Compute False Positive Rates for Histological Evaluations

Description

Computes the proportion of histological evaluations (biopsies or lumpectomies) that did not lead to a breast cancer diagnosis, stratified by trial arm and screening round (first round versus beyond first round).

Usage

false_positives(
  long_data,
  hist_data,
  id_col = "id",
  arm_col = "arm",
  month2_col = "month2",
  bc_month_col = "monthBC",
  hist_month2_col = "month2",
  first_round_months = 9L,
  window_months = 6L
)

Arguments

long_data A data frame in long format (one row per participant-arm-month), as produced by expand_to_long(). Must include id_col, arm_col, month2_col, and bc_month_col.
hist_data A data frame with one row per histological evaluation (biopsy or lumpectomy), containing id_col and hist_month2_col (0-indexed follow-up month of the evaluation, on the same scale as month2_col).
id_col Name of the participant identifier column. Default: “id”.
arm_col Name of the trial arm column. Default: “arm”.
month2_col Name of the 0-indexed month-from-entry column in long_data. Default: “month2”.
bc_month_col Name of the column containing the month2 of breast cancer diagnosis (NA if no diagnosis). Default: “monthBC”.
hist_month2_col Name of the 0-indexed evaluation month column in hist_data. Must be on the same scale as month2_col. Default: “month2”.
first_round_months Month2 threshold for defining the first screening round. Default: 9L.
window_months Number of months (in month2 units) after a histological evaluation during which a breast cancer diagnosis counts as a true positive. Also used for deduplication: repeat evaluations within this many months of a prior one are dropped. Default: 6L.

Details

A histological evaluation is classified as a true positive if a breast cancer diagnosis is recorded within window_months months of the evaluation (in month2 units, inclusive). All other evaluations are false positives.

Before classification, the following pre-processing steps are applied:

  • Censoring filter: Only evaluations that fall within the observed follow-up period for the participant-arm are counted. Evaluations that occur after the last observed month2 for a given arm are excluded.

  • Deduplication: Repeat evaluations within window_months months of a prior evaluation (per participant-arm) are dropped so that one diagnostic episode is not counted multiple times.

Evaluations are stratified by period based on the 0-indexed follow-up month (month2):

  • first_round: Evaluation month2 is at or below first_round_months.

  • beyond_first_round: Evaluation month2 exceeds first_round_months.

Value

A data frame with one row per arm-period combination that is observed in the data (arm-period pairs with no histological evaluations are omitted), containing:

  • arm: Trial arm (“STOPBASE” or “CONTINUE”).

  • period: Screening period (“first_round” or “beyond_first_round”).

  • n_hist: Total number of histological evaluations (after deduplication).

  • n_positive: Number that led to a breast cancer diagnosis.

  • fpr: False positive rate (1 - n_positive / n_hist).

References

García-Albéniz X, Uno H, Bhatt DL, McArdle PH, Joffe MM, Hernán MA. Continuation of Annual Screening Mammography and Breast Cancer Mortality in Women Older Than 70 Years: A Prospective Observational Study. Ann Intern Med. 2020;172(6):381–389. doi:10.7326/M18-1199

See Also

expand_to_long()