Introduction to Recurrent Observations in R

Recurrent observations refer to the phenomenon where an individual returns for multiple visits within a specified time period. In this article, we’ll explore how to add a column that indicates the earliest recurring observation within 90 days, grouped by patient ID, using the popular R programming language.

Prerequisites: Understanding Key Concepts

Before diving into the code, let’s cover some essential concepts:

• Date class in R: The Date class represents dates and allows for easy manipulation of date-related operations.
• dplyr: A popular data manipulation library in R that provides a powerful grammar for data transformation.
• lubridate: Another essential library in R for working with dates, times, and intervals.
• tidyr: A versatile library for tidying and transforming data in R.

Loading Libraries

To begin our analysis, we need to load the necessary libraries:

library(dplyr)
library(lubridate)
library(tidyr)

These libraries provide us with the tools needed to manipulate and transform our data efficiently.

Data Preparation

Next, let’s assume we have a dataframe df1 containing the person ID, visit date, and other relevant information:

# Create sample data
person_ID     visit_date       i1         date      
1               2/25/2001           
1               2/27/2001           
1               4/2/2001            
2               3/18/2004           
3               9/22/2004             
3               10/27/2004          
3               5/15/2008 

# Convert 'visit_date' to Date class
df1 %>% mutate(visit_date = mdy(visit_date))

By converting the visit_date column to the Date class, we can perform date-related operations more efficiently.

Grouping by Person ID and Calculating Recurrence

Now that our data is ready, let’s group it by person ID and calculate the recurrence:

# Group by 'person_ID' and create a binary column 'i1'
df1 %>% 
  mutate(visit_date = mdy(visit_date)) %>% 
  group_by(person_ID) %>% 
  mutate(i1 = replace_na(+(difftime(lead(visit_date), visit_date, units = 'day') < 90), 0))

Here’s what’s happening:

• We calculate the difference between each subsequent visit date using difftime. This returns a numerical value representing the number of days.
• We then compare this value with 90 using < to determine if there is an observation within 90 days.
• If the condition is met, we assign 1 to the new column 'i1'; otherwise, we assign 0.
• Finally, we use replace_na() to handle any missing values and ensure that our results are accurate.

Finding the Earliest Recurring Observation

With our binary column 'i1', we can now identify the earliest recurring observation for each person ID:

# Use 'i1' to find the corresponding next visit_date
df1 %>% 
  mutate(i1 = case_when(as.logical(i1) ~ lead(visit_date), i1 = NULL)) %>% 
  ungroup

Here’s what’s happening:

• We use case_when() to evaluate the value of 'i1'. If it is true, we assign the next visit date using lead().
• If the condition is not met, we set 'i1' to NULL, effectively excluding that observation from our results.

Final Result

After executing these steps, we get the following output:

# Output
  person_ID visit_date       i1          date      
1         1 2001-02-25 2001-02-27 2001-02-27  
2         1 2001-02-27 2001-04-02 2001-02-27  
3         1 2001-04-02          NA            
4         2 2004-03-18          NA            
5         3 2004-09-22 2004-10-27 2004-09-22  
6         3 2004-10-27          NA            
7         3 2008-05-15          NA            

As expected, the new column 'date' contains the earliest recurring observation within 90 days for each person ID.

Last modified on 2024-08-23