Ambulatory infusion pumps are not a single device category with uniform capabilities. Primary documentation from manufacturers and FDA guidance shows meaningful differences in delivery modes, indicated routes, fluid restrictions, and mobility design. A procurement or clinical engineering team evaluating these devices should treat the operator's manual and premarket submission expectations as the core evidence base, not marketing summaries. This review examines documented specifications from the Moog Curlin 6000, ICU Medical CADD-Solis 2110, Eitan Medical Avoset and Sapphire, and FDA's total product life cycle guidance. It identifies what the documents establish, what remains unknown, and how a hypothetical evaluation could be structured without inventing test results or clinical outcomes.

Delivery Modes Define the Workflow, Not the Device Name

The term ambulatory infusion pump describes a form factor and use environment, but the programmed therapy modes determine what a specific pump can do. The Moog Curlin 6000 manual lists five delivery modes: continuous infusion, patient controlled analgesia, TPN with automatic ramping, intermittent, and a mode for variable-rate therapies [1]. The CADD-Solis 2110 manual describes continuous rate, intermittent bolus, PCA or PCEA dose, and clinician bolus, with each method programmable individually or in combination [6]. The Avoset manual states medication is delivered at continuous rate, with an intermittent bolus, with a patient bolus, or with taper [5].

These mode lists are not interchangeable. A pump that supports PCA demand dosing requires lockout and bolus interval logic that a simple continuous-rate pump may not include. The CADD-Solis documentation illustrates two distinct behaviors for intermittent bolus and PCA interaction: one setting treats PCA lockout as the controlling interval, while another treats the programmed bolus interval as controlling [6]. The Sapphire manual lists continuous, intermittent, TPN, PCA, multi-step, and epidural modes [7]. A team comparing devices should map each required therapy protocol to a documented mode, rather than assuming the ambulatory label covers the workflow.

Route and Fluid Restrictions Are Explicit and Consequential

Documented indications and contraindications create hard boundaries for device selection. The CADD-Solis 2110 manual states the pump is indicated for intravenous, intra-arterial, subcutaneous, intraperitoneal, near-nerve, intraoperative site, epidural, and subarachnoid infusion, but it is not to be used for blood or cellular blood products because the pumping mechanism would damage them [6]. The same manual warns against epidural or subarachnoid administration of drugs not indicated for those spaces [6]. The Sapphire manual, by contrast, lists blood and blood products among the fluids the pump is designed to deliver [7].

These differences matter for formulary and service-line planning. A hospital system that needs a single ambulatory platform for both epidural analgesia and blood product infusion would find the CADD-Solis contraindication disqualifying for the blood use case, while the Sapphire documentation supports that fluid class. The Avoset manual lists parenteral routes including intra-arterial, intravenous, subcutaneous, perineural, and epidural, plus enteral infusions, and includes TPN, enteral nutrition, and lipids among fluids [5]. Route and fluid documentation should be checked against the specific drug labeling, because FDA guidance expects manufacturers to identify a legally marketed drug or biologic for each proposed route and dosage [2].

Documentation Checklist for an Ambulatory Pump Evaluation

Build a comparison table with one row per required capability and one column per candidate device. For each row, cite the exact manual section or premarket submission page. Required rows include: delivery modes, indicated routes, contraindicated fluids, environmental operating ranges, power source and battery requirements, accuracy test conditions, and software elements such as drug library and alarm handler. Mark any row where the documentation is missing or incomplete as unknown, and request the missing evidence from the manufacturer before proceeding.

Mobility Evidence Requires Environmental and Power Specifications

FDA guidance states that if an infusion pump is intended for transport or ambulatory use, the manufacturer should describe how it is designed for mobility, environmental conditions such as water exposure, altitude, and electromagnetic interference, and ruggedness [4]. This is a documentation requirement, not a performance claim. The Avoset manual provides operating conditions of 5 to 40 degrees Celsius, 15 to 90 percent relative humidity, and 70 to 106 kPa atmospheric pressure, with a note that operation was tested in transient conditions of minus 20 and 50 degrees Celsius for 10 minutes [5].

The Moog Curlin 6000 manual describes a small, compact, lightweight design that operates accurately in any position and can be carried in soft-packs, pole mounted, or placed in locked safety shells [1]. The manual also includes a safety note requiring an AC adapter and extra C cell batteries when the pump is used in a life-sustaining situation [1]. These are design and operational constraints, not independent verification of battery life or accuracy. A procurement team should request the specific environmental test conditions and power management documentation for each candidate device, because the manuals do not provide uniform test protocols across manufacturers.

A Hypothetical Evaluation Workflow for Two Candidate Pumps

Consider a hypothetical evaluation of the CADD-Solis 2110 and the Sapphire pump for a hospital-based ambulatory infusion program. The first step would be to extract the documented indications, contraindications, delivery modes, and fluid classes from each operator's manual. The CADD-Solis manual documents a blood product contraindication and epidural anesthetic limitation to short-term indwelling catheters [6]. The Sapphire manual documents blood and blood product delivery and lists continuous, intermittent, TPN, PCA, multi-step, and epidural modes [7]. These documented differences would be recorded without assigning a winner.

The second step would be to request missing evidence. Neither manual provides a complete, comparable accuracy test protocol in the supplied excerpts, so the team would ask each manufacturer for the test conditions, fluid viscosity, and syringe or set configurations used. The third step would be to map the institution's required therapy protocols to documented modes and route indications. This hypothetical workflow is a suggestion for organizing evidence, not a claim that either device has been tested or selected in any real procurement. The manuals alone cannot establish comparative clinical performance or total cost of ownership.

What the Documentation Cannot Establish

Primary manuals and FDA guidance leave several evaluation questions unanswered. The supplied excerpts do not include complete accuracy test results for all devices, so no comparative accuracy claim can be made. The Moog manual includes a footnote that Moog cannot guarantee the plus or minus 5 percent accuracy claim when using syringes [1], but the full test conditions are not present in the excerpt. The Avoset manual lists pump accuracy as plus or minus 5 percent per IEC 60601-2-24 in its technical specifications appendix, but the supplied excerpt does not include the complete test protocol [5].

The documentation also does not establish real-world reliability, service life across all models, or interoperability with hospital information systems. The Sapphire manual states a service life of 7 years from the date of manufacture [7], but no comparable service life figure appears in the other supplied excerpts. FDA guidance describes software elements such as drug libraries, real-time clocks, on-board memory, pump logs, alarm handlers, and watchdog timers as items manufacturers should describe [4]. Whether a specific pump includes all of these elements requires the full premarket submission or technical file, which is not present in the retrieved excerpts.