Drug shortages have been defined as a lack of supply that causes either a need for an alternative source of the same medication or a therapeutic alternative to the original medication.1 National and local shortages of medications commonly used by anesthesiologists in the perioperative period have been occurring with increasing frequency.2,3 In 2010, the American Society of Health-System Pharmacists identified shortages in 39 medications directly affecting anesthetic care, many of which remained unresolved over the course of the year.1
In this study, we examined anesthesia provider practices and patient responses during a shortage of pharmacy-prepared syringes of the vasopressor ephedrine at a large tertiary care academic institution, Massachusetts General Hospital in Boston, Massachusetts. This shortage was a result of the closure of the pharmaceutical compounding company Ameridose (Westborough, Massachusetts) due to concerns of the Food and Drug Administration over the company’s sterility-testing process. The concerns of the Food and Drug Administration were triggered by the detection of contaminated injectable steroids at New England Compounding Pharmacy (Framingham, Massachusetts). This action abruptly eliminated the supply of pharmacy-prepared syringes of dilute ephedrine and phenylephrine to anesthesia providers at Massachusetts General Hospital. Although the hospital pharmacy began producing prepared syringes of phenylephrine for use in the operating room, ephedrine was not prepared by the pharmacy. Thus, individual anesthesia providers were required to dilute their own ephedrine syringes. This scenario induced a natural experiment to study anesthesia provider behavior and patient outcomes.
The purpose of this study was to determine whether anesthesia providers changed their practice of medication administration in response to a shortage of pharmacy-prepared syringes of ephedrine. We hypothesized that, because of the lack of pharmacy-prepared syringes, providers would be less likely to use ephedrine and would increase their use of phenylephrine.
Human Subjects Protections
This retrospective study was reviewed and approved by the Partners IRB.
Setting and Patient Selection
The first shortage of pharmacy-prepared syringes occurred on October 1, 2012. To study the effects of the shortage, we included consecutive patients from September 1, 2012, to September 30, 2012, as a reference group and patients from October 1, 2012, to October 31, 2012, as the group cared for during the medication shortage.
In both groups, consecutive patients undergoing elective surgery were identified by Current Procedural Terminology codes for colectomy (44140–44147, 44150, 44151, 44155, 44156, 44160, 44204–44212, 45110–45116, 45119–45121, 45123, 45395, 45397,45499, 45550, 48140), ventral and inguinal hernia repair (49505, 49507, 49565, 49659, 49560, 49650), parathyroidectomy(60500, 60502, 60505), thyroidectomy (60200, 60210, 60212, 60220, 60225, 60240, 60260, 60252, 60254, 60270, 60271), and total knee arthroplasty (27437–27447, 27486, 27487) or hip arthroplasty (27125, 27130, 27132, 27134, 27138). Only patients receiving general anesthesia, defined as having had an endotracheal tube or a laryngeal mask airway placed, were included in the analysis. Patients were excluded if they met any of the following criteria: received other vasopressors such as norepinephrine or vasopressin; underwent emergency surgery designated by an “E” classification of their ASA physical status or with afterhours surgery (starting after 7 PM or before 6 AM) or were ASA physical status class IV or greater. It was our intention to exclude patients who had arterial lines placed; however, none of the remaining patients fit this criterion.
Definitions of Outcome Measures
We examined the frequency of receiving at least 1 bolus of ephedrine and phenylephrine, as well as the total dosage administered of both these medications as our primary outcome measures.
Additional patient information was extracted from the anesthesia information management system (AIMS) including age, sex, and ASA physical status classification, as well as information related to the patient’s anesthetic care such as anesthetic technique, anesthesia provider, and medications administered. The AIMS included primary anesthesia provider type (anesthesia residents/fellows, certified registered nurse anesthetists or attending anesthesiologists) who managed only 1 patient at a time.
After completing the initial testing as described earlier, we undertook several additional exploratory analyses in an attempt to further characterize medication use patterns before versus during the shortage. These analyses included collecting heart rate, mean arterial blood pressure, and systolic blood pressure immediately before the administration of ephedrine and phenylephrine, which we refer to as threshold values. If there were multiple bolus administrations, then the average of the threshold variables was calculated and recorded as the threshold value. We also examined heart rate and arterial blood pressure throughout the entire anesthetic period for all cases in the study regardless of vasopressor administration. These values were determined using the median values during nonoverlapping 10-minute intervals and taking the resultant minimal value as described in a previous article.4 The AIMS samples hemodynamic variables in 1-minute intervals. The dose of propofol (mg) administered during the first half of the total anesthetic time was recorded. Data regarding the use of glycopyrrolate and atropine administration were also collected. Finally, the fraction of the age-adjusted minimal alveolar concentration was recorded, and the median value was reported.
Associations were assessed using the t test or χ2 test. We tested the primary hypothesis that providers would use less ephedrine and more phenylephrine during the shortage using multivariate regression methods adjusting for possible confounders. For the primary outcome, mixed-effects models were used for adjustment where each individual provider was included through an identification number as a repeated measure. Other covariates including provider type, age, sex duration of surgery, airway type, surgery type, and ASA physical status were included as fixed effects within the model. This model was specified before the initiation of analysis. After analysis was completed, we performed a post hoc analysis where provider type was modeled as a random effect to account for the hierarchal relationship between provider type and an individual anesthesia provider. The results were unchanged. Furthermore, we generated plots of residuals from the logistic regression model by provider type and did not find any obvious trends (Fig. 1). An unstructured covariance model was used for the regression model. There were <5 providers that took care of multiple patients on the same day. We plotted residuals for each provider across the study time period and did not detect correlation among successive patients.
Continuous variables, specifically age and duration of surgery, were categorized into quintiles and then included into the regression model to adjust for nonlinear associations. Given the high incidence of the outcome, odds ratios were converted to risk ratios using a previously described method for logistic regression models.5 Exploratory analyses also used multivariate mixed-effects regression methods to test adjusted associations. Changes in provider type, that is, resident becoming an attending, did not occur over the study time period. Pairwise interactions of each covariate and difference in time were tested and found to be nonsignificant (all P > 0.053).
We also performed a sensitivity analysis by restricting the cohort to providers who cared for patients both before and during the shortage and repeated the same analysis. To assess for heterogeneity among providers, we calculated the proportion of patients who received a bolus of ephedrine before and after the shortage for each individual provider and created a histogram of the results. All analyses were performed with Stata (version 12; StataCorp, College Station, Texas), with statistical significance defined as P < 0.05.
Six hundred two records were eligible for analysis; of which, 304 (50.5%) records were gathered from surgery before the medication shortage and 298 (49.5%) records were included during the shortage. Baseline characteristics between the 2 study groups were comparable with the exception of the distribution of provider type (Table 1).
Differences in Ephedrine and Phenylephrine Use
The proportion of patients receiving ≥1 boluses of ephedrine was significantly larger before the shortage (48.7% vs 39.3%, P = 0.0199). In multivariate analyses adjusting for age, sex, ASA physical status, surgery type, primary anesthesia provider, and surgical duration, patients were significantly less likely to receive at least 1 ephedrine bolus during the shortage (relative risk [RR] = 0.78 [95% CI, 0.61–0.96]; P = 0.0198). Patients were also more likely to receive ≥1 phenylephrine boluses (RR = 1.27 [95% CI, 1.02–1.51]; P = 0.0357). There was no change in the proportion of patients receiving any vasopressor or receiving both phenylephrine and ephedrine. There was an increase in the dose of phenylephrine administered (+49.85 μg [95% CI, 12.63–87.07]; P = 0.0087) accompanied by a decrease in the dose of ephedrine (−3.49 mg [95% CI, 5.68 to −1.30]; P = 0.0018; Tables 2 and 3).
Differences in the Use of Other Medications
We further investigated differences in administration patterns for other vasoactive medications. There was no difference in glycopyrrolate administration (P = 0.2749), propofol dosage (P = 0.4698), or median minimal alveolar concentration (P = 0.2303; Table 4). Only 2 patients received atropine in the study population, with one in each subgroup, and thus, it was not included in our analysis.
Patient Hemodynamic Patterns
Given the significant differences in vasopressor administration patterns, we investigated differences in the hemodynamic measures. The slowest heart rate was greater during the shortage in unadjusted analysis (56.5 vs 58.5, P = 0.0235), and this difference was statistically significant in regression results (P = 0.0303). There was no difference detected in minimal mean arterial blood pressure. There was also no statistically significant difference in threshold hemodynamic variables for either vasopressor. These results are outlined in Table 4.
When restricting the cohort to only providers who cared for patients both before and during the shortage, there was no change in the effect estimate size in any of the analyses. A histogram showing the change in the proportion of patients receiving ephedrine by individual providers is displayed in Figure 2. Heterogeneity in the response to the shortage was observed between providers. Fifty-eight percent of providers reduced the use, whereas 34% increased the use of ephedrine.
The events at our institution created a natural experiment to study changes in anesthesia provider behavior and patient hemodynamics in response to a shortage of pharmacy-prepared syringes. Our findings confirm that some providers altered their medication administration patterns because patients were significantly less likely to receive ephedrine and more likely to receive phenylephrine during the shortage. These findings confirm that even relative medication shortages can affect practice patterns and that relative medication shortages may be considered as potential sources of quality variance along with absolute shortages. The change in practice was observed in the majority of providers; however, there were certain providers who increased their use of ephedrine during the shortage.
The choice or substitution of vasopressor medication for bolus use during anesthesia is not clearly evidence based for most surgical scenarios. Although there have been randomized trials comparing these 2 vasoactive medications in parturients undergoing cesarean delivery,6 there are no studies examining the comparative effectiveness in the general surgical population. Our analyses suggested that providers substituted larger doses of phenylephrine for ephedrine during the shortage and achieved similar arterial blood pressure values with this selection. We expected that this substitution would have resulted in slower heart rates, but we found no evidence of this with the methods applied here, whether defined as use of rescue vagolytic or slowest heart rate assessed by sampling windows. This observed lack of difference may have been because of our chosen sampling method for vital signs,4 a lack of power or a true lack of difference.
If there were no clinically meaningful differences in patient hemodynamics despite differences in vasopressor administration patterns, one explanation was that anesthesia providers altered their anesthetic practice to maintain hemodynamics. We investigated the potential evidence of such a modification, including examining the depth of anesthesia, dosage of propofol, and the use of vagolytic drug such as atropine or glycopyrrolate to counteract bradycardia. These exploratory analyses showed no meaningful differences or suggestions of differences in any of these practice patterns. Again, this may have been because of inadequate measurement, limited power, or no real effects on patient hemodynamics when medication patterns for ephedrine and phenylephrine differed.
Several previous studies have attempted to examine the effect of prepackaged syringes of emergency drugs on patient safety and costs, but effects on drug use were not examined.7–10 Although no patient-level effects were demonstrated here, the findings of the present study demonstrate that even a relative shortage of a medication, otherwise remedied by a moment’s work, can result in changes in practice patterns.
Conclusions from our study may be limited by features of the study design. First, we focused on several common surgeries performed at a single institution, and conclusions from this study are most relevant for the study site and case mix. We limited our inclusion criteria to specific common procedures and to nonemergency cases to create relatively homogeneous groups within the study population. We also excluded ASA physical status IV patients because we felt that these patients have stricter hemodynamic goals and should be analyzed separately. The change in practice was observed in the majority of providers; however, there were certain providers who increased their use of ephedrine during the shortage. Given this heterogeneity, the effect seen in this study may not be generalizable to other settings.
Second, the study institution has limited capacity to perform compounding on-site, which also may not be available in other settings. Third, we did not examine postoperative complications for study patients, although the differences in heart rate and arterial blood pressure provided no suspicion of attributable harm.4,11 Fourth, we do not have data on the prevalence of hypertension as a diagnosis or data for oral antihypertensives taken by patients in each study period. However, we suspect no systematic differences based on the closely matched Current Procedural Terminology codes and ages between groups. Our study was retrospective and thus susceptible to unmeasured confounding. However, the shortage was an exogenous event, and there is no reason to suspect a difference in patient characteristics before and during the shortage. There was a difference in provider type between groups. The exact reason for this discrepancy is unclear. Provider type was adjusted for in the multivariate analysis. We also included individual providers in the adjusted analyses using mixed-effects models.
Unique features of anesthesia practice at the study institution may differ from patterns nationally. The rate of ephedrine use was 48.5% and more than the rate of 30% published by Lejus et al.12 in a retrospective study. Possible explanations for the difference include varying case mix between hospitals or simply differences in clinical practice. We also did not have data regarding whether providers prepared ephedrine in anticipation of need before a case and if this changed during the shortage. Previous research has shown that there is significant variation in this practice between providers, and thus, there is a potential impact on drug waste which was not examined.13
Irrespective of the lack of clinically significant impact, this study provides valuable insight into anesthesia provider behavior. Our findings suggest that providers can be induced or incentivized to change their practice by altering the barrier to administration of a particular medication. This relative shortage also highlights a potential vulnerability in the drug supply chain. Further efforts are needed to ensure that compounding pharmacies adhere to standards that will ensure a stable and safe drug supply so that providers can focus on providing anesthetic care to our patients.
Name: Karim S. Ladha, MD.
Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.
Attestation: Karim S. Ladha has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is the author responsible for archiving the study files.
Name: Karen C. Nanji, MD, MPH.
Contribution: This author helped design the study and write the manuscript.
Attestation: Karen C. Nanji approved the final manuscript.
Name: Eric Pierce, MD, PhD.
Contribution: This author helped design the study and write the manuscript.
Attestation: Eric Pierce approved the final manuscript.
Name: K. Trudy Poon, MS.
Contribution: This author helped analyze the data and write the manuscript.
Attestation: K. Trudy Poon has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
Name: Joseph A. Hyder, MD, PhD.
Contribution: This author helped design the study, conduct the study, and write the manuscript.
Attestation: Joseph A. Hyder has seen the original study data, reviewed the analysis of the data, and approved the final manuscript.
This manuscript was handled by: Sorin J. Brull, MD, FCARCSI (Hon).
1. De Oliveira GS, Theilken LS, McCarthy RJ. Shortage of perioperative drugs: implications for anesthesia practice and patient safety. Anesth Analg. 2011;113:1429–35
2. Jensen V, Rappaport BA. The reality of drug shortages—the case of the injectable agent propofol. N Engl J Med. 2010;363:806–7
3. Ventola CL. The drug shortage crisis in the United States: causes, impact, and management strategies. P T. 2011;36:740–57
4. Hyder JA, Kor DJ, Cima RR, Subramanian A. How to improve the performance of intraoperative risk models: an example with vital signs using the surgical Apgar score. Anesth Analg. 2013;117:1338–46
5. Zhang J, Yu KF. What’s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280:1690–1
6. Lee A, Ngan Kee WD, Gin T. A quantitative, systematic review of randomized controlled trials of ephedrine versus phenylephrine for the management of hypotension during spinal anesthesia for cesarean delivery. Anesth Analg. 2002;94:920–6
7. Jensen LS, Merry AF, Webster CS, Weller J, Larsson L. Evidence-based strategies for preventing drug administration errors during anaesthesia. Anaesthesia. 2004;59:493–504
8. Vipond A, de Mello W. Drugs used in anaesthetic emergencies: current practice and a cost analysis of prefilled syringes. Anaesthesia. 2000;55:303–4
9. Gurung AM, Tomlinson AA. Pre-preparation of succinylcholine: significant waste for questionable benefit. Anaesthesia. 2004;59:211–2
10. Wikner MA, Tighe SQ. Pre-filled ephedrine syringes. Anaesthesia. 2001;56:712
11. Gawande AA, Kwaan MR, Regenbogen SE, Lipsitz SA, Zinner MJ. An Apgar score for surgery. J Am Coll Surg. 2007;204:201–8
12. Lejus C, Blanloeil Y, Oudot M, Le Teurnier Y, Lepage JY, Loutrel O, Asehnoune K. Atropine and ephedrine: a significant waste in the operating theatre. Anaesthesia. 2012;67:300–1
© 2015 International Anesthesia Research Society
13. Ducat CM, Merry AF, Webster CS. Attitudes and practices of New Zealand anaesthetists with regard to emergency drugs. Anaesth Intensive Care. 2000;28:692–7