Merge branch '5-inference' [#12]

report/SAR-2023-017-BH-v01.Rmd

@@ -62,6 +62,10 @@ source("scripts/results.R", encoding = 'UTF-8')
 
 ## Objectives
 
+1. To describe the missingness in zip codes at each follow up collection;
+1. To impute missing Zip codes with data available in previous follow up collections.
+1. To assess the sensitivity of the association between mortality and socioeconomic status to the imputation of participant missing location.
+
 <!-- ## Data reception and cleaning -->
 
 # Methods
@@ -162,6 +166,12 @@ sch.df %>%
 
 # Conclusions
 
+Simple imputation on zip codes do not extend the range of observations available for modeling in this dataset when using a single observation per individual.
+
+When using multiple observations per individual, there is a minute increment in the number of events, but there is different imputation approaches do not yield different datasets.
+
+When using multiple observations per individual in the model specification evaluated, the time-varying exposure allows for the inclusion of terms that violated the proportional hazards assumption in the constant exposure.
+
 # References
 
 - **SAP-2023-017-BH-v02** -- Analytical Plan for Sensitivity of mortality rates to the imputation of missing socioeconomic data: cohort study

report/SAR-2023-017-BH-v01.md

@@ -48,6 +48,10 @@ output:
 
 ## Objectives
 
+1. To describe the missingness in zip codes at each follow up collection;
+1. To impute missing Zip codes with data available in previous follow up collections.
+1. To assess the sensitivity of the association between mortality and socioeconomic status to the imputation of participant missing location.
+
 <!-- ## Data reception and cleaning -->
 
 # Methods
@@ -168,6 +172,12 @@ Table: **Table 4** Schoenfeld test for both datasets.
 
 # Conclusions
 
+Simple imputation on zip codes do not extend the range of observations available for modeling in this dataset when using a single observation per individual.
+
+When using multiple observations per individual, there is a minute increment in the number of events, but there is different imputation approaches do not yield different datasets.
+
+When using multiple observations per individual in the model specification evaluated, the time-varying exposure allows for the inclusion of terms that violated the proportional hazards assumption in the constant exposure.
+
 # References
 
 - **SAP-2023-017-BH-v02** -- Analytical Plan for Sensitivity of mortality rates to the imputation of missing socioeconomic data: cohort study