A Novel 3D-Printed Gravity-Independent Air-Eliminating Filter for Rapid Intravenous Infusions
INTRODUCTION: The absence of a consistent downward G vector can make separation of gases from liquids challenging, such as in field medicine without stable upright equipment or during spaceflight. This limits the use of medical equipment and procedures like administration of intravenous (IV) fluids in microgravity and can make field medicine hazardous. Administering IV fluids and medications in microgravity requires a technique to separate air from the liquid phase. Current commercial filters for separation of gases are incompatible with high flow and blood. We present a novel filter designed to provide adequate air clearance without a consistent downward G vector. METHODS: Inline air-eliminating filters were designed for use with IV fluid tubing in microgravity using computer-aided design software and printed using nylon 12 on an EOS Selective Laser Sintering 3D printer. A 0.2-μm membrane filter was adhered around a central, hollow pillar with external spiral baffles allowing separation and venting of air from the fluid. Results were compared against commercially available inline air-eliminating filters. RESULTS: The 3D-printed filters outperformed the commercial filters in both percentage of air removed and flow rates. The centrifugal, baffled filter had flow rates that far exceeded the commercial filters during rapid transfusion. DISCUSSION: IV fluid administration is an often underappreciated and a necessary basic requirement for medical treatment. An air-eliminating filter compatible with blood and rapid transfusion was developed and validated with crystalloid solutions to allow the successful administration of IV fluid and medication without a consistent downward G vector. Formanek A, Townsend J, Ottensmeyer MP, Kamine TH. A novel 3D-printed gravity-independent air-eliminating filter for rapid intravenous infusions. Aerosp Med Hum Perform. 2024; 95(6):327–332.
Field medicine can make equipment dependent upon a stable downward G vector challenging, especially when relying on upward flotation of bubbles in a container of liquid; if a gravity-driven air trap loses its intended orientation with respect to G, air may be infused into the patient. Human spaceflight takes place in an austere, remote, and physiologically challenging environment. Medical provisions are severely limited by considerations of power, weight, volume, and the available skill of the crew. Further, in a 2.5-yr Mars mission involving six crewmembers, one significant medical event could be expected per Mars mission.1 Microgravity induces profound cardiovascular changes which extend from volume distribution, solute handling, and cardiorenal effects, including a profound diuresis and decrease in plasma volume by 10–17%.2 While the physiological effects of spaceflight are known in healthy spaceflight participants, compounded changes in physiology from further stress to the cardiovascular system are currently unknown. This hypovolemia may have profound interactions with other pathophysiology such as sepsis, burns, and trauma, and, therefore, intravenous fluid resuscitation during long-duration spaceflight may at one point be needed.3
Unfortunately, there are multiple challenges involved in intravenous fluid administration during spaceflight. These include transportation, storage, and the administration of intravenous fluids. Common and recommended practice terrestrially is to place an upright (gravity driven) air trap distal to a blood/fluid warmer to collect bubbles that have outgassed.4,5 In microgravity, liquids and gases do not undergo buoyancy-driven separation, and the miscibility of air and liquid results in trapped air pockets which can cause air emboli.6 In microgravity environments, the intravenous fluid takes on a formation akin to foam, which significantly complicates its administration.6
Inline air-eliminating filters have been tested which prove to be effective in overall gas removal from crystalloid solutions at low flow and transfusion pressure.7 However, no gravity-independent air-eliminating device compatible with high flow and blood products currently exists on the market. Further, the cost, weight, and one-time use nature of current commercially available air-eliminating filters complicate their use in microgravity or long-duration spaceflight. 3D-printing has been successfully demonstrated in microgravity,8 which presents the possibility of 3D-printing an intravenous (IV) inline air filter in flight, using generic raw materials and a small-volume supply of filter membrane, that would allow higher flow rates with improved air filtration in the microgravity environment. We created a 3D-printable inline intravenous air-eliminating filter for use in microgravity that has the potential to address the weight, cost, and flow issues of using commercially available filters, and we hypothesized that such a device would have improved flow rate and bubble control compared with commercial filters.
METHODS
Equipment
An air-eliminating filter was developed for use with IV fluid tubing in microgravity using computer-aided design (CAD) software and printed using nylon 12 polyamide powder on an EOS Selective Laser Sintering 3D Printer. The air eliminating filter used a Sterlitech (Auburn, WA, United States) 0.2-μm oleophobic membrane filter around a central, hollow pillar allowing venting of air. Fluid was accelerated centrifugally around the hollow core, forcing the less dense air into contact with the filter (Fig. 1).
Citation: Aerospace Medicine and Human Performance 95, 6; 10.3357/AMHP.6376.2024
A second baffled version of the filter included helical baffling about the central pillar, requiring that more turns be completed by the fluid and increasing the time available for gas to pass through the filter before the liquid exited the chamber. Fluid enters the chamber, opening a potential space between an expandable latex membrane and the hydrophobic membrane filter. The fluid-air mixture is then forced to rotate around the hollow core by helical baffles in the housing (Fig. 2).
Citation: Aerospace Medicine and Human Performance 95, 6; 10.3357/AMHP.6376.2024
Procedure
The 3D-printed air-eliminating filters were tested against a commercially produced B. Braun (B. Braun Medical, Bethlehem, PA, United States) 5-μm air eliminating filter. A liter bag of 0.9% commercially prepared Baxter (Deerfield, IL, United States) saline solution at 20°C and sea level was manually de-aired and run through IV tubing suspended at 125 cm (92 mmHg of pressure). For the control, no filter was placed in the IV line. Flow rates were calculated by recording the average time to drain the IV bag. The commercial filter flow rates were recorded with one filter in line, two in parallel, and three in parallel. The centrifugal filter and the baffled centrifugal filter were recorded in the same manner. All filters were primed to remove air in the filter. The times were then recorded for the same configurations with the IV bag pressurized to 150 mmHg and 300 mmHg (upper limit of pressure bag) via a pressure bag.
Air eliminating ability was recorded by injecting 60 mL of air into the flowing IV line via a three-way stopcock and syringe. A one-way valve was placed in the line just distal to the IV bag to prevent air from flowing back into the bag. The experiments were repeated using one B. Braun filter in line, two in parallel, three in parallel, and finally the centrifugal filter and baffled centrifugal filter, all primed to remove air in the filter. In a first test branch, the air bolus was injected slowly over 60 s into the line and agitated to create numerous small bubbles. The air entrained in the degassed, closed collection bag downstream of the filters was then measured. The slow air bolus was conducted at pressures of 92 mmHg, 150 mmHg, and 300 mmHg. The experiment was then repeated using a fast air bolus: 60 mL of air injected over 5 s to simulate a large pocket of air entering the filter. Again, all five filter configurations were tested at pressures of 92 mmHg, 150 mmHg, and 300 mmHg. Resistance and timing of transfusion were assessed with a 1-L bag of saline without addition of the air bolus. Overall time was calculated from when the flow started to when the bag was empty and flow rate was calculated in mL · min−1. Two runs were completed for each configuration with the average of both runs calculated.
Statistical Analysis
The primary measure of this study was the performance in time and percentage of air removed for each type of filter. We conducted an analysis of variance (ANOVA) applied to the variable filter configurations in each pressure run and tested for flow rate, air removal % with a slow bolus, and air removal % with a fast bolus. Statistical analysis was performed in JMP (SAS, Cary, NC, United States).
RESULTS
The flow rate data is displayed in Fig. 3. Infusion at 92 mmHg for the control with no filters was 449 mL · min−1, 153 mL · min−1 for the single B. Braun filter, 232 mL · min−1 for two filters in parallel, 279 mL · min−1 for three filters in parallel, 81 mL · min−1 for the centrifugal filter, and 57 mL · min−1 for the baffled centrifugal filter. Maximum flow rate (infused at 300 mmHg) for the control with no filters was 917 mL · min−1, 316 mL · min−1 for the single B. Braun filter, 481 mL · min−1 for two filters in parallel, 575 mL · min−1 for three filters in parallel, 204 mL · min−1 for the centrifugal filter, and 847 mL · min−1 for the baffled centrifugal filter. Flow rates for the different filter configurations were significantly different for each flow pressure (92 mmHg ANOVA: F = 293.043, DOF = 11, P < 0.0001; 150 mmHg ANOVA: F = 162.687, DOF = 11, P < 0.0001; 300 mmHg ANOVA: F = 358.106, DOF = 11, P < 0.0001).
Citation: Aerospace Medicine and Human Performance 95, 6; 10.3357/AMHP.6376.2024
In the slow and fast air bolus tests (Fig. 4), the volumes of air that were extracted by the filters were significantly different for each flow pressure (92 mmHg Slow Air Bolus: ANOVA F = 9.000, DOF = 9, P = 0.0166; 92 mmHg Fast Air Bolus: ANOVA F = 545.219, DOF = 9; P < 0.0001; 150 mmHg Slow Air Bolus: ANOVA F = 16.910, DOF = 9, P = 0.0041; 150 mmHg Fast Air Bolus: ANOVA F = 1243.955, DOF = 9, P < 0.0001; 300 mmHg Slow Air Bolus: ANOVA F = 47.968, DOF = 9, P = 0.0004; 300 mmHg Fast Air Bolus: ANOVA F = 58.158, DOF = 9, P = 0.0002). In the 92-mmHg slow administration group, 85% of air injected was eliminated with the single B. Braun filter, and 100% was eliminated with the two and three filters in parallel, centrifugal filter, and baffled centrifugal filter. In the 300-mmHg slow administration group, 3.3% of administered air was eliminated by the single B. Braun filter, 35% by the two filters in parallel, 80% by the three filters in parallel, 91% by the centrifugal filter, and 92.1% by the baffled centrifugal filter.
Citation: Aerospace Medicine and Human Performance 95, 6; 10.3357/AMHP.6376.2024
In the 92-mmHg fast air bolus administration group, 4% of air injected was eliminated with the single B. Braun filter, and 27% with the two filters in parallel, 39% with the three filters in parallel, 68% with the centrifugal filter, and 99.6% with the baffled centrifugal filter. In the 300-mmHg fast administration group, 2% of administered air was eliminated by the single B. Braun filter, 3% by the two filters in parallel, 12% by the three filters in parallel, 86% by the centrifugal filter, and 78% by the baffled centrifugal filter.
DISCUSSION
Air embolisms are rare and are often misdiagnosed. While more frequently associated with invasive vascular procedures and mechanical ventilation, case reports exist of peripheral IVs as the source of air embolism.9 Small volumes of venous air are generally benign and the pulmonary system can normally remove venous air by diffusion into the alveoli, but can be exceeded with 50 mL of air.10 In addition, a right-to-left shunt may introduce arterial air. Overall incidence of a patent foramen ovale (PFO) is estimated to be 27.3% based on autopsies.11 Unfortunately, noninvasive screening via transthoracic echocardiography has limited sensitivity; sensitivity of detecting an interatrial shunt with color Doppler is 12%, bubble study (agitated saline contrast) sensitivity is close to 50%, and bubble study with Valsalva increases sensitivity to only 93%.12,13 Furthermore, functionally closed PFOs by higher left sided pressures may hence open and create a right-to-left shunt if right atrial pressures are elevated. Hence, even if a known PFO is a disqualifying condition for spaceflight, current noninvasive screening cannot guarantee absence of a right-to-left shunt.14 As such, introduction of venous air may be problematic and the centrifugal baffled filter we have created has potential applications where reliable de-airing of IV solution is challenging.
The centrifugal and baffled filters did exhibit adequate flow rates, but further optimization should be conducted to minimize internal resistance, especially in low driving pressure states. While the baffled centrifugal filter design was slower at 92 mmHg driving pressure, at 150 mmHg and 300 mmHg it was the fastest filter configuration, reaching 847 mL · min−1. The centrifugal filter was also comparable to the commercial inline filter at low flow rates and pressure but performed significantly better at higher driving pressures. In terms of flow rate analysis vs. infusion pressure, though the centrifugal filter did not outperform the commercial filters at low pressures, the flow rate significantly exceeded the commercial filters at higher infusion pressures.
The centrifugal and baffled filters performed best at high flow and eliminating fast air boluses. Interestingly, the centrifugal configuration had better performance at higher driving pressures, likely due to the increased centrifugal forces resulting from higher angular velocity about the core of the filter, which would increase the buoyancy of the gas phase. We were able to determine a statistically significant difference between the different filter configurations, all of which favored the centrifugal and centrifugal baffled filters. In a microgravity setting, it is reasonable to expect that fluids will be administered through a pressure system, which suggests that the centrifugal baffled filter will perform better than the commercially available filters. Given the significant flow rates that we have found in our study, this could serve as an excellent mediator for massive transfusions for hemorrhagic shock.
The primary limitation inherent to this preliminary study is the lack of testing in a microgravity environment, which must be conducted before transitioning to operational use. The next phase of testing should be conducted, for example, under microgravity conditions during parabolic flight; the 20+ seconds of zero-g are sufficient for filter testing, though it would require slight modifications to the protocol to limit transfusion time and to truncate the slow air bolus time to 20 s. Further, such a parabolic study could add data regarding the +2 Gz portion of the flight. Another limitation is the relatively small sample size. Although the analysis has determined a statistically significant difference in the performance parameters of the filter groups in favor of the 3D-printed filters, further testing is needed. Our experiment was tested with crystalloid solutions; colloid solutions including blood products have not been tested thus far. However, the oleophobic membrane filter was selected for its compatibility with blood. Furthermore, this study does not address the issue of sterility. However, in the private sector, sterile additive manufacturing is actively being explored and these techniques could be easily applied to this study. In addition, the dimensions and performance of the air-eliminating filter can continue to be optimized to reduce flow resistance while maintaining high filtering capabilities. A further limitation relates to how the testing was done. Air injected into an IV line is not the same as air that is intermixed in an IV bag in microgravity. However, we are limited by our study being performed in a 1-G environment to validate the filter prior to testing in microgravity.
The benefit of producing a filter through 3D-printing allows for better efficiency in “upmass,” as a generic raw material can be shared among multiple applications.8 Further, the mass and volume of medical devices that may never be used can be saved for other purposes, while their functions are not foregone, as unlaunched devices can be printed as needed. This provides an excellent modality to change the amount and type of available health-related equipment based on the situation. However, the time taken to 3D-print the filter would likely necessitate it to be used to replenish supplies as opposed to be primary printed when the device is needed and not as a “just-in-time” solution.
Furthermore, the proposed centrifugal air-eliminating filter has potential for use in terrestrial settings such as field medicine in which it is impractical or unsafe to have an upright air trap; a gravity air trap would have to be secured to a stable platform or risk tipping and transfusing a bolus of air. The centrifugal air-eliminating filter could more safely facilitate transfusion of blood products in the field. All components of the filter were selected for their compatibility with blood. Specifically, the oleophobic membrane filter was selected with the advice of the Sterlitech engineers for blood compatibility. However, further testing, especially red blood cell sheering, would need to be conducted before compatibility with blood could be confirmed.
This 3D-printed centrifugal filter represents an important first step in solving the significant challenge of infusion of IV fluid, especially for colloids and under rapid transfusion conditions, in conditions without a consistent downward G vector. In addition, there is utility in austere terrestrial environments where an upright air trap is impractical. Further research is necessary to confirm in vivo utility and flow in microgravity conditions.
CAD drawing of 3D-printed centrifugal filter. Fluid enters the filter housing around a hollow core surrounded by membrane filter. The less-dense gas is forced into contact with the membrane filter and is vented from the device.
CAD diagram of baffled centrifugal filter. Fluid enters the chamber, opening a potential space between an expandable latex membrane and the hydrophobic membrane filter. The fluid-air mixture is then forced to rotate around the hollow core by helical baffles in the housing. Air passes through the hydrophobic membrane filter and exits through openings on either side of the hollow core.
Flow rate vs. infusion pressures.
Infusion pressure vs. percentage of air removed in slow and fast air bolus cohorts.
Contributor Notes
