In a sample of over 300 subjects who report a history of physician-diagnosed asthma recruited directly from the community, we found that 27% had a negative methacholine challenge test, a negative test frequency that is within the 25-40% range noted by others [6
]. Compared to the methacholine positive subgroup, the clinical characteristics of the methacholine negative subgroup included an older age of asthma onset and better lung function.
One interpretation of our data is that the methacholine negative subgroup had false negative results. For example, methacholine challenge is acknowledged as a highly sensitive test for asthma, but it can be negative in subjects who are not having active symptoms, in subjects who have normal lung function, or in elite athletes with exercise-induced asthma [10
]. In particular, it is well recognized that athletes with good lung function can have exercise-induced asthma and negative methacholine tests [13
]. Although the subgroup here with a negative methacholine challenge test was characterized by well-preserved lung function, they were not characterized as having asthma in remission, because their histories revealed significant symptoms and health care utilization. In addition, methacholine challenge sensitivity can be reduced by using forced maneuvers to measure airflow (as we did), since the deep breath can reverse bronchoconstriction [14
]. It is possible, therefore, that subjects with negative methacholine challenge in our study could have had a positive test result using a different methacholine protocol or that they could have had a positive result with an indirect bronchoprovocation challenge with exercise, eucapnic voluntary hyperpnea, or adenosine monophosphate. It is also possible that adult onset asthma is characterized by lack of methacholine responsiveness; certainly, children (who also have short asthma durations) can have normal lung function, have exercise induced bronchospasm, and yet have negative methacholine tests [10
]. All of these possibilities mean that it is not possible to fully exclude a false negative result for asthma in the subgroup identified here as methacholine negative. It has recently been reported that methacholine challenge testing does not have a high negative predictive value for a clinical diagnosis of asthma [12
]. The study on which this conclusion was based required subjects to have signs and symptoms suggestive of asthma but no firm diagnosis of asthma. These unusual entry criteria were necessary to optimize the protocol to determine the usefulness of mannitol as a bronchoprovocation test in this kind of patient population. Only 42% of the enrolled subjects were positive for methacholine, a percentage far lower than in other studies that had less unusual enrollment criteria. In addition, the average FEV1
was over 90% and less than 10% of subjects had beta-agonist reversibility. So although clinicians thought it likely that these subjects had asthma, the normal lung function and normal methacholine results suggest some of the clinician-based diagnoses of asthma were incorrect and that the negative methacholine test was a true negative for the diagnosis of asthma.
Although we acknowledge the possibility of a false negative methacholine test for the diagnosis of asthma in a subgroup of our study population, we consider it very likely that a large fraction of the negative methacholine subgroup did not have asthma. This interpretation is based on the findings that the methacholine negative subgroup were characterized by a high frequency of asthma diagnosis as an adult and by normal or near normal spirometry. As discussed above, it is possible that adult onset asthma is a peculiar asthma phenotype that is less likely to be characterized by methacholine responsiveness. But we consider it more likely that adult onset asthma is a subgroup more likely to be misdiagnosed. Similarly, although spirometry can be normal in stable asthma (and it was indeed normal in over one third of subjects in the methacholine positive subgroup), and methacholine challenge tests are more likely to be negative when lung function is normal [16
], the fact that 72% of the methacholine negative subgroup had normal spirometry is unusual and suggests an incorrect diagnosis of asthma in many of these subjects.
Dysanapsis is a term used to describe disproportionate airway growth compared to lung parenchyma growth and dysnapsis has been proposed as a determinant of bronchial hyperresponsiveness [17
]. Since direct measures of airway and lung size are not feasible in vivo
, the ratio of FEF25-75
to FVC is used as a surrogate [20
]. Although a low ratio is known to predict airway hyperresponsiveness [18
], it is not the main determinant, accounting for just 7.6% of variability in airway reactivity in a previous study of 764 pediatric and adult subjects presenting to a clinical pulmonary function laboratory for methacholine testing [19
]. We found a similar finding here - the FEF25-75
to FVC ratio was significantly lower in the methacholine positive group than in the methacholine negative group but the overall relationship between FEF25-75
/FVC and methacholine reactivity was relatively weak.
Much of the literature focuses on missed opportunities to diagnosis and treat asthma, with little research devoted to indentifying reasons for over-diagnosis of asthma [21
]. This may be contributing to a trend towards diagnosing any persistent cough, wheeze, or dyspnea as asthma, in an effort to err on the side of caution. There is evidence that the diagnosis of asthma is often made in absence of any diagnostic studies. A Swedish study recently reported that of 499 individuals with a confirmed diagnosis of asthma, only 20% had ever received spirometric assessment [23
There was significant morbidity referable to airway disease in the subjects with a negative methacholine test. Many had airway symptoms and reported regular use of albuterol or ICS. More impressively, some reported emergency department visits or hospitalizations and close to a third had been prescribed oral steroids for an exacerbation of asthma-like symptoms in the previous 2 years. It is possible that the subjects with the emergency room visits and hospitalizations represent the fraction of the methacholine negative subgroup that indeed had asthma. It is also possible, however, that non-asthmatic airway disease due to vocal cord dysfunction or to viral or bacterial bronchitis was the correct diagnosis in these patients.
Our study had several limitations. For example, while participants had to have a physician-diagnosis of asthma, this inclusion criterion was by self-report. We were not able to verify physician-diagnosis against the patient’s medical record nor were we able to determine the method by which the physician made the diagnosis. This could have allowed some subjects into the study who may have either received an unfounded diagnosis, or who may have incorrectly recalled a positive diagnosis of asthma. In addition, subjects were asked to hold SABA and LABA before methacholine testing, but they were not asked to hold ICS except when in combination preparations with LABA. There is some research to suggest that asthmatics with negative PC20
can have a positive PC20
when retested after holding ICS for 2-3 weeks [6
]. While this is not a consistent finding across the literature [24
], it is possible that not asking subjects to hold ICS could have inappropriately categorized some individuals, who may have otherwise had a positive PC20
. Given that 82% of the subjects with a negative PC20
were not prescribed an ICS and that the remaining 18% were tested using a higher PC20
threshold, it is only a small fraction of subjects who could have been vulnerable to this type of misclassification. Even so, adult onset asthma and FEV1
≥90% predicted remained significant predictors of negative PC20
when the model was restricted to steroid naïve subjects; further strengthening the validity of our findings. Using a PC20
of 8mg/mL as the cutoff for subjects not on ICS may have been a limiting factor in this study. ATS guidelines suggest that a PC20
between 4mg/mL and 16mg/mL indicates borderline airway hyperresponsiveness, whereas a PC20
less than 4mg/mL is evidence of airway hyperresponsiveness [5
]. Thus, there is a possibility that subjects with a negative PC20
would have subsequently had a positive PC20
if the 16mg/mL threshold was used for all subjects. However, the vast majority of those with a positive PC20
reacted to methacholine at or below the 4mg/mL dose, before the 8mg/mL cutoff was reached.
In a review of methacholine challenge tests in a large group of subjects who report physician-diagnosed asthma, we find a sizeable percentage with negative tests. These subjects are characterized by a diagnosis of adult-onset asthma and by normal or near normal spirometry. Based on these findings, we urge caution in the assessment and diagnosis of adults presenting with asthma-like symptoms, because they may not have asthma. Methacholine bronchoprovocation testing should be considered in this patient subgroup, especially if their spirometry is normal.
Inappropriately diagnosing individuals with asthma is not without consequences. Those individuals with an improper diagnosis are burdened with the cost and exposed to the side effects of medications they may not need. Providing asthma-related care to individuals who do not have asthma is a misuse of resources and diminishes the opportunity to examine the underlying pathology behind the individual’s asthma-like symptoms, which may increase future risk associated with untreated disease. In addition to the implications for individual patient outcomes, misdiagnosing asthma has considerable implications on the validity and reliability of asthma research.