Paramedic Initiation of Neuroprotective Agent Infusions
Successful Achievement of Target Blood Levels and Attained Level Effect on Clinical Outcomes in the FAST-MAG Pivotal Trial (Field Administration of Stroke Therapy – Magnesium)
Background and Purpose—Paramedic use of fixed-size lumen, gravity-controlled tubing to initiate intravenous infusions in the field may allow rapid start of neuroprotective therapy for acute stroke. In a large, multicenter trial, we evaluated its efficacy in attaining target serum levels of candidate neuroprotective agent magnesium sulfate and the relation of achieved magnesium levels to outcome.
Methods—The FAST-MAG phase 3 trial (Field Administration of Stroke Therapy – Magnesium) randomized 1700 patients within 2 hours of onset to paramedic-initiated, a 15-minute loading intravenous infusion of magnesium or placebo followed by a 24-hour maintenance dose. The drug delivery strategy included fixed-size lumen, gravity-controlled tubing for field drug administration, and a shrink-wrapped ambulance kit containing both the randomized field loading and hospital maintenance doses for seamless continuation.
Results—Among patient randomized to active treatment, magnesium levels in the first 72 hours were assessed 987 times in 572 patients. Mean patient age was 70 years (SD±14 years), and 45% were women. During the 24-hour period of active infusion, mean achieved serum level was 3.91 (±0.8), consistent with trial target. Mg levels were increased by older age, female sex, lower weight, height, body mass index, and estimated glomerular filtration rate, and higher blood urea nitrogen, hemoglobin, and higher hematocrit. Adjusted odds for clinical outcomes did not differ by achieved Mg level, including disability at 90 days, symptomatic hemorrhage, or death.
Conclusions—Paramedic infusion initiation using gravity-controlled tubing permits rapid achievement of target serum levels of potential neuroprotective agents. The absence of association of clinical outcomes with achieved magnesium levels provides further evidence that magnesium is not biologically neuroprotective in acute stroke.
Recanalization of occluded arteries, by intravenous fibrinolysis, endovascular mechanical thrombectomy, or both, is a well-proven strategy for treatment of acute ischemic stroke care.1–5 However, recanalization therapies can only be administered after completion of brain imaging, and mechanical thrombectomy performed only after transport of a patient to an advanced stroke center with a neurocatheterizaiton laboratory. Because neural infarction progresses quickly, there is a need for stroke therapies that can be initiated by paramedics in the pre-hospital setting to stabilize and salvage brain tissue in the first minutes after stroke onset. Early neuroprotection is a promising treatment strategy that can decrease time to therapy for patients with stroke and provide a complementary treatment potentiating recanalization interventions.
Most compounds with neuroprotective properties being developed for acute stroke treatment require intravenous infusion, often with a loading dose followed by a maintenance dose, at set rates. However, programming and maintaining infusion pumps to deliver loading doses of intravenous infusions at controlled rates is not a standard element of paramedic scope of practice in the United States. Similarly, seamlessly transitioning from a pre-hospital loading infusion to an immediately after emergency department (ED) maintenance infusion is not a regular aspect of combined Emergency Medical Services and receiving ED care. If the many promising neuroprotective agents that require controlled intravenous infusions are to be tested in the in the pre-hospital setting, when they can be given with the greatest opportunity to ameliorate outcomes, methods need to be devised for safe paramedic administration of intravenous infusions.
This study investigates the effectiveness of the fixed-lumen, gravity-controlled infusion approach for pre-hospital initiation of intravenous agents deployed in the FAST-MAG phase 3 trial (Field Administration of Stroke Therapy – Magnesium), which tested the efficacy of magnesium sulfate as a neuroprotective agent initiated by paramedics in the field.6–8 The 2 aims of this report are to evaluate the technical efficacy of the infusion approach by analyzing achieved serum levels of magnesium and to delineate the relation between achieved serum magnesium levels and clinical outcomes.
The FAST-MAG trial was a multicenter, randomized, blinded, placebo-controlled, pivotal clinical study conducted of pre-hospital initiation of magnesium sulfate conducted in 315 enrolling ambulances and 60 receiving stroke centers. The detailed Methods of the trial have been described previously.6,7
For the active magnesium arm, the drug administration regimen aimed to rapidly double the serum magnesium concentration and maintain the elevated level for the first 24 hours after enrollment.8,9 The study treatment regimen consisted of a loading dose of 4 g Mg (or matched placebo) in 54 mL of saline infusing over 15 minutes, started by paramedics in the field, followed by a maintenance infusion of 16 g Mg (or matched placebo) over 24 hours, started by ED nurses after hospital arrival. The maintenance dose was to be initiated by ED nurses immediately after completion of loading dose.
A research pharmacy prepared loading dose infusion bags for paramedic use in the study. In the active arm, the pre-hospital dose bag contained 4.81 g of magnesium sulfate in 60 mL of normal saline, allowing 6 mL for priming and 54 mL (containing 4 g Mg) for administration. The placebo pre-hospital dose bag consisted of 60 mL of normal saline only. To deliver the pre-hospital dose, paramedics used gravity-controlled tubing with fixed-lumen size, selected so that when hung at the standard in-ambulance bag height (216 cm), infusion would occur at, and no faster than, 3.6 mL/min (Figure 1). The trial protocol was approved by the institutional review board at each pre-hospital and hospital study site.
Magnesium blood levels were not required by study protocol. However, attending physicians were permitted to order serum magnesium levels if their usual practice was to do so in patients with acute stroke or if the patient developed a condition, such as altered mental status, for which they felt serum magnesium evaluation was clinically indicated. For this study, all magnesium (Mg) levels that were obtained within 72 hours of admission were abstracted from the hospital record.
To preserve study blinding, final, 90-day outcome disability and functional evaluations were performed by raters who had no contact with the hospital chart; the final outcome raters were unaware of whether magnesium levels had been drawn and, if drawn, resulting values. Magnesium serum levels were abstracted from the hospital record only after the final outcome evaluation had been performed. In addition, at the level of data handling, a separate, firewalled process was implemented to insulate magnesium level data from all other trial data. A separate, dedicated group of research staff extracted the magnesium serum level data from the hospital record; these personnel performed no other data abstraction. They entered the magnesium levels into a local database, separate from the main trial database. The magnesium level database was not unlocked until after all prespecified primary and secondary analyses had been conducted and reported from the main trial database.
To assess whether the combined pre-hospital and in-hospital infusion regimen was successful in achieving the targeted rapid and sustained doubling of serum magnesium levels during the 24-hour active infusion period, we analyzed the serum levels in active arm patients, with the placebo arm patients serving as controls. To determine post-infusion washout of magnesium, we also analyzed the subsequent 48 hours after the end of infusion. A sigmoid model for the relation between serum Mg versus time in minutes was constructed, treating multiple observations in the same patient as non-independent and observations from different patients as independent. To specifically analyze the effect of the pre-hospital loading dose, we also analyzed serum magnesium levels in the active and placebo arms measured during the first 30 minutes after completion of the field bolus. To determine the representativeness of patients in whom serum magnesium levels were clinically obtained, we compared patient-level and hospital-level variables for patients with and without a magnesium level ascertainment, in active and control groups combined, using multivariate analysis. To delineate biological factors affecting attained magnesium levels, we analyzed patient demographic and clinical features associated with achieved magnesium levels in the active Mg treatment group.
To assess the effect of achieved magnesium levels on efficacy and safety outcomes, 2 analyses were performed. First, among all patients regardless of treatment arm, we analyzed for differences in serum magnesium levels within the first 24 hours for patients with and without nondisabled (modified Rankin Scale score of 0–1) outcome at 90 days, death at 90 days, and symptomatic intracerebral hemorrhage, using t tests. Symptomatic intracerebral hemorrhage was defined as occurrence of either (1) hemorrhagic transformation in the area of a qualifying ischemic stroke, adjudicated as causally related to neurological deterioration (≥4 points worsening on the National Institutes of Health Stroke Scale (NIHSS) compared with the previous examination) in the first 4 days after onset or (2) intracerebral hemorrhage appearing in a different vascular territory than a qualifying ischemic stroke and adjudicated as causally related to any new neurological deficit in the first 4 days after onset.
Second, we grouped active arm magnesium patients in 5 quintile groups based on achieved magnesium level during the first 24 hours. In forest plot analysis, each actively treatment magnesium quintile group was compared with all placebo patients in whom a magnesium level had been drawn. Outcomes at 3 months analyzed included freedom from disability (modified Rankin Scale score of 0–1), independence in activities of daily living (Barthel Index≥90), minimal neurological deficit (NIHSS≤1), and median neurological deficit (median NIHSS). Safety outcomes analyzed were symptomatic intracranial hemorrhage within the first 4 days and death by 3 months. These efficacy and safety end point analyses were adjusted for age, baseline stroke severity (Los Angeles Motor Scale score), pre-stroke disability, and geographical region of enrolling ambulance. The criterion for statistical significance was set at an α level of 0.05.
Among the 1700 patients enrolled in the FAST-MAG trial, 1130 (66.5%) had magnesium serum levels drawn during the first 24 hours after study entry, including 569 patients in the magnesium group and 561 in the control group. Among the patients with an early magnesium level drawn, mean age was 70 years (SD±14 years), 45% were women, and mean Los Angeles Motor Scale (LAMS) score was 3.8 (SD±1.3; Table 1). Among these patients, during the first 72 hours after study entry, 570 (50.4%) had 1 magnesium level drawn, 263 (23.8%) had 2 levels, and 298 (26.4%) ≥3. Among the patients with serum magnesium levels drawn, time from last known well to study agent start was median 45 minutes (interquartile range, 35–60 minutes) and time from paramedic arrival on scene to study agent start was median 23 minutes (interquartile range, 18–27 minutes).
Patients having magnesium levels drawn in the first 24 hours were similar to those not having levels drawn in age, sex, body mass index, pre-stroke disability, history of vascular risk factors, and renal function, but were more often nonwhite, had 9-minute faster start of study infusion, had more severe presenting neurological deficits, and had more often intracranial hemorrhage as their presenting stroke type (Table 1). Hospital characteristics associated with having an early magnesium level drawn were larger hospital size, academic affiliation, and being a certified Primary Stroke Center.
Figure 2 shows the serum magnesium levels in the active Mg and the placebo groups during the first 72 hours after enrollment. During the 24-hour period of active infusion, the mean magnesium serum level in the active treatment arm was 3.91 (SD±0.8) and the mean level in the placebo group was 1.92 (SD±0.3). Magnesium levels rapidly increased with the field bolus and climbed minimally during next 24 hours. Magnesium levels declined steadily during 25 to 48 hours and were normal or close to normal in 49 to 72 hours. The paramedic loading dose was effective in achieving target serum levels rapidly. In the 30-minute interval immediately after field bolus completion, serum magnesium levels were assessed in 152 magnesium group and 132 placebo group patients. The mean magnesium serum level in the active treatment arm was 3.8 (SD 1.1), and the mean level in the placebo group was 1.9 (SD 0.3).
In unadjusted analysis, serum magnesium levels, analyzed as a continuous variable in both active and placebo arm patients combined, did not differ among patients with good and bad outcomes. Patients with and without independent functional outcome (modified Rankin Scale score of 0–1) at 3 months had similar early Mg levels, 3.0 (±1.6) versus 3.0 (±1.5; P=0.53); as did patients without and with symptomatic hemorrhage, 3.0 (±1.5) versus 2.7 (±1.3; P=0.31); and patients without and with mortality by 3 months, 3.0 (±1.6) versus 3.1 (±1.4; P=0.51).
When patients in the active Mg arm were grouped in quintiles by highest serum Mg level measured, several characteristic differed among the 5 groups (Table 2). Features associated in a graded manner with higher magnesium levels were as follows: older age; female sex; lower weight, height, and body mass index; lower glomerular filtration rate; and nonstroke mimic final diagnosis. Features differing among the quintiles in a nongraded manner were as follows: blood urea nitrogen, hemoglobin and hematocrit, and tobacco use.
In adjusted analysis, patient quintile of highest achieved blood serum magnesium level did not modify the likelihood of achieving an excellent day 90 functional outcome (modified Rankin Scale score of 0–1), either in comparison to control patients or among the different magnesium level groups (Figure 3). Similar results were noted for the additional 90-day efficacy outcomes of independence in activities of daily living (Barthel Index ≥90), minimal neurological deficit (NIHSS≤1), and final neurological deficit (NIHSS score; Table 3). Safety outcomes, symptomatic intracranial hemorrhage, and mortality also did not differ by quintile of achieved magnesium levels in treatment arm patients (Table 3).
This study confirms the technical efficacy of gravity-controlled tubing and joint packaging of field and hospital dose to enable paramedic initiation of intravenous agents in the field. This infusion strategy succeeded in rapidly doubling the serum magnesium level and maintaining this elevation throughout the planned first 24 hours. In addition, the results provide reassurance that the neutral effect of magnesium on stroke outcome in the FAST-MAG trial was not because of inadequate intravenous delivery of the study agent. Not only were target levels of Mg achieved by the infusion but also, in analyses of multiple outcomes, no relation of achieved Mg level to efficacy or safety outcomes was observed. Magnesium blood levels were accessed in patients with hospital blood serum levels drawn under standard care conditions. Results indicate that patients treated in larger hospitals with Academic status and Primary Stroke Center status were more likely to have magnesium serum blood levels drawn. Furthermore, patients with more profound neurological deficit on hospital arrival were more likely to have blood serum magnesium levels accessed during hospitalization.
The determinants of variations in achieved serum Mg level observed in this study conform to physiological expectations. The FAST-MAG trial used the same fixed loading and fixed maintenance doses of magnesium sulfate for all active arm patients. A fixed-dose regimen had been found to have acceptable range of variation in dose optimization studies9 and had been used in a previous international pivotal trial of in-hospital start of magnesium sulfate for stroke.10 The fixed dose approached assured regimen simplicity for EMS use, but did lead to some variation in achieved serum Mg levels according to patient demographic and physiological characteristics. Among patients receiving active magnesium infusion, patients with lower body mass index, weight, and height had higher magnesium levels, because of drug distribution in lower blood volume. As magnesium is renally cleared, patients with decreased renal function—lower estimated glomerular filtration rate and higher blood urea nitrogen—had higher blood serum levels.11
Several different analyses of the relation of achieved magnesium blood levels to efficacy and safety end points showed no evidence of outcome modification by magnesium. The lack of association of higher serum magnesium levels with an increase in symptomatic hemorrhagic transformation of ischemic stroke or worse outcomes in primary intracranial hemorrhage provides reassurance that weak antiplatelet effects of magnesium observed in some pre-clinical assays did not translate into untoward antihemostatic cerebral effects. However, the absence of an association of achieved serum magnesium levels with improved disability, activities of daily living, and survival through 3 months provides further evidence that magnesium did not confer beneficial neuroprotection, at low, intermediate, or high serum levels.
Intravenous infusions using infusion pumps are not a standard element of paramedic practice in United States. This constraint is a substantial obstacle to the development of pre-hospital therapies, as many promising pharmacological agents require nonbolus, intravenous administration. Adding infusion pumps universally to paramedic practice is an option, but has substantial drawbacks, including the cost of purchasing and maintaining the new equipment and difficulty in maintaining paramedic expertise in pump operation when the apparatus would be used only infrequently. Accordingly, development and validation of simpler approaches to pre-hospital intravenous infusion is desirable. One option is slow syringe infusion by hand, which was used in the previous FAST-MAG pilot trial.12 However, slow manual infusion requires prolonged paramedic attention, potentially distracting them from other duties and is vulnerable to overly fast infusion because of vehicle jostling or human error.13 Another option is use of roller-clamps to adjust drop rate. However, drop counts are known to be not fully reliable, rate-setting is subject to human error, and drop sizes vary.14,15 For the FAST-MAG pivotal trial, we selected the strategy of gravity-controlled infusion used a fixed-lumen size tubing to automatically control the infusion rate to target. This approach is simple for paramedics to use—just hang the bag, attach the tubing, and initiate the infusion. It does have the drawback that slower than target infusion rates may occur without alarms to providers if the line kinks or there is intravascular obstruction. Faster than target infusion rates cannot occur, assuring safety, although slower than target infusion rates may reduce efficacy. It is reassuring that slower than target infusion rates occur with any substantial frequency in the FAST-MAG trial, as demonstrated by the within target serum levels of magnesium on first post-arrival blood draws.
An additional challenge, in randomized trials of agents requiring an out-of-hospital loading infusion and immediately subsequent in-ED maintenance infusion, is assuring the rapid availability of matched active or placebo doses. If the ambulance carries only the loading dose, then systems would need to be put in place for the receiving pharmacy to be made aware before arrival of which maintenance dose (active or placebo) to prepare, while maintaining blinding of all communicating personnel, and would need to immediately mix and make available in the ED the maintenance dose. To address this issue in FAST-MAG, each ambulance carried a shrink-wrapped kit containing both the out-of-hospital and in-hospital infusions. Paramedics started the loading dose and handed the matched maintenance dose to receiving ED nurses for administration. The sustained in-target range of magnesium levels in FAST-MAG during the first hours after patient arrival indicates that this system succeeded in allowing for rapid transition from field to hospital matched doses.
This study has limitations. This was not a formal pharmacokinetic study with standardized draws of magnesium levels at prespecified intervals. However, the opportunistic analysis of clinically obtained Mg levels permitted probing of infusion success in a large number of patients. Sites of blood draws were based on pragmatic clinical practice; in some cases, sampling from above the infusion line may have occurred.16 However, the rarity of extremely high outlier Mg levels suggests that this was rare. Serum magnesium levels were measured in the multiple clinical laboratories of participating sites rather than a central core facility. This may have led to some variation in observed magnesium levels because of assay variation rather than because of blood level variation. However, commercially available magnesium assays generally show good congruence.17
In this large-scale ambulance trial, paramedic infusion initiation using gravity-controlled tubing enabled rapid achievement of target serum levels of a potential neuroprotective agent. Combined pre-hospital and hospital treatment regimen achieved targeted levels for the magnesium intervention. The absence of an association of efficacy and safety outcomes with achieved magnesium levels provides further evidence that magnesium is not biologically neuroprotective in acute stroke.
Sources of Funding
This study was supported by National Institute of Health - National Institute of Neurological Disorders and Stroke (NIH-NINDS) Award U01 NS 44364.
Presented in part at the International Stroke Conference, Nashville, TN, February 11, 2015, and at the American Neurological Association Annual Meeting, Chicago, IL, September 27, 2015.
Guest Editor for this article was Eric Smith, MD, MPH.
- Received October 7, 2016.
- Revision received March 21, 2017.
- Accepted April 20, 2017.
- © 2017 American Heart Association, Inc.
- Adeoye O,
- Hornung R,
- Khatri P,
- Kleindorfer D
- Campbell BC,
- Mitchell PJ,
- Yan B,
- Parsons MW,
- Christensen S,
- Churilov L,
- et al
- Saver JL,
- Starkman S,
- Eckstein M,
- Stratton S,
- Pratt F,
- Hamilton S,
- et al
- Saver JL,
- Starkman S,
- Eckstein M,
- Stratton S,
- Pratt F,
- Hamilton S,
- et al
- Muir KW,
- Lees KR
- Saver JL,
- Kidwell C,
- Eckstein M,
- Starkman S
- Crass RE,
- Vance JR
- Flack FC,
- Whyte TD
- Bailey D,
- Martens P,
- Mah W,
- Yip PM