A comparison of approaches to identify possible cases of local anesthetic systemic toxicity in the FDA Adverse Event Reporting System (FAERS) database

ABSTRACT

Background: Local anesthetic systemic toxicity (LAST) is a rare but potentially serious adverse event .

Methods: Data from the US Food and Drug Administration Adverse Event Reporting System were examined for liposomal bupivacaine (LB), bupivacaine, or other injectable local anesthetics. Possible LAST cases were identified based on MedDRA system organ classes (Approach 1), a recent publication (Approach 2), and a novel approach based on LAST literature (Approach 3). Disproportionality analyses compared possible LAST cases for LB and bupivacaine with other injectable local anesthetics.

Results: Approaches 1, 2, and 3 identified 75, 42, and 29 possible LAST cases associated with LB, respectively, compared with 9,595, 3,422, and 549 for other injectable local anesthetics. The proportional reporting ratios (95% CI) for LB versus other injectable local anesthetics for the 3 approaches were 1.9 (1.6–2.3), 2.9 (2.2–3.9), and 1.6 (1.1–2.2), respectively. Based on sales data, the estimated incidence of possible LAST with LB was 0.1 per 10,000 uses; literature estimates for LAST with other injectable local anesthetics were 0 to 18 per 10,000 uses.

Conclusions: Our findings suggest the estimated incidence of possible LAST cases with LB is similar to, or less than, the reported incidence with other injectable local anesthetics.

KEYWORDS:

• Bupivacaine
• disproportionality analysis
• FAERS
• local anesthetic systemic toxicity
• liposomal bupivacaine
• pharmacovigilance

1. Introduction

To manage postsurgical pain and mitigate harms from opioids (e.g. nausea, vomiting, constipation, itching, dizziness, drowsiness, fatal respiratory depression, misuse, abuse, dependence, or addiction) [1–6], recent evidence-based clinical practice guidelines from the American Society of Regional Anesthesia and Pain Medicine recommend multimodal analgesia [7]. Multimodal analgesia is broadly defined as using two or more analgesic medications with different mechanisms of action (e.g. nonsteroidal anti-inflammatory drugs, acetaminophen, gabapentinoids, and other adjunctive analgesics) and often includes controlling pain at the surgical site with local anesthetics administered by wound infiltration, peripheral nerve block, or neuraxial anesthesia. Most local anesthetics offer rapid onset of analgesia, but their effect subsides within a few hours, limiting their impact on acute postsurgical pain, which can last several days or longer.

Although generally considered safe, all local anesthetics carry the risk of local anesthetic systemic toxicity (LAST), which can occur when local anesthetics are administered by injection (e.g. wound infiltration, peripheral nerve block, or neuraxial anesthesia) and inadvertently enter a blood vessel, or are absorbed very quickly at a highly vascular injection site, causing a rapid rise in systemic levels that results in toxicity [8–10]. The clinical presentation of LAST is varied and encompasses numerous signs and symptoms related to the cardiovascular system and central nervous system, including tinnitus, dizziness, hypotension, seizures, and arrhythmia, which can lead to cardiac arrest and death [9–11]. The number, type, timing, and severity of signs and symptoms with LAST are variable, and in published reports, ‘atypical’ presentations of LAST account for an estimated 40% of such cases [9,10]. This variability in clinical presentation can make it difficult to accurately identify cases of LAST, particularly when local anesthetics are used in the perioperative setting, in which patients are exposed to numerous medications with potentially overlapping adverse events. The difficulty is amplified when using publicly available data sources for pharmacovigilance such as the US Food and Drug Administration’s (FDA) Adverse Event Reporting System (FAERS).

Estimates of the incidence of LAST reported in the scientific literature have varied widely (e.g. 0.7 to 11 cases per 10,000 uses for epidural anesthesia [11,12], 0 to 18 cases per 10,000 uses for peripheral nerve blocks [13]), depending on the study methods, definition of LAST used, technique of administration, and the different local anesthetics examined. The incidence of LAST appears to have decreased in the past few decades [14], which may be related to the use of ultrasound-guided peripheral nerve blocks and reduced inadvertent intravascular placement of local anesthetic injections [10,11,15]. Despite its low and apparently declining incidence, LAST remains a concern for both clinicians and manufacturers of local anesthetics [10].

Liposomal bupivacaine (LB; EXPAREL® [bupivacaine liposome injectable suspension], Pacira Pharmaceuticals, Inc., Parsippany, NJ) is a prolonged-release formulation of bupivacaine that can provide postsurgical analgesia and reduce opioid consumption for up to 72 h after a single administration into the surgical site [16,17]. Both preclinical and human data suggest that compared with traditional, immediate-release bupivacaine hydrochloride (hereafter referred to as ‘bupivacaine’), LB may have a reduced risk of acute systemic toxicity because of its liposomal formulation and resulting pharmacokinetic profile (i.e. lower peak plasma concentration with a sustained release over a longer duration). However, since LB is a relatively new drug (marketed in the United States since 2012), and data on LAST associated with LB are limited, continued pharmacovigilance is imperative; a recent publication by Aggarwal has underscored the importance of better understanding this association [18]. The objectives of this study, therefore, were to compare different methods that examine the possible association between LAST and LB using the FAERS database and explore how this association may be affected by the specific definition of LAST used in the analysis.

2. Methods

2.1. Data source

Data were downloaded electronically from the FAERS database website (https://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Surveillance/AdverseDrugEffects/ucm082193.htm) for the period from 1 January 2012 through 31 March 2017. The FAERS database is maintained by the FDA and includes voluntary reports – related to adverse events, medication errors, and product quality complaints for approved (i.e. not investigational), over-the-counter, and prescription drugs – made by health-care providers, patients, and drug manufacturers [19]. These voluntary reports to the FDA can be made by health-care providers or patients using a MedWatch form, or by contacting the drug manufacturer, which then completes a MedWatch form and sends it to the FDA for inclusion in FAERS.

The MedWatch form is a two-page document that contains basic patient information (e.g. age, sex, race), a description of the adverse event (e.g. date, seriousness, outcome), suspected product(s) (e.g. name, dose, frequency, route), and concomitant product(s) (e.g. medications). If multiple products are reported on a MedWatch form, the reporter can identify a primary suspect drug which they feel is most responsible for the adverse event(s). Drug manufacturers can also submit information to FAERS based on adverse events reported in published literature and serious adverse events reported in clinical trials.

The adverse events reported on MedWatch forms are standardized in FAERS using the Medical Dictionary for Regulatory Activities (MedDRA), which classifies adverse events broadly according to the system organ class (SOC) involved (e.g. cardiac disorder), as well as more specifically according to the preferred term (PT) for an adverse event (e.g. the PT for ‘low blood pressure’ is ‘hypotension’) [20]. The use of SOCs and PTs helps to standardize heterogeneous data from voluntary reports for use in postmarketing surveillance analyses.

2.2. Identification of local anesthetics in FAERS

The FAERS database drug file was searched to identify reports involving LB, bupivacaine, and other local anesthetics commonly administered by injection, using both complete drug names and partial drug names with wildcards (e.g. exp*, bup*, *cain, *cane, and lipo*) to allow for misspellings of relevant brand names and generic names (note: search query available upon request). Medications identified by these queries were reviewed by both a clinical pharmacist and an anesthesiologist to confirm that they were injectable local anesthetics , although the specific route of administration used for cases in FAERS was not always reported. This search identified injectable local anesthetics other than LB and bupivacaine, including articaine, chloroprocaine, dibucaine, etidocaine, levobupivacaine, lidocaine, mepivacaine, prilocaine, procaine, and ropivacaine; other drugs included in a previously published analysis of LB and LAST by Aggarwal (e.g. cocaine) were also included to compare our results with the most current literature on this topic [18].

2.3. Identification of possible cases of LAST

After identifying FAERS reports involving LB, bupivacaine, or other injectable local anesthetics, these reports were further evaluated to determine if they were possibly related to LAST using three different approaches to define LAST, as described below.

2.3.1. Approach 1: broad MedDRA definition

Approach 1 used a broad definition of LAST intended to mirror a previous analysis conducted by the manufacturer of LB for the FDA (private correspondence, 2016). With this approach, any of the hundreds of signs, symptoms, or conditions related to the SOCs for cardiac disorders, nervous system disorders, or respiratory, thoracic, and mediastinal disorders in MedDRA were considered as possibly related to LAST.

2.3.2. Approach 2: Aggarwal definition

Approach 2 attempted to replicate the methods used in a recent analysis of LAST in FAERS [18] with a narrower subset of signs or symptoms related to the cardiovascular system and central nervous system. Since the specific methods used in that analysis were not clear, two variations were developed based on our understanding of the published approach. One variation identified possible LAST based on signs or symptoms related to the cardiovascular system (i.e. bradycardia, hypotension) or central nervous system excitation (i.e. agitation, dizziness, tinnitus, seizure, convulsion), as well as respiratory depression or arrest if subsequent to central nervous system excitation and cardiovascular system depression/collapse if subsequent to other signs or symptoms related to the cardiovascular system. The second variation identified possible LAST based on reporting of any of the aforementioned signs or symptoms related to the cardiovascular system and central nervous system, regardless of order.

2.3.3. Approach 3: proposed new definition based on signs or symptoms related to the cardiovascular system and central nervous system

Approach 3 was developed by identifying signs and symptoms related to LAST reported in the literature, including two reviews of published cases of LAST [21,22], recommendations to help anesthesiologists recognize typical and atypical cases of LAST [10,11], and two reports that commented on LB and LAST (including the FAERS analysis cited in the preceding section) [18,23]. All signs or symptoms related to LAST were extracted, removing duplicates (e.g. arrhythmia and arrhythmias) and those deemed too vague to be useful (e.g. pain, neurologic symptom). The remaining signs or symptoms were then assigned to one of the following five categories: (1) central nervous system excitement, (2) central nervous system depression, (3) cardiovascular system excitement, (4) cardiovascular system depression, and (5) seizure. Most signs and symptoms were assigned to only one of these categories, but some were assigned to two categories, as deemed appropriate (e.g. ‘anesthetic complication cardiac’ was assigned to both cardiovascular system excitement and cardiovascular system depression) (Supplementary Online Resource 1). Cases were considered to have a clinical presentation possibly related to LAST if they involved signs or symptoms related to both the cardiovascular system (excitement or depression) and the central nervous system (excitement or depression) categories or if they involved seizure. For example, a FAERS report with ‘dizziness’ (i.e. central nervous system excitement) and ‘cardiac arrest’ (i.e. cardiovascular system depression) would be deemed possibly related to LAST, but one with ‘dizziness’ alone would not.

A further examination of cases possibly related to LAST was undertaken to determine if they had a higher or lower likelihood of being solely attributable to LB. Cases where multiple local anesthetics were reported (e.g. if a case reported both LB and bupivacaine) and where other medications were reported and those medications were known to have adverse events related to the five categories of cardiovascular system and central nervous system signs or symptoms (i.e. as determined by a clinical pharmacist based on the prescribing information and mechanism of action) were considered to have a lower likelihood of being solely attributable to LB. For example, a FAERS report with ‘seizure’ would be identified as possibly related to LAST but determined to have a lower likelihood of being solely attributable to LB if bupivacaine and ropivacaine (which are independently associated with LAST) were also reported. Similarly, a FAERS report with ‘agitation’ (categorized as central nervous system excitement) and ‘bradycardia’ (categorized as cardiovascular system depression) would be identified as possibly related to LAST (since the case included signs or symptoms for both the cardiovascular system and the central nervous system) but determined to have a lower likelihood of being solely attributable to LB if the drug phentermine was also reported, since phentermine is a stimulant known to cause potential central nervous system excitement. This review was undertaken to provide a clinical perspective on the attribution of LAST based on the limited information available in FAERS.

2.4. Disproportionality analyses

Disproportionality analysis is often used to determine if the number of cases for an adverse event of interest (e.g. LAST) reported in FAERS for a specific medication (e.g. LB) differs markedly from the number reported with other medications (e.g. other injectable local anesthetics) [24]. This can be evaluated using standard pharmacovigilance metrics such as proportional reporting ratio, relative odds ratio, and relative reporting ratio [24,25]. These measures are derived from standard 2 × 2 contingency tables that summarize the number of possible cases for the adverse event of interest (i.e. LAST) and all other adverse events reported for the drug of interest (i.e. LB or bupivacaine) as the primary suspect drug versus all other injectable local anesthetics.

An example is provided below to illustrate how these three ratios are derived from a hypothetical 2 × 2 table for drug A and drug B, where a total of 100 adverse events were identified in FAERS for drug A and 1000 adverse events were identified for drug B, with 20 adverse events of interest for drug A and 50 adverse events of interest for drug B.

Table

	Drug A	Drug B	Total
Adverse event of interest	20	50	70
Other adverse events	80	950	1,030
Total adverse events	100	1,000	1,100

The proportional reporting ratio would be calculated as follows: drug A (20/100 = 0.20) divided by drug B (50/1,000 = 0.05) equals 4.0. The relative odds ratio would be calculated as follows: drug A (20/80 = 0.25) divided by drug B (50/950 = 0.05) equals 4.75. The relative reporting ratio would be calculated as follows: (20*1100 = 22,000) divided by (100*70 = 7000) equals 3.14.

Since disproportionality analysis builds on the case definition used for the adverse event of interest (i.e. signs and symptoms related to LAST), three sets of disproportionality analyses were conducted to examine the number of cases possibly related to LAST reported in FAERS using the three approaches to identify LAST described in the preceding section. For each of the three approaches, standard 2 × 2 contingency tables with Yates continuity correction were constructed. These included the number of possible cases of LAST and reports of all other adverse events not related to LAST for LB as the primary suspect drug compared to all other injectable local anesthetics, as well as for bupivacaine as the primary suspect drug compared to all other injectable local anesthetics, respectively.

For each approach and each drug (i.e. LB and bupivacaine), standard statistical measures were calculated for each disproportionality analysis, including proportional reporting ratio, relative odds ratio, and relative reporting ratio, each with their respective 95% confidence intervals. All analyses were conducted using SAS version 9.4 (SAS Enterprises, Cary, NC). The unit of measurement for possible cases of LAST was the number of unique patients meeting the three definitions, whereas adverse events not related to LAST were counted as individual reports. For example, a patient in FAERS with ‘coughing,’ ‘back pain,’ and ‘fever’ (and no other PTs possibly related to LAST) would be counted as three adverse events not related to LAST in the disproportionality analyses.

3. Results

3.1. Reports of LB and other local anesthetics in FAERS

For the period analyzed, there were 123 reports in FAERS where LB was listed as the primary suspect drug (including one where LB was listed as both the primary and secondary suspect drug, as allowed in FAERS) and two reports where LB was listed as an interacting or concomitant drug, yielding a total of 125 reports for any adverse event (i.e. not limited to LAST). Over the same period, there were 17,529 reports of adverse events involving bupivacaine (as primary suspect drug) or other injectable local anesthetics (regardless of status as primary suspect drug).

3.1.1. Approach 1: broad MedDRA definition

Using Approach 1, based on adverse events for SOCs related to cardiac disorders, nervous system disorders, or respiratory, thoracic, and mediastinal disorders, 75 cases possibly related to LAST (and 241 other adverse events not related to LAST) were identified for LB, compared with 9595 cases possibly related to LAST (and 66,851 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 1.9 (95% CI, 1.6–2.3) and a chi-squared value of 34.7 (Table 1). As mentioned above, a proportional reporting ratio of 1.9 indicates that the proportion of adverse events identified in FAERS that were related to LAST was approximately twice as high for LB (75/316 = 0.24) as for other injectable local anesthetics (9595/76,446 = 0.13). It should be noted that this ratio may fluctuate over time based on the number of adverse events reported in FAERS for LB and other injectable local anesthetics, and only represents a strong signal if it remains elevated.

Table 1. Disproportionality analysis results for LB and bupivacaine using Approach 1: broad MedDRA definition.

Event	Local anesthetic		Total	PRR (95% CI)	ROR (95% CI)	RRR (95% CI)	Chi-squared
	LB	All other local anesthetics
LAST	75	9595	9670	1.9 (1.6–2.3)	2.2 (1.7–2.8)	1.9 (1.5–2.3)	34.7
All other AEs	241	66,851	67,092
Total	316	76,446	76,762
	Bupivacaine	All other local anesthetics
LAST	228	9442	9670	1.5 (1.3–1.7)	1.6 (1.4–1.9)	1.5 (1.3–1.7)	41.0
All other AEs	993	66,099	67,092
Total	1221	75,541	76,762

AE: adverse event; CI: confidence interval; LAST: local anesthetic systemic toxicity; LB: liposomal bupivacaine; MedDRA: Medical Dictionary for Regulatory Activities; PRR: proportional reporting ratio; ROR: relative odds ratio; RRR: relative reporting ratio.

By comparison, 228 cases possibly related to LAST (and 993 other adverse events not related to LAST) were identified for bupivacaine, compared with 9442 cases possibly related to LAST (and 66,099 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 1.5 (95% CI, 1.3–1.7), with a chi-squared value of 41.0 (Table 1). For both LB and bupivacaine, the relative odds ratio and relative reporting ratio estimates were similar to the proportional reporting ratio.

3.1.2. Approach 2: Aggarwal definition

Using the first variation of the Aggarwal-based approach applying specific combinations of signs or symptoms related to the cardiovascular system and central nervous system reported in a previous analysis of the FAERS database, no cases possibly related to LAST were identified for LB if PTs for both the cardiovascular system and the central nervous system were required. If the second variation was used to identify LAST based on any of the cardiovascular system and central nervous system signs or symptoms specified, 42 cases possibly related to LAST (and 344 other adverse events not related to LAST) were identified for LB, compared with 3422 cases possibly related to LAST (and 87,759 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 2.9 (95% CI, 2.2–3.9) and a chi-squared value of 51.7 (Table 2).

Table 2. Disproportionality analysis results for LB and bupivacaine using Approach 2: Aggarwal Definition.

Event	Local anesthetic		Total	PRR (95% CI)	ROR (95% CI)	RRR (95% CI)	Chi-squared
	LB	All other local anesthetics
LAST	42	3422	3464	2.9 (2.2–3.9)	3.1 (2.3–4.3)	2.9 (2.2–3.8)	51.7
All other AEs	344	87,759	88,103
	Total	386	91,1181
	Bupivacaine	All other local anesthetics
LAST	104	3360	3464	2.0 (1.6–2.4)	2.1 (1.7–2.5)	2.0 (1.6–2.4)	49.7
All other AEs	1310	86,973	88,283
	Total	1414	90,333

AE: adverse event; CI: confidence interval; LAST: local anesthetic systemic toxicity; LB: liposomal bupivacaine; PRR: proportional reporting ratio; ROR: relative odds ratio; RRR: relative reporting ratio.

Among the 42 cases possibly related to LAST for LB, 27 (64%) had only one relevant sign or symptom related to the cardiovascular system or central nervous system (most commonly hypotension [n = 5], cardiac arrest [n = 4], respiratory arrest [n = 3], and seizure [n = 3]), and 25 (60%) involved >1 drug (14 of which involved another injectable local anesthetic).

By comparison, 104 cases possibly related to LAST (and 1310 other adverse events not related to LAST) were identified for bupivacaine, compared with 3360 cases possibly related to LAST (and 86,973 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 2.0 (95% CI, 1.6–2.4), with a chi-squared value of 49.7 (Table 2). For both LB and bupivacaine, the relative odds ratio and relative reporting ratio estimates were similar to the proportional reporting ratio.

3.1.3. Approach 3: proposed new definition based on signs or symptoms related to the cardiovascular system and central nervous system

Of the 123 cases in FAERS with LB listed as the primary suspect drug, 85 (69%) reported at least one sign or symptom related to the cardiovascular system or central nervous system that could be associated with LAST. Upon further examination, 56 (66%) of these 85 cases were identified as not related to LAST, in many instances because only one sign or symptom related to the cardiovascular system or central nervous system (excluding seizure) was present. The remaining 29 (34%) of these 85 cases (i.e. 24% of the 123 total cases reporting LB in FAERS) were identified as possibly related to LAST based on the proposed new definition requiring signs or symptoms related to both the cardiovascular system and the central nervous system, or seizure.

Therefore, 29 cases possibly related to LAST (and 409 other adverse events not related to LAST) were identified for LB, compared with 549 cases possibly related to LAST (and 12,305 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 1.6 (95% CI, 1.1–2.2) and a chi-squared value of 5.1 (Table 3). It should be noted that among these 29 cases, 8 (28%) were determined to have a lower likelihood of being solely attributable to LB because other local anesthetics were also reported or because other medications with adverse events in the five categories of signs or symptoms related to the cardiovascular system and central nervous system were also reported, while 21 cases (72%) were deemed to have a higher likelihood of being solely attributable to LB.

Table 3. Disproportionality analysis results for LB and bupivacaine using Approach 3: proposed new definition based on signs or symptoms related to the cardiovascular system and central nervous system.

Event	Local anesthetic		Total	PRR (95% CI)	ROR (95% CI)	RRR (95% CI)	Chi-squared
	LB	All other local anesthetics
LAST	29	549	578	1.6 (1.1–2.2)	1.6 (1.1–2.3)	1.5 (1.1–2.2)	5.1
All other AEs	409	12,305	12,714
	Total	438	12,854
	Bupivacaine	All other local anesthetics
LAST	74	504	578	1.0 (0.7–1.2)	0.9 (0.7–1.2)	1.0 (0.8–1.2)	0.16
All other AEs	1713	11,001	12,714
	Total	1787	11,505

AE: adverse event; CI: confidence interval; LAST: local anesthetic systemic toxicity; LB: liposomal bupivacaine; PRR: proportional reporting ratio; ROR: relative odds ratio; RRR: relative reporting ratio.

By comparison, 74 cases possibly related to LAST (and 1713 other adverse events not related to LAST) were identified for bupivacaine, compared with 504 cases possibly related to LAST (and 11,001 other adverse events not related to LAST) for all other injectable local anesthetics; disproportionality analysis yielded a proportional reporting ratio of 1.0 (95% CI, 0.7–1.2), with a chi-squared value of 0.16 (Table 3). For both LB and bupivacaine, the relative odds ratio and relative reporting ratio estimates were similar to the proportional reporting ratio.

4. Discussion

Managing acute postsurgical pain while minimizing the use of opioids and risk of opioid-related adverse events requires the use of local anesthetics targeting the source of pain through wound infiltration, peripheral nerve block, or neuraxial analgesia. All local anesthetics are associated with LAST, which can be a complex and challenging condition to identify under certain circumstances because its clinical presentation varies widely, with approximately 40% of the cases reported in the literature having an ‘atypical’ presentation [10]. The difficulty in identifying LAST is compounded when attempting to do so using the relatively limited information available in the FAERS database rather than undergoing a full medical chart review, the current ‘gold standard’ for patient safety research [26].

As mentioned above, a recent publication based on an analysis of the FAERS database reported 130 cases of LAST associated with LB as the primary suspect drug and 346 cases of LAST associated with bupivacaine as the primary suspect drug [18]. That publication reported for LAST a proportional reporting ratio of 6.23 (95% CI 5.41–7.18) for LB compared with ‘all other drugs’ and 3.36 (95% CI 3.05–3.69) for bupivacaine compared with ‘all other drugs,’ suggesting an increased risk of LAST with LB. However, we were unable to replicate these results; moreover, our analyses not only identified fewer possible cases of LAST associated with LB than were reported in the earlier publication (42 vs. 130) but fewer than 130 total cases in FAERS involving LB (n = 125), even when examining 9 additional months of FAERS data. Compared with the previous analysis, we also found far fewer possible cases of LAST associated with bupivacaine in FAERS (104 vs. 346). These findings suggest the possibility of systematic errors within the previous analysis, whose author has recently submitted a journal erratum with a revised conclusion (personal communication, 2018).

It is also worth noting that the Office of Surveillance and Epidemiology (OSE) at the FDA recently completed their own analysis of FAERS to identify possible cases of LAST with LB and other local anesthetics [27]. Although OSE methods were not fully detailed, they were based on the same three SOCs as our Approach 1 and searched from 28 October 2011 to 8 November 2017 for LB and from 1 July 2006 to 8 November 2017 for other local anesthetics. The OSE analysis identified 39 possible cases of LAST with LB, which is consistent with our findings based on both Approach 2 and Approach 3, and further underscores the discrepancies noted with the previously published analysis on this topic [18].

To shed further light on this topic, our analysis also considered two other approaches to defining LAST in the FAERS database. One very broad (i.e. sensitive) approach using adverse events for three different SOCs (consistent with a previous analysis submitted by the manufacturer of LB to the FDA) was tested, as was a new approach intended to be more specific, to limit false-positive cases that can artificially inflate the estimated incidence of LAST with LB and thus confound efforts to mitigate it. This new definition was developed after reviewing literature summarizing the clinical presentation of LAST, including specific signs or symptoms related to the cardiovascular system and central nervous system that were commonly present in case reports of LAST. Consistent with the literature on LAST [10,11], rather than accepting that the presence of any of these individual signs or symptoms could, in isolation, suggest LAST, they were used as components in a definition of LAST that begins with central nervous system excitement, followed by central nervous system depression, cardiovascular system excitement, and cardiovascular system depression, as systemic toxicity increases. Recognizing that not every case of LAST would follow this specific progression, the new definition was modified to allow for central nervous system excitement or central nervous system depression, combined with cardiovascular system excitement or cardiovascular system depression, to identify possible cases of LAST; the presence of seizure was also added to this definition since it has previously been used in isolation as a proxy for LAST [10].

Beyond proposing a new definition of LAST for use in the FAERS database, our analysis also considered the likelihood – based on the limited information available in FAERS – that possible cases were solely attributable to LB (or any other local anesthetic). This additional step was deemed important since many of the signs or symptoms related to LAST are not specific to any one local anesthetic, and the presence of multiple local anesthetics in a case decreases the likelihood that LAST can be attributed to a single medication. This rationale was also extended to concomitant medications other than local anesthetics since signs or symptoms related to central nervous system excitement (e.g. agitation), central nervous system depression (e.g. lethargy), cardiovascular system excitement (e.g. hypertension), or cardiovascular system depression (e.g. hypotension) are common adverse events for a variety of medications.

Although defined a priori, the three different approaches to defining LAST followed an expected pattern, with the broadest approach (i.e. Approach 1 based on adverse events for three SOCs) identifying the highest number of possible cases of LAST reporting LB (n = 75), and the more specific approach (i.e. Approach 3 that requires involvement of both the cardiovascular system and the central nervous system, or seizures) identifying the lowest number of cases of LAST reporting LB (n = 29). In pharmacovigilance, a strong disproportionality signal (e.g. proportional reporting ratio >2.0) can suggest a possible association between a specific drug and a specific adverse event of interest that may warrant further investigation [24,25]. The point estimates for proportional reporting ratios for LB compared with other injectable local anesthetics for those two approaches were near or below the commonly accepted minimum threshold of 2.0, though some of the upper bounds of the confidence intervals exceeded this level. Although these proportional reporting ratios were higher than those estimated using the same approaches with bupivacaine compared with other injectable local anesthetics (1.5 and 1.0, respectively), reporting bias (e.g. tendency to report adverse events more frequently with newer medications, also known as the Weber effect [28]) may be a confounder since bupivacaine has been available in the United States since the 1970s and LB was only launched in April 2012 [16,29]. The possibility of reporting bias in the FAERS database for older (i.e. bupivacaine) and newer (i.e. LB) injectable local anesthetics may preclude a direct comparison of these proportional reporting ratios.

One of the main limitations of the FAERS database is that all cases identified as possibly related to LAST represent only the numerator for the incidence rate. To provide additional perspective, the number of vials of LB sold during the period of interest can be used to approximate the denominator, since the manufacturer records all vials sold, LB has a limited shelf life and is unlikely to be purchased and stored for long periods of time, and the recommended dose of LB is one vial per patient. From its commercial launch in April 2012 until the end of March 2017, the manufacturer sold 2,924,424 vials of LB (data on file). Based on the 29 cases identified in FAERS using Approach 3 with the proposed new definition of LAST, the estimated incidence of LAST with LB is 1 case per 100,842 uses or 0.1 case per 10,000 uses, if it is assumed that all cases of LAST associated with LB are consistently reported to FAERs. By contrast, the reported incidence of LAST with injectable local anesthetics other than LB ranges from 0.7 to 11 cases per 10,000 uses for epidural anesthesia [11,12] and 0 to 18 cases per 10,000 uses for peripheral nerve block [13].

The occurrence of LAST is thought to be related to systemic exposure and absorption of local anesthetics, as broadly measured by C_max [30,31]. Whereas plasma concentrations of bupivacaine peak within 1 to 2 h of administering immediate-release bupivacaine by wound infiltration, LB releases bupivacaine more slowly, with an initial (lower) peak in plasma concentration within 1 h of administration and a second (lower) peak after 12 to 36 h [32,33]. Although C_max may vary based on the methods of drug administration and measurement and may not accurately reflect bupivacaine concentrations in local tissues throughout the body, it is commonly reported as a proxy for systemic concentration in studies on LAST. The administration of LB has been found to produce a lower C_max for bupivacaine, and data from healthy volunteers suggest that C_max with LB remains well below previously proposed C_max thresholds for cardiovascular and central nervous system toxicity (e.g. 2000–4000 ng/mL), even after a repeated dose of LB within 72 h (note: the current indication allows for only one dose of LB per patient) [32,34,35]. Despite the limitations of C_max as a measure of systemic toxicity, these data suggest that the risk of LAST associated with peak C_max concentrations for bupivacaine may be reduced with LB.

In preclinical studies examining intra-arterial or intravenous administration of LB in dogs to understand the impact of inadvertent intravascular injections in humans, the maximum tolerated doses for bupivacaine were 0.1 mg/kg (intra-arterial; carotid artery) and 1.8 mg/kg (intravenous; saphenous/cephalic veins), whereas the maximum tolerated doses for LB were 4.5 mg/kg for both intra-arterial and intravenous [34], which is equivalent to administering a full 20-mL vial of LB (266 mg; i.e. the highest dose available) to a 60-kg adult. Although LB is not approved for intra-arterial or intravenous administration and intravascular administration of LB should be avoided, these findings suggest that the pharmacokinetic profile of LB may reduce the risk of acute systemic toxicity and thereby the risk of LAST. However, further data (e.g. prospective clinical registries, such as those established to monitor the safety and effectiveness of peripheral regional anesthesia) are needed to confirm these findings in other real-world settings [36–38].

4.1. Limitations

This study has numerous limitations, including those inherent to any analysis of the FAERS database, such as duplicate reports, reporting accuracy and quality, incomplete or insufficient details about drug administration (e.g. site, route, dose, timing), and the absence of important patient characteristics (e.g. medical history, comorbidities). In addition, disproportionality analysis is unable to adjust for all of these potential confounding factors and must therefore be interpreted with caution and considered with other factors to determine the need for additional investigation [24].

The proposed new definition of LAST in FAERS, based on a combination of signs or symptoms related to the cardiovascular system and central nervous system, or seizure, has not been validated against complete medical records for cases of LAST and may overestimate or underestimate the number of cases possibly related to LAST. Future analyses could further explore the specific signs or symptoms included in each category, rate them by relative severity, and explore the sensitivity and specificity of using different combinations. The concept that concomitant medications in FAERS other than injectable local anesthetics could also result in the same signs or symptoms related to the cardiovascular system and central nervous system designated as possibly related to LAST should also be evaluated further, since this exploratory analysis was based on common adverse events reported in the prescribing information or known to a clinical pharmacist.

Another important source of confounding in this analysis is that LB is used for postsurgical analgesia, and therefore all patients exposed to LB are also exposed to numerous other interventions (e.g. fasting, anesthesia, surgery, surgical devices, physical therapy) over a short time period, making it difficult to isolate the cause for signs or symptoms possibly related to LAST. In addition, the indication for surgery may itself be responsible for some of these signs or symptoms related to the cardiovascular system or central nervous system (e.g. seizures may not be unexpected in a patient undergoing a neurosurgical intervention for a neurological disorder). Furthermore, it is challenging to identify some of the signs or symptoms of LAST in patients receiving general, regional, or neuraxial anesthesia, which are commonly used in many surgical procedures for which LB is administered to manage postsurgical pain.

5. Conclusions

Managing acute postsurgical pain is critical to patient recovery after surgery, and the use of multimodal analgesia with two or more medications and interventions targeting the source of pain is recommended to enhance efficacy and reduce the harms associated with opioids. The use of injectable local anesthetics is associated with LAST, which can be challenging to recognize using retrospective administrative records, particularly when relying on the limited information available in the FAERS database. Our results suggest that including all adverse events for three SOCs based on the MedDRA hierarchy may overestimate the number of cases possibly related to LAST, whereas a proposed new definition that requires reporting of signs or symptoms related to both the cardiovascular system and the central nervous system (or seizures) may be more appropriate. Disproportionality analyses based on three approaches to defining LAST in the FAERS database showed little evidence of a disproportionate signal for LB compared with other injectable local anesthetics. However, these results are exploratory, and this method does not replace full medical adjudication using a complete medical record to determine whether the signs or symptoms observed truly represent an occurrence of LAST and, if so, whether a single medication can be identified as its sole cause, particularly when this rare event occurs in the busy perioperative setting.

Author Contributions

All authors participated in the study conception and design and interpretation of the data, were responsible for review and critical revision of the manuscript for intellectual content, gave final approval of the version to be published, and agree to be accountable for all aspects of the work. ARJ and CCV collected and analyzed the data.

Declaration of interest

S Dagenais and R Scranton are employees of Pacira Pharmaceuticals, Inc. AR Joyce and CC Vick are Principals at Venebio Group, LLC, and served as consultants to Pacira Pharmaceuticals, Inc. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors wish to acknowledge assistance provided by Brian Faley, PharmD, and Aaron Shiraz, MD, when developing some of the methods used in this manuscript. Editorial support for development of this manuscript was provided by Vandana Sharma, PhD, at C4 MedSolutions, LLC (Yardley, PA), a CHC Group company, and was funded by Pacira Pharmaceuticals, Inc.

Supplementary data

Supplementary data can be accessed here.

Additional information

Funding

This paper has been funded by Pacira pharmaceuticals.