Add/subtract date from date time format in dbplyr

I'm using dbplyr to access data in my database and I'm trying to calculate rolling sum amounts that would include all amounts from the previous 1 year. My Dates are in dttm format. I'm trying to get a table that shows 1 year prior to the Start_Date similar to below:

I would like to subtract exactly 1 year from the Start_Date to get a rolling one-year window for each row/observation for rolling sum calculation.

I've done this in dplyr before and I would normally use the following code to get this:

output = df %>%
  mutate(Previous_1year = Start_Date %m+% years(-1))

However, this method doesn't seem to be compatible with dbplyr and my database, assuming this is because dbplyr doesn't know how to translate the code back into SQL.

I've also tried running the following code in dbplyr which I found from another question in SO:

output = df %>%
   mutate(Previous_1year = sql('DATEADD(y, -1, Start_Date)'))

but I'm getting the following error and not sure why I'm getting a invalid datetime:

! nanodbc/nanodbc.cpp:1752: XX000: [Amazon][Amazon Redshift] (30) 
Error occurred while trying to execute a query: [SQLState XX000] ERROR:  Invalid datetime part for DATEADD().

Because the data is so large, I'm unable to collect all of it and do the data wrangling through dplyr to subtract 1 year from each of my Start_Date observations. So, hoping to do this within the database first so then I can do my calculations before collecting the results.

This works on Postgres, so might work on Redshift:

library(dplyr, warn.conflicts = FALSE)
library(DBI)

pg <- dbConnect(RPostgres::Postgres())

df <- tbl(pg, sql("SELECT '2021-07-02 08:03:29'::timestamp AS start"))
df %>%
  mutate(previous = start - years(1))
#> # Source:   SQL [1 x 2]
#> # Database: postgres  [iangow@/tmp:5432/iangow]
#>   start               previous           
#>   <dttm>              <dttm>             
#> 1 2021-07-02 08:03:29 2020-07-02 08:03:29

^{Created on 2023-04-03 with reprex v2.0.2}