Given the rarity of PAP, there are very few epidemiologic studies that have been performed on this patient population. To our knowledge, the largest independent cohort of patients examined remains a study by Inoue et al. that examined a total of 248 patients from a national registry in Japan15. Since then, there have been other, smaller cohorts looking at the demographics of this condition in Italy, Germany and China that have performed similar analyses of smaller groups20–22. From these studies, it appears that PAP affects males at a higher rate than females and has an association with smoking and possibly dust exposure.
It is difficult to draw a direct comparison between the data reported from our study and prior studies that have been performed by nature of the method of data collection. Given that these instances of hospitalization from the NIS do not represent individual patients, we are unable to use this data to determine the inherent incidence or prevalence of PAP in the United States23. It does, however, provide a large wealth of information regarding the nature of hospitalization in PAP patients, including demographic data surrounding hospitalized patients as well as the overall healthcare costs associated with the disease.
Based on previous cohorted studies from other countries, the median age of diagnosis of PAP patients ranged from 39 – 51 years8,15,20,21, whereas the mean age of hospitalized patients in our study was on the lower end at 41.45 years. One reason for this may be that patients who are at the beginning of their diagnosis tend to have symptoms requiring hospitalization and subsequent therapy, whereas older patients tend to have a lower likelihood of recurrences following therapy24. Given the description of a small number of patients with “spontaneous clinical remission” or entering a “quiescent phase” seen in some studies25 following initial diagnosis, it is also possible that the younger age of hospitalization compared to cohorted studies is consistent with the natural progression of disease. As reported in prior epidemiologic studies, we also found a predominance of hospitalized male PAP patients as compared to female PAP patients (ratio of 1.22). Of note, however, the ratio favoring men was much smaller than other prior studies have seen, which usually range from 2-2.6. Instead, our finding was more consistent with the study from Germany by Bonella et al., who reported a ratio of 1.3,21 as well as more recent US-based data5 suggesting even less of a male predominance in this condition with a M:F ratio of .95.
The socioeconomic breakdown of inpatient hospitalization had a relatively uniform distribution across income ranges. As noted above, the patients were equally distributed across income classes from < $40,000, $40-50,999, $51-59,999, and > $60,000. The majority of patients that were admitted with PAP were White (58%) which is slightly less than the proportion of the U.S population that self-identified as White (72%) around that time, according to the 2010 U.S Census26. Instead, disproportionately represented were Black or African American patients who made up 24.18% of the population hospitalized, which is higher than the proportion of Black or African Americans noted in the U.S Census (12.6%) at that time. However, this data comes with the caveat that we did not control for confounding factors in this analysis, as detailed clinical information such as smoking history, are not provided by the NIS.
Interestingly, 50% of the admissions for PAP were elective admissions – which suggests that many patients may have been hospitalized with the purpose of a planned procedure, such as a therapeutic whole lung lavage10,11,27. In our study, we found that whole lung lavage was performed 195 times, while bronchoalveolar lavages were performed 240 times throughout the three-year period. Although whole lung lavages are considered the definitive therapeutic intervention, only 79 institutions world-wide have been clearly identified that perform the procedure, and there is great variability in the execution of the procedure itself11,28,29. Our data suggests that bronchoalveolar lavages are performed more frequently, possibly due to waning experience with whole lung lavages. Of note, one study in China noted that BAL was interchangeably used with WLL to treat the condition22, and it is plausible that other less experienced centers have used this strategy as well. Unfortunately, given the limited nature of our database, we are unable to comment on the indications for which the BALs were performed.
There has been one study performed by McCarthy et al. in 2018 that similarly used a large-scale insurance claims database that looked at approximately 5% of the US population and found an average of 109 patients yearly with the diagnosis of pulmonary alveolar proteinosis between 2008 and 20125 within their database. This study found the annual per patient healthcare costs was found to be $54,865 in this time period, and adjusted for inflation to 2014, comes to approximately $56,571.69. This is significantly higher than the healthcare cost that we calculated from the NIS database, which was approximately $29,932. There is a multitude of explanations for this discrepancy in data; the initial being that while McCarthy’s data looked at total healthcare costs over a year, whereas the NIS data only includes the cost of a hospitalization. Not included in the $29,932 is the cost of prescription medications, outpatient visits, as well as emergency room visits, which may contribute to the large difference. Further, the McCarthy data was comprised of insurance claims for one healthcare provider, the UnitedHealth Group, whereas our data is representative of the insurance costs from all payers, ranging from private insurance groups to federally funded Medicare, which may skew the reported costs.
McCarthy’s study also found that their PAP patients had a longer hospital stay, with a mean length of stay (LOS) of 15.96 days. Based on our analysis of the NIS data, patients with a diagnosis of PAP had an average hospital stay length of 6.24 days (CI: 3.9 – 8.5). Similarly, their data suggested a higher risk of comorbidity for PAP patients, with a CCI of 1.84, whereas our data found our hospitalized PAP patients to have an average CCI of 0.72 (0 -2.6). Of note, while the McCarthy estimation of CCI is within the confidence interval of our study, the LOS that we found was not. These findings together tend to suggest that the patient population that we studied was overall less sick – with fewer comorbidities and a shorter hospitalization course. Although the studies were performed in different years and utilized separate insurance claims databases, there is a significant difference in LOS during hospitalization that should be closely examined in the future. The yearly mortality rate in patients that were hospitalized with PAP averaged 5.03% from 2012-2014 based on the NIS data. Although there have not been any studies to our knowledge looking at mortality rate, there have been studies looking at survival trends with PAP and have suggested a disease specific survival rate of > 80% at 5 years30 with 70% of patients remaining free from recurrent PAP manifestations 7 years after initial therapy12. As referenced earlier, our data found 50% of total admissions to be elective admissions, which could explain the lower LOS and mortality rate.
Our study was limited by the nature of the database that we used. Unfortunately, as the NIS records instances of hospitalization rather than individual patient cases (i.e. one patient may have accounted for several admissions), we are unable to calculate the incidence or prevalence of disease, thus restricting our overall findings. Another significant limitation of our study was the inability to distinguish autoimmune PAP from other causes of PAP, such as secondary PAP, which has a significantly worse prognosis. Unfortunately, the ICD-9 code used for billing PAP (previously 516, now J84.01) does not differentiate between these conditions. However, because autoimmune PAP accounts for >90% of cases with a low incidence of comorbid conditions, it can be hypothesized that our select NIS subgroup does accurately reflect the general population of PAP at large. Finally, we were unable to analyze other clinical factors of interest, such as smoking status, laboratory work, or diagnostic studies, because the diagnosis of PAP relied heavily on the accuracy of medical billing rather than confirmatory levels of GM-CSF autoantibodies.
Still, there are several benefits to utilizing this rich database, including the fact that it contains one of the largest collections of PAP hospitalizations in existence. By sheer sample size alone, the NIS database allows us to study rare conditions such as PAP and its associated treatment costs, thus allowing us to calculate the economic impact of the disease.
Based on the descriptive characteristics of this cohort, we found that our cohort of PAP admissions were associated with a shorter LOS and mortality rate than other studies have found. This information may be useful for clinicians as they consider elective admissions for complex medical procedures, such as whole lung lavage and weigh the risks and benefits of this decision in a rare condition such as PAP.