Merge fragmented records into a single observation

Merges multiple rows representing fragments of the same observation (e.g., participant visit, assessment session, study encounter, or questionnaire administration) into a single row by selecting the first non-missing value within each variable.

Usage

MergeFragmentedRecords(
  data,
  id_var = "subject",
  date_var = "date",
  session_var = "session",
  keep_session = TRUE,
  session_name = "first_session",
  n_rows_name = "n_rows_collapsed",
  arrange_desc_session = FALSE,
  empty_strings_to_na = TRUE,
  df = lifecycle::deprecated()
)

Arguments

data

A data frame containing fragmented records.

id_var

Character string specifying the participant identifier variable. Default is "subject".

date_var

Character string specifying the visit or assessment date variable. Default is "date".

session_var

Character string specifying the session identifier variable used to order fragmented records. Default is "session".

keep_session

Logical. If TRUE, the first session value encountered within each group is retained. Default is TRUE.

session_name

Character string specifying the name of the retained session variable. Default is "first_session".

n_rows_name

Character string specifying the name of the variable recording the number of rows merged. Default is "n_rows_collapsed".

arrange_desc_session

Logical. If TRUE, records are ordered by descending session number before merging. Default is FALSE.

empty_strings_to_na

Logical. If TRUE, empty character strings are converted to missing values prior to merging. Default is TRUE.

df

Deprecated (since 19.15.0). Use data instead.

Value

A data frame containing one row per unique combination of id_var and date_var.

Additional variables may include:

n_rows_collapsed

Number of fragmented rows merged.

first_session

First session value retained, if keep_session = TRUE.

Details

This function is useful when data collection is interrupted and restarted, resulting in multiple partial records for the same participant and date. Examples include tablet-based cognitive testing, mobile applications, REDCap surveys, wearable device uploads, and electronic assessments where internet connectivity or software issues may split a single visit across multiple records.

Rows are grouped by the combination of id_var and date_var. Within each group, observations are ordered by session_var, and the first non-missing value encountered for each variable is retained.

For each variable within a participant-date grouping, the function returns the first non-missing value after sorting by session order.

For example:

subject   date         session   Stroop   Trails
101       2024-01-01   1         50       NA
101       2024-01-01   2         NA       100

becomes

subject   date         Stroop   Trails
101       2024-01-01   50       100

No attempt is made to resolve conflicting non-missing values across sessions. If multiple non-missing values exist for the same variable, the first value encountered after sorting is retained.