Healthcare / Operations
Healthcare / Operations case study — how Meritt Dev built and runs Anesthesia staffing group: OR analytics and payroll without the spreadsheets Anesthesia staffing group: OR analytics and payroll without the spreadsheets
An anesthesia staffing group provides CRNAs and anesthesiologists to nine hospital sites of a large US health system. Its operating-room reporting arrived as a 70-page PDF two to three weeks late, and every month about 10,500 timesheet rows were cleaned by hand before payroll could be run.
Overview
We built two connected products: OR Analysis, a live utilisation and financial dashboard with a board-ready deck export, and Data Resolution, which turns raw timesheets into clean hours and a payroll upload file. Delivered by Meritt's founders as engineering leads on the project, working with our US partner studio.
- [01]
The challenge
An unused OR block costs the group roughly $1,000 a day, but the only utilisation view was a static PDF with no financial analysis, delivered weeks after the fact. Decisions about block schedules and coverage were made on stale numbers.
Timesheets came out of the scheduling system with duplicates, non-work rows, missing times and shifts that spanned two sites. One sample went from 10,572 raw rows to 7,071 after manual cleaning. Payroll meant one person reviewing every provider line by line and colour-coding a workbook before uploading it to the payroll provider.
- [02]
What we built
OR Analysis ingests billing exports, block schedules, program schedules and quality workbooks and shows block, surgeon and room utilisation, heatmaps, coverage, financial summaries and a quality scorecard with an approval workflow. One click exports a board deck as PDF or PowerPoint.
Data Resolution cleans the timesheets: removes non-work rows, resolves duplicates, extracts times and splits multi-site shifts, then generates the 1099 CRNA payroll as a ready-to-upload CSV. AI is used where judgement helps and nowhere a number depends on it: suggesting answers for ambiguous rows, resolving surgeon names, reading block schedules from images and answering questions over the dashboard data.
A Next.js and TypeScript monorepo on Azure with PostgreSQL, covered by 1,515 automated tests.
- [03]
Where it is now
Started as a free proof of concept demonstrated in January and is now in production on Azure, with billing data live across all nine sites. A September payroll run matched all 108 lines of the manually reviewed upload it replaced.
The group's leadership gets utilisation and financial figures as the month happens rather than weeks later, and the monthly timesheet clean-up is a review of exceptions instead of a rebuild of the whole workbook.
Related