Efficacy of Antitoxin Therapy in Treating Patients With Foodborne Botulism: A Systematic Review and Meta-analysis of Cases, 1923–2016

We conducted a systematic review and meta-analysis of botulism treatment. Published studies generally were low quality, but support timely administration of antitoxin and high quality supportive care to reduce mortality.

Abstract

Background

Botulism is a rare, potentially severe illness, often fatal if not appropriately treated. Data on treatment are sparse. We systematically evaluated the literature on botulinum antitoxin and other treatments.

Methods

We conducted a systematic literature review of published articles in PubMed via Medline, Web of Science, Embase, Ovid, and Cumulative Index to Nursing and Allied Health Literature, and included all studies that reported on the clinical course and treatment for foodborne botulism. Articles were reviewed by 2 independent reviewers and independently abstracted for treatment type and toxin exposure. We conducted a meta-analysis on the effect of timing of antitoxin administration, antitoxin type, and toxin exposure type.

Results

We identified 235 articles that met the inclusion criteria, published between 1923 and 2016. Study quality was variable. Few (27%) case series reported sufficient data for inclusion in meta-analysis. Reduced mortality was associated with any antitoxin treatment (odds ratio [OR], 0.16; 95% confidence interval [CI], .09–.30) and antitoxin treatment within 48 hours of illness onset (OR, 0.12; 95% CI, .03–.41). Data did not allow assessment of critical care impact, including ventilator support, on survival. Therapeutic agents other than antitoxin offered no clear benefit. Patient characteristics did not predict poor outcomes. We did not identify an interval beyond which antitoxin was not beneficial.

Conclusions

Published studies on botulism treatment are relatively sparse and of low quality. Timely administration of antitoxin reduces mortality; despite appropriate treatment with antitoxin, some patients suffer respiratory failure. Prompt antitoxin administration and meticulous intensive care are essential for optimal outcome.

Botulism is a rare neuroparalytic illness characterized by bilateral cranial nerve palsies and descending paralysis, which may lead to respiratory failure. Disease can be fatal, particularly without treatment. Botulinum neurotoxin (BoNT) mediates the effects of disease [1, 2]. Foodborne botulism occurs following ingestion of toxin found in contaminated foods; outbreaks occur periodically, and contamination of a widely consumed food could cause many illnesses. BoNT is the most toxic substance by weight known, and various countries with biologic warfare programs have weaponized it as a biological weapon. The Centers for Disease Control and Prevention classifies BoNT as a category A biological agent [3, 4].

Seven botulism toxin serotypes, A–G, were described between 1919 and 1970. Most human botulism is caused by serotypes A, B, E, and F, but others have pathogenic potential as well. Paralysis induced by BoNT can last weeks to months. In cases in which the respiratory tract or respiratory muscles are impaired, mechanical ventilation is life-saving. Botulinum antitoxins can neutralize circulating botulinum toxin, preventing toxin binding to the neuromuscular junction, but does not reverse existing paralysis. The mainstays of treatment are supportive care and botulinum antitoxin. These measures have been credited with reducing botulism mortality in the United States from >60% in the early 20th century to <5% at present [5]. Additionally, other targeted treatments have been attempted to ameliorate the effects of botulism, such as cathartics and enemas to clear toxin from the gastrointestinal tract, and guanidine and 3,4-diaminopuridine to stimulate acetylcholine release [6].

Equine-source botulism antitoxins of varying valencies (anti-A, anti-B, anti-AB, etc) and neutralizing capacities have been used over the past century. Currently, the sole botulinum antitoxin product licensed for treatment of noninfant botulism in the United States is equine-source botulism antitoxin heptavalent [7]. A human-source antitoxin, botulism immune globulin intravenous (BIG-IV), is licensed in the United States solely for the treatment for infants with botulism type A or type B.

Botulism is rare, with <100 noninfant cases and approximately 100 infant cases reported annually in the United States [8]. Consequently, peer-reviewed studies on the efficacy of botulinum antitoxin are sparse. In nearly a century, only 1 study, involving a retrospective cohort, examined the effectiveness of equine-origin botulinum antitoxin, finding reduced mortality and other measures of severity associated with early treatment [9]. For infant botulism, 1 randomized controlled trial found that use of BIG-IV was associated with reduced duration of mechanical ventilation and hospitalization without reported adverse events [10–13].

A large botulism outbreak, either naturally occurring or intentionally created, may strain resources, and at present, no evidence-based guidance exists to prioritize antitoxin administration. Data-based treatment recommendations are needed to ensure an effective and efficient public health response. We sought to systematically review all published reports and studies on the treatment of foodborne botulism. The questions that guided our systematic review were (1) what benefit should be expected from botulinum antitoxin? (2) Is there a window of time beyond which administration of antitoxin is no longer beneficial? (3) Do any patient demographic or clinical characteristics predict greater benefit from antitoxin administration?

METHODS

This study was registered on the International Prospective Register of Systematic Reviews (PROSPERO; CRD42015024327). Our systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [14].

Search Strategy

With the assistance of an expert research librarian, we queried PubMed via Medline, Web of Science, Embase, Ovid, and Cumulative Index to Nursing and Allied Health Literature (CINAHL) using keywords such as botulism, Clostridium botulinum, botulinum antitoxin, botulism, botulinum, BoNT, antibacterial agents, antitoxins, botulinum antitoxin, respiration artificial, activated charcoal, cholinergic antagonists, drugs investigational, disaster planning, immunoglobulins, cholinergic, guanidine, anticholinergic, ventilation, antitoxin, antisera, antiserum, antibody, antibiotic, charcoal, pharmaceutical, germine, immunoglobulin, intubation, experimental animal, and nonhuman; the search strategy is detailed in Supplementary Appendix 1. Embase, Scopus, Medline, CINAHL contain conference proceedings and dissertations). Additionally we searched National Technical Information Service, Defense Technical Information Center, and Google Scholar for government documents, and manually included article references for additional studies.

Inclusion and Exclusion Criteria

We included any article in the English language including case report, series, cohort, controlled trial, and animal study that reported patient-level data on the natural history of foodborne botulism or the effect of a directed treatment such as antitoxin. Botulism was defined by clinical diagnosis, epidemiologically (reported exposure known source with neurologic symptoms of botulism), or based on laboratory assays. Because of different pathophysiology, human cases of iatrogenic, infant, and wound botulism were excluded. Animal cases were excluded if they did not describe both controlled toxin exposure and antitoxin treatment. Case reports and series of <3 cases were abstracted for descriptive analysis, including specific rare treatments and outcomes, and excluded from meta-analysis and formal statistical analyses.

Article Screening and Abstraction

Articles were screened independently by 2 investigators (J. C. O. and P. K. T.) using Covidence (www.covidence.org), an online tool for systematic reviews. Initial screening entailed abstract review, followed by review of all potentially relevant full text articles. Reviewers determined if cases reported a case of foodborne botulism with discussion of both treatment strategies and clinical outcomes; if these criteria were met, the study was included. In both phases, disagreements were resolved by discussion and consensus was reached on all articles.

Articles were abstracted using a standardized form developed on Research Electronic Data Capture (REDCap) (Supplementary Appendix 2). Study data were collected and managed using REDCap electronic data capture tools hosted at Mayo Clinic. REDCap is a secure, web-based application designed to support data capture for research studies, providing (1) an intuitive interface for validated data entry; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for seamless data downloads to common statistical packages; and (4) procedures for importing data from external sources [15]. Primary abstraction was completed by 1 of 8 reviewers (E. H., A. G. E. L., R. A., D. D., A. S., O. A., J. M., or E. T.) trained in use of the abstraction tool and definitions by 1 investigator (J. C. O.). All articles were independently abstracted and reviewed by a second reviewer (J. C. O. or P. K. T.). Disagreements were resolved through discussion at weekly meetings.

Data Abstraction

Articles were reviewed for outcomes including mortality, hospital length of stay, duration of mechanical ventilation, and long-term neurologic sequelae. Articles were grouped first by the article type (case series, cohorts and trials, animal studies) and then by the intervention (eg, monovalent, bivalent, heptavalent antitoxin).

Meta-analysis

For meta-analytic purposes, we combined cohort studies, outbreak reports, or case series with N ≥3 to estimate the effect of antitoxin on mortality, ventilation, and hospital length of stay. In studies in which both the toxin type and antitoxin type were reported, only patients who received antitoxin which matched identified toxin exposures were considered “treated.”

For mortality analysis, the odds of death were calculated as a proportion of the total number of treated vs not-treated patients in a given report. Because in smaller studies the binomial distribution may not approximate the normal distribution, we performed a double arcsine Freeman-Tukey transformation to stabilize the variance and restore assumptions of normalcy. We calculated estimated variance by adding a proportionality constant to the observed number of deaths to adjust for zero values; this was equal to the reciprocal of the total number of subjects in the study. Resultant data were combined using the random-effects model prescribed by DerSimonian and Laird [16]. Subgroup analyses were performed by antitoxin type and toxin types. The odds of death for each type was calculated and compared. Heterogeneity was assessed using I², where 0% indicates low heterogeneity and 100% high heterogeneity. Publication bias was assessed using visual inspection of a funnel plot using the method prescribed by Egger et al [17]. Statistical analysis was performed using Stata (StataCorp, College Station, Texas) and Review Manager 5 (Nordic Cochrane Center, Copenhagen, Denmark) software.

Duration of hospitalization and ventilation were modeled using mean and standard deviation data from cohorts and case series providing sufficient data to determine patients’ duration of treatment. On meta-analysis, only studies including both treated and untreated patients were analyzed, as interhospital ventilation and extubation practices were too great for indirect comparison.

Early vs late administration of antitoxin, defined by report authors, was analyzed relative to mortality, duration of ventilation, and duration of hospitalization using the methods described above. Subgroups were defined by (1) definitions given in the study for early vs late (Supplementary Appendix 8) or (2) patients who received antitoxin within 24 hours of presenting for medical care vs all others. Additional sensitivity analyses were performed on studies with sufficient patient-level data to exclude those who died when calculating length of stay and duration of hospitalization. This was performed to determine the effect of any survivor bias on these time-dependent variables. Finally, using study evidence ratings, we conducted an additional sensitivity analysis excluding the 50% lowest-scoring studies as rated by the evidence-rating tool described below. Results from these sensitivity analyses were compared to the main outcomes to identify confounding by reporting and publication bias.

Meta-regression was performed on all studies by year of publication to determine the role that evolving supportive care techniques have played over time in each of these domains.

Individual Study Evidence Rating

Each publication was assessed by both reviewers for quality of evidence using a tool developed by the authors (Supplementary Appendix 3). This tool evaluates articles for design quality, level of certainty, and completeness of reporting. Study design and evidence were modified from the criteria used by the US Preventive Services Task Force. The risk-of-bias tool was modified from the Newcastle-Ottawa Scale. Completeness was assessed based on the presence of 15 pieces of information deemed pertinent to a botulism treatment article (Supplementary Appendix 3).

RESULTS

The search strategy yielded 13 055 distinct abstracts for screening. One abstract could not be obtained; 13069 were screened, of which 12455 studies were excluded. Full text evaluation was performed on 599 articles, of which 223 met inclusion criteria (Figure 1). Characteristics of the comparative studies included are described in Supplementary Appendix 7. Outcomes are summarized in Table 1. Additional details on the method of diagnosis are shown in Supplementary Appendix 8. Individual studies and case reports unsuitable for quantitative analysis are summarized in Table 2.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) selection flow diagram for studies addressing botulism antitoxin therapy outcomes, 1948–2016.

Table 1.

Study Outcomes

Study	Mortality	Ventilation	Hospitalization	Long term deficits	Adverse effects of antitoxin
Dolman et al, 1948[56]	100%	0%	0%	N/A	N/A
Gray et al, 1948 [57] bNorthern Queensland	27%	NR	NR	NR	NR
Gray et al, 1948[57] b Northern Territories	20%	NR	NR	NR	NR
Johnson and Style, 1949[39]	25%	0%	100%	NR	NR
Dolman et al, 1954[58]	60% mortality	NR	NR	NR	NR
Dolman et al., 1963[18]	19.1% overall 28.9% without antitoxin 3.5% with antitoxin	NR	NR	NR	NR
Eadie et al., 1964[37]	66% mortality; 50% with antitoxin	1 patient ventilated for three days before death	Mean 2.3 days hospitalization	NR	1 patient developed shock and vomiting after antitoxin, but recovered
Minter et al, 1964[42]	0%	66% required ventilation for a mean of 11 (1.4)days	Mean of 14.7 (3.1) days	None (duration of followup unspecified)	NR
Roger, 1964[49]	60% mortality among those not receiving antitoxin 0% with dose who did	2 out of 8 required mech ventilation	NR	NR	NR
Koenig et al., 1967[41]	0% mortality	1 ventilated for 4 days	Mean 9.8 (6.9) days	No residual effects at 6 weeks	NR
Armstrong et al, 1968[50]	33%; the single case given antitoxin died	1; the case receiving antitoxin was ventilated for 3 days before expiring	Two cases were hospitalized for four days	NR; notably one untreated case was pregnant and gave birth to a healthy child 2 months after recovery	NR
Fukuda et al, 1970[38]	14.3%; two received type E antitoxin, one no antitoxin	NR	NR	NR	1 receiving bivalent AB had a strong skin reaction
Faich et al, 1971[24]	0%	100%, ventilated from 3–12 weeks	Hospitalized for 3–15 weeks	NR	NR
Dolman et al., 1974[55]	46% overall	NR	NR	NR	NR
Cherington et al., 1975[59]	14% mortality	6 patients required ventilation (duration unspecified)	NR	Survivors were largely symptom free at 90 days; one patient had constipation	Fever following antitoxin administration (1 patient) Skin rash due to guanidine (1 patient)
Horwitz et al, 1975[27]	66% mortality (1 treated with antitoxin, one untreated)	2 required mechanical ventilation until death	NR	NR	Serum sickness due to trivalent antitoxin in survivor
Koenig et al., 1975[60]	0% mortality	NR	NR	NR	NR
Barrett et al., 1977[22] b Chefornak	0% mortality	NR	NR	NR	NR
Barrett et al., 1977[27] b New Stuyahok	66% mortality	1 patient required MV	NR	NR	NR
Puggiari and Cherington, 1978[61]	0% mortality	1 required six weeks of respiratory assistance	Hospitalized for 8 weeks weesk and 57 days	Recovery at 3 months	NR
Ball et al. 1979[62]	50% mortality	NR	47.5 (31.8) days	At 3 months, one had recovered completely and the other had residual weakness attributed to muscle wasting.	NR
Boselli et al, 1979[63]	33%mortality	1 required 87 days of ventilation, the other 16 months	NR	NR	NR
Nightingale et al., 1980[64]	83% mortality	NR	NR	NR	NR
Morris Jr. and Hatheway, 1980[65]	0% mortality	7/8 required ventilator for mean of 27 days	NR	NR	NR
Tacket et al, 1984[9]	14.3% of patients died in the hospital 42% of these were directly attributable to botulism	Median 25 days of support	Median 40 days of hospitalization	Not reported	None reported
Macdonald et al, 1985[66]	1 died while still hospitalized six months after exposure	NR	NR	NR	1/20 had serum sickness
Shih et al., 1986[67]	12.5% of patients died overall 50% in pre-antitoxin era, 7.6% in post antitoxin era	NR	NR	NR	NR
Lecour et al., 1988[68]	0%	NR	27 days (range 8–37)	NR	NR
Critchley 1989 [69]	4% mortality	20.5 (14.5) days	19.3 (17.7) days	4% were not discharged within 60 days	None reported
Slater 1989[31]	14% mortality	3/7 required veintaltor support for a mean (SD) of 11.5 (2.1) days	NR	None reported in survivors	None reported
Barrett et al, 1991[21]	6% attributed mortality	14 patients were nasotracheally intubated for a mean 8.6 days (range 3–20). Nine tracheostomy 15.5 days (range 7–24).	NR	NR	None reported
Hibbs et al, 1996[70]	NR	NR	NR	NR	NR
Angulo et al, 1998[20]	0 deaths	NR	NR	NR	One hypersensitivity reaction, one case of pruritis
Chaudhry et al, 1998[2]	3 deaths, all in untreated arm	2 intubated duration unspecified	NR	NR	None reported

Table 2.

Individual Case Report of Botulism Exposure, Treatment and Outcomes, 1946–2015

Report	Patient characteristics	Method of confirmation	Treatment	Outcomes
Miley 1946[71]	27 year old female who consumed canned beets	Clinical appearance; no laboratory confirmation	Ultraviolet blood irradiation 24 hours after admission	Drastic improvement within 48 hours; complete resolution by hospital discharge on day 13
Kendall 1949[72]	33 year old female who consumed canned tongue	Clinical appearance; no laboratory confirmation	Supportive care	Eye and facial weakness persisted at 7 week evaluation
Bennet et al, 1964[73]	51 year old female who consumed canned cantaloupe	Mouse bioassay confirmed type A botulism	Supportive care only; 21 days of ventilation	Death in hospital 26 days after admission
Cherington and Ryan, 1968[74]	57 year old female who consumed canned string beans	Clinical appearance	Supportive care, 27 days of ventilation, guanidine	Death 34 days after admission EMG and strength improved following each administration of guanidine transiently
Haller et al., [75]	53 year old male consuming canned tomato juice	Clinical appearance and serum/stool testing (assay unspecified)	Bivalent AB antitoxin, IV procaine penicillin	Discharge 9 days after admission; 10 weeks after exposure, no residual symptoms
Cherington and Ryan, 1970[76]	43 year old female exposed to type A botulism by canned potato salad	Mouse bioassay	Bivalent AB antitoxin, guanidine, 28 days of ventilation	Hospitalized for 42 days, continued to have proximal muscle weakness 6 weeks after discharge
Craig et al, 1970[77]	31 year old male with type E botulism from izushi exposure	Mouse bioassay	Monovalent E antitoxin	Death 109 hours following contaminated meal
Oh and Halsey, 1975[78]	54 year old male exposed to type B botulism from preserved tomatoes	Mouse bioassay	AB antitoxin on day 4 after hospitalization; guanidine for two prolonged courses; intubated x 26 days	Strength improved on guanidine infusion.
Rosenthal and Belafsky, 1975[79]	34 year old male exposed to type B botulism from canned vegetables (beets)	Testing of food jar; assay unspecified	Trivalent ABE antitoxin	Patient discharged 12 days after admission had sluggish pupils on discharge
Blake et al., 1977[80]	Two women exposed to type A botulism from canned beef	Mouse bioassay and culture	Surviving patient received trivalent ABE antitoxin; guanidine; penicillin V; edrophonium	One died within 2 hours of hospitalization; the other had a prolonged course but survived to discharge 64 days later with mild generalized weakness
Cherington and Schultz, 1977[59]	40 year old female exposed to canned vegetables	Clinical diagnosis Stool + for toxin (assay unspecified)	Guanidine Germine-3-acetate Bivalent antitoxin Required intubation	Improved symptomatically after administration of guanidine Recovered completely
Maroon and Bisonnette., 1977[81]	57 year old male with foodborne type B exposure	Mouse bioassay	Trivalent ABE	Residual weakness was noted at 1.5 year follow up
Puggiari and Cherington, 1978[61]	41 year old female and 18 year old female exposed to canned peppers	Clinical diagnosis	Guanidine	Ocular improvement and non-respiratory muscle improvement after guanidine
Riddle, 1985[82]	23 year old female exposed to type A botulism from canned tomatoes	Stool/serum and food positive for toxin (assay unspecified)	Trivalent ABE Ventilation	Improvement in respiratory strength after antitoxin administration Discharged 23 days later with residual weakness
Colebatch et al. 1989[83]	49 year old male with type A botulism after eating rice & vegetables	Mouse bioassay	Unspecified polyvalent antitoxin Required intubation	Residual deficits persisted two years later
Martinez-Castrillo et al, 1991[84]	43 year old female with type B botulism after eating canned beans	Clinical appearance Food/serum positive for toxin B (assay unknown)	Trivalent ABE	Still had dry eyes/mouth 30 days later Full recovery within 60 days
Davis et al., 1992[85]	31 year old male with type A botulism after eating canned green chilli	Stool/food jar tested positive (assay not reported)	3,4 DAP Trivalent ABE Ventilation	Survival and improvement at 90 days; no improvement was noted around time of 3,4 DAP administration
Paterson et al., 1992[86]	33 year old male after eating home preserved asparagus	Clinical appearance	Trivalent ABE, vancomycin and plasmapheresis Required intubation	All symptoms resolved within 6 weeks
Mackle et al, 2001[87]	62 year old male with type E botulism after eating reheated cold chicken	Mouse bioassay	Supportive care	Recovery over several months
Cengiz et al, 2004[88]	40 year old female with canned green bean exposure	Clinical appearance	Bivalent AB and monovalent E antitoxin	Complete recovery within 3 months
Bhutani et al., 2005[89]	35 year old male with type A botulism from potatoes	Cultures, Toxin assays (assay not reported)	Trivalent antitoxin Intubation	Complete resolution after 6 ½ months
Al Nassar et al., 2012[90]	28 year old female with botulism from Faseikh	Stool culture	Trivalent ABE antitoxin Intubation	Complete resolution
Vasa et al., 2012[91]	69 year old male with type A botulism after eating green beans/ tomatoes	Unknown assay	Monovalent A antitoxin	Recovery 5 months after exposure
Centers for disease control, 2013[92]	39 year old male exposed to type B botulism from home fermented tofu	Mouse bioassay	Antitoxin, unspecified Intubated IVIG and edrophonium were also administered for suspected myasthenia gravis	Discharge from hospital 23 days later
Gasparini et al., 2013[93]	78 year old female with submandibular sialadenitis, exposure unknown	PCR on stool and rectal swab	Trivalent antitoxin ABE Antibiotics	Complete resolution
Kotan et al, 2013[94]	43 year old female after eating homemade canned beans	Clinical Appearance	Bivalent antitoxin A	Resolution by 6 weeks
Pender et al, 2013[95]	91 year old female with type A foodborne botulism after eating home canned borscht	Mouse bioassay	Unspecified antitoxin and edrophonium Required intubation	Residual dysphagia persisted beyond hospital discharge
Zhang et al., 2013[96]	33 year old male with type E botulism after eating dried crude beef	Food specimen culture, assay unknown	Gastric lavage, antibiotics, supportive care Required intubation	No residual symptoms at time of hospital discharge
Arora et al, 2014[97]	35 year old male after eating stale Dahi vada	Clinical diagnosis	Antibiotics, cathartics, hydrocortisone	Survival to discharge
Anniablli et al., 2015[98]	21 year old male with type B botulism from canned turnips	Mouse bioassay	Unspecified antitoxin, metoclopramide Required intubation	Resolution of symptoms by discharge

Patient Characteristics

Data tying individual outcomes to age, gender/sex, comorbidities, and severity of illness at time of initial contact with medical care were reported inconsistently. While we cannot exclude that some populations may receive greater benefit from antitoxin treatment, we could not identify such a subpopulation on the basis of the highly limited data set.

Supportive Care

All articles that described survival alluded to the importance of high-quality supportive care, particularly respiratory critical care. However, insufficient detail (eg, modes of mechanical ventilation and nutritional support) in these publications precluded evaluation of specific intensive care components on clinical outcomes.

Relationship Between Type of Toxin and Clinical Outcome

Mortality among patients treated and not treated with antitoxin, by toxin type, is presented in Figures 2–5. Overall, antitoxin reduced mortality (odds ratio [OR], 0.22; 95% confidence interval [CI], .17–.29); a high degree of heterogeneity was observed (I² = 69.3%). Heterogeneity was largely due to studies not reporting toxin type (Figure 2). Subset analyses by toxin type were significantly less heterogeneous as described below. On meta-regression modeling, study year, design, selection, comparability, and completeness variables were not substantial contributors to heterogeneity (P = 0.86). Egger test identified no evidence of publication bias (P = 0.26).

Figure 2.

Deaths among botulism patients, antitoxin treatment vs no antitoxin treatment, unspecified toxin type, 1948–2016. Odds ratio and 95% confidence intervals (CIs) demonstrated by point and lines extending to either side. Effect size (ES) and weighting illustrated by gray squares. Overall effect estimates provided by diamonds, centered on the odds ratio with points extending to the 95% CI.

Figure 3.

Deaths among botulism patients, antitoxin treatment vs no antitoxin treatment, toxin type A, 1948–2016. Subset of figures reporting administration of antitoxin containing antitoxin A to confirmed toxin type A exposures. Odds ratio and 95% confidence intervals (CIs) demonstrated by point and lines extending to either side. Effect size (ES) and weighting illustrated by gray squares. Overall effect estimates provided by diamonds, centered on the odds ratio with points extending to the 95% CI.

Figure 4.

Deaths among botulism patients, antitoxin treatment vs no antitoxin treatment, toxin type B. Subset of figures reporting administration of antitoxin containing antitoxin B to confirmed toxin type B exposures. Odds ratio and 95% confidence intervals (CIs) demonstrated by point and lines extending to either side. Effect size (ES) and weighting illustrated by gray squares. Overall effect estimates provided by diamonds, centered on the odds ratio with points extending to the 95% CI.

Figure 5.

Deaths among botulism patients, antitoxin treatment vs no antitoxin treatment, toxin type E, 1963–2015. Subset of figures reporting administration of antitoxin containing antitoxin E to confirmed toxin type E exposures. Odds ratio and 95% confidence intervals (CIs) demonstrated by point and lines extending to either side. Effect size (ES) and weighting illustrated by gray squares. Overall effect estimates provided by diamonds, centered on the odds ratio with points extending to the 95% CI.

The greatest reduction in botulism-related mortality was associated with the use of type E antitoxin (OR, 0.13; 95% CI, .06–.30; I² = 10%; Figure 5), followed by type A antitoxin (OR, 0.57; 95% CI, .39–.84; I² = 81.2%; Figure 3); reduction in mortality was not statistically significant for type B antitoxin (OR, 0.74; 95% CI, .27–1.97; Figure 4). Key data on the impact of type E antitoxin come from Dolman and Iida, who examined Canadian mortality rates of botulism type E before (baseline) and after introduction of type E antitoxin therapy [18]. Baseline mortality in this study was derived from secondary data and was thus excluded from the meta-analysis. Dolman and Iida’s baseline was derived from 75 outbreaks with 374 patients in 6 countries; case fatality rate was 30%. In Canada, in 28 outbreaks, 85 of 220 patients were treated with anti-E antitoxin. Among patients not treated with antitoxin, 28.9% died, compared with 3.5% of those treated [18].

Use of Different Antitoxin Formulations

Studies that reported both toxin type and administered antitoxin type were less heterogeneous overall and demonstrated a survival benefit to antitoxin (OR, 0.16; .09–.30; I² = 0%; Figure 6). Fifteen studies reported use of trivalent antitoxin, 10 bivalent antitoxin, and 2 heptavalent antitoxin.

Figure 6.

Survival of botulism-exposed patients by antitoxin valence, 1948–2015.

Only studies reporting specific antitoxin type included. Odds ratio and 95% confidence intervals (CIs) demonstrated by point and lines extending to either side. Effect size (ES) and weighting illustrated by gray squares. Subgroup and overall effect estimates provided by diamonds, centered on the odds ratio with points extending to the 95% CI.

Use of anti-ABE trivalent antitoxin was most commonly reported [19–34]. Trivalent antitoxin, when administered in cases of botulism types A, B, or E, reduced mortality (OR, 0.13; 95% CI, .04–.38; I² = 0%). Side effects from this formulation were less commonly reported than all other antitoxins with side effects reported. Anti-ABE antitoxin was generally reported as effective and well tolerated. Case series reported occasional residual neuromuscular deficits persisting up to several months after therapy (Table 2).

Bivalent AB antitoxin was the second most commonly reported formulation [25, 35–43]; its use was not significantly associated with reduction in mortality (OR, 0.37; 95% CI, .10–1.31; I² = 0%). Information on heptavalent antitoxin includes 2 reports of foodborne botulism outbreaks. One was caused by toxin type A–contaminated canned bamboo in Thailand [43–45] and affected 137 patients, of whom 20 were treated with heptavalent antitoxin, and the others with bivalent AB or quadrivalent ABEF antitoxin. No deaths were reported. The second outbreak, caused by type A toxin, affected 8 prison inmates in Utah [46]. All received heptavalent antitoxin; none died. Low mortality in both outbreaks was attributed to excellent critical and supportive care of the patients [45].

The largest available data source on heptavalent antitoxin was the unpublished Centers for Disease Control and Prevention’s expanded-access Investigational New Drug application. Data included mostly foodborne botulism cases and several cases of wound botulism, infant botulism, and other syndromes. Of 249 patients treated under this protocol, 105 were confirmed as having botulism. One child experienced hemodynamic instability after administration, comprising the only serious adverse event seen with heptavalent antitoxin. Allergic reactions, typically rash, were noted in 6 patients, all of whom recovered without sequelae. Seven deaths were observed in confirmed botulism cases treated with heptavalent antitoxin; none were attributed to the antitoxin [47].

A variety of other antitoxin combinations are reported in Figure 6, including quadrivalent ABEF [43], monovalent A [48], monovalent E [18, 49], and a combination of bivalent AB formulation and monovalent E formulation [50, 51]. The small number of deaths from botulism reported in patients treated with these agents precludes a statistical assessment of their impact on clinical outcomes.

Effect of Timing of Antitoxin Administration

Kongsaengdao et al [44] reported on a subset of 18 severely ill patients during a type A outbreak associated with bamboo shoots in Thailand examined in Wongtanate et al [43] and Wintoonpanich et al [45]. In this subset, patients who received antitoxin on day 4 had significantly shorter duration of ventilator dependence compared with those receiving it on day 6 [44].

Sheth et al reported on 6 patients in a botulism type B outbreak associated with bottled carrot juice. Five patients received antitoxin: 3 within 24 hours, 1 on day 13, and 1 on day 45. At the time of publication, 2 patients had been ventilator dependent for >1 year [52]. We treated these patients as having 365 ventilator-days, demonstrating some reduction in ventilator time in those who had early treatment.

Yu et al also reported a benefit from heptavalent antitoxin given within 48 hours compared with later administration [47]. In patients treated early, the proportion of patients requiring mechanical ventilation was significantly reduced and there were no deaths, compared with 7 deaths among patients treated later. Overall, earlier antitoxin administration reduced mortality, compared with later administration (OR, 0.12; 95% CI, .03–.41; I² = 0%; Figure 7).

Figure 7.

Survival for botulinum neurotoxin–exposed patients with early vs late exposure to antitoxin, 1984–2016. Four studies which reported “early” vs “late” groups included. Definitions varied between studies; Sheth et al and Tacket et al defined “early” as within administration within 24 hours of presentation and late as all others. Yu et al and Oriot et al reported outcomes for before and after 48 hours postpresentation to care. Abbreviations: CI, confidence interval; ES, effect size.

Animal Studies

In general, animal studies showed some benefit to antitoxin treatment (Supplementary Appendices 5 and 9).

Therapeutic Agents Other Than Antitoxin

We identified no improvement with several agents reported in the literature (Supplementary Appendix 6).

Study Quality Assessments

Most studies were rated as level III evidence (case reports and series, opinions of expert authorities) (Supplementary Appendix 4). Outcome assessments were not consistently high quality, with a mean of 5/12 on our scoring rubric. Completeness of reporting was generally good, with an average score of 11/16.

Study design was not a major contributor to interobserver variability (P = .50). Study outcome assessments were similarly not a major contributor (P = .54). Completeness was also not significantly associated on meta-regression (P = .31).

DISCUSSION

In this systematic, comprehensive literature review of botulism treatment, we examined all relevant publications since the early 20th century. Our findings support the routine administration of botulinum antitoxin to botulism patients. Antitoxin treatment reduces mortality, and the available data show that earlier antitoxin administration reduces both mortality and ventilation time, compared with later administration. However, there are reports of benefit even with late antitoxin administration, defined by the authors as anywhere from >24 hours after illness onset to 48 hours after presentation. We found no clear indication of a point in the course of illness at which antitoxin administration was no longer beneficial. Although the nature of the data did not allow for quantitative analysis of the benefits of supportive care, this modality is doubtless essential to survival of patients with severe botulism, as evidenced by decreased mortality since the introduction of ventilator care in the 1960s. Despite some early promising studies for guanidine, we did not identify clear or sustained benefit to any alternative treatment modalities applied to botulism patients.

Qualitative studies and case reports broadly supported treatment with botulinum antitoxin. Several older studies dating to the first half of the 20th century reported dramatic clinical resolution following antitoxin administration with all types (Table 2); the significance of these observations is unclear. It is worth noting that while mortality in patients not treated with antitoxin was high, often these patients did not have access to advanced life support. Therefore, these poor outcomes were likely influenced by the lack of supportive care.

Available data do not suggest any patient characteristics that predict a response to therapy. Because the data do not provide evidence of demographic or clinical indicators for predictors of better outcome, we cannot recommend any specific criteria for prioritizing antitoxin treatment when its availability is limited. Further study is needed to risk-stratify patients and identify patients likeliest to benefit from antitoxin treatment.

Correspondence between toxin types and antitoxin serotype corresponded with clinical outcome. The currently licensed heptavalent antitoxin provides appropriate treatment for botulism caused by serotypes A–G. Data available to us were limited to unpublished prelicensure surveillance. Therefore, treatment outcomes should be monitored and analyzed on an ongoing basis.

Recently, reports of new toxin types have been published, including a novel toxin subsequently shown to be a hybrid type A/F fully neutralized by HBAT [53, 54] and a novel toxin identified and assembled from the published gene sequence of a C. botulinum isolate [55]. These reports illustrate important challenges in the field of botulism. New botulinum toxins of clinical significance may be discovered. The routine analysis and dissemination of clostridial gene sequences can accelerate such discoveries, and likely facilitate assembly of toxins, for purposes hopefully benign but possibly nefarious. Preparedness requires careful laboratory investigation of all suspected botulism cases and ongoing research and development of new countermeasures.

Appropriate supportive care is considered a cornerstone in survival and recovery from botulism [56, 57]. It must be borne in mind that a substantial proportion of botulism patients suffer respiratory compromise, some despite prompt diagnosis and early antitoxin treatment; therefore, survival of some botulism patients, treated or untreated with antitoxin, depends on high-quality intensive care. Longstanding experience shows that deaths from botulism can be averted by providing meticulous intensive care, including ventilator care when required [52]. Antitoxin should be given as soon as possible along with meticulous intensive care. Likewise, mortality in case series and reports was highest in patients remote from hospital or emergency care facilities. Early recognition and best intensive care unit practices are an important part of the comprehensive care of the botulism patient. In a situation of limited antitoxin availability, the availability of critical supportive care will determine survival for patients experiencing respiratory compromise. Critical care is an essential component of emergency preparedness for botulism events, both naturally occurring and intentionally created.

Our study has several limitations. We did not include reports of wound botulism in the review, so our findings may not fully apply to that syndrome. Although we made every effort to separate outbreaks and case reports to avoid analyzing the same individual twice, the way that outbreaks are reported and analyzed, particularly in older literature, makes this nearly impossible. For example, several patients are likely reported in each of 2 case series reported by Dolman [18, 58] likely, but it is impossible to determine which. We attempted to account for this with several sensitivity analyses serially excluding these and other high overlap studies; we did not see a substantial change in our overall results.

Our review includes publications spanning 8 decades. Aside from improvements in antitoxin manufacture, types, and valence, this period encompasses dramatic improvements in supportive care technology and techniques and the rise of critical care as a specialty. The effect of these on outcomes is profound, but we could not methodologically account for it in our results. We must therefore accept that some of the benefit attributed to antitoxin may in fact be due to some of these advances. Thus, we cautiously interpret our results to suggest that appropriate antitoxin therapy in conjunction with high-quality supportive care produce the best outcomes in botulism patients.

A final limitation is our inability to adjust for the effect of disease progression and timing. Although we did observe some benefit from early antitoxin administration as defined by the report authors, several confounders are inherent in this finding. First, early administration may have been a marker for early recognition and initiation of other interventions. Second, early administration in several studies may have reflected early treatment of clinically milder cases that presented after delayed diagnosis of a severely ill outbreak index case. However, a beneficial effect of early antitoxin administration was reported in nearly all studies examining this measure. Overall, this would support a benefit of earlier administration, but we may have overestimated the degree of benefit.

Another significant limitation is our focus on foodborne botulism. This was done primarily because of the different physiology involved in foodborne, infant, and iatrogenic botulism, and allowed us to focus on a less heterogenous population experiencing the most common form of botulism. We do not know how these findings may apply to other forms of botulism.

In conclusion, we found that early administration of antitoxin of serotype-appropriate antitoxin, along with high-quality supportive care, was consistently associated with reduced mortality in botulism patients. No demographic or clinical predictors of the response to antitoxin were identified.