Clinical Characteristics and Ancillary Test Results Among Patients With Botulism—United States, 2002–2015

Abstract

Background

Botulism is classically described as a bilateral, symmetric, descending flaccid paralysis in an afebrile and alert patient without sensory findings. We describe the reported spectrum of clinical findings among persons >12 months of age in the United States during 2002–2015.

Methods

The Centers for Disease Control and Prevention collects clinical findings reported by physicians treating suspected cases of botulism nationwide. We analyzed symptoms and signs, and neuroimaging and cerebrospinal fluid (CSF) results. A case was defined as illness compatible with botulism with laboratory confirmation or epidemiologic link to a confirmed case, and presence or absence of at least 1 sign or symptom recorded. Physicians’ differential diagnoses were evaluated.

Results

Clinical information was evaluated for 332 botulism cases; data quality and completeness were variable. Most had no fever (99%), descending paralysis (93%), no mental status change (91%), at least 1 ocular weakness finding (84%), and neuroimaging without acute changes (82%). Some had paresthesias (17%), elevated CSF protein level (13%), and other features sometimes considered indicative of alternative diagnoses. Five of 71 (7%) cases with sufficient information were reported to have atypical findings (eg, at least 1 cranial nerve finding that was unilateral or ascending paralysis). Illnesses on the physician differential included Guillain-Barré syndrome (99 cases) and myasthenia gravis (76 cases) and, rarely, gastrointestinal-related illness (5 cases), multiple sclerosis (3 cases), sepsis (3 cases), and Lyme disease (2 cases).

Conclusions

Our analysis illustrates that classic symptoms and signs were common among patients with botulism but that features considered atypical were reported by some physicians. Diagnosis can be challenging, as illustrated by the broad range of illnesses on physician differentials.

Botulism is a life-threatening illness first recognized centuries ago; it results from blockade of acetylcholine release into neuromuscular junctions [1]. Timely antitoxin administration and outbreak response depend on physician suspicion of the illness [2], but most physicians have never seen a case [3, 4]. Botulism is described as a bilateral, symmetric, descending flaccid paralysis in an afebrile and alert patient without sensory findings [5, 6]. Divergences such as asymmetric weakness and ataxia have led to delayed diagnoses and incorrect alternative diagnoses [5, 7]. Abnormal results on neuroimaging, cerebrospinal fluid (CSF) testing, and edrophonium testing are not considered features of botulism [4, 5, 8, 9]; their frequency in botulism has not been assessed. Few opportunities exist to determine the spectrum of clinical findings physicians may encounter because botulism is rare and many countries do not have botulism surveillance systems [10, 11].

In the United States, botulism is a nationally notifiable disease. Antitoxin treatment for patients >12 months of age is only available from federal stockpiles and after clinical consultation with experts at the Centers for Disease Control and Prevention (CDC) or state health departments in Alaska and California. CDC records epidemiologic information obtained through consultations in its National Botulism Surveillance System. In recent years, clinical data have been included among the data systematically recorded. We performed a descriptive analysis of symptoms, signs, and ancillary test results reported among cases of botulism in the United States to improve recognition of cases.

METHODS

The CDC’s National Botulism Surveillance System database includes clinical data about illnesses treated with botulinum antitoxin—that is, those deemed by public health experts to be suspicious for botulism and within the window to benefit from antitoxin treatment. We analyzed data from 2002–2015, consecutive years with the most complete data.

Case Definitions

A botulism case was an illness (1) deemed sufficiently suspicious after public health consultation to warrant antitoxin release, (2) laboratory-confirmed or that occurred in a person who ate the same food as a laboratory-confirmed case patient, and (3) for which the presence or absence of at least 1 sign or symptom of botulism was recorded in the database. A nonconfirmed illness was defined as illness (1) deemed sufficiently suspicious after public health consultation to warrant antitoxin release, (2) without laboratory confirmation or epidemiologic link to a laboratory-confirmed case, and (3) which presence or absence of at least 1 sign or symptom of botulism was recorded.

Data

Data recorded during consultations included (1) demographics; (2) presence, absence, or unknown status of a checklist of symptoms and signs (eg, dysphagia, extraocular palsy, and respiratory failure resulting in mechanical ventilation); (3) presence, absence, or unknown status of bilaterality and symmetry of symptoms and signs; (4) proximal and distal muscle strengths per the Medical Research Council scale [12]); (5) deep tendon reflexes (DTRs) graded as hyporeflexic (ie, <2), hyperreflexic (>2), or normoreflexic (ie, 2) per the National Institute of Neurological Disorders and Stroke’s Myotatic Reflex Scale [13]; (6) results of diagnostic tests and procedures including lumbar puncture, edrophonium chloride (Tensilon) test, and imaging; (7) differential diagnoses per attending physician; (8) dates of illness onset, hospital admission, care in intensive care units, and consultation with public health department(s); and (9) conclusions from electrodiagnostic testing.

Data collected also included botulinum toxin type and transmission categories. Transmission categories are botulism acquired by consumption of preformed toxin in food (foodborne botulism), by toxin produced in wounds contaminated with Clostridium botulinum (wound botulism), or by toxin produced by other routes. The “other” category includes botulism acquired after colonization of the adult gut by C. botulinum or Clostridium baratii (adult intestinal colonization), after high-dose botulinum toxin administration (iatrogenic botulism), and by an indeterminate mode.

New variables have been added to the surveillance system over time. Methods of soliciting some data have changed over the years; for example, abnormal reflexes were solicited with a question inquiring whether reflexes were abnormal until it was replaced in 2009 with numerical grading scores. Effort was taken during data cleaning to standardize data collected at different periods; this was not possible for questions about bilaterality and symmetry of cranial nerve signs and comparisons of muscle weakness; for these variables, data were restricted to periods during which data were collected in a uniform way.

Analysis

Botulism cases were analyzed for frequency of demographic characteristics, toxin type, and transmission categories. Clinical data, including frequency of symptoms and signs, and data on reflexes, laboratory tests, and other diagnostic tests were summarized. We used the Fisher’s exact test to assess for differences in the frequency of signs and symptoms between foodborne and wound transmission categories; this was limited to cases due to toxin type A. Rank order of illnesses on the differential diagnosis at the time of public health consultation was recorded. Consultation records were reviewed in detail for cases with symptoms or signs that deviated from the majority. Patients who developed symptoms of botulism while hospitalized for unrelated complaints were excluded from the calculation of days from symptom onset to hospital admission. We evaluated aggregate clinical presentations for bilaterality, descending paralysis, and mental status to determine if some did not meet these aspects of the classic description; symmetry and bilaterality of extremity weakness were not included in our aggregate analysis because of the low number of patients without missing data for each of these. Conclusions from electrodiagnostic testing were not analyzed because these were often oversimplified to phrases such as “consistent with botulism” without explanation or included details that our independent electrodiagnostic consultants believed were erroneously concluded from the minimal data available to us.

Nonconfirmed illnesses were evaluated for frequency of demographic characteristics. Number of days from symptom onset to hospital admission and days to public health consultation were also examined, and time intervals were compared between nonconfirmed illnesses and botulism cases using the Wilcoxon rank-sum test.

All statistical analyses were performed using SAS version 9.4 software (SAS Institute, Cary, North Carolina).

RESULTS

Botulism Cases

Demographic and Epidemiologic Characteristics

During 2002–2015, botulinum antitoxin was released from government stockpiles for 1249 patients >12 months of age (Figure 1). Among 517 illnesses for which (1) antitoxin was released and (2) or that occurred in a person who ate the same food as a laboratory-confirmed case patient 64% (332) had presence or absence of at least 1 sign or symptom recorded, ie, were botulism cases. Median age was 48 years (range, 3–92 years), and 70% (231 of 332) were male. Toxin type A caused botulism in 82% (273 of 332) of cases (144 wound and 114 foodborne), type E in 7% (23 of 332; all foodborne), type B in 6% (19 of 332; 7 wound and 11 foodborne), type F in 3% (10 of 332; 2 foodborne and 8 “other”), and an undetermined toxin type in 2% (7 of 332 cases). Cases occurred in 35 states, with the greatest number in California (131), Washington (32), Ohio (29), and Alaska (24).

Frequency of Reported Symptoms

Among 322 cases, 316 (98%) reported at least 1 of the following symptoms: dysphagia, blurred vision, slurred speech, double vision, and change in the sound of voice. Shortness of breath was reported for 65% (184 of 332). Abdominal pain, constipation, paresthesias, diarrhea, and urinary retention were each reported among <25% (Table 1). The only statistically significant differences in symptom frequency for foodborne and wound botulism cases caused by toxin type A were that having a gastrointestinal complaint (nausea, vomiting, constipation, abdominal pain, or diarrhea) (65% vs 30%; P < .0001) and dizziness (66% vs 48%; P = .03) were significantly more common among foodborne than wound cases and subjective weakness or fatigue (92% vs 74%; P = .0009) was significantly more common among wound than foodborne cases.

Frequency of Reported Signs

Absence of fever, descending paralysis, alert mental status, ptosis, and extremity muscle weakness on examination were each reported for >75% of patients (Table 2). Findings from detailed review of consultation records for cases with atypical presentations of body temperature (fever), mental status (not alert), and paralysis (not descending) are noted in Table 2. Facial palsy was reported for 47% (128 of 272 cases). Respiratory failure leading to mechanical ventilation occurred in 66% (204 of 309) of cases. Among 177 cases with shortness of breath, 81% (144) required mechanical ventilation sometime during their hospitalization; among 170 cases that required mechanical ventilation, 85% (144) had preceding shortness of breath. Extremity weakness (ie, muscle strength <5) was noted in 77% (137 of 178) with reported descending paralysis.

Among 323 cases, 89% (289) exhibited at least 1 of the following signs often considered features of botulism: descending paralysis, ptosis, extraocular palsy, facial paralysis, and unreactive pupils. However, pupils were dilated among 37% (92 of 252), unreactive among only 25% (25 of 198), and constricted among 3% (8 of 237); all but 2 cases involving constricted pupils had botulism due to to contaminated black tar heroin. Sensory deficits were reported in 7%. There were no significant differences in frequency of signs reported between foodborne and wound botulism cases due to toxin type A.

Bilaterality and Symmetry of Signs

Bilaterality of signs ranged from 85% to 100% depending on the assessed sign. Facial paralysis, extraocular palsy, and ptosis were unilateral in 15% (9 of 61), 12% (9 of 76), and 9% (11 of 122 patients), respectively. Dilated and unreactive pupils were bilateral for all patients. Symmetry of weakness was evaluated for patients with bilateral signs: Symmetry was reported for 86% (19 of 22) with bilateral facial paralysis, 80% (35 of 44 cases) with bilateral ptosis, and 68% (19 of 28 cases) with bilateral extraocular muscle palsy (Table 2).

Weakness Patterns

Weakness was reported for 61% (78 of 126) of patients with a recorded strength for all proximal and distal muscles of upper and lower extremities. Among these patients, proximal muscles were weaker than distal muscles for 51% (40 of 78 cases), proximal and distal muscle strengths were the same for 42% (33 of 78 cases), and distal muscles were weaker for 6% (5 of 78 cases).

Among 262 cases with subjective muscle weakness, only 63% (166) had recorded strength <5 in an extremity. Among the 187 cases with any recorded strength <5 on the Medical Research Council scale, 89% (166) reported subjective muscle weakness.

Deep Tendon Reflexes

Among 163 cases with at least 1 DTR that could be categorized, 48% (78) were normoreflexic throughout, 41% (67) were hyporeflexic in at least 1 location, and 8% (13) were hyperreflexic in at least 1 location. Three percent (5 of 163 cases) had both hypo- and hyperreflexic DTRs; of these, 4 had diffuse hyporeflexia in the upper extremities and hyperreflexia in the lower extremities and 1 had hyporeflexia and hyperreflexia in both the upper and lower extremities.

CSF Analysis

CSF was tested from 39% (108 of 277 cases). White blood cell count was >5 cells/μL in 7% (6 of 86 cases; median, 50 cells/μL; range, 7–139 cells/μL; normal, 0–5 cells/μL); for these cases with elevated white blood cell count, the median CSF red blood cell count was 4 (range, 1–23 cells/μL) and the median glucose was 65 (range, 21–119 mg/dL; normal, 40–80 mg/dL). Protein was >60 mg/dL in 13% (11 of 85 cases; median, 77 mg/dL; range, 65–132 mg/dL; normal, 15–60 mg/dL). CSF glucose was <40 mg/dL in 5% (4 of 76 cases; median, 14.5 mg/dL; range, 2–22 mg/dL); 1 of these patients had a CSF white blood cell count of 54 cells/μL.

Edrophonium Testing

Edrophonium testing was performed and results documented for 14% (36 of 249 cases): 72% (26 of 36) were reported as negative, no response, or normal, 14% (5 of 36) as positive or mild improvement, and 14% (5 of 36) as indeterminate.

Neuroimaging

An abnormal result of head imaging (magnetic resonance imaging or computed tomography) was recorded for 18% (29 of 162 cases). Eighteen cases had nonacute or nonspecific findings such as chronic microvascular changes, atrophy, and subtle white matter disease; 6 had sinusitis; 2 reportedly had acute findings without additional information (1 with acute brain disease and abscesses throughout and 1 with punctate acute left frontal cortical infarct). Among the 92 patients without neuroimaging, 98% (90) had epidemiological risk factors for botulism, including injection drug use, consumption of Alaska Native foods, or foods associated with an outbreak of which they were part.

Differential Diagnosis

Physicians reported a range of 0–6 illnesses other than botulism on their differential diagnosis at the time of public health consultation for 274 of 332 botulism cases (83%). The most common illnesses were Guillain-Barré syndrome (GBS) (99 cases) and myasthenia gravis (76 cases) (Table 3). For 160 botulism cases occurring during 2009–2015, botulism was listed first for 90% (144) of cases, second for 6% (10), third for 3% (5), and fourth for 1 case.

Atypical Findings

Among 71 botulism cases for which presence or absence of bilateral cranial nerve deficits and descending paralysis were reported, 7% (5 of 71) had at least 1 cranial nerve sign that was unilateral, altered mental status, or ascending paralysis.

Nonconfirmed Illnesses

Among 732 cases for which (1) antitoxin was released and (2) was not laboratory confirmed or epidemiologically-linked to a laboratory confirmed case, 69% (502) had presence or absence of at least 1 sign or symptom recorded, ie, were nonconfirmed illnesses. Among these 502, the median age was 49 years (range, 2–89) and 65% (324) were males. Among 245 nonconfirmed illnesses with this information, botulism was listed first for 83% (203), second for 12% (30), third for 4% (10), and fourth and fifth for 1 illness each.

Comparison of Time Intervals to Consultation for Botulism Cases and Nonconfirmed Illnesses

The time from symptom onset to public health consultation was shorter (P < .001) among 307 botulism cases (median, 2 days; range, 0–24 days; interquartile range [IQR], 1–5 days) than among 455 patients with nonconfirmed illness (median, 4 days; range, 0–56 days; IQR, 2–7 days). The time from hospitalization to public health consultation was shorter (P < .0001) among 308 botulism cases (median, 1 day; range, 0–24 days; IQR, 0–3 days) than among 450 patients with nonconfirmed illness (median, 2 days; range, 0–33 days; IQR, 1–4 days).

DISCUSSION

Consistent with classic descriptions, we found that the most common clinical features identified among patients with botulism, each in more than three-quarters of patients, were the symptoms dysphagia, blurred vision, slurred speech, and diplopia, and the signs descending paralysis, ptosis, and muscle weakness of extremities. Our findings confirm previous reports that GBS and myasthenia gravis are commonly considered alternative diagnoses [4–6]. We found that, although paresthesias were reported by a considerable proportion of patients (17%), physicians rarely reported sensory deficits (7%), consistent with the expectation that botulinum toxin does not affect sensory nerve fibers [14–16]. About two-thirds of patients with foodborne botulism were found to have gastrointestinal complaints. As has previously been reported [17–18], we found that edrophonium chloride testing for myasthenia gravis was sometimes positive among persons with botulism.

Some patients reportedly exhibited clinical deviations from the classic description of botulism. These (1) may truly have been atypical presentations or (2) may have been incorrectly interpreted as atypical. As an example to the first instance, a small proportion of patients had weaknesses described as asymmetric, possibly because circulating toxin affected 2 sides of their bodies differently (eg, because of anatomic asymmetry in cranial nerves or because of atherosclerosis); asymmetric weaknesses have previously been reported [7]. As an example to the second instance, for some, paralysis was reported as “not descending” possibly because physicians suspected an ascending paralysis if patients reported symptom onset in an inaccurate chronologic order or, alternatively, if no directionality of paralysis was observed because paralysis had not progressed from cranial nerves to extremity weaknesses.

Similarly, other expected findings were not always reported by treating physicians, possibly because this truly occurred or because presence of these can be difficult to discern. Half of our cases exhibited the stereotypical proximal muscle weakness greater than distal muscle weakness [8], but nearly an equal number reportedly had strengths equal in proximal and distal muscles and physicians reported distal muscles weaker than proximal in a few. Some patients had CSF findings suggestive of meningitis (above normal white blood cell count) or of GBS (above normal protein), but no further information is available to confirm or explain the unexpected results. Unreactive pupils, which have been described as common among botulism patients and a way of distinguishing botulism from myasthenia gravis [19–21], were uncommon among our cases. Constipation was rarely reported during consultations, although it is a hallmark of infant botulism [22].

Botulism is a public health emergency because of the severity of illness, because 1 case may herald an outbreak, and because many persons may be ill during an intentional exposure [23]. Diagnosis must be made from symptoms and signs because the delay to obtain laboratory results would diminish the effectiveness of antitoxin treatment. Findings from outbreaks that have identified mild and misdiagnosed cases indicate that the threshold for considering botulism should be low [24–30]. Regardless of the reasons, that several unexpected findings were reported by physicians nationwide is notable and have implications for public health preparedness and response; physicians should be aware of these symptoms and signs so that they do not discount botulism when they observe similarly atypical features. Other findings that have implications for preparedness and response are the following: (1) Clinical features such as descending paralysis, a buzzword associated with botulism, may not be evident or may be delayed until patients are hospitalized for days; awaiting observation of these before suspecting botulism could result in a missed diagnosis, (2) Shortness of breath has sometimes been considered an indicator of impending respiratory failure; however, some botulism patients with shortness of breath complaints did not require mechanical ventilation but others without shortness of breath did, indicating that it may not always be a reliable indicator. Close respiratory monitoring may be needed to help avoid mechanical ventilation for some patients.

Physicians obtained CDC consultation later for nonconfirmed illnesses than botulism cases. It is likely that some illness included in the nonconfirmed illnesses were botulism that was not laboratory confirmed because specimens were obtained outside the window for confirmation [31]. Moreover, a low level of botulinum toxin that is still sufficient to cause paralysis may not be detected in serum by the mouse bioassay [32]. Delay in CDC consultation can result in a delay in administering botulinum antitoxin, which is most effective early in illness [33–35]. That illnesses dissimilar to botulism (eg, multiple sclerosis and sepsis) were included in physician differentials at the time of consultation is further evidence that botulism can be a difficult diagnosis.

Our analysis has several limitations. CDC consultations were generally with 1 treating physician for each case. These physicians were often newly involved in the affected patients’ care or did not have immediate access to the medical record, which could have resulted in inaccuracies in reports. Physicians ranged from interns to attendings and primary care doctors to subspecialists; expertise may have affected the quality and completeness of data we received. Physicians could report “unknown” for symptoms and signs; they likely did this if they did not evaluate a patient for a symptom or sign, did not recall the result or find it on the chart, or were unable to discern during the history and physical whether the feature was present. For example, physicians have sometimes reported that paralysis was so severe they could not determine whether it was descending. In calculating percentages, we chose to exclude responses of “unknown” as well as responses that were “missing” (ie, blank) from the denominator; this may have resulted in an overestimate of the frequency of some symptoms and signs. Moreover, we assessed the frequency of only symptoms and signs known to be associated with botulism or illnesses confused with botulism.

We cannot quantify the number of atypical results that might have occurred from concomitant illness (eg, meningitis in addition to botulism) or incorrect technique and interpretation of tests (eg, interpretation of edrophonium testing results). Physicians treating adult patients may not closely monitor signs such as constipation, particularly when patients have more serious medical issues, affecting the frequencies we reported. (Anecdotally, constipation is often reported to CDC several days after consultation when physicians are unable to obtain stool specimens). However, the fact that atypical findings have been reported for specific symptoms, signs, and ancillary test results, and most for >1 case, may assist physicians with weighing unexpected findings with findings supportive of botulism.

In conclusion, we report one of the largest analyses of the clinical characteristics of noninfant botulism cases. To our knowledge, the largest published cohorts were 59 outbreak cases reported to CDC in 1977 [36], 706 reported in the Republic of Georgia during 1980–2002 [37], and 163 outbreak cases in Thailand during 2006 [38]. One of our most important observations is that some laboratory-confirmed cases have features that were reported to diverge from the classic presentation and may even wrongly suggest an alternative diagnosis. Regardless of the reasons for these, physicians should be aware of the spectrum of observations reported by colleagues for botulism cases. Standardized prospective evaluation of future cases, including of electrodiagnostic findings, are needed to further elucidate the clinical presentation of botulism. Public health officials are available 24/7 to perform clinical consultations no matter where botulism is on the physician differential; they can assist with weighing atypical findings with the benefit of early antitoxin administration for botulism.

Notes

Supplement sponsorship. This article appears as part of the supplement “Botulism,” sponsored by the Centers for Disease Control and Prevention.

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Acknowledgments. The authors acknowledge Viggy Parr, BA; Beau Bruce, MD, PhD; and Jeremy Sobel, MD, MPH.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

References