Pulmonary Illness Related to E-Cigarette Use in Illinois and Wisconsin — Final Report

On July 10, 2019, the Children’s Hospital of Wisconsin notified the WDHS of five previously healthy adolescents who had been admitted during the previous 30 days with progressive dyspnea, fatigue, and hypoxemia. Two patients underwent intubation and mechanical ventilation. Computed tomography (CT) of the chest in four patients revealed bilateral ground-glass opacities, predominantly in the lower lobes. Extensive infectious, rheumatologic, and oncologic workups were unrevealing. All patients reported a history of e-cigarette use in the days or weeks before symptom onset.

On July 25, the WDHS issued an alert to clinicians describing the clinical syndrome and requested that similar cases of lung injury associated with e-cigarette use be reported to public health authorities. The WDHS was subsequently contacted by a physician in Illinois seeking clinical treatment guidance for a patient with similar clinical presentation and e-cigarette product exposure, and the WDHS promptly notified the IDPH on July 31. A joint WDHS–IDPH public health investigation was initiated on August 1 to identify additional case patients and to characterize the pulmonary clinical syndrome related to the use of e-cigarettes and related products. The CDC was consulted for technical assistance early in the investigation, and a CDC epidemiologic assistance field team (Epi-Aid) was deployed to the WDHS and IDPH on August 20, 2019.

An outbreak case definition was initially developed by the WDHS and IDPH and was further refined in coordination with the CDC and the Council for State and Territorial Epidemiologists. Table 1 provides specific details of the probable and confirmed case definitions. All cases described in this article were classified according to and met the working outbreak definitions of confirmed or probable cases that have been mutually accepted by the CDC and the jurisdictions that had been affected earliest in the outbreak (Table 1).

The WDHS and IDPH released their first health alert notices on July 25 and August 2, respectively, to inform clinicians of the initial cases and to request reporting of possible cases to their local health departments. Medical records were requested for all patients with reported cases. A standardized medical record abstraction form was developed in Research Electronic Data Capture software (REDCap, Vanderbilt University) to systematically collect demographic data on the case patients, the signs and symptoms at presentation, laboratory results, imaging findings, reported drug exposures, clinical course, treatments, and medical outcomes. Infectious disease and pulmonary consultations and discharge notes were reviewed to determine whether causes other than vaping were potentially responsible for the clinical syndrome. Medical charts were reviewed by a group of clinicians from the investigative teams (six of the authors) with at least two reviewers for each chart, and discrepancies between reviewers were adjudicated by means of consensus. A standardized interview was administered to case patients to characterize the use of e-cigarettes and related products in the 3 months before symptom onset. Information regarding such use came from clinical documentation and interviews with patients.

To assess whether this cluster represented an increase in severe respiratory illness, the IDPH established a baseline rate of severe unexplained respiratory illness by using a syndromic surveillance definition (in contrast to a working outbreak disease-specific case definition). Syndromic surveillance monitors near-real-time prediagnostic data sources (largely from emergency department [ED] visits) to provide early detection of potential public health threats by means of validated algorithms.¹⁴ Data from the National Syndromic Surveillance Program¹⁵ were searched with the use of the Electronic Surveillance System for the Early Notification of Community-Based Epidemics (ESSENCE) tool, which includes information on vital signs, reported symptoms, ED discharge destination, and diagnosis.¹⁵

All ED visits from January 1, 2018, to August 15, 2019, among persons 14 to 30 years of age were searched in order to identify encounters for severe unexplained respiratory illness (defined in Table S2 in the Supplementary Appendix, available with the full text of this article at NEJM.org); searches were limited to counties in Illinois where cases had already been identified. The syndromic definition was designed to capture data on patients who had presented to EDs in Illinois with severe respiratory symptoms and initial pulse oximetry of no more than 96% (or if pulse oximetry was not recorded), who were admitted to the hospital at which they presented or if the discharge disposition was not to home, and who had a discharge diagnosis that was not consistent with a known cause (e.g., bacterial or viral pneumonia or a chronic respiratory disease such as asthma). The formulated query was validated to ensure that it captured data on ED visits by all patients with confirmed or probable cases in Illinois.

Descriptive analyses were performed for all the patients with confirmed or probable cases that were reported in Wisconsin and Illinois. Results were reported as proportions or median values. All analyses were conducted with the use of Stata software, version 16 (StataCorp). Proportions regarding e-cigarette use were limited to the 81 patients who underwent extensive interviews regarding vaping. To calculate the rate of visits for severe unexplained respiratory illness according to syndromic surveillance, a denominator of all ED visits for the same time period, age range, and hospitals was used. The mean monthly rates of ED visits for severe unexplained respiratory illness per 10,000 ED visits for June 1 through August 15, 2018, and for June 1 through August 15, 2019, were compared with a two-sided Student’s t-test.

As of September 6, 2019, a total of 60 patients had cases reported to the WDHS, and 82 to the IDPH. Of the 142 cases reported, 30 were excluded after chart review, and 14 cases were pending case classification because of incomplete medical records or interviews. A total of 98 case patients met the definition of a probable case (48 total, with 21 in Wisconsin and 27 in Illinois) or a confirmed case (50 total, with 20 in Wisconsin and 30 in Illinois). The dates of symptom onset ranged from April 21, 2019, through August 30, 2019, which is outside the typical influenza season. Comprehensive interviews had been conducted for 83% of the 98 patients with a confirmed or probable case.

The median age of the case patients was 21 years (range, 15 to 53) (Table 2); 26% of the patients were younger than 18 years of age. The majority of case patients were male (79%). Most of the patients had no documented relevant medical history, with no underlying chronic lung disease except for asthma (which was noted in 22% of the patients). Patients resided in multiple urban and rural, noncontiguous counties across both states.

Case patients presented with a combination of respiratory, gastrointestinal, and constitutional symptoms (Table 2). The median duration of symptoms before hospital presentation was 6 days, with a wide range (0 to 155 days) of symptom duration; 71% of patients presented within 7 days after symptom onset. A total of 66% of the hospitalized patients had been seen in outpatient settings (EDs, urgent care clinics, or other outpatient clinics) before admission for related symptoms, and 45% of all the patients who were seen in the outpatient or ED setting received antibiotic agents for presumed respiratory tract infection, primarily oral azithromycin (in 20 of the 44 patients who received antibiotics on an outpatient basis), amoxicillin–clavulanate (in 6), levofloxacin (in 5), or doxycycline (in 4). All the patients who received antibiotics on an outpatient basis had reported progression of respiratory symptoms, which had prompted subsequent hospital admission.

A total of 97% of the patients had respiratory symptoms at hospital presentation. The most common respiratory symptoms were shortness of breath (85%), cough (85%), and chest pain (52%). Reported gastrointestinal symptoms included nausea (66%), vomiting (61%), diarrhea (44%), and abdominal pain (34%). All patients had one or more constitutional symptoms, with the most common being subjective fever (84%). Upper respiratory symptoms such as rhinorrhea, sneezing, or congestion were not commonly reported.

Details of the vital signs at presentation and laboratory findings are shown in Table 2. According to the initial recorded vital signs, 63% of patients had tachycardia (heart rate range, 55 to 146 beats per minute), and 43% had tachypnea (respiratory rate range, 14 to 59 breaths per minute). At presentation, 33% of the patients had oxygen saturation between 89% and 94% while they were breathing ambient air, and 25% had oxygen saturation of less than 89% while they were breathing ambient air. A total of 33% of the patients had a documented fever (temperature, ≥38°C) at triage, and 54% had a fever recorded at some point in their admission, but medical records were incomplete and this may be an underrepresentation of the true proportion with an objective fever.

A high percentage of patients had leukocytosis (83%), defined as a white-cell count of more than 11,000 per cubic millimeter, with a median white-cell count of 16,000 per cubic millimeter (interquartile range, 12,000 to 19,000). A total of 91% of the patients had a neutrophil predominance (neutrophil percentage, >80%). Among the 27 patients who had an eosinophil percent listed, none had a value greater than 2%. The erythrocyte sedimentation rate was more than 30 mm per hour in 90% of the 30 patients in whom it was checked. The median procalcitonin value was 0.53 μg per liter (reference ranges differed among the hospitals). Mildly elevated serum aminotransferase values were noted in 46% of the patients and were transient. Just under one third of the patients had mild hyponatremia, hypokalemia, or both. Acute renal insufficiency was observed in 1 patient, which resolved with intravenous hydration.

A total of 43 patients underwent bronchoalveolar lavage; the majority of patients received antibiotics, glucocorticoids, or both before the procedure. Of the 26 bronchoalveolar-lavage specimens with reported cell counts, the median values were as follows: eosinophils 1% (range, 0 to 18), neutrophils 58% (range, 10 to 91), lymphocytes 8.5% (range, 1 to 40), and macrophages 22% (range, 2 to 68). A total of 13 of the 23 available cytology reports on bronchoalveolar-lavage specimens noted lipid-laden macrophages (7 were reported with oil red O stain, and 6 mentioned no specific staining); the remaining 10 reports did not comment on the use of oil red O stain. Of the 13 samples with noted lipid-laden macrophages, 4 reports listed moderate lipid-laden macrophages, 8 reports listed scant to minimal, and 1 report did not quantify.

Six patients underwent transbronchial lung biopsy, and two of these patients also underwent an open lung biopsy, which was performed during clinical workup; three of these patients were receiving both antibiotics and glucocorticoids at the time, two were receiving antibiotics, and one was receiving neither before the procedure. Pathologists reported a range of findings, including mild and nonspecific inflammation, organizing pneumonia and bronchiolitis with organizing pneumonia, acute diffuse alveolar damage and foamy macrophages, and interstitial and peribronchiolar granulomatous pneumonitis. Infectious disease evaluations for possible viral, bacterial, and fungal pathogens were negative in nearly all case patients in whom the testing was performed (Table S1 in the Supplementary Appendix).

At presentation, 83% of the patients had an abnormal chest radiograph (Table 2). A total of 91 of the 98 case patients underwent CT, and the findings were abnormal for 100% of these patients. Opacities in both lungs were present in 100% of the patients. Ground-glass opacities in both lungs were characteristically observed on CT, sometimes with subpleural sparing. Of the 91 patients who underwent CT imaging, 6 cases of pneumomediastinum, 11 of pleural effusion, and 2 of pneumothorax were present (in 15 patients). Two patients had both a pneumomediastinum and a pneumothorax, and two patients had both a pneumomediastinum and pleural effusion.

All but 5 patients were hospitalized (93 patients [95%]), and the median duration of hospital stay was 6 days (Table 2). Intensive care unit admission for respiratory failure was common (53% of all patients; 56% of hospitalized patients), and 26% of all patients underwent intubation and mechanical ventilation (27% of hospitalized patients). No patient received a tracheostomy.

A total of 25 case patients had documentation in clinical notes of having acute respiratory distress syndrome (ARDS). Of these 25 patients, the investigative team was able to independently verify that 12 patients (48%) met the Berlin Criteria for ARDS,¹⁶ with an average index of partial pressure of arterial oxygen (Pao₂) to fraction of inspired oxygen (Fio₂) of 189 mm Hg. For the remaining patients, medical records and documentation were insufficient to verify the diagnosis independently. Two patients underwent extracorporeal membrane oxygenation, and one of these patients died. For both patients, there was clinical documentation that the ARDS criteria were met.

Most patients received systemic glucocorticoids (intravenous or oral) during admission (84% of the patients overall; of these, 81% received intravenous administration). Documentation by the clinical team that the respiratory improvement was due to the use of glucocorticoids was found in approximately half the patient notes (51%). All but five patients who began receiving systemic glucocorticoids were treated with at least 7 days of glucocorticoid therapy.

All patients had a history of use of e-cigarettes and related products within the 90 days before symptom onset, and 92% of those with data (37 of 40 patients) regarding the date of last use reported vaping in the week before symptom onset. Most patients (69 of 78 patients [88%]) reported at least daily e-cigarette use. Of the 81 patients who were extensively interviewed, 73% reported use of nicotine products and 89% reported use of THC products; of the 78 patients who answered the question about CBD products, 9% reported use of CBD products (Table 2). A total of 27% of the patients reported using THC products only, whereas 11% reported using nicotine-containing products only. A total of 60% of the patients reported using both nicotine and THC products. Patients reported using 27 distinct brands of THC products and 25 brands of nicotine products in a wide range of flavors. The most common THC product that was reported was marketed under the “Dank Vape” label (reported by 49 of 73 interviewed patients [67%]). Patients reported use of a number of different e-cigarette devices to aerosolize these products. Of the 72 patients who were extensively interviewed, 12 reported smoking combustible cigarettes as well.

Figure 1 shows the clinical course of five patients who met the confirmed case definition and who underwent intubation and mechanical ventilation. Representative of the case series in general, most of the patients had previous outpatient visits before admission and ultimately began receiving systemic glucocorticoids.

A 17-year-old male patient with no clinically significant medical history presented to a hospital ED with a 2-day history of shortness of breath, nonproductive cough, and generalized weakness as well as a 1-week history of fever before presentation, nausea, vomiting, abdominal pain, and diarrhea. He sought care with his primary care provider and at multiple EDs for his gastrointestinal symptoms during the week before his admission and was treated with intravenous fluids and given metronidazole, levofloxacin, and an antiemetic agent. His condition worsened, and new respiratory symptoms developed, so he presented again to the ED, where he was found to have hypoxemia with an oxygen saturation of 85% while he was breathing ambient air, tachycardia with a heart rate of 112 beats per minute, and a temperature of 37.9°C. A radiograph of the chest showed opacities in both lungs that were suggestive of infection or acute lung injury (Figure 2A). His laboratory results showed leukocytosis (white-cell count, 18,000 cells per cubic millimeter) with a neutrophil predominance (94%) and no eosinophils (0%), as well as elevated inflammatory markers with a C-reactive protein level of 32 mg per deciliter and an erythrocyte sedimentation rate of 68 mm per hour.

The patient began receiving amoxicillin and azithromycin as empirical therapy for community-acquired and atypical pneumonia and was admitted to the hospital while he was receiving 3 liters of oxygen per minute through a nasal cannula. Within hours, he was transferred to the pediatric intensive care unit owing to respiratory deterioration. He was intubated and mechanically ventilated, receiving a high Fio₂ and positive end-expiratory pressure, and met the criteria for moderate ARDS. A radiograph of the chest that was obtained approximately 12 hours after presentation showed rapid worsening of diffuse lung opacities (Figure 2B), and a high-resolution CT image of the chest showed diffuse hazy ground-glass opacities with subpleural sparing, findings consistent with pneumonitis (Figure 2C through 2E). On day 2 of hospitalization, the patient underwent bronchoscopy, which showed normal-appearing bronchi. Cytologic testing of bronchoalveolar-lavage specimens showed a neutrophil predominance (78%) with no eosinophils (0%) and a moderate number of lipid-laden macrophages on oil red O staining. Infectious workup, including blood cultures, testing for the human immunodeficiency virus (HIV), testing for urinary histoplasma and blastomyces antigens, polymerase-chain-reaction (PCR) panel for nasopharyngeal respiratory virus, PCR panel for enteric pathogens, and bronchoalveolar-lavage studies including bacterial and fungal cultures and pneumocystis stain, was negative. He began receiving high-dose intravenous glucocorticoids on day 2 of the hospitalization, and antibiotics were discontinued on day 4 because an infectious cause was deemed to be unlikely. The patient’s clinical condition improved, and he was extubated after receiving intravenous glucocorticoids for 3 days. The patient was discharged home on hospital day 6 with instructions to continue an oral glucocorticoid–tapering regimen for 6 weeks.

The mean monthly rate of visits to the ED for severe respiratory illness as identified by syndromic surveillance between June 1 and August 15, 2019, was twice the mean monthly rate that occurred between June 1 and August 15, 2018 (7.4 cases per 10,000 visits vs. 3.8 cases per 10,000 visits), in Illinois counties. This difference was significant for both male and female patients (P<0.05 for both comparisons; P<0.001 for the combined comparison) (Figure 3).