Work Sampling in Hospitals and Services
What work sampling can contribute in hospitals and services, with limits on interpreting quality, workload and care outcomes.
Updated:
Sampling estimates time distribution through instantaneous readings. In hospitals it may guide investigations of supplies, documentation or coordination. It does not independently measure care quality, clinical need, safe staffing or burnout.
Define a specific question
For example: what share of a unit team's time involves searching for material, and on which shifts? This differs from patient waiting time or staffing requirements. Patient waits need pathways and timestamps; staffing needs demand, acuity, competencies and safety requirements.
Documentation, medication preparation and coordination can be essential. Time without direct contact is not automatically waste or deficient care. Do not use an unsupported sector percentage without comparable source and population.
Classify with the clinical team
An initial framework might separate direct care, documentation, clinical preparation, coordination, logistics, confirmed resource waiting, breaks and other. Define simultaneous-activity rules and avoid counting the same instant twice. Waiting reasons may need later verification.
Do not record diagnoses, clinical conversations, screens or patient identities when unnecessary for activity description. Observers must follow access, confidentiality and unit protocols and withdraw when care requires it.
Design and precision
Define units, professionals, shifts and days, covering the conditions represented. Select instants with known probabilities and record missed readings, especially when workload or privacy prevents observation. Such missingness may be informative and is not fixed by replacing it with a quieter moment.
Under independent Bernoulli records, n=ceiling[z²p₀(1−p₀)/e²]. With z=1.96 and p₀=0.5, plan 385 readings for five absolute points or 1,068 for three. People observed together and repeated episodes may be correlated; adapt the analysis.
Teaching example
Assume 600 valid unit readings:
| Activity | Records | Proportion |
|---|---|---|
| Direct care | 240 | 40% |
| Documentation | 120 | 20% |
| Preparation and coordination | 120 | 20% |
| Logistics | 60 | 10% |
| Confirmed equipment waiting | 30 | 5% |
| Breaks/other | 30 | 5% |
The approximate normal 95% logistics margin is 2.4 percentage points under the simple model. This does not establish all that time is avoidable. Check routes, shortages and replenishment before a trial. Also monitor incidents, quality, care time and team workload: transferring a task does not eliminate its cost.
Data protection and governance
Review the protocol with clinical and data-protection staff before collection. Determine purpose, legal basis and, where health data are involved, special-category requirements. Do not presume legitimate interest applies to public authorities performing their tasks.
An impact assessment is required where processing is likely to create high risk under Article 35 and applicable criteria, not automatically for every study. Aggregating results does not necessarily anonymise source records. These require case-specific assessment; software does not certify compliance.
Results and tools
Report proportions, coverage and limitations without promising universal wait reductions or clinical improvements. Observer reactivity can vary; randomisation does not remove it.
Do not call a simple activity proportion hospital OEE. Equipment with defined output needs specific OEE inputs; direct-contact time does not replace clinical outcomes.
WorkSamp supports state planning and recording. Verify suitability and configuration for healthcare before entering data.
Sources
NIST: proportion intervals. GDPR: Articles 6, 9 and 35. Spain's Occupational Risk Prevention Act.