Orthopedic and rehabilitation devices tend to get into trouble in the same places. Implants carry cyclic load for years, sensors sit next to metal, wearables slide over soft tissue, and patients use the device at home with nobody watching.
Orthopedics is one of the main areas at OVA Solutions, a medical device engineering group of 62 engineers working under an ISO 13485:2016 certified quality system. We have built a load-sensing rehabilitation crutch, a wearable knee monitor, a post-operative cold therapy device, electronics for a rehabilitation chair and a cranial fixation tool.
Load and fatigue testing set the schedule
An orthopedic implant is loaded with every step for years, so the bench program is built around cyclic fatigue tests on the worst-case size and configuration. The methods are well established:
- ASTM F1717 for spinal implant constructs in a vertebrectomy model
- ASTM F2077 for intervertebral body fusion devices
- ASTM F543 for metallic bone screws and ASTM F382 for metallic bone plates
- ISO 14243 for wear of total knee replacements and ISO 7206-4 for the endurance of hip stems
A fatigue run goes to millions of cycles, so one specimen can hold a test frame for days. A common approach is to run finite element analysis first (ASTM F2996 for hip stems, ASTM F3161 for knee femoral components) to find the worst case and put only that configuration on the frame. For fracture fixation plates, non-spinal metallic bone screws and washers, spinal plating systems and facet screw systems, FDA publishes performance criteria that can replace direct comparison testing with a predicate, under the Safety and Performance Based Pathway, a form of 510(k).
Instruments need the same discipline. On our cranial fixation tool, a robotic arm ran 100 standardized insertions under constant torque to measure durability and bit wear before design for manufacturing. The note on validating a cranial fixation interface explains why no fatigue standard gives a driver’s cycle count and why the count should follow from the reprocessing claim.
Sensors on or near metal
Wearables over an implant
An inertial sensor fuses a gyroscope, an accelerometer and a magnetometer. On a knee monitor for arthroplasty patients, the magnetometer sits above a mass of cobalt chromium and titanium. In a 2023 case series with prosthesis users, one participant’s discrete-point error reached 29.21 degrees because a microprocessor-controlled prosthetic knee interfered with the magnetometer. Knee flexion is a sagittal rotation, so it can be recovered from the gyroscope bounded by the gravity vector, with magnetic heading kept out of the clinical axis.
The error budget matters as much. Published knee flexion error of body-worn inertial sensors is 4.4 to 9.4 degrees, and gait deficits after knee replacement are 4.8 to 6.6 degrees. Our note on designing a wearable knee monitor explains why the defensible claim is change within one patient over weeks: most of the error is a constant bias from mounting and soft tissue, and it largely cancels between week two and week eight.
Implants and the MRI scanner
If any part is implanted or stays on the patient during imaging, plan MR safety testing early. FDA’s guidance on testing and labeling devices for the MR environment (October 2023) points to ASTM F2052 for displacement force, ASTM F2213 for torque, ASTM F2182 for radio frequency heating, ASTM F2119 for image artifacts and ASTM F2503 for marking. It counts screws, plates and hip implants as fully implanted passive devices, and orthopedic external fixators as partially implanted ones.
Load sensing
A crutch measures load through the crutch, while the surgeon prescribes load through the limb. The link changes with the gait pattern, which is why a 2016 comparison of three commercial biofeedback devices against a force plate found weighted kappa of 0.76, 0.58 and 0.19. Our rehabilitation crutch note recommends writing the claim against the measured variable and validating it across gait patterns.
Fixation and mounting
For implants, fixation to bone is the design itself: thread form and pull-out strength, plate geometry, press-fit or cement. A change here ripples through the test plan, because spinal methods such as ASTM F1717 test the assembled construct.
For wearables, mounting is the measurement. Straps sit over skin and muscle that move relative to bone, and every millimeter of donning variation shows up as degrees. Strap design, placement against bony landmarks and a short guided posture at the start of each session matter more than the choice between two sensor parts.
For instruments, the interface with the implant has a two-sided specification. A press-fit screw-to-blade interface must hold the screw during placement and release it cleanly afterward. Wear moves both limits, so retention and release are measured at intervals across the claimed life.
Materials and biocompatibility
FDA clears devices, not materials. Its guidance on ISO 10993-1 (September 2023) asks for a biological evaluation of the final finished device, including processing, sterilization and manufacturing residues. ISO 10993-1 itself was revised in 2025 as a sixth edition.
The evaluation scales with the type and duration of contact: limited (under 24 hours), prolonged (24 hours to 30 days) or long-term (over 30 days). A bone screw is long-term contact with tissue and bone, which brings in endpoints such as genotoxicity and implantation. A strap or a crutch grip touches intact skin, and Attachment G of the same guidance lets certain devices made from common polymers and fabrics submit material information instead of test reports.
Make test samples with production processes, because a new supplier, surface treatment or cleaning agent can reopen the evaluation. For additively manufactured implants, FDA’s 2017 technical considerations guidance still applies: porous structures make residual powder harder to remove and the device harder to sterilize, so validation uses the worst-case geometry.
Usability at home and in the operating room
Rehabilitation happens at home. Lisa Voronkova, our CEO, put it this way on the BoneChat podcast: the clinic sees a few minutes, and recovery happens in the other 167 hours of the week. Published compliance with partial weight bearing is 37.5% after lower-extremity surgery and 22% in patients 65 and over with femur or pelvic fractures. Feedback has to arrive within the step, and a display the patient must look at while walking on a partially loaded limb is a fall risk. Haptic feedback performed better in published studies.
Surgeons are users too. On one of our projects surgeons liked the technical specifications, but the handle was wrong for how they grip instruments during a four-hour procedure, and only watching them work revealed it. FDA’s human factors guidance, revised on August 3, 2026, still recommends at least 15 validation participants per distinct user population, and it counts healthcare providers and lay users as separate populations. Electrical devices for home use are also designed to IEC 60601-1-11, the collateral standard for the home healthcare environment.
Clinical evidence starts with the claim
Most orthopedic 510(k) submissions rest on bench data and a predicate comparison. A PMA usually relies on a clinical study, and a De Novo request often needs clinical data. For sensing devices, the wording of the claim decides the study: on a knee monitor, a claim of change within one patient turns validation into a test-retest design instead of a comparison with motion capture. Confirm the plan with FDA through a Pre-Submission before the protocol is locked.
Typical FDA pathways for orthopedic devices
Orthopedics was the second-largest 510(k) review panel in 2025, with 404 of 3,198 clearances, behind radiology. Physical medicine, which covers most rehabilitation equipment, had 111. A 510(k) is cleared, a De Novo request is granted and a PMA is approved. The class follows the classification regulation:
| Device | Class and usual pathway | Source |
|---|---|---|
| Crutch | Class I, exempt from 510(k) | 21 CFR 890.3150 |
| Metallic bone plates and screws | Class II, 510(k) | 21 CFR 888.3030, 888.3040 |
| Intervertebral body fusion device | Class II, 510(k); Class III and PMA if it contains a therapeutic biologic | 21 CFR 888.3080 |
| Many knee and hip replacement designs | Class II, 510(k) | 21 CFR 888.3560, 888.3353 |
| Intervertebral disc prosthesis | Class III, PMA (15 original approvals, 2004 to 2026) | openFDA, product code MJO |
| Non-invasive bone growth stimulator | Class II, 510(k), since May 18, 2026 (Class III before) | 91 FR 20352 |
Exemptions have limits. Under 21 CFR 890.9, a 510(k) is required when a device has a different intended use or a different fundamental scientific technology than devices already sold in its type, so a crutch that senses load and gives feedback needs a regulatory opinion early. A new device type of low to moderate risk with no predicate goes through De Novo. For fiscal year 2027, FDA user fees are $28,653 for a 510(k) ($7,163 for a small business), $191,020 for a De Novo request ($47,755) and $636,732 for a PMA ($159,183). The FDA class and pathway finder and 510(k) clearance time by product code show where similar devices landed.
We do the engineering side of these submissions: the design history file, risk management inputs and verification reports. For a rehabilitation chair, that meant firmware under software safety Class B, five pre-production units and a 510(k) documentation package. The regulatory strategy and the submission stay with the client or their regulatory consultant. Powered surgical tools raise their own questions, covered in our article on surgical instrument development.
Frequently asked questions
Which standards cover fatigue testing of orthopedic implants?
Device-specific ASTM and ISO methods, such as ASTM F1717 for spinal constructs, ASTM F543 for bone screws, ASTM F382 for bone plates and ISO 7206-4 for hip stems. For some plates and screws, FDA also publishes performance criteria.
Can a wearable sensor measure accurately over a metal implant?
Yes, if the clinically meaningful axis does not depend on the magnetometer: knee flexion can be estimated from the gyroscope bounded by gravity. Validate the device with patients who have implants.
Do orthopedic devices need a clinical trial for FDA?
Usually not for a 510(k), which typically rests on bench testing and a predicate. Intervertebral disc prostheses reached the market through PMA, which relies on clinical studies.
How long does an orthopedic device take to develop?
At OVA, idea to production usually takes 6 to 12 months. The rehabilitation chair electronics took 2,100 hours in 16 weeks, and the load-sensing crutch took about 1,660 hours over eight months, ending with 100 pairs produced. FDA review and any clinical study come on top.
If you are building an orthopedic or rehabilitation device, our orthopedic device development page lists what we have built and the engineering decisions behind it. To go through your load case, sensor choice or test plan with an engineer, contact us.
More on this topic: Orthopedics & Rehabilitation