INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There has been concern in recent years regarding rising rates of asthma morbidity and mortality (1). The reasons for this are unclear. Case-control studies using retrospectively collected data have suggested that excess use of short-acting -agonists (4, 5) and underuse of inhaled corticosteroids (6) may be associated with this phenomenon, although more recent data suggest that patterns of -agonist use may be a marker of more severe asthma rather than a causal factor (6). Recently, near-fatal asthma (NFA) has been used as a surrogate marker of patients at risk of fatal asthma (5). Because of retrospective data collection from chart review in previous studies, the types and numbers of risk factors that could be evaluated were limited. We prospectively enrolled asthma patients admitted to our hospital to more comprehensively assess risk factors for severe life-threatening asthma. We used hospitalized patients with asthma as a control group, to better evaluate possible risk factors for NFA rather than merely risk factors for hospitalization for severe asthma. The primary risk factors for NFA to be evaluated in this study were: (1) relative overuse of -agonists; (2) relative underuse of inhaled corticosteroids; (3) a prior history of mechanical ventilation; and (4) poor perception of bronchoconstriction. Our a priori hypotheses included the following: (1) NFA patients were more likely to be taking lower doses of inhaled corticosteroids and using more -agonists; (2) a prior history of NFA would be a strong predictor of a current episode of respiratory failure; (3) NFA patients would more likely be atopic; and (4) NFA patients would be more likely to have impaired perception of dyspnea. We also evaluated patient responses to deteriorating asthma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We prospectively identified all patients admitted to the Vancouver General Hospital with a primary diagnosis of acute asthma, as well as patients admitted to St. Paul's Hospital with NFA for 19 mo from 1991 to 1993. Patients requiring mechanical ventilation or with a Pa_CO2  45 mm Hg met our a priori case definition for respiratory failure, and were classified as subjects with NFA. Patients between the ages of 18 and 55 yr who had a confirmed diagnosis of asthma were eligible.

Clinical and demographic data were obtained through the use of structured questionnaires during a personal interview. The epidemiologic data were obtained from responses to the European Health Survey Questionnaire (9). Data obtained included duration of asthma, previous history of NFA, duration of the current attack, and prescribed therapy. Compliance with prescribed therapy was assessed by a question about whether the patient was taking his or her prescribed medications. We recorded smoking status and occupational history, and precipitating factors for the current asthma attack. Arterial blood-gas measurements, chest radiography results, and electrocardiography results at the time of arrival in the emergency department (ED) were recorded from admission hospital records. If the patient was already endotracheally intubated, the first available arterial-blood-gas measurement was used. The results of lung-function testing performed in the recovery phase of hospitalization were recorded, and patients were offered the opportunity to return for a methacholine (MCh) inhalation challenge test after discharge. MCh testing was done with standard methodology (10). In addition to measuring FEV₁, after each doubling dose of MCh, subjects were asked to rate their sensation of dyspnea at each MCh dose, using the modified Borg scale (11), a validated instrument for assessing perception of dyspnea. Skin-prick tests were done with a panel of common allergens (Bencard Ltd.). A positive result was defined as a reaction of  3 mm to one or more allergens. All patients gave written consent to participate in the study, and the study protocol was approved by the University of British Columbia and the relevant hospital ethics committees.

Statistical Analysis

Subjects were grouped and analyzed as those who were mechanically ventilated or had a Pa_CO2  45 mm Hg (NFA cases) and those who were hospitalized with severe asthma without ventilatory failure (controls). Patients were only enrolled once during the study period. Data was analyzed using SPSS statistical software (SPSS Inc., Chicago, IL). Clinical characteristics of cases and controls were compared through Student's t test for continuous variables, and proportions were compared with chi-square tests. Where appropriate, we applied Bonferroni's correction for multiple comparisons. Risk factors for NFA were evaluated by logistic regression in univariate and multivariate analyses. NFA was the dependent variable, and a stepwise model was used to assess clinically relevant and statistically significant variables identified in the univariate analyses. We did not include a history of previous ICU admission or mechanical ventilation in the same multivariate models because of collinearity. All data are expressed as mean and SD unless otherwise specified. Statistical tests were two sided, and significance was accepted at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We enrolled 99 patients during the study period, 80 in the control group and 19 cases of NFA. Ten patients with NFA were mechanically ventilated. There were no deaths, although one patient with NFA suffered hypoxic brain injury. There were no significant differences in the case and control groups with respect to demographic variables, smoking status, duration of asthma, or presence of atopy (Table 1). Cases were more likely than controls to have had prior ICU admissions and prior mechanical ventilation (p < 0.001).

Clinical characteristics, chest-radiographic findings, arterial-blood-gas measurements, and medication use by the study subjects are presented in Table 2. Inhaled -agonists were used by 95% (76 of 80) of controls and by all cases before admission. Salbutamol was used by 95% (72 of 76) of controls and by 90% (17 of 19) of cases. Fenoterol was used by only three patients (two controls and one case). Cases were more likely than controls to be prescribed prednisone at the time of admission and to have had prescriptions for theophylline (53% of cases versus 27% controls, p < 0.05). Oral corticosteroids had been used more frequently in the past by cases (95%, versus 71% of controls, p < 0.05). Of patients who had used oral corticosteroids previously, 22% had had one treatment course, 24% had had between two and four courses, and 54% had been treated five or more times.

There was a significantly greater prevalence of hyperinflation on the chest radiograph for NFA patients. Spirometry done at admission and before hospital discharge, and follow-up PC₂₀ MCh results, are shown in Table 3. Although the PC₂₀ for MCh did not differ for cases and controls, there was a significant difference for perception of dyspnea between male and female cases (p < 0.05). Male cases had less perception of dyspnea (Borg score: 1.9 ± 1.4 at PC₂₀, compared to 3.9 ± 1.2) than did female cases. However, there was no overall gender difference between cases and controls in perception of dyspnea.

Duration of symptoms before hospital admission is shown in Figure 1. The majority of subjects had worsening asthma symptoms for more than 48 h (85% of cases and 82% of controls), and more than 50% of subjects in each group had worsening asthma for longer than 7 d. There was no difference in the occurrence of specific symptoms such as cough, wheezing, and chest tightness. Control subjects were more likely to report a sensation of a sour taste in the mouth at night (p = 0.05). Less than one third (31%) of all patients had a history of hospital admission within the year before the study admission, although 57% had been to the ED with acute asthma at least once; 21% of patients had made three or more ED visits.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Duration of symptoms among cases and controls during the period prior to hospitalization.

Subjects' Responses to Worsening Asthma

Only 43% of patients contacted either the family physician (n = 40) or the ED for advice about worsening asthma (n = 3); 39 subsequently visited their family physician, and one attended an allergist. Control patients were more likely to visit their physicians before eventually being admitted to the hospital (46% versus 16% of cases; p = 0.02). The first contact with the health-care system was the ED for 60% of patients.

All patients who had had physician contact before hospital admissions were asked to recall the advice and instructions given to them. Ten patients (nine controls and one case) were advised to increase their bronchodilator use. Only 11 subjects (all controls) were instructed to increase their use of inhaled corticosteroids. Nine controls and one NFA case had a peak flow meter at home, but none was asked to make a measurement with it. One patient in the control group was instructed to call an ambulance, and of the 13 patients instructed to go to the ED, 12 were controls.

The usual daily dose of -agonist and inhaled corticosteroids, and the patient responses to exacerbations, are shown in Figures 2 and 3. Controls used larger doses of -agonists in the 24 h before admission (p = 0.040), Figure 2. There was a trend for cases to use less inhaled corticosteroid, but this was not statistically significant.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Mean daily dose of inhaled -agonists prescribed, dose usually taken, dose typically taken during asthma exacerbations, and dose taken for the 24 h prior to the study hospital admission. SEM = standard error of the mean; mcg = micrograms.

View larger version (15K): [in this window] [in a new window]   Figure 3.   Mean daily dose of inhaled corticosteroids prescribed, dose usually taken, dose typically taken during asthma exacerbations, and dose taken in the 2 wk prior to the study hospital admission. SEM = standard error of the mean; mcg = micrograms.

Electrocardiography Results

There were no life-threatening arrhythmias at the time of presentation. The most common electrocardiographic pattern was sinus tachycardia.

Multivariate Analyses

Multivariate analysis showed significant differences between groups (Table 4). The magnitude varied somewhat, depending upon the variables included in each model. A history of ICU admission, previous mechanical ventilation (OR: 99.2; 95% CI: 8.7 to ) and having air conditioning at home were significant risk factors. A history of prescriptions for prednisone or theophylline and a history of asthma being worse in January and February were associated with increased risk in the univariate analysis, but in the multivariate analysis only prior prescription for theophylline was significant (OR: 11.3; 95% CI: 1.1 to 116 in the model with a history of mechanical ventilation only). The compliance data and medication doses for inhaled -agonists and inhaled corticosteroids were not significant in either univariate or multivariate analyses.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The design of this case-control study of hospitalized asthma patients allowed us to more accurately identify risk factors for NFA, and to do so in a more standardized way than has previously been possible. Our findings confirmed results in previous studies (12, 13) that identified a history of prior mechanical ventilation and ICU admissions for severe asthma as strong predictors for NFA. These data emphasize the importance of history taking to assess asthma severity, and of risk stratification to aggressively manage patients at risk for NFA. Prescriptions for oral corticosteroids and theophylline were more common for NFA patients in our study. This is not surprising, since these medications are likely markers for more severe asthma (5, 14). We found that most patients had evidence of worsening asthma over a period of 2 to 7 d before hospitalization, rather than sudden rapid deteriorations. Therefore, for most asthma patients who are hospitalized, there is a window of opportunity to implement more aggressive therapy and follow-up with the objective of preventing progression to fatal or near-fatal events.

A potential criticism of any study such as this is always the selection of a control group. We selected hospitalized patients with asthma as the control group, to isolate the variables of interest: risk factors for NFA in severely asthmatic individuals. A control group of community asthma patients could have been another option. Because most of these patients would have mild asthma, and would never have been hospitalized, using a community control group would probably increase the observed magnitude of association for NFA risk factors. However, the ability to discern between risk factors for NFA and simple hospitalization for severe asthma would be possible only with a three-way comparison between groups. The significant historical and clinical characteristics that we identified with this study design also allow for their practical application when assessing risks for asthmatic patients in the ED. The prospective accrual of patients was advantageous for standardizing data collection and minimizing dependence on data extraction from chart review. The reliability of the information obtained from both cases (15) and controls is also improved, and allowed us to better determine the individual responses to a severe attack.

Although 89% of the cases were prescribed inhaled corticosteroids, the compliance with therapy with these medications was poor (29%), and was almost identical to that in the control group (28%). Cases were more likely to be prescribed higher doses of -agonists and lower doses of inhaled corticosteroids than the controls (Figures 2 and 3). Several case-control studies have implied (4, 16) that excessive use of -agonists is associated with NFA and asthma deaths, especially in the absence of inhaled corticosteroids (6). This risk appeared especially to be associated with a rapid increase in -agonist use (16). Our findings do not support this hypothesis, but show that increasing use of -agonists reflects increasing severity of asthma in the period prior to presentation with NFA.

The final common pathway leading to death in patients with asthma has been uncertain. Several studies have reported an association between the use of inhaled -agonists and asthma deaths (4, 16). It was hypothesized that administration of high doses of -agonists was associated with cardiotoxicity and hypokalemia, causing a predisposition to malignant cardiac arrhythmias (17). We confirm and extend Molfino's observations in his retrospective study (18) by showing no serious arrhythmias in patients admitted through the ED with NFA, and in particular, none in those requiring intubation. Molfino and colleagues studied relatively few patients, and there was no control group of severe asthma patients without respiratory failure. Our data reinforce and emphasize that asthma patients die of hypoxic respiratory failure. Therapy for acute severe asthma should focus on using supplemental oxygen, with aggressive bronchodilation with -agonists and early administration of corticosteroids (19).

Atopy is a risk factor for more severe asthma (20). In a report from the midwest of the United States, sensitization to the aeroallergen Alternaria alternata (21) was associated with respiratory failure in asthma. We found no difference in between our cases and controls in the degree of atopy. This lack of difference was consistent even when we examined absolute wheal size (data not shown).

Home exposure to air conditioning was a strong risk factor for NFA. This, to our knowledge, is the first time that such exposure has been identified as a risk factor. However, these data should be interpreted with caution, given the small number of subjects in our study. The observed association may be directly related to air conditioning or may be an indirect result of increased exposure to aeroallergens because air-conditioned homes are better sealed and insulated. A recent report from Bavaria showed that children in homes with wood-burning and coal stoves were less likely to be atopic than were children in homes using other heating sources, such as forced air (22). Cases in our study were more likely to report their asthma to be worse in the months of January and February. This finding also suggests that indoor exposures may contribute to poorer asthma control and to episodes of NFA. Patients with NFA reported more exposure to dusty conditions (data not shown), which again supports environmental exposures as important risk factors for NFA.

Our findings suggest a gender difference between male and female NFA cases in perceiving dyspnea. Although our numbers are small, this observation is supported by other work suggesting that males may be at greater risk because of precipitous declines in lung function (23). In a recent ED study, we showed that males are more likely to have severe airflow obstruction than are females at the time of presentation (FEV₁ % predicted: 49 ± 20% [mean ± SD] in females, versus 33 ± 15% in males; p < 0.001) (24). Kikuchi and colleagues (25) have identified differences in perception among NFA patients, non-NFA patients, and controls, but they did not report a gender difference. These observations have important implications for developing strategies to prevent deaths from acute asthma.

The overall reliance on crisis management in the ED for NFA patients (59.8% of patients in the present study), and the lack of specialist involvement in the outpatient care of these patients, are important features to emphasize. Controls showed a greater trend toward increased doses of medication than did cases in response to worsening asthma. However, the general lack of self-management skills in our patients, combined with suboptimal medical advice, is consistent with other reports (26, 27) that patients are not confident about changing therapy when asthma control deteriorates. Ideally, asthma education should address these needs and encourage a greater partnership of patients with their primary care physicians (28, 29). Unfortunately, we had little success in trying to educate the NFA patients identified in this study (30), and these results also highlight the need to develop new strategies capable of changing behavior (31).

A retrospective study in France has shown reduced survival in asthma patients with an NFA episode who were older than 40 yr of age, and also an independent association of NFA with smoking (32). Only 10 (52.6%) of our NFA patients were older than 40 yr, which may account for our inability to show age-related risk for this group. We also found a relatively high proportion of smokers in both of our study groups. Of interest in this regard is a recent report suggesting that current smoking may reduce the efficacy of inhaled corticosteroids in asthmatic individuals (33).

Not surprisingly, patients in our study who had more severe asthma (NFA cases) were more likely to show hyperinflation on their chest radiographs. It has been hypothesized that hyperinflation, as measured by increased lung volumes during the recovery phase, may be a marker for life-threatening asthma (1). However, these data await confirmation by prospective studies designed to test this hypothesis. The many clinical similarities between hospitalized patients and those who have had an episode of NFA confirms our experience in imaging the airways of these patients. We found no difference in airway caliber among patients with severe asthma and those with NFA, but when compared with nonhospitalized community-control asthmatic individuals (34), the airway caliber of NFA patients did differ.

In summary, our findings confirm and extend previous observations of the characteristics of patients with NFA. A history of mechanical ventilation or ICU admission for acute asthma is a strong risk factor for NFA. Patients with NFA do not present with life-threatening cardiac arrhythmias. We have shown that males with NFA, when compared with female cases, have a reduced perception of dyspnea. Although a number of guidelines designed to improve the management of asthma patients (35) have been published, there remains the challenge of implementing them (36) in high-risk patients, and especially of overcoming patient and physician barriers (28) to effective asthma education. Our findings clearly identify a failure of patients to respond to worsening asthma as a potential area for major physician and patient intervention.