Visit Tracker: Examples

This report demonstrates the visit_tracker SAS macro for operational monitoring of longitudinal studies. The macro creates expected schedules of participant visits, applies visit windows, matches actual to expected visits, and summarizes completion and retention metrics. Each example below uses synthetic data to showcase a different scheduling logic or windowing rule.

Table of Contents

Example 1 — Semiannual (18 months)
Example 2 — Dense early
Example 3 — LOOKUP + custom order
Example 4 — LAG windows
Example 5 — Weekly
Summary and Next Steps

Example 1 — Semiannual schedule (baseline, 6m, 12m, 18m) with a 90-day post window

In this example, participants are expected every 6 months (baseline, 6m, 12m, 18m).

A fixed 90-day window after each target date allows visits to be considered valid if late.

This mimics a classic longitudinal cohort with semiannual follow-up.

Example 2 — Dense early schedule (D7, D14, D28, M2, M3)

This design front-loads follow-up soon after baseline.

Visits are scheduled on Day 7, Day 14, Day 28, then monthly at 2 and 3 months.

Windows are narrow (±3 to 7 days), so protocol deviations appear quickly if visits slip.

Example 3 — LOOKUP schedule with custom order mapping

Here, each participant has their own target schedule, stored in a lookup table.

Visit names (e.g., D3, D10, W4, M2) are not tied to simple day offsets, so the macro uses a mapping to preserve the intended order.

This reflects studies where calendars differ per subject (e.g., post-surgery, post-delivery, or custom event-based follow-up).

Example 4 — Asymmetric post-window (LAG policy)

The LAG policy allows the post-window to extend longer than the pre-window.

In this case, quarterly visits (Q1–Q4) have a nominal ±7/21-day window, but the LAG logic expands the post-allowance further based on interval length.

This mimics operational rules where late arrivals are tolerated, but very early visits are not.

Example 5 — Weekly for 8 weeks

style='text-align:center; font-size:16px; line-height:1.4;'>Weekly visits (baseline through Week 8) are scheduled tightly with only a ±2–3 day tolerance.

style='text-align:center; font-size:16px; line-height:1.4;'>Because windows are narrow and cadence is high, the model naturally produces more out-of-window flags.

style='text-align:center; font-size:16px; line-height:1.4;'>This design mirrors intensive follow-up protocols such as early treatment monitoring.

Summary and Next Steps

Across these examples, visit_tracker generated expected schedules, applied windows, greedily matched actuals to targets, and summarized on-time/early/late/missed outcomes at both participant and visit levels. The plots visualize completion/retention with an optional target line.

Choose an INTERVAL schedule for simple baseline-offset calendars or a LOOKUP schedule for per-ID targets and custom visit naming. Use order_map_ds=or lookup_order_var= to control non-chronological display order.

What to customize next

• Windows: Tune window_def, window_pre, window_post to match protocol tolerance.
• Visit ordering: Provide a small map via order_map_ds= for labels like BL, D3, W4, M2.
• Target rate: Draw a reference line with target_rate= (e.g., 0.80 for 80).
• Outputs: Rename out_* tables to integrate with your reporting pipeline.
• Plots: Aggregate to per-visit (overall) before plotting to avoid percentages > 100 when multiple sites exist.

Report generated on August 29, 2025.