Perioperative Respiratory Complications in Patients with Asthma

This investigation was approved by the Mayo Institutional Review Board. The study population was originally drawn from all residents of Rochester, Minnesota, in whom the initial diagnosis of asthma had been made from 1 January 1964 through 31 December 1983. This cohort has been the subject of other investigations, and details regarding the methods used in its identification have been reported. [2,7,8]To summarize, diagnostic indexing data from the Rochester Epidemiology Project were used to identify the medical records of Rochester residents with asthma or related diagnoses. These records were then carefully reviewed by trained nurse abstractors according to defined diagnostic criteria to determine if the resident qualified as an incidence case of asthma. Patients were classified as having a single wheezing episode, probable asthma, or definite asthma based on the criteria shown in Table 1.

As a result of this process, 1,547 incidence cases of definite asthma were identified. Using the Mayo Surgical Index and Surgical Information Retrieval System, which contain information on every surgical procedure performed at Mayo Rochester (in one of two hospitals), it was determined that 1,036 of these patients had undergone at least one surgical procedure at Mayo Rochester between the index date of their diagnosis and October 1994. Surgical procedures performed at the other medical facility in Rochester, Minnesota (Olmsted Community Hospital) were not studied. The medical record of each of these patients was reviewed to determine if the patient had received a general anesthetic or central neuroaxis block (epidural or subarachnoid block) for at least one of these surgical procedures. According to this criterion, surgical episodes involving sedation or peripheral nerve block were not analyzed. The first surgical procedure meeting these criteria after the index date of asthma diagnosis was selected for further review. Thus only one surgical episode was examined for each patient.

The perioperative medical, surgical, nursing, and anesthesia records for each surgical episode were reviewed by trained abstractors. Preoperative information recorded included preoperative use of bronchodilators and corticosteroids (defined as antiasthma medications), smoking history, American Society of Anesthesiology (ASA) physical status (a preoperative assessment scale from 1 to 5, with 1 representing healthy patients and 5 representing patients with severe morbid disease), and recent history of respiratory symptoms, including bronchospasm, respiratory infections, and therapy in any medical facility for asthma symptoms. Intraoperative information included the type and duration of anesthesia, induction and primary anesthetic agent, and the use of tracheal intubation. Measures of outcome in the intraoperative period (up to admission to the postanesthesia recovery unit) included the use of bronchodilators (other than as part of the patient's usual regimen), notation in the anesthesia record of signs or symptoms of bronchospasm (wheezing, increased airway pressure during positive-pressure ventilation, or prolonged expiratory phase), notation of laryngospasm, pneumothorax as documented by subsequent chest roentgenograph or thoracocentesis, hemoglobin oxygen saturation by pulse oximetry or arterial blood gases less than 90% (when available), and death. Measures of outcome in the postoperative period included the notation of bronchospasm, the use of bronchodilators (other than the resumption of the patient's preoperative regimen), arterial hemoglobin oxygen saturation less than 90% by pulse oximetry or arterial blood gases, the need for postoperative mechanical ventilation of the lungs for respiratory insufficiency (i.e., unanticipated mechanical ventilation), oral temperature > 38.3 degrees Celsius lasting longer than 12 hours, other respiratory outcomes such as pneumonia, and hospital death.

This was a nested case-control study restricted to the subpopulation of persons with asthma undergoing surgery at the Mayo Clinic in Rochester. The cases were patients who experienced at least one outcome, whereas the controls were patients who did not experience an outcome. Potential factors contributing to complications were assessed by comparing cases and controls using Fisher's exact test for categorical variables and the rank-sum test for continuous variables. Exact confidence intervals for complications frequencies were used as appropriate. [9]In all cases, tests were two sided with P less or equal to 0.05 considered statistically significant.

Of the 1,547 residents of Rochester, Minnesota, who were newly diagnosed with definite asthma between 1 January 1964 and 31 December 1983, 706 patients received a surgical procedure at the Mayo Clinic involving a general anesthetic or central neuroaxis block subsequent to the date of diagnosis. The median age at asthma diagnosis was 8.2 yr (2.3, 28.9 interquartile values for the 25th and 75th percentiles, respectively), and the median age at the time of surgery was 18.3 yr (9.9, 36.4 interquartile values) (Figure 1). The median time from diagnosis to surgery was 6 yr (2.7, 11.6 interquartile values). Fifty-four percent (383 patients) had experienced asthma symptoms and 41% (289 patients) had received at least one prescription antiasthma medication during the 1-yr interval before surgery. Sixty-one patients (8.6%) had received corticosteroid therapy for asthma within the 30 days before surgery.

Bronchospasm developed in 12 patients (1.7% [exact 95% CI, 0.9, 3%]) in the perioperative period (Table 2). In four patients, the episode of bronchospasm began during anesthesia. It persisted into the postoperative period in one of these patients. In the other eight patients, the onset of bronchospasm occurred after operation. All of these 12 patients received additional bronchodilators; among other patients, no use of additional bronchodilators was noted. Laryngospasm developed in two additional patients during operation. All episodes of bronchospasm and laryngospasm in the immediate perioperative period were successfully treated. No episodes of pneumothorax, prolonged postoperative fever, unanticipated mechanical ventilation, pneumonia, or death in the hospital were noted. Thus a perioperative respiratory complication developed in a total of 14 patients (2% [95% CI, 1.1, 3.3%]). In 13 patients, these complications did not cause significant morbidity. Respiratory failure, probably caused by refractory bronchospasm, developed on the second postoperative day in one 65-yr-old patient who had emergency repair of a ruptured abdominal aortic aneurysm. This patient had also experienced bronchospasm during and immediately after operation, which had been initially successfully treated. This patient eventually recovered and was discharged from the hospital.

Univariate analysis was performed to identify patient characteristics associated with complications (Table 3). Characteristics not associated with an increased frequency of complications included patient sex, smoking history, emergency surgery, and the presence of symptoms of an upper respiratory infection at the time of surgery. Characteristics associated with an increased frequency of complications included higher ASA physical status, recent use of antiasthmatic drugs, recent asthma symptoms, and recent therapy in a medical facility for asthma symptoms. Thus complications were more common in patients with more active disease, as defined by these latter three characteristics. The frequency of complications did not depend on the type of anesthesia; complications occurred in 12 of 628 (1.9%) patients receiving general anesthesia and 2 of 78 (2.6%) patients receiving regional anesthesia. In those patients receiving general anesthesia, complications occurred more frequently in patients in whom the trachea was intubated compared with those in whom it was not (12 of 462 vs. 0 of 166, P = 0.013). There were no significant differences in complication frequencies among anesthetic agents used to produce and maintain general anesthesia (data not shown); a volatile anesthetic was used as the primary maintenance agent in 95% of patients receiving general anesthesia.

The characteristics of patients who did or did not develop complications were compared (Table 4). The mean time from asthma diagnosis to surgery did not differ between the two groups. Patients in whom complications developed were significantly older, both at the time of asthma diagnosis and the time of surgery. The durations of anesthesia and hospitalization were higher in patients who experienced complications.

The major finding of this study is that the frequency of perioperative bronchospasm and laryngospasm in this cohort of persons with asthma was surprisingly low. These events did not lead to severe morbidity in most patients.

Several retrospective reviews of the perioperative medical records of asthmatics are available. Shnider and Papper [5]and Gold and Helrich [4]both reported that intraoperative bronchospasm developed in approximately 6% of the patients with asthma who were reviewed. The latter study found that 11% of patients experienced a postoperative respiratory complication. Two more recent reviews of patients with asthma who received preoperative corticosteriod therapy reported a similar frequency of perioperative bronchospasm, [10,11]as has a series from Japan. [12]In a recent preliminary report, Vener and associates [3]found that 30% of 206 children with asthma suffered a perioperative respiratory complication, including a 23% frequency of bronchospasm. Postoperative respiratory complications developed in 7% of their patients. The frequency of complications did not depend on the severity of asthma symptoms or the chronic use of bronchodilators before operation. In a review of approximately 156,000 anesthetics based on quality-assurance data, Olsson [13]reported a 0.80% frequency of intraoperative bronchospasm in 3,210 patients with asthma and a 0.16% frequency of bronchospasm in patients without asthma.

Two prospective studies of anesthetic complications that included patients with asthma have been performed. In the Multicenter Study of General Anesthesia, [14]a prospective, randomized clinical trial of four general anesthetic agents that included assessment of perioperative outcomes, 486 patients reported a history of asthma. Of these persons, 1.7% experienced a severe respiratory outcome, with a 0.81% frequency of bronchospasm. Among patients without asthma (n = 16,535), 0.94% experienced severe respiratory outcomes and bronchospasm developed in 0.13%. In a univariate analysis, these frequencies differed between persons with or without asthma for bronchospasm, but not for severe respiratory outcomes. However, these researchers found that a history of asthma was not predictive of a severe respiratory outcome or bronchospasm in logistic regression analysis. Pizov and colleagues [15]found that 25% of 56 patients with asthma wheezed after the intravenous induction of anesthesia, compared with 6% of 96 persons without asthma. This study was unique in that a blinded observer auscultated the chest to determine the presence of wheezing, whereas other studies (including the present one) relied on notation in the anesthetic record, presumably a less-sensitive method. They did not report postoperative outcomes.

Comparison of our results with those of these other studies is complicated by several factors. Retrospective studies have well-recognized limitations. A primary concern is the sensitivity and specificity of data notation. It is assumed that a clinically significant outcome, such as marked bronchospasm, would have been noted in the patient record. However, it is likely that less-severe events, such as mild, transient wheezing, may not have been noted, so ours may be a conservative estimate of the true frequency of such complications. This may account for the higher frequency of wheezing noted by Pizov and colleagues, [15]who collected data prospectively, and may illustrate a limitation of our retrospective design. On the other hand, the sensitivity of data notation may have been increased in patients perceived as having increased risk (e.g., patients recently hospitalized for asthma). Furthermore, both asthma therapy and anesthetic practice have changed considerably since earlier studies, and both factors may modify the frequency of perioperative complications.

One critically important factor in comparisons among studies is the patient populations examined. In many studies the diagnostic criteria for asthma are not specified; rather patients were identified by a notation of "asthma" in the patient record or by patient report. Such a definition of asthma may be inadequate. [6]In a report of the methods used to identify the population that formed the basis for our study, [2]approximately 15% of cases of definite asthma were not coded as having a form of "asthma" on the patient record. Furthermore, approximately 50% of those patients noted on the medical record as having a form of "asthma" did not meet the diagnostic criteria for definite asthma. We noted during our review of records that some patients previously diagnosed as having asthma had no notation of this diagnosis on preanesthetic examination, especially those patients with inactive disease. Thus, by using only diagnoses coded on the medical record, it is possible that patients may be incorrectly assigned to asthmatic and nonasthmatic groups. This factor may have biased the asthmatic groups of previous studies toward including more severe asthmatics.

Comparisons among studies must also account for variations in severity of disease. In some patients, especially children, the symptoms of asthma may diminish or disappear over time. We found that patients without recent symptoms of asthma had a very low frequency of complications. Indeed, laryngospasm, not bronchospasm, developed in our two asymptomatic patients experiencing perioperative complications. Laryngospasm may or may not be related to reactive disease of the lower airways. The absence of bronchospasm in our asymptomatic patients (95% confidence interval, 0.0, 1.2%) is consistent with frequencies reported by Forrest and coworkers [14]and by Olsson [13]for a nonasthmatic population (0.13% and 0.16%, respectively). Thus the frequency of bronchospasm in persons with asthma but no recent symptoms may approach that of the nonasthmatic population. Conversely, patients with factors indicative of active disease experienced a higher frequency of complications. In contrast, neither the studies of Gold and Helrich [4]or Vener and associates [3]found an association between the severity of asthma symptoms and the frequency of complications, although it is not clear whether these studies included asthmatic persons with no symptoms.

Because all studies (including this one) have been performed in tertiary care centers, referral bias is also possible, such that the patients with asthma studied may not be representative of a more general population of asthmatics having surgery. They may have more severe disease, may have more comorbid conditions, or may undergo more extensive surgery. This factor may explain the high frequencies of complications noted in some series. [3]We tried to minimize such biases by including only patients who were incidence cases of asthma in the population of Rochester, Minnesota. Our findings do not represent the incidence of perioperative complications in all asthmatic residents of Rochester, Minnesota, because (1) prevalence cases of asthma (those diagnosed before 1964) were not included, (2) the residency of patients in Rochester at the time of their surgery was not a study requirement, and (3) although approximately 95% of the residents of Olmsted County, Minnesota (in which Rochester is located) receive their surgical care at the Mayo Clinic, Rochester (unpublished data), there are other sources of care, and these surgical episodes were not reviewed. Because patients with more severe disease would presumably be referred to Mayo Rochester for their surgical care, we assume that our results represent the upper limit of complication frequencies in this population.

The degree to which the Rochester asthma cohort is representative of the general asthmatic population has been discussed. [2,7,8]To summarize, although the incidence and severity of asthma in the Rochester population is similar to that found in other studies, there may be considerable geographic variability among populations, [16,17]and complications may be more frequent in some of these other populations. Similarly, the general level of medical and surgical care is good in the Rochester population, which may have also contributed to our low observed frequency of complications.

Another factor associated with an increased frequency of perioperative complications was more advanced age, as demonstrated by the higher mean age of patients in whom complications developed (Figure 1). Of the 211 patient younger than 12 yr at surgery, bronchospasm developed in none of them. The number of complications in our series was insufficient to perform meaningful multivariate analysis to determine if there were multiple independent risk factors. However, the median age of patients with and without asthma symptoms within 1 yr before surgery was identical (18 yr), suggesting that the more advanced age of patients who experienced complications could not be explained by age-related differences in the proportion of patients with active disease. Factors responsible for the apparent increase in complication frequency in older patients are unknown.

Even if bronchospasm occurs relatively infrequently, it may be clinically important if it often leads to severe outcomes. Cheney and associates [18]reported from the American Society of Anesthesiologists Closed Claims Project that of 40 cases of bronchospasm resulting in settled malpractice claims, 35 (88%) involved brain damage or death. Only approximately one half of these 40 patients had a history of asthma, chronic obstructive pulmonary disease, or smoking. Thus significant bronchospasm may occur in patients with no history of reactive airways disease. These data do not permit calculation of complication frequencies, nor do they provide information regarding what proportion of episodes of bronchospasm lead to such severe outcomes. In our series, most cases of bronchospasm did not produce severe morbidity. The single exception was a critically ill patient (ASA physical status 5) requiring emergent repair of a ruptured abdominal aortic aneurysm, in whom postoperative respiratory failure developed but who was eventually discharged from the hospital. In contrast to the findings of Vener and associates, [3]who found a 5.8% frequency of major respiratory complications, such as severe respiratory acidosis and respiratory failure, in asthmatic children having surgery, a major respiratory complication developed in no patient younger than 18 yr in our study.

Our study was not designed to discern a causal relationship between such factors as tracheal intubation, anesthetic agent, and anesthesia type, and the subsequent occurrence of complications. For example, the status of asthma may have influenced the decision to intubate, whereas the lack of intubation may have reflected a shorter or less-involved surgical procedure. It also cannot be determined if there was a causal relationship between the occurrence of complications and increased duration of hospitalization, or whether other factors such as age and variations in the acuity of the surgical procedure may have contributed. Similarly, although the lack of association between the presence of upper respiratory infection and complications is consistent with some previous work in a general population, [19]our data do not exclude the presence of such symptoms as a risk factor for complications. Finally, because the presence of asthma may have caused anesthesia providers to alter care in an attempt to reduce risk (by avoiding tracheal intubation or by the prophylactic use of bronchodilators, for example), it is not known whether the relatively low frequency of complications may reflect these precautions or whether patients with asthma would still be at low risk even in the absence of such precautions.

Recommendations urge physicians to prescribe routine preoperative corticosteroid therapy to all persons with asthma to reduce perioperative risk. [10,11]Although the design of our study is not appropriate to specifically evaluate this recommendation, we note that of the 645 patients with asthma who did not receive corticosteroids within the 30 days preceding surgery, only seven (1.1%) experienced a complication. Thus the efficacy of such routine preoperative corticosteroid therapy for persons with asthma remains to be established.

These results suggest that most persons with asthma can safely undergo anesthesia with modern techniques. The frequency of complications was particularly low in patients with inactive asthma. Even when complications occurred, events such as bronchospasm were rarely associated with serious morbidity.

The authors thank Maria Lindahl, Joan LaPlante, and Pam Maxson for assistance with data abstraction, Pam Schmidt for computer programming, and Cathy Nelson and Janet Beckman for secretarial support.