PIV insertion

Each encounter is one patient who needed a peripheral IV from me. Success means IV access was obtained. First attempt success means access on the first stick. Logged October 2025 to September 2026.

Encounters—patients attempted

Success rate—— placed

First attempt success—— on the first stick

Ultrasound use—of — individual attempts

Encounters by month

Each bar is the month's encounters, split by how access was reached.

Placed on first attemptPlaced on a later attemptNo access obtained

Which attempt succeeded

Share of all encounters, by the attempt that obtained access.

Where access was obtained

Successful placements only, by landmark and by arm.

SBFT placement

Small bore feeding tubes placed at the bedside, January 26 to September 27, 2026. A placement counts as successful on insertion. The tip location is the outcome that matters, because the goal is the small bowel.

Placements—— new, — repositioned

First attempt success—— placed on the first pass

Tip in small bowel—— ended in the small bowel

Placements in order

Letter shows the nare used. Color shows where the tip ended.

Small bowelStomach

Tip location by nare

Placements in each nare, split by where the tip ended.

Compass rounding

Rounding on staff is one of the most important things a charge nurse, resource, or CSN can do, but presence without structure is easy to let slip on a busy MICU night. So I built an app called Compass to hold myself accountable to it. I established three check-in windows per shift as my rounding goal, and each time I completed an interaction, I logged it and categorized what the conversation was about, whether clinical, social, emotional support, education, or even a surface-level check-in. Over three months and 367 rounding windows, here's what the data says.

Presence rate—— of — windows

Missed opportunity—— did not round

Check in rate—— check ins

Substantive—interactions with a category

Presence by month

Share of addressable windows where I rounded.

Presence by time of shift

Rounding windows at the start, middle and end of the night.

What the substantive interactions covered

Share of category tags. One interaction can carry more than one category, so the shares describe the mix and not a count of interactions.

Lunch times

SARA has a built in module that can predict a nurse's lunch time based on a nurse's lunch time preferences and history of going to lunch on time. Over the course of a month, I observed the times selected for themselves and then ran the lunch times scheduling algorithm. I then compared what staff self-selected versus what the algorithm predicted. Data from August 25 to September 27, 2026.

Exact match—— of — data points

Within 30 minutes—— data points

Within 1 hour—— data points

Median error—mean —

Data points by accuracy band

How far SARA's predicted time was from the actual time.

Hit the preferred time

If the algorithm set the schedule, compared with the times staff entered themselves.

When the algorithm moved someone

How far the predicted time sat from the preferred time, for those that moved off it.

Rows without a single preferred time are left out of the preferred time comparison. Only closeness to the preferred time is measured, so this says nothing about coverage. Which staff were working together mattered more than preference in how lunches landed, but that is not yet captured or recorded in the data.