updated group work solution

sessions/group_work_day2.qmd

@@ -4,8 +4,7 @@
 ## Objective of this session
 
 In this session, We will get hands on practice using the Framingham
-dataset. You will conduct a proper causal mediation analysis,
-considering:
+dataset. You will conduct a causal mediation analysis, considering:
 
 -   Conduct causal mediation analyses with exposure induced
     mediator-outcome confounder or multiple mediators
@@ -31,6 +30,8 @@ can either use the **mediation** package or **CMAverse** package.
 
 2.  Perform a causal mediation analysis
 
-3.  Conduct a sensitivity analyses and discuss the results
+3.  Conduct a sensitivity analyses for uncontrolled confounding and
+    discuss the results
 
-4.  Write a short report of your mediation models
+4.  Write a short report of your mediation models. Focus on item
+    6,9,10,11,14,15,18,19,20,21,22 in the long Agrema checklist.

sessions/group_work_day2_solution.qmd

@@ -0,0 +1,94 @@
+# Day 2: Group work
+
+::: callout-note
+## Objective of this session
+
+In this session, We will get hands on practice using the Framingham
+dataset. You will conduct a causal mediation analysis, considering:
+
+-   Conduct causal mediation analyses with exposure induced
+    mediator-outcome confounder or multiple mediators
+-   Conduct sensitivity analyses
+-   Report the results following the guidelines
+:::
+
+We will continue working with the obesity and cardiovascular disease
+example.
+
+First, reflect on the group work from yesterday and discuss the
+following questions with your group members:
+
+-   Review the assumptions of causal mediation, and discuss the
+    robustness of your estimates
+-   Are there other mediators than systolic blood pressure?
+-   Are there interactions between obesity and the mediators?
+
+Re-conduct the mediation analysis and address the discussion points. You
+can either use the **mediation** package or **CMAverse** package.
+
+1.  Re-do the DAG, and investigate the potential pathways
+
+2.  Perform a causal mediation analysis
+
+Load data:
+
+```{r setup}
+library(here)
+library(tidyverse)
+
+load(here::here("data/frmghamdata.RData"))
+```
+
+Clean the data:
+
+```{r}
+frmgham_clean <- frmgham %>%
+  select(
+    id = randid, 
+    w1 = sex,
+    w2 = age,
+    w3 = cursmoke,
+    w4 = educ,
+    a = bmi, # this is the exposure
+    m = sysbp, # this is the mediator
+    y = cvd # this is the outcome
+  ) %>% 
+  na.omit()
+
+frmgham_clean <- frmgham_clean %>%
+  mutate(a = case_when(a >= 30 ~ 1,
+                       TRUE ~ 0))
+
+```
+
+Do causal mediation analysis:
+
+```{r}
+library(CMAverse)
+
+res_rb <- cmest(
+  data = frmgham_clean, model = "rb", outcome = "y", exposure = "a",
+  mediator = "m", basec = c("w1", "w2", "w3", "w4"), EMint = TRUE,
+  mreg = list("linear"), yreg = "logistic",
+  astar = 0, a = 1, mval = list(120),
+  estimation = "paramfunc", inference = "delta"
+)
+
+summary(res_rb)
+```
+
+3.  Conduct a sensitivity analyses for uncontrolled confounding and
+    discuss the results
+
+```{r}
+uc_sens <- cmsens(
+  object = res_rb,
+  sens = "uc"
+)
+
+uc_sens
+
+```
+
+4.  Write a short report of your mediation models. Focus on item
+    6,9,10,11,14,15,18,19,20,21,22 in the long Agrema checklist.

sessions/reporting-guidelines.qmd

@@ -50,10 +50,12 @@ report. See @fig-agrema-checklist-sf.
 ![AGReMA checklist from Lee et
 al.](/images/agrema-checklist-sf.PNG){#fig-agrema-checklist-sf}
 
-## Exercise
+::: callout-note \## Exercise - 15 min
 
--   how many of the items are reported in the mediation analysis papers
+-   How many of the items are reported in the mediation analysis papers
     you reviewed in the beginning of the course?
--   what do you think are the most important things missing?
+-   What do you think are the most important things missing?
+
+:::
 
 ## References