Our study focused on clinicians practicing in a tertiary medical center in Southeast United States, a region near Atlantic Ocean with high humidity favoring mold growth. The prevalence of cHP in this region is unknown. The prevalence of HP varies considerably depending upon case definitions, intensity of exposure to inciting antigens, climate condition, local practice patterns and host risk factors [15-18]. Using a large insurance database (150 million subjects), Fernández Pérez et al. estimated the one-year prevalence rate of cHP ranged from 1.67 to 2.71 per 100,000 persons [17]. In our study, more than half of the 261 patients were in the Research Triangle area that had an average population of approximately 2 million during the study period (2008-2013). This would give a prevalence rate of about 6.5 per 100,000 persons. The higher humidity in the southeast regions of the United States that promotes mold growth is likely an important factor because the main exposure source in our cohort was environmental molds. The 261 cases over 6 years in our study (43.5/year) also represented a greater case encounter than that of the reported cohorts. For example, the Japanese cohort that was compiled by questionnaire had 222 patients over 10 years (22.2/year) [11]. The Denver cohort that included only patients with pathology had 142 patients over 27 years (5.3/year) [4]. The higher case encounter rate in our medical center could be in part due to the referral bias although the higher prevalence of cHP is also contributory. Clinicians practicing in regions that have high prevalence of environmental mold growth should have high suspicion of cHP when evaluating patients with interstitial lung diseases.
Our study found that the most common criterion used by the clinicians in the diagnosis of cHP was environmental exposure (74.5% of the cohort). The importance of exposure history in the diagnosis of cHP has been repeatedly demonstrated in the literature [9, 19, 20]. In the studies by Johannson et al. and Salisbury et al., exposure history was one of the two most common criteria for the diagnosis of cHP [19, 20]. A multitude of causative agents can be found in the workplace and home environment [19]. Identifying a clear inciting antigen based on clinical history with a definitive timeline of exposure preceding symptoms helps the clinicians diagnose cHP. Removal of the exposure is also considered the cornerstone of cHP diagnosis [21]. When the exposure source can be pinpointed, such as birds, hay, antigen avoidance by removal from exposure or wearing protective devices tends to be more effective in halting or reversing the disease progression [4]. An extensive search, however, may not reveal a clear source because the latency between exposure and onset of disease varies widely from months to years and occult or low-level persistent exposure to unknown source makes it challenging to discern the type of antigen [21]. If inciting agents cannot be identified by clinical history or laboratory tests, like in many cases of fibrotic HP, the diagnostic confidence may decrease and clinicians may more likely resort to lung biopsy. The inability to identify an inciting antigen was independently associated with shortened survival even after controlling for important variables such as the presence of pulmonary fibrosis (4).
The most characteristic features are air-trapping or the mosaic attenuation predominantly in the upper lobes. There may also be non-specific findings including airway-centric disease, centrilobular nodules and ground glass opacities. Imaging may also overlap with radiologic features of other ILD such as linear densities, honeycombing that make it challenging for clinicians to diagnose cHP [22-24]. It is important to distinguish cHP from other ILD, such as idiopathic pulmonary fibrosis IPF as the non-fibrotic HP is reported to have a better prognosis [3, 8]. Having a more defined pattern for radiologists to discern cHP from other forms of ILD may also enhance the specificity of HRCT. Salisbury et al. derived and validated a diagnostic model for cHP based solely on radiologic findings when the extent of mosaic attenuation or air trapping is greater than reticulation and the disease has diffuse axial distribution with a specificity <90% [25].
CT scan is considered the most useful non-invasive tool in the diagnosis of cHP [19, 20]. The most characteristic features are air-trapping or the mosaic attenuation predominantly in the upper lobes. There may also be non-specific findings including airway-centric disease, centrilobular nodules and ground glass opacities. Imaging may also overlap with radiologic features of other ILD such as linear densities, honeycombing that make it challenging for clinicians to diagnose cHP [22-24]. It should be noted that not all patients had CT reports in our system when our specialists made the diagnosis of HP. Conceivably most if not all patients should have CT chest reports before the referral. The specialists may have seen them, but we did not have access to these outside reports.
In our study, supportive CT findings were only present in 31.6% of the patients, and almost half of these patients also had undergone biopsy. The majority of these patients (92) had VATS, 13 patients had bronchoscopy and 15 patients had both (Table 1). In the Delphi survey, bronchoscopy with bronchoalveolar lavage (BAL) and transbronchial biopsy was recommended to increase diagnostic confidence [9]. BAL as the diagnostic test is poorly characterized in patients with fibrotic cHP and a wide range conditions can be associated with lymphocytosis [26-28]. Adams et al. also demonstrated the combination of transbronchial biopsy and BAL increases the likelihood that the procedure will give adequate information to allow a confident diagnosis of cHP as well as possibly reducing the need for more invasive surgical lung biopsy [28]. However, this study also showed that the diagnostic yield of BAL was low [28]. Therefore, while not diagnostic on its own, BAL can helpful in the diagnosis of HP when combined with other clinical findings. During the study period, clinicians in our institution did not rely on BAL or transbronchial biopsy to diagnose cHP. The findings in our study that pathology from VATS biopsy was used more frequently than bronchoscopy to diagnose cHP reflected this practice pattern. More robust bronchoscopy data with better discriminant power would help decrease regional variations in clinical practice.
It is also notable that 6.9% of patients who were given the diagnosis of cHP did not have any of the three criteria. Among these 18 patients, 50% were diagnosed based on responsiveness to steroids, 66.7% were diagnosed based on nonspecific imaging findings without mosaic attenuation, and 16.7% were diagnosed based on eosinophilia (Table 3). The first two items (steroid responsiveness and nonspecific imaging) did not meet consensus in the modified Delphi survey [9]. Allergic manifestations, such as wheezes, reached unimportant threshold [9]. Another finding in our study was almost 50% of the patients who carried an ICD-9 code of cHP (495) actually did not have cHP as evaluated by pulmonary or allergy specialists. Among these patients, 29% had no underlying lung disease. For those patients who did have lung disease, asthma is most common (18.5%) followed by non-HP ILD (16.5%), COPD (12.5%) and pneumonia (7.5%) (Table 4). The causes for the unusually high percentage of miscoding are unclear but may be related in part to the unfamiliarity of clinicians in the diagnostic criteria of cHP and/or ICD coding [29, 30].