A weapons-screening system can detect a knife and still fail as a hospital entrance.

Technology matters. The people, space, procedures, and tradeoffs around it matter at least as much. That is what controlled testing and live checkpoint observation continue to show.

Why hospital leaders are deciding now

California hospital leaders are making consequential decisions about hospital weapons screening technology. The California Hospital Association represents more than 400 hospitals and health systems, and many are preparing for Assembly Bill 2975.

The law directs the Occupational Safety and Health Standards Board to amend the hospital workplace-violence standard by March 1, 2027, including automatic body screening at specified entrances. Hospitals outside California also face high volumes of workplace violence, some involving weapons. The testing lesson applies beyond one statute.

Security and executive leaders are receiving polished presentations, performance claims, and throughput estimates. Those claims are useful starting points. They are not acceptance criteria for a procurement decision.

What field testing and live checkpoints showed

I have witnessed 28 individual controlled weapons-screening evaluations. During BPS’ 2026 benchmarking work, the firm observed more than 50 active weapons-screening checkpoints across the United States.

BPS has used a written weapons-screening test methodology for years. A recent controlled evaluation of five systems, plus those live observations, underscored the same finding: several prominent systems struggled in rigorous testing under protocols that commonly occur in busy healthcare entrances.

Those results do not mean the systems could never work. They mean equipment can fail in real operating conditions, and any path to acceptable performance will involve tradeoffs in detection scope, nuisance alerts, divestment, staffing, throughput, footprint, or workflow.

Hospitals may not discover those tradeoffs until after procurement or deployment, when changes are more expensive and operationally disruptive. Testing is more than determining whether a device can detect a knife. Phones, keys, implanted devices, bags, and personal belongings are part of the operational landscape.

How to read the results

Across several years of testing, some systems have not demonstrated enough stability, clear-passage performance, detection reliability, or component reliability to continue without change. Observed problems include instability, excessive alerts on clean entrants, missed threats, inconsistent localization, and component failure.

Hospitals should expect residual errors.  Even capable systems may produce nuisance alerts, misses or localization errors under some conditions. 

Completing thorough testing creates a fuller evidence record. It is not a guarantee. Even with a strong configuration, residual risk remains.

Testing must expose the tradeoffs

No technology setting simultaneously maximizes detection, eliminates nuisance alerts, avoids divestment, preserves throughput, and minimizes staffing. Testing reveals how the hospital chooses to manage that balance.

Detection scope versus operational burden

A system configured to detect smaller knives or less distinct objects may generate more alerts on common items. Each alert creates secondary screening, slows throughput, affects patient and visitor experience, consumes staff time, and uses space. Persistent nuisance alerts can also weaken attention and create fatigue among screeners.

A hospital may choose a narrower target threat. Leadership should make that decision explicitly rather than inherit it from product settings.

Divestment

Allowing entrants to keep phones, keys, and other approved belongings can reduce friction and preserve flow. Requiring those items to be removed may improve clear-passage performance, but it adds time, trays, property-control issues, and staff work. Testing should measure both the alert benefit and the checkpoint cost.

Coverage, bags, and entrance design

Additional sensors may improve body coverage, and X-ray may provide better information about bags. Both add equipment, trained operators, cost, power, queue space, and person-to-bag controls. Entrance consolidation can reduce equipment and staffing, but it may concentrate volume, complicate clinical access, and leave other perimeter pathways to control.

Clinical access

A primary lane that works for ambulatory entrants may not clear wheelchair users, people with implanted devices, or urgent patients. An equivalent alternate-screening process preserves access to care, but it requires trained staff, space, privacy, alternative screening technology, and time. Those resources belong in the total operating cost.

These points are not arguments against the technology. They explain why a system can be capable and still be a poor match for a particular hospital entrance. Device throughput is not checkpoint throughput. Detection performance cannot be separated from the people, space, and procedures needed to resolve the alerts the device produces.

Test the configuration, not the roadmap

Give credit only for capability demonstrated by the hardware, software, and sensor arrangement submitted for testing. Planned upgrades, future detection capabilities, and promised site-specific improvements may later have value. They are not current performance.

The same principle applies to physical configuration. If a system requires multiple sensors for full body coverage, test the proposed multi-sensor arrangement. If cameras support attribution or localization, test their failure and obstruction. If a bag scanner is included, test each device independently and then test the handoff.

The hospital is buying an operating checkpoint, not an isolated demonstration.

What a disciplined hospital evaluation should establish

A credible protocol begins with the hospital’s risk and operating requirements. Security, facilities, emergency department and patient-experience leaders, and occupational health should set those requirements together. An independent protocol keeps the evaluation from becoming a vendor demonstration. A physical security risk assessment is often the right place to lock those requirements before any equipment is scored.

Learn more about the BPS approach to risk assessment

At minimum, the protocol should address the following:

• Define the target threat. Identify the range of critical objects, materials, and sizes the hospital may wish to detect. Leadership may not have made that policy decision yet, so flexibility is essential. Report important article types separately so strong results do not conceal weaknesses.
• Set written acceptance gates. Establish limits for critical-threat misses, nuisance alerts, equipment instability, and component failures before testing begins.
• Test more than one operating profile. Record the exact sensitivity setting, software version, threat library, calibration, and lane geometry. Repeat affected tests after changes, because greater sensitivity may improve detection while increasing nuisance alerts.
• Use realistic people and movement. Vary body build, clothing, item placement, orientation, walking speed, and position within the lane. Test the passage rules the hospital will enforce, including close spacing and arrival surges.
• Evaluate the whole checkpoint. Measure person-to-alert attribution, localization, secondary-screening time, staffing demand, queue formation, bag control, accessibility, and alternate screening. Device throughput alone is not checkpoint throughput.
• Challenge failure and recovery. Interrupt power and network service where it is safe, and simulate component loss. Require clear fault notification and documented verification after repair, relocation, or software change.

Use failure as information

A failed test does not always require permanent rejection. It may reveal a correctable configuration problem, an environmental limitation, or a mismatch with the operating concept. The vendor may explain the result, make a controlled correction, and return for a documented retest.

The hospital should not lower the acceptance standard after seeing the result.

Any correction should be evaluated for its operational consequence. If sensitivity is lowered to reduce nuisance alerts, repeat the critical-threat matrix. If divestment is introduced, measure throughput, staffing, and property handling. If added sensors, local model training, or vendor calibration are required, include the footprint, time, cost, and dependence in the decision. A better retest matters only when the hospital understands what produced it.

Procurement momentum should not turn a pilot into an approval. A conference-room demonstration does not establish performance beside automatic doors, wheelchairs, dense bags, clinical traffic, and simultaneous alerts. Site acceptance and stress testing still matter.

The decision before an award

The lesson is not that hospitals should distrust manufacturers. Trust should be earned through evidence. Independent, repeatable testing protects hospitals and capable vendors by separating demonstrated performance from claims.

Before an award is made, leadership should be able to answer one practical question: Has the exact proposed configuration passed a written, site-relevant protocol at the settings and staffing levels the hospital intends to use, or is it adaptable enough to flex if screening objectives change?

A hospital that learns the answer during a controlled test avoids learning it in a crowded lobby. That diligence will not eliminate weapons risk. It will help select technology that fits a complete program the hospital can operate, measure, and improve. Hospitals that want a planning path before they test can start with BPS’s AB 2975 weapons screening guide for California hospitals.

Read our AB 2975 Weapons Screening Guide

Article Note

This article is operational guidance for hospital leaders considering weapons-screening technology. It is not legal advice and is not a determination of compliance with Assembly Bill 2975, Cal/OSHA, or any other standard. Requirements, effective dates, and entrance criteria should be confirmed with qualified legal, occupational-health, and regulatory counsel. BPS does not sell weapons-detection systems, represent manufacturers, or benefit from a specific technology decision. Testing observations described here are not product ratings.