Tresserra 2026 — Technical Workload Unit (SEAP pilot)
A Spanish multi-centre pilot that timed each technical process in the histopathology laboratory and proposed an 11-minute Technical Workload Unit as a common currency for workload.
Read status: abstract only. The Elsevier full text could not be retrieved. Methods detail — who timed what, sample sizes per process, any dispersion measures — is therefore unknown, and several criticisms below are provisional for that reason. Marked
[unverified]where that matters.
What they did
Timed technical procedures over 5 days across 10 hospitals, and took the median time for each process. No statement in the abstract about who recorded the times, how many specimens per process, or whether any dispersion was reported. [unverified]
Headline numbers
Median technician time per specimen, by process:
| Process | min/specimen | TWU |
|---|---|---|
| Large specimen embedding | 10.91 | 1.00 |
| Liquid-based cytology | 4.02 | ~0.37 |
| Registration | 3.92 | ~0.36 |
| Immunohistochemistry | 2.13 | ~0.19 |
| Small specimen embedding | 1.96 | ~0.18 |
| Microtomy | 1.01 | 0.09 |
| Archiving | 0.49 | ~0.04 |
The Technical Workload Unit (TWU) is set at 11 minutes, rounded from the 10.91 min of large-specimen embedding because that is the most time-consuming task. All other processes are expressed as fractions of it, from 0.09 TWU (block preparation and sectioning) up to 1 TWU.
Why it is interesting here
It is an external benchmark for exactly the kind of operational measurement the group already does internally, and it comes from a national society working group rather than a single department — so it is a defensible number to compare against rather than a local anecdote. It also proposes a unit, which is the part that makes cross-department comparison possible at all.
Statistical and methodological problems
Several, and they matter if these numbers are going to be used for staffing arguments.
- Medians with no dispersion. The abstract reports point estimates only. Time-and-motion data is right-skewed and heteroscedastic — a median with no IQR or range cannot support a staffing calculation, because the tail is where the staffing pressure actually lives.
[unverified]whether the full text reports spread. - Ten hospitals, and no between-site variance reported. This is the biggest missed opportunity. With 10 sites the single most useful output would be how much sites differ from one another — that is what tells you whether the benchmark transfers. Pooling to one median discards precisely the information a reader needs.
- No case-mix adjustment. “Large specimen embedding” spans a Whipple and a simple excision. Without complexity stratification, 10.91 min is an average over an unstated specimen mix, and a department with a different mix cannot use it directly.
- Per-specimen normalisation ignores batching. Microtomy and IHC are batched processes: per-specimen time falls as batch size rises. A per-specimen figure without the batch size it was measured at is not portable between laboratories with different run sizes.
- Anchoring the unit to the noisiest estimate. Defining 1 TWU as the most time-consuming task ties the entire scale to the process with the widest expected variance. A more robust anchor would be a high-volume, low-variance process such as microtomy.
- Precision beyond what the design supports. Reporting 0.09 TWU implies two significant figures from a 5-day pilot; the rounding of 10.91 to 11 then propagates into every ratio.
- Observation effect. Staff who know they are being timed do not work at their usual pace. Not addressed in the abstract.
[unverified] - Self-described pilot, single country. Spanish laboratory staffing models, grossing conventions, and technician scope of practice are not universal. Transfer to a Turkish department needs local re-measurement, not adoption.
What I would want before using these numbers
Between-site range per process; batch sizes for microtomy and IHC; a complexity breakdown of “large specimen”; and the number of specimens behind each median.