Determines which months are invalid based on a low contribution of data.
This is based upon the long term daily average for admissions. Days that
fall below 2 SD of the long term mean are flagged. If more than the
threshold number of flagged days occur in a single month, then the
month is invalidated and removed from further analysis.
verify_episodes(episode_length = NULL, threshold = 10)
| episode_length | a table created by |
|---|---|
| threshold | threshold number of days to use in calculation |
a table of the same form as episode_length with an additional
column indicating if the episode comes from a valid period of time.
This procedure removes episodes that occur during particularly sparse periods (as this is likely that these months are contributing poor data) and return only episodes that have a logical consistency in non-sparse months. The analyst should consider if the denominator for the number of study months should be changed following the use of this function.