Demographic Characteristics of Asthma Subjects Associated With Stepping Down From Fluticasone Propionate/Salmeterol Combination Therapy
Patients on fluticasone propionate/salmeterol maintenance therapy who have more frequent physician interactions, especially interactions with asthma specialists, may be more likely have their controller medication reduced.
Combination therapy of an inhaled corticosteroid (ICS) and a long-acting beta-agonist (LABA) is recommended for patients not adequately controlled on a low or moderate dose of ICS.1 Once asthma control is achieved and maintained on ICS/LABA combination therapy, asthma guidelines suggest reducing the dose of the combination or removing the LABA.1,2
According to a recently conducted clinical trial, the particular step-down strategy used may also influence a patient’s level of asthma control.3 Specifically, this trial concluded that stepping down to a lower dose of fluticasone propionate/salmeterol combination (FSC) is more effective than switching to ICS alone. Although this trial helped offer insight into the clinical benefits of stepping treatment down to a lower dose versus switching to ICS alone, a real-world understanding of patient characteristics that could impact the likelihood to step down or not step down is essential. A thorough understanding of patient characteristics may help differentiate patients more likely to follow either of the step-down strategies. Patient-specific treatment approaches may be utilized to prevent future unfavorable outcomes. This is crucial, as the type of step-down strategy followed has been shown to impact asthma-related healthcare utilization and costs in the future.4
The objective of this study was to identify factors that are associated with a step-down from FSC, either to a lower dose of FSC or to a switch to ICS monotherapy.
This study used the MarketScan Commercial Claims and Encounters Database, which represents the medical experience of insured active employees and early retirees, and those insured by employer-sponsored plans (ie, non-Medicare eligible) and their dependents. The database includes detailed cost, use, and outcomes data for healthcare services performed in inpatient and outpatient settings, prescription drug claims, and information on patient enrollment. The MarketScan data are compliant with the Health Insurance Portability and Accountability Act, thus all subjects are anonymized. Over 34 million lives per year were covered through 2010. MarketScan is a nationally representative database of subjects enrolled within private commercial plans. Since this was a retrospective study using existing patient data, institutional review board approval was not needed.
Patient Selection Criteria
Subjects needed to satisfy certain criteria to be eligible for inclusion in the final sample. These criteria were evaluated in the 12-month period prior to the index date. The criteria were: 1) a minimum of 2 prescriptions of similar-dose FSC prior to the index date, including the first prescription. (This condition served as an indication that subjects were being stabilized using FSC.) 2) Being diagnosed with asthma, defined as having at least 1 medical claim with a diagnosis of asthma (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 493.xx) in any field; 3) being aged at least 12 years or below 65 years at index date; and 4) having continuous insurance coverage.
Subjects were excluded if they had evidence of chronic obstructive pulmonary disease based on diagnosis (ICD-9-CM codes 491.xx, 492.xx, 496.xx) or if they received an anticholinergic (ie, ipratropium or tiotropium) in the 12-month period prior to the index date, or if they had: 1) a higher-dose FSC prescribed (as compared with the first FSC); 2) any other ICS and LABA combination therapy (ie, mometasone/formoterol, fumarate dihydrate, or budesonide/formoterol fumarate dihydrate); or 3) a controller medication other than FSC from first FSC prescription through the index date. Controller medications included beclomethasone, budesonide, flunisolide, fluticasone, triamcinolone, mometasone, formoterol, salmeterol, omalizumab, montelukast, zafirlukast, zileuton, cromolyn, nedocromil, aminophylline, dyphylline, oxtriphylline, and theophylline.
This study implemented a retrospective study design (Figure). The initial population from which the final sample was derived consisted of subjects with at least 1 prescription claim for FSC during the study period of January 1, 2006, to December 31, 2010. To permit adequate follow-up and allow for a 12-month pre-period, their first FSC Rx claim had to occur during the identification period of January 1, 2007, to December 31, 2009.
After the first FSC Rx, each patient was followed for a period of 12 months to assess the presence of step-down. Step-down was defined as reduction in dose of the FSC prescription (low-dose FSC) or removal of the LABA to ICS monotherapy (ICS monotherapy) (Table 1). Subjects who did not step down within the 12-month period following the date of the first FSC Rx were placed into the FSC group.
The date of step-down was termed the index date for each step-down group. The FSC group had no change in dose of FSC or presence of ICS within 12 months of the first FSC Rx. For the FSC group, the date of fill of the last FSC Rx was termed the index date.
After meeting study criteria, each patient in the 2 step-down groups were matched on a 1:3 ratio to the FSC group on index dates (± 5 days) between the 2 groups. FSC subjects were randomly selected to be matched to low-dose FSC subjects first and then the remaining FSC subjects were again selected randomly to be matched to the ICS monotherapy subjects. This 2-step matching process resulted in 2 mutually exclusive pair-wise comparison study groups: 1) low-dose FSC to FSC, and 2) ICS monotherapy to FSC. Matching on index date allowed for seasonal variations in asthma therapy use to be accounted for in the study.
The overall study objective was to identify the factors associated with a step-down. A set of patient characteristics was first generated to gain a thorough understanding of the patient population. Patient characteristics included demographics, comorbidity characteristics, and proxies for asthma control (short acting β-agonist [SABA] and oral corticosteroid [OCS] use), asthma-related healthcare utilization, and costs. These patient characteristics were assessed during the period from date of first FSC Rx to index date (not including index date). This period was specifically chosen as it was assumed that events occurring in the period prior to receipt of the first FSC may not be related to a patient’s likelihood to step down since they were not receiving FSC during that time period. Additionally, since this period varied for each patient, factors dependent on time period of assessment, such as use of asthma-related medical and pharmacy services, were standardized on a monthly basis.
Patient’s age (as of index date), gender, region, and type of health plan (health maintenance organization [HMO], preferred provider organization [PPO], point-of-service organization, and others) were the key demographic factors assessed. Asthma specialist care was also evaluated. The medical claims during the period from date of first FSC Rx to index date were used to assess the presence of an asthma specialist visit. If at least 1 medical claim was identified with a specialty of interest, the patient was considered to receive asthma specialist care. Two types of asthma specialties were defined: 1) allergy and immunology, and 2) pulmonary disease.
Overall comorbidity was measured using the Charlson comorbidity index (CCI) and the number of unique 3-digit diagnosis codes. Furthermore, the number of unique drugs and number of unique drug classes were evaluated to better understand each patient’s overall pharmaceutical use. Lastly, presence of allergic rhinitis (ICD-9-CM code 477.xx) was evaluated.
According to the Expert Panel of the National Heart, Lung, and Blood Institute guidelines,1 frequent use of SABA or OCS to treat asthma symptoms may be considered a sign of asthma that is not well controlled. Therefore, SABA and OCS use was obtained as a proxy for asthma control; specifically, presence and monthly number of SABA canisters and OCS Rxs were evaluated.
Asthma-related healthcare utilization—a potential factor associated with a step-down event—was also evaluated in this study. An asthma-related visit was defined as a medical claim with a primary diagnosis of asthma (ICD-9-CM 493.xx). Types of visits included hospitalizations (primary discharge diagnosis of ICD-9-CM 493.xx), emergency department (ED) visits, combined end point of hospitalization or ED visits, physician visits, outpatient visits, and other visits. Both presence and number of asthma-related visits were measured and standardized on a monthly basis to account for the difference in time period of assessment between subjects
In addition to the above, others factors of interest that were entered into the statistical model were: 1) duration of time between first FSC date and index date; 2) number of FSC dispensings (same dose as first FSC) before the index date; 3) year of index date; and 4) season at index, categorized as spring (March-May), summer (June-August), fall (September-November), and winter (December-February).
Demographic and other patient characteristics were presented for the 4 study groups using standard summary statistics (means and percentages). Inferential statistics were used to analyze differences between groups. Two comparisons were made: 1) low-dose FSC vs FSC (reference category), and 2) ICS monotherapy versus FSC (reference category). Categorical characteristics were compared using univariate conditional logistic regression, and characteristics that were continuous in nature were compared using paired t tests.
Similarly, multivariate analyses for evaluating the factors associated with step-down were done using conditional logistic regression models with relevant covariates. For evaluating factors associated with type of step-down, only a logistic regression model with relevant covariates was used. Selected covariates included in the models based on clinical relevance and a priori hypotheses about their association with step-down included: demographic information (ie, age, gender, United States geographic region, physician specialty, and health plan); overall comorbidity (ie, CCI and presence of allergic rhinitis); asthma-related utilization (ie, use of SABA, OCS, physician visits, and presence of asthma-related hospitalization or ED visit); and others (ie, time to index date and prior number of FSC dispensings). Additionally, the model evaluating the type of step-down (low-dose FSC or ICS monotherapy) also included season of step-down. Odds ratios (ORs) along with 95% CIs are reported. A factor was considered to be significantly associated with a step-down if P ≤.05 for the OR.
A bootstrap resampling technique was used to confirm the factors associated with step-down. This involved repeated random sampling with replacement from the original data; each resampling is called a bootstrap sample. A separate multivariate conditional logistic regression analysis was performed on each bootstrapped sample. A single 95% CI around the estimate was calculated by using the percentile interval method.5 Original sample ORs were reported for each factor as estimates from the original sample.6
To ensure accuracy, the data was reviewed for consistency and checked for influential outliers. All statistical tests performed tested a 2-sided hypothesis of no difference between study groups at a significance level of P = .05.
A total of 685,869 subjects were identified with first FSC Rx during the identification period. Subjects who did not satisfy the study criteria were excluded.
displays the number of subjects excluded due to each criterion (not mutually exclusive). A majority of subjects were excluded due to lack of asthma diagnoses, not having at least 2 FSC dispensings, and no continuous health plan eligibility in the 12-month period before the index date.
After applying all criteria, the sample consisted of 42,024 subjects. A total of 3251 (7.7%) of these subjects stepped down to a lower dose of FSC (n = 1461) or switched to ICS monotherapy (n = 1790). After matching, almost all subjects were retained in both step-down groups. The final sample had 1459 low-dose FSC subjects matched to 4377 FSC subjects, and 1775 ICS monotherapy subjects matched to 5325 FSC subjects.
Patient baseline characteristics and univariate analysis for both step-down groups are listed below in Table 2 and
respectively. The mean age of the groups was 40 years (SD = 15) and females comprised 59% of the sample.
Multivariate Analysis Results
Low-dose FSC group. The conditional logistic regression found subjects visiting an asthma specialist had twice the odds of stepping down to low-dose FSC compared with subjects not visiting an asthma specialist (adjusted OR, 1.924; 95% CI, 1.772-2.518) (
). Also, each additional physician visit per month was associated with 16% higher odds of stepping down to low-dose FSC (OR, 1.16; 95% CI, 1.11-1.28). However, region (south vs northeast; OR, 0.763; 95% CI, 0.587-0.914), time to index date (OR, 0.992; 95% CI, 0.989-0.992), and number of FSCs dispensed (OR, 0.929; 95% CI, 0.883-0.955) were associated with significantly lower odds of stepping down to low-dose FSC. Use of SABA, OCS, or presence of hospitalization/ED visits did not predict stepping down to low-dose FSC.
ICS monotherapy group. The trends found in the univariate comparisons were retained in the multivariate analysis for the most part. The conditional logistic regression found subjects visiting an asthma specialist had 47% greater odds of stepping down to ICS monotherapy compared with subjects not visiting an asthma specialist (OR, 1.472; 95% CI, 1.312-1.846) (Table 4). Further, being female, having a diagnosis of allergic rhinitis, and increasing number of physician visits were associated with higher odds of stepping down to ICS monotherapy. Factors associated with reduced odds of stepping down to ICS only included age, number of FSC Rxs, time to index date, health plan (PPO vs HMO), and region (south vs northeast). Use of OCS was found to be associated with a higher odds of stepping down to ICS monotherapy; however, other measures of asthma control, such as number of SABA canisters and presence of hospitalization/ED visits, did not predict stepping down to ICS monotherapy.
This study aimed to use medical and pharmacy claims data to provide a description of asthma subjects who stepped down from FSC therapy to low-dose FSC, stepped down to ICS alone, or remained on the same dose of FSC. Comparative assessments among these subjects helped us to gain insight about the differences among these groups in terms of demographic information, comorbidity burden, asthma-related healthcare utilization, and other relevant characteristics. This study informs payers and providers of patient demographics and characteristics that are related to reducing asthma therapy as suggested by the National Asthma Education and Prevention Program, Expert Panel Report 3.1
The results of this study demonstrated that a low percent of subjects who were taking FSC at a stable dose were eventually stepped down to lower doses of FSC or ICS monotherapy. In addition, patients with more specialist care, more physician visits, and less prior FSC use had greater odds of stepping down from FSC to a lower dose of FSC. This may indicate appropriate physician care, since asthma management guidelines recommend that patients be regularly monitored and that they step down when asthma is adequately controlled; furthermore, studies have shown that asthma control can be maintained when patients are stepped down from a high-dose ICS/LABA to a lower-dose ICS/LABA.3,7-10 However, measures of asthma control were not found to be associated with step-down to low-dose FSC, while oral corticosteroid use was found to be associated with stepping down to ICS only. This finding is contradictory since patients who are adequately controlled are recommended to step down.10-12
Other studies have shown that several factors influence the success of step-down therapy. Clinical markers, such as a normalized acetylcholine PC20, longer time free of asthma symptoms, lower fractional exhaled nitric oxide, and higher forced expiratory volume in 1 second (FEV1), have been shown to increase odds of success of step-down.13 However, older age, gastroesophageal reflux disease, rhinitis, acetaminophen prescriptions, greater number of SABA fills, higher ICS dose, higher body mass index, current smoking, and increased numbers of primary care consultations predicted loss of asthma control after step-down.14
Use of claims data provided several advantages, including access to large study populations with demographic diversity in a real-world setting. However, interpretation of results will require consideration of certain limitations. First, this study evaluated only those patient characteristics for which information was available in the current database. Additional data about patient’s clinical status (ie, FEV1, symptom scores, etc) could have helped us to gain an understanding of factors associated with step-down. Second, the current database is representative of the commercially insured population, and results from this study may not be applicable to the general population. Lastly, this study possesses inherent limitations of a typical retrospective database analysis, including possible coding errors, selection bias, lack of clinical data, and no randomization.
In conclusion, the results of this analysis using a large claims database suggest that subjects on FSC who have more frequent physician interactions, especially interactions with asthma specialists, may be more likely to have their controller medication reduced. Measures of asthma control were not significant predictors of therapy step-down, and this suggests that the decision to step down asthma therapy may be determined more from the patient and provider interactions and less from medication history. Payers need to consider the importance of physician assessment of asthma control and how clinical assessment may impact treatment and treatment patterns. In addition, it is also important for physicians to closely monitor asthma control over time, especially after changing therapy, since this may help in making subsequent decisions on reducing either the ICS dose or removing the LABA per asthma treatment guidelines.
The authors wish to acknowledge the following individuals for their contributions: special thanks to Amol Dhamane, MS, who was the lead analyst on this project; Manan Shah, PharmD, PhD, for being a consultant; and Michael Bogart, PharmD, for manuscript writing and editing.