sensitive_totals_mild_full <- df_input_sensitive %>%
+    mutate(
+      RSV_Mild_Total = sum(rsv_primary_inf),
+      Flu_Mild_Total = sum(flu_primary_inf),
+      COVID_Mild_Total = sum(covid_primary_inf),
+      Overall_Resp_Mild_Total = sum(overall_resp_primary_inf)
+    )
+  sensitive_totals_severe_full <- df_input_sensitive %>%
+    mutate(
+      RSV_Severe_Total = sum(rsv_secondary_inf),
+      Flu_Severe_Total = sum(flu_secondary_inf),
+      COVID_Severe_Total = sum(covid_secondary_inf),
+      Overall_Resp_Severe_Total = sum(overall_resp_secondary_inf)
+    )
+} else {
+  sensitive_totals_mild_full <- df_input_sensitive %>%
+    mutate(
+      RSV_Mild_Total = sum(rsv_primary_inf),
+      Flu_Mild_Total = sum(flu_primary_inf),
+      Overall_Resp_Mild_Total = sum(overall_resp_primary_inf)
+    )
+  sensitive_totals_severe_full <- df_input_sensitive %>%
+    mutate(
+      RSV_Severe_Total = sum(rsv_secondary_inf),
+      Flu_Severe_Total = sum(flu_secondary_inf),
+      Overall_Resp_Severe_Total = sum(overall_resp_secondary_inf)
+    )
+}
+
+#reformat data 
+if (study_start_date >= covid_season_min) {
+  specific_totals_mild <- specific_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total, COVID_Mild_Total)
+  specific_totals_severe <- specific_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total, COVID_Severe_Total)
+  sensitive_totals_mild <- sensitive_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total, COVID_Mild_Total)
+  sensitive_totals_severe <- sensitive_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total, COVID_Severe_Total)
+  sensitive_overall_totals_mild <- sensitive_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total, COVID_Mild_Total,
+           Overall_Resp_Mild_Total)
+  sensitive_overall_totals_severe <- sensitive_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total, COVID_Severe_Total,
+           Overall_Resp_Severe_Total)
+} else {
+  specific_totals_mild <- specific_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total)
+  specific_totals_severe <- specific_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total)
+  sensitive_totals_mild <- sensitive_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total)
+  sensitive_totals_severe <- sensitive_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total)
+  sensitive_overall_totals_mild <- sensitive_totals_mild_full %>%
+    select(RSV_Mild_Total, Flu_Mild_Total, Overall_Resp_Mild_Total)
+  sensitive_overall_totals_severe <- sensitive_totals_severe_full %>%
+    select(RSV_Severe_Total, Flu_Severe_Total, Overall_Resp_Severe_Total)
+}
+
+specific_totals_m <- specific_totals_mild %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "mild", codelist_type = "specific")
+specific_totals_s <- specific_totals_severe %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "severe", codelist_type = "specific")
+specific_totals <- rbind(specific_totals_m, specific_totals_s)
+
+sensitive_totals_m <- sensitive_totals_mild %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "mild", codelist_type = "sensitive")
+sensitive_totals_s <- sensitive_totals_severe %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "severe", codelist_type = "sensitive")
+sensitive_totals <- rbind(sensitive_totals_m, sensitive_totals_s)
+
+sensitive_overall_totals_m <- sensitive_overall_totals_mild %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "mild_overall", codelist_type = "sensitive")
+sensitive_overall_totals_s <- sensitive_overall_totals_severe %>%
+  unique() %>%
+  gather(key = "combo", value = "n") %>%
+  mutate(outcome_type = "severe_overall", codelist_type = "sensitive")
+sensitive_overall_totals <- rbind(sensitive_overall_totals_m, sensitive_overall_totals_s)
+
 #count number of patients in each category for specific phenotypes - separately for mild and severe 
 patients_specific_mild <- rlang::duplicate(df_input_specific) %>% 
   group_by("combo" = mild_combined) %>%
@@ -180,6 +299,7 @@ patients_specific_severe <- rlang::duplicate(df_input_specific) %>%
   ungroup() %>% 
   mutate(outcome_type = "severe", codelist_type = "specific")
 patients_specific <- full_join(patients_specific_mild, patients_specific_severe)
+patients_specific <- rbind(specific_totals, patients_specific)
 
 #count number of patients in each category for sensitive phenotypes - separately for mild and severe
 patients_sensitive_mild <- rlang::duplicate(df_input_sensitive) %>% 
@@ -193,6 +313,7 @@ patients_sensitive_severe <- rlang::duplicate(df_input_sensitive) %>%
   ungroup() %>% 
   mutate(outcome_type = "severe", codelist_type = "sensitive")
 patients_sensitive <- full_join(patients_sensitive_mild, patients_sensitive_severe)
+patients_sensitive <- rbind(sensitive_totals, patients_sensitive)
 
 #count number of patients in each category for sensitive phenotypes now including overall resp - separately for mild and severe
 patients_sensitive_mild_overall <- rlang::duplicate(df_input_sensitive) %>% 
@@ -207,6 +328,7 @@ patients_sensitive_severe_overall <- rlang::duplicate(df_input_sensitive) %>%
   mutate(outcome_type = "severe_overall", codelist_type = "sensitive")
 patients_sensitive_overall <- full_join(patients_sensitive_mild_overall, 
                                         patients_sensitive_severe_overall)
+patients_sensitive_overall <- rbind(sensitive_overall_totals, patients_sensitive_overall)
 
 #combine as one tibble 
 patients_combined <- patients_specific %>%