Characterizing the severe asthma population in the United States: claims-based analysis of three treatment cohorts in the year prior to treatment escalation.

Abstract Objectives: Little is known about the disposition of severe patients prior to treatment escalation. To classify patients by treatment step using pharmacy data and describe their economic and healthcare utilization, insurance status, and sociodemographic characteristics in the year prior to escalation to Global Initiative for Asthma (GINA) steps 4 and 5. Methods: This was a retrospective claims cohort study of asthma patients (age 12–75 years) newly initiated on “stable therapy” (three consecutive months of therapy) with omalizumab, high intensity corticosteroids (HICS; ≥1000 µg/d inhaled fluticasone equivalent or oral prednisone), or high-dose inhaled corticosteroid (HDICS; ≥500–<1000 µg/d fluticasone equivalent) from 2002 to 2011. Other asthma treatments were compared as a reference. Results: Of 25 297 patients, 856 initiated omalizumab, 6926 initiated HICS, and 11 445 initiated HDICS. In the year prior to treatment escalation to omalizumab, HICS, and HDICS, respectively, individuals had high annual mean medical expenditures ($14 071, $12 030, and $7570), utilization (27 outpatient and 10 specialty care visits; 19 outpatient and three specialty; 15 outpatient and two specialty), asthma-related prescription drugs (11.74, 7.8, and 5.17) and chronic comorbidities (2.68, 2.67, and 2.19). Prior to omalizumab treatment, patients were more likely to be salaried, full-time employees with commercial PPO/POS insurance. Conclusions: Prior to escalating treatment to GINA steps 4 and 5, individuals experienced significant annual medical expenditures, healthcare resource utilization and polypharmacy burden, which may reflect poorly controlled asthma and the need to escalate treatment. Medical claims data and utilization-based measures may be helpful in classifying individuals by GINA treatment step.


Introduction
Asthma, a chronic inflammatory disorder of the airways, is characterized by recurring episodes of wheezing, breathlessness, chest tightness, and coughing, that affects 15-20 million individuals in the United States [1]. Despite guidelinedirected care of asthma, there remains significant uncontrolled disease [2][3][4]. This lack of control can be due to poor medication adherence or to refractory disease that is difficult to control [2][3][4]. The extent of the burden of asthma is associated with the level of severity and control [5]. Failure to control the symptoms of asthma can result in exacerbations, hospitalization, and potentially fatal outcomes. Thus, the principal objective of asthma treatment is to achieve symptom control and to reduce the risk of future complications of the disease.
Asthma severity is assessed from the level of treatment required to control symptoms and exacerbations [5]. Severe asthma is asthma that requires higher doses or more longterm controller medications for control, or asthma that remains uncontrolled despite high-dose treatment [5]. Because asthma severity and control can fluctuate over time, recommended asthma treatment takes a stepwise approach; as severity and control fluctuate, so do medication and treatment, with an escalation in therapy considered if asthma is poorly controlled [1,5]. Both the Expert Panel Report 3: Guidelines for the Diagnosis and Treatment of Asthma (EPR-3) [5] and the Global Initiative for Asthma (GINA) [1,6] guidelines suggest that patients who are poorly controlled on steps 1-3 be titrated to steps 4-6 (EPR-3) or steps 4 and 5 (GINA). EPR-3 steps 4 and 5 are comparable with GINA step 4 and include medium-to high-dose inhaled corticosteroids with long-acting beta agonists (LABA). EPR step 6 is comparable with GINA step 5 and includes adding oral corticosteroids with previous step therapies and considering the addition of omalizumab.
High-dose corticosteroids for treatment of severe asthma are often administered as high-dose inhaled corticosteroid (HDICS; 500-51000 mg/d fluticasone equivalent) or highintensity corticosteroids (HICS; 1000 mg/d fluticasone equivalent or oral prednisone). Omalizumab (Xolair) is a humanized anti-IgE monoclonal antibody, which inhibits the IgE-mediated inflammatory process in allergic asthma and has been shown to reduce the rate of asthma exacerbations when added to inhaled corticosteroids and long-acting beta agonists (LABA) [7][8][9]. Although omalizumab and high-dose corticosteroids can be effective treatment options for severe asthma, little is known about the characteristics of individuals who end up receiving these treatments. EPR-3 and GINA guidelines only recommend that patients who are poorly controlled on steps 1-3 be titrated to higher steps of therapy [1,5,6]. Based on this, we hypothesized that individuals needing an escalation in treatment to GINA steps 4 and 5 would exhibit poorly controlled asthma requiring more intensive treatment. But are there differences in outcomes that reflect poorly controlled asthma among individuals prior to being escalated to higher dose inhaled corticosteroids compared with oral corticosteroids or omalizumab? Because omalizumab is an expensive biologic therapy with significant reimbursement restrictions, we hypothesized that patients escalated to omalizumab therapy would have characteristics of better drug access and insurance. It is unclear if there are national patterns of the types of insurance or demographic characteristics more likely to have access to omalizumab compared with corticosteroids. An examination of these characteristics in the year prior to treatment escalation may help to understand if they affect access to treatment. This research aims to understand which types of patients end up receiving or having access to omalizumab vs. HICS or HDICS.
In addition, the authors are not aware of previous research that has attempted to classify patients by these treatment steps in the larger population. Successful methods of classifying patients by GINA/EPR-3 treatment step would provide important information to those who manage asthma at the population level. The objective of this study was to (1) explore the feasibility of classifying patients by treatment step using pharmacy claims; (2) characterize patients by treatment step (insurance coverage, comorbidity burden, healthcare resource utilization, and medical expenditures) in the year prior to treatment escalation or ''step up''; and (3) explore the relationship between access and treatment selection.

Data source
This retrospective cohort study used longitudinal, integrated medical, and pharmacy claims data from the MarketScan Õ Commercial Claims and Encounters database. The MarketScan Õ Claims database, inclusive of years 2002 through 2011, captures person-specific and longitudinal clinical and expenditure data occurring at the hospital, clinic, and pharmacy. These administrative data originate from a selection of large employers, health plans, and public organizations. The MarketScan Õ databases link paid claims and encounter data to detail patient information across sites and types of providers, and over time. The annual medical databases include private sector health data from approximately 100 payers. Historically, more than 500 million claim records are available in the MarketScan Õ databases. These data represent the medical experience of insured employees and their dependents for active employees, early retirees, COBRA continues, and Medicare-eligible retirees with employer-provided Medicare Supplemental plans. The medical claims are linked to outpatient prescription drug claims and individual-level enrollment data through the use of unique enrollee identifiers. All personal identifiers were deidentified and the data are Health Insurance Portability and Accountability Act (HIPAA) compliant.

Study population
Individuals (age 12-75 years) were extracted from the MarketScan Õ database ( Figure 1) if they had a diagnosis of asthma in at least two outpatient claims with primary or secondary diagnoses of asthma (ICD-9 code 493.xx) or at least one emergency department or hospitalization claim with a primary diagnosis of asthma. Children 512 years of age were excluded because omalizumab is only indicated for those 12 years of age. Individuals were excluded if they had a diagnosis of chronic obstructive pulmonary disease (or at least one claim for an anticholinergic medication), emphysema, or cystic fibrosis (ICD-9 codes 491.2, 493.2, 496, 506.4, 492.x, 506.4, 518.1, 518.2, and 277.0 x). All individuals were required to have continuous enrollment 12 months before and after the index date. The index date was defined as the date of the first prescription claim with 28 days' supply for newly initiated therapy during the study period.
These individuals were then categorized into four mutually exclusive treatment groups: (1) omalizumab; (2) high-intensity corticosteroids (HICS); (3) high-dose inhaled corticosteroids (HDICS); and (4) other asthma treatments (OAT). The primary focus of the analysis is the three severe cohorts (omalizumab, HICS, and HDICS). The OAT cohort is included to provide a comparison group or reference point to provide context for the main focus on asthma severity. The following definitions were applied to allocate subjects to the treatment groups based on the prescription drug utilization of at least 28 d supplied: (1) omalizumab cohort -patients with any use of omalizumab, (2) HICS cohort -patients not in the omalizumab cohort with use of 1000 mg/d fluticasone powder (or dose equivalent; Supplementary Table 1) or oral prednisone (or dose equivalent), (3) HDICS cohort -patients not in the omalizumab or HICS cohorts with use of 500-999 mg/d fluticasone powder (or dose equivalent), and (4) patients not in any of the three previous treatment groups but with the use of any other asthma medication. Patients could only be assigned to one treatment group. Although patients may have initiated several different treatments over the course of the available years, they were allocated only to the most severe cohort based on the inclusion criteria.
Each injection of omalizumab was assumed to be a 28-d supply (any claims with both an NDC and a J code within a 10-d window were considered a single administration and the index date was assigned to the first date of such claims). For all other drugs, days' supply was extracted from the days' supply field in the claims database and the index dates were assigned to the date of the first prescription with at least 28 d supplied.
Subjects were then further divided into those on ''stable therapy'' and those not. Stable therapy was defined as a minimum of three consecutive months of therapy. Hence, individuals were included in the omalizumab stable therapy cohort if they had three claims within the first 90 d after the index date of new therapy with omalizumab. Because of the potential for differential timing of prescription refills within 3 months, individuals were included in HICS, HDICS, or OAT stable therapy cohorts if they had a medication possession ratio 0.70 for the first 3 months (which suggests that they had medication for at least 70% of the expected 90 d).
Although subjects were allocated to four cohorts based on treatment for at least 3 months after the index date, this analysis focuses only on the year prior to the index date.

Outcomes
Outcomes included medical expenditures, healthcare resource utilization, insurance status, and demographic characteristics. Geographical division was defined based on US Censusdefined regions [10]. Medical expenditures reflected the amount paid to the provider; total medical expenditures include the amount paid by the health plan and the patient. The number of comorbid conditions for each patient was calculated as the total number of unique ICD-9 codes (excluding asthma) reported in the claims (this method has been used extensively in previous research) [11][12][13]. Although patients were identified based on stable therapy on one of the three regimens, this analysis focuses only on year prior to treatment escalation.

Statistical analysis
Unadjusted statistical values for all outcomes during the preindex (baseline) period were calculated. Comparisons among the four treatment cohorts for pre-index demographic characteristics, co-morbidities, rate of healthcare resource use, and expenditures were examined by testing for statistically significant associations with a p value 5 0.05 required for significance using F-statistics. To compare the likelihood of assignment to each respective cohort, a multinomial logistic regression model was estimated with treatment cohort as a categorical dependent variable. All expenditures were expressed in $US 2011 using the medical component of the CPI to inflate prior years to 2011. The primary purpose of the MarketScan Õ data source is tracking payments and, therefore, partial records (missing data) are not common and no imputation was performed. All analyses were performed using STATA 12 (StataCorp LP; College Station, TX).

Study population
Of the 965 829 individuals who had any asthma diagnosis, 0.4% (4129) were excluded for insufficient diagnosis and a further 38% (369 259) were excluded for diagnosis of COPD, emphysema, or cystic fibrosis ( Figure 1). Of those individuals with adequate diagnosis of asthma who were prescribed omalizumab, HICS, or HDICS (n ¼ 265 501), 26% (70 343) were 12-74 years of age and had continuous enrollment data 12 months before and after the index date ( Figure 1). Of those with adequate diagnosis of asthma who were prescribed OAT (326 940), 19% (62 935) were 12-74 years of age and had continuous enrollment data 12 months before and after the index date ( Figure 1).

Unadjusted baseline characteristics of patients with severe asthma
In the year prior to treatment initiation, individuals initiating omalizumab were more likely to be younger (mean age 41.6 years), active full-time, salaried employees from the South with commercial insurance (Table 1). Individuals initiating omalizumab were younger than the other severe cohorts (HICS and HDICS), but the OAT cohort was the youngest (mean age 39 years) ( Table 1). The insurance type for individuals receiving omalizumab was more likely to be point of service/preferred provider organization (POS/PPO) and less likely to be comprehensive or health maintenance organization (HMO) coverage (Table 1). Although the mean age of individuals using omalizumab was lower and they were less likely to be 455 years old, individuals receiving OAT or HDICS were more likely to be in the younger age groups (age 12-24 years) than those receiving HICS or omalizumab ( Table 1). The number of comorbidities was highest in individuals receiving omalizumab, followed by HICS, HDICS, and OAT (Table 1).

Unadjusted health resource utilization and comorbidities of patients with severe asthma
In the year prior to treatment escalation, health resource utilization (HRU) was generally high in this severe asthma population and was worse for individuals prior to treatment step up to omalizumab than to HICS followed by HDICS ( Table 2). As expected, the OAT cohort had consistently lower levels of HRU compared with the three more severe cohorts ( Table 2). Outpatient visits and asthma prescriptions were high in the year prior to step up; higher in those titrated to omalizumab than to HICS or HDICS ( Table 2). Emergency room (ER) visits, inpatient visits, home health visits, and regular office visits were higher for individuals prior to treatment with omalizumab than with HICS or HDICS ( Table 2). Specialty care visits (mostly allergist) were substantially higher for individuals before treatment with omalizumab, but still significantly greater for the HDICS and HICS cohorts compared with OAT (Table 2). Asthma-specific HRU trends were similar to general HRU (Table 2). Examining procedures specific to severe asthma, substantially more individuals received bronchodilator response spirometry in the year prior to treatment escalation to omalizumab than to HICS or HDICS, all of which were greater than OAT ( Figure 2). In contrast, nitric oxide monitoring was very low in all groups ( Table 2). Chronic comorbidities common to patients with asthma were evident in this severe population and, in general, there were a greater number of these comorbidities in individuals in the year prior to receiving omalizumab than the other groups (Table 2).

Unadjusted healthcare expenditures of patients with severe asthma
Unadjusted annual healthcare expenditures were high for all three severe groups compared with OAT (Table 3). Medical expenditures, pharmacy expenditures, inpatient expenditures, and outpatient expenditures were all higher in the year prior to treatment with omalizumab than for HICS, followed by HDICS and OAT ( Figure 3). Similar trends of increased expenditures with increasing treatment step were observed for asthma-specific expenditures (Table 3). On an average, outpatient expenditures made up the majority of healthcare expenditures incurred by individuals with severe asthma, followed by pharmacy expenditures and then inpatient expenditures ( Figure 4). Outpatient expenditures were much higher and pharmacy expenditures lower in the OAT cohort (outpatient 64%; pharmacy 19%; inpatient 18%). In the year prior to treatment escalation, patients contributed a significant amount annually out of pocket on medical expenditures, with increasing expenditures associated with increasing asthma severity ($936 prior to escalation to OAT, $1174 prior to escalation to HDICS, $1561 to HICS, and $1916 prior to escalation to omalizumab), similar to the increasing  (12) 1972 (17) 1069 (15) 162 (19) 10 311 (14) 0.000 Salary, union 601 (2) 307 (3) 199 (3) 23 (3) 1668 (2) 0.000 Salary, other 527 (2) 218 (2) 144 (2) 33 (4) 1352 (2) 0.000 Hourly, non-union 1618 (5) 736 (6) 443 (6) 50 (6) 4308 (6)  contribution by payers across the four groups (Table 4). Copayments made up the largest amount of out of pocket pharmacy expenditures for patients while out of pocket medical expenditures were similar for copayments, coinsurance, and deductibles (Table 4).

Polypharmacy burden of patients with severe asthma
In general, patients were taking several different medications in the year prior to treatment step up with more individuals taking multiple asthma medications prior to treatment with omalizumab than HICS followed by HDICS ( Figure 5). Patients were most commonly taking short-acting beta agonists (SABA), oral prednisone, and inhaled fluticasone (or equivalent: with or without LABA) ( Figure 5). Polypharmacy was common among individuals with severe asthma and significant percentages of individuals were taking other asthma medications in the year prior to treatment escalation ( Figure 5). Note that these utilization patterns comprise the year prior to treatment escalation; hence, the near zero percentage use of fluticasone 500-1000 mg in the HDICS cohort -which by definition had no record of use of fluticasone 500-1000 mg until the index date. This may likewise explain the very low use of leukotriennes and fluticasone 5500 mg in the OAT cohort in the year prior to treatment.

Multivariable regression results
In the multivariate model, baseline healthcare utilization, expenditures, age, geographic location, employment, and insurance type were predictive of treatment escalation (Table 5). (Note that the analysis in Table 5 does not include the OAT cohort because the primary focus is on the three severe cohorts) Individuals with higher baseline ER visits, prescriptions, and health care expenditures were more likely to be stepped up to omalizumab or HICS than HDICS (Table 5). Patients older than 55 years of age were 0.1-0.6 times less likely to be titrated to omalizumab than HDICS and those of 25-64 years of age were 1.2-1.4 times more likely to  be titrated to HICS than HDICS (Table 5). In the year prior to treatment escalation to omalizumab, individuals were 1.5 times more likely to be from the South and were more likely to have commercial insurance and to be active full time employees (Table 5). In contrast to the higher likelihood of patient initiating omalizumab having POS and PPO insurance, patients initiating HICS were less likely to have comprehensive and POS insurance than HMO compared with those initiating HDICS (Table 5). Multiple comorbidities were associated with severe asthma; patients with 1 comorbidities were 1.2-3 times more likely to be escalated to HICS than HDICS and 2.4-5.5 times more likely to be escalated to omalizumab than HDICS (Table 5).

Discussion
GINA guidelines suggest that patients exhibiting increasing asthma risk or lack of control should be titrated to the next step of therapy [14]. The higher steps require increasing intensity of therapy including increased corticosteroid dose and/or adding new medications. To shed light on the national burden of asthma among these patients, we used a novel means of categorizing patients by treatment step in order to identify characteristics associated with treatment escalation in the year prior to step up. We found that individuals initiating therapy reflecting GINA steps 4 and 5 experienced increased healthcare expenditures and utilization consistent with expectations of poorly controlled asthma in the year prior to escalation. Patients in these cohorts had a heavy asthma polypharmacy burden and exhibited significantly higher medical and pharmacy utilization and expenditures before treatment escalation compared with average asthma patients (OAT), with increasing treatment steps associated with higher baseline healthcare expenditures and utilization. In addition, our results show that patients escalated to omalizumab exhibit characteristics consistent with better drug access and very poorly controlled asthma. Outcomes in the year before treatment escalation were consistent with more intensive treatment and poorly controlled asthma, with a relatively greater burden experienced by those patients ending up on omalizumab. Most patients who received omalizumab received bronchodilator responsiveness testing in the year preceding treatment escalation. Patients in all three of these asthma treatment cohorts paid a significant amount of out of pocket expenditures ($1916 for omalizumab, $1561 for HICS, and $1174 for HDICS), including copayments for office visits and prescriptions. The three treatment cohorts represent patients with a heavy asthma polypharmacy burden that appeared to increase with increasing treatment step (HDICS5HICS5omalizumab). Consistent with these results, previous literature has shown that omalizumab is infrequently used as a single agent or without concomitant inhaled corticosteroids, often a leukotriene receptor antagonist, a short-acting beta-agonist inhaler, and at least one course of oral corticosteroids [15]. These cohorts also had high utilization (outpatient and specialty care visits), asthma-related prescription drugs, and chronic comorbidities. An earlier study of the MarketScan Õ database found that a high proportion of patients initiating omalizumab had 1 or more asthma-related ED visits and asthma-related hospitalizations [16].
Our results also help to provide insight into the types of patients who are more likely to receive omalizumab compared with HDICS and HICS. Certain characteristics of severe asthma patients are associated with treatment escalation to omalizumab. Some of these characteristics may reflect more severe asthma such as higher baseline ER visits, prescriptions, health care expenditures, and more comorbidities. In addition, it appears that individuals titrated to omalizumab were more likely to have characteristics consistent with better drug access such as commercial insurance and being an active full time employee [17]. It is also interesting to note that individuals from the South were more likely to be escalated to omalizumab than the Northeast and other regions. The reason behind this association is unclear. It is possible that the South region has a greater incidence of atopic asthma. It is also possible that there are geographic differences in treatment patterns. Future research into what is driving this geographic correlation would be beneficial. Results from our study show a consistent trend of higher costs associated with greater degrees of severity (or increasing step of therapy). In addition, the OAT cohort provided a less severe reference group. Results were consistent with expectations and the OAT cohort exhibited lower HRU and expenditures than all three severe cohorts.
The three intensive treatment cohorts identified in the current study also experienced greater utilization and expenditures than those with or without asthma in a separate study [13]. In a nationally representative, general population study, Sullivan et al. found medical expenditures for adults were $4198 for those without asthma and $6850 for those with asthma ($US 2008). The asthma population in this study included all levels of severity in a general asthma population. This is substantially less than the $7500-$14 000 found for the three severe treatment cohorts in our study. The fact that our estimates are significantly higher than the general asthma population in the US underscores the burden and expense associated with severe asthma. Furthermore, Sullivan et al. included only adults while our study included adolescents. Because asthma expenditures are higher in adults than in adolescents [18][19][20], our results would have been even greater if we had included only adults.
The three treatment cohorts identified in this study aligned with GINA steps 4 and 5 and all outcomes appeared to be consistent with expectations for these steps. In addition, this research provides important information about the disposition of individuals just prior to escalation to GINA steps 4 and 5. The deleterious outcomes found in the results are consistent     with poorly controlled asthma and reflect the need for treatment escalation.
In addition to the value of empirical results presented, this research demonstrates a preliminary methodology for classifying patients according to GINA/EPR-3 treatment steps using pharmacy claims data. This methodology may be useful to those who manage asthma at the population level to better understand their patients and target programs to improve population health. Our research shows that medical claims data and utilization-based measures may be helpful in classifying individuals by the GINA treatment step. This methodology for classifying patients by GINA treatment step may be useful for analyses of population health and large health organizations. The methodology could be applied to classify patients before and after treatment escalation to determine if outcomes improved after an increase in GINA step therapy. It could also be used to help identify the percentage of a given population (national, state, and HMO)

Limitations
There are limitations associated with the study design. This was a cohort study with no direct measures of asthma control in the MarketScan Õ database. Hence, the results only allow for speculative associations with indirect measures of control (such as asthma-related utilization and expenditures). The MarketScan Õ claims provide a complete listing of reimbursed services for a large insured population, but the accuracy of diagnostic information and the timing of drug exposure information have not been validated. An explicit comparison of the MarketScan Õ claims diagnostic information with direct medical records is beyond the scope of this study. However, many factors provide confidence in the methodological approach of this study. First, patients in this study were required to have several asthma prescription claims consistent with asthma; and second, the clear synergy between the expected association of the results with asthma severity (for example the direct association between HRU and expenditures with increasing asthma severity) provides face validity that the diagnostic and classification algorithm used in these claims were valid. Nonetheless there are limitations. A record of prescriptions filled at the pharmacy does not ensure that the patient used the drug, which could result in overestimation of true adherence for patients who filled prescriptions but did not use the medication as prescribed. The claims data do not provide a record of samples given by the physician, which could result in underestimation of medication use. Some patients may have had more than one health insurance provider and filled one or more prescriptions using other or no insurance. Of note, because of the descriptive nature of the study, causal relationships cannot be determined by statistical testing.
Our study design required patients to be on ''stable therapy'', defined as having three consecutive months of therapy after treatment initiation. The rationale behind requiring 3 months of therapy was to ensure that patients could be reasonably categorized as being stable on the new treatment. In claims data analyses, it is challenging to be sure that patients are taking the prescription drugs that they have filled. Requiring only one prescription would likely include a significant number of patients who filled the first prescription but never took it. Some patients may take two prescriptions only to realize that they have side effects or are not responders. Requiring three prescriptions further increases the odds that they are taking the medication and continuing on therapy. For example, a certain percentage of omalizumab patients are known to not respond to treatment. Clinically, treatment response for omalizumab is typically defined after 16 weeks of treatment. Response for ICS therapy is typically defined by shorter durations of treatment. Requiring 3 months of therapy was assumed to be a good compromise for all three treatment cohorts. However, it is possible that some patients were not actually on stable therapy and our results would be limited to a commensurate degree. It is also possible that requiring ''stable therapy'' could have influenced the study outcomes. The MarketScan Õ claims databases are based on a large convenience sample of mostly large employers that may fail to generalize well to other populations, in particular the uninsured: It was challenging to find data that were adequate to address the research questions delineated above. Because the research focused on the severe asthma population, the data source had to be large enough to capture relatively rare use of omalizumab and oral corticosteroids. In order to categorize patients by the higher treatment steps, the data had to contain comprehensive pharmacy use. To address study outcomes, the data had to contain comprehensive medical claims, expenditure information, insurance status, and sociodemographic characteristics. Unfortunately, there was no data available that met the above criteria and captured the uninsured (largely because the uninsured do not have centralized claims data). However, MarketScan data did meet all other criteria and were determined to be the most appropriate. A number of individuals with severe asthma do not have insurance coverage and would not be captured in the MarketScan Õ data. It is not clear how the inclusion of uninsured patients would impact our results. It is highly unlikely that those without insurance would have access to omalizumab. It is possible that uninsured patients would be more likely to be in the HDICS or the HICS group, but this research was unable to provide concrete evidence of this relationship. Our study did include Medicare patients and found that having Medicare did not have a significant impact on allocation to treatment cohort (Table 5). It is unclear if Medicaid would have similarly been insignificant. Future research in the Medicaid and uninsured populations would be beneficial.
Although sociodemographic characteristics are included in this analysis (such as employment status and employment classification), income was not available. Income may be an important predictor of treatment access and insurance status among the severe asthma population, particularly for omalizumab. In addition, the information about employment status and employee classification is limited with many individuals being classified as ''other'' or ''unknown''. As a result, results regarding employment and employee classification should be interpreted with caution. Despite these limitations, the age and the gender distribution of the current sample appears to be similar to that of other nationally representative samples of asthma patients [13]. While conclusions should be tempered by these limitations, the results provide important information about the severe asthma population.

Conclusion
In recent national data, individuals who were escalated to GINA steps 4 and 5 experienced significant medical expenditures and healthcare burdens in the year prior to step up. The high medical expenditures, utilization, and polypharmacy burden of these patients prior to step up may reflect poorly controlled asthma consistent with the need to escalate treatment. Medical claims data and utilization-based measures may be helpful in classifying individuals by the GINA treatment step.