Disease Burden of Patients with Asthma/COPD Overlap in a US Claims Database: Impact of ICD-9 Coding-based Definitions

ABSTRACT

The inclusion of an asthma/chronic obstructive pulmonary disease (COPD) overlap syndrome (ACOS) population in the 2015 Global Initiative for Chronic Obstructive Lung Disease strategic documents has raised questions about the profile of these patients in clinical practice, as they are mostly excluded from asthma and COPD clinical trials. We estimated the disease burden, co-morbidities, and respiratory treatments of patients with asthma/COPD overlap, utilizing the Truven MarketScan commercial and Medicare databases. Patients with ≥1 COPD or chronic obstructive asthma diagnostic code were identified between January 1, 2008, and December 31, 2011. The asthma/COPD overlap group was defined and stratified based upon type and frequency of asthma diagnostic code (chronic obstructive asthma only, COPD and chronic obstructive asthma, and COPD and ≥1 asthma code). 1,488,613 patients were identified; of these, 1,171,626 were diagnosed with COPD alone and 316,987 with asthma/COPD overlap. Patients with asthma and COPD had higher disease burden indicators and inhaled corticosteroid/long-acting beta-agonist use compared with COPD alone. This trend was consistent for all definitions of asthma/COPD overlap. Patients with obstructive asthma and COPD tended to be older, with greater disease burden compared with other definitions; this population may represent a more severe form of asthma/COPD overlap. Disease burden and treatment also varied based on the codes defining asthma/COPD overlap, indicating possible phenotypic differences. More clinical insight and detailed phenotyping is needed to determine the reasons for coding variation in asthma/COPD overlap, with implications for further research to address unmet needs.

KEYWORDS:

• ACOS
• corticosteroid
• long-acting beta-agonist
• diagnostic codes
• treatment pattern

Introduction

Asthma and chronic obstructive pulmonary disease (COPD) are both characterized by airflow obstruction. While these diseases have distinct features and pathogenic attributes, many patients exhibit characteristics of both diseases. Patients who exhibit characteristics of both diseases may represent a phenotype called Asthma COPD Overlap Syndrome (ACOS) (1–4). Broadly, ACOS can be described as an obstructive airflow condition with clinical characteristics of both asthma and COPD, such as forced expiratory volume in one second (FEV₁)/forced vital capacity ratio <0.70 and evidence of airflow reversibility (post-bronchodilator FEV₁ increase of 12% and 200 mL).

Recent strategic documents published by Global Initiative for Chronic Obstructive Lung Disease and Global Initiative for Asthma acknowledge this overlap (5,6), yet ACOS does not yet have a well-defined universal diagnostic criteria. Developing criteria for ACOS has been difficult considering that the clinical presentations of asthma and COPD can be similar with respect to symptomology (2,7). Additionally, whether ACOS is simply the coexistence of asthma and COPD, or a distinct phenotype with fundamental pathogenic mechanisms related to both asthma and COPD, has yet to be determined (8). A variety ofdiagnostic criteria have been proposed for identifying overlap disease, yet a specific and clinically relevant definition of ACOS does not exist, and proposed definitions vary widely. Some definitions only require a history of asthma, whereas others are more stringent requiring features of current asthma, such as reversible lung function (1,9).

Studies of ACOS have also found conflicting results regarding outcomes and disease burden compared with patients with only asthma or COPD. Some studies have shown that patients diagnosed with ACOS tend to have more comorbid conditions, more frequent severe exacerbations, and greater healthcare utilization than patients with COPD or asthma alone (10–16); however, others have shown no difference (17,18). These conflicting results may be due to different definitions of ACOS used across studies applied to a range of patient populations recruited from the community or clinic and highlight the importance of understanding the impact of different disease definitions on these characteristics. However, a recent study found that patients with ACOS classified based on a previous history of asthma exhibited similar characteristics to those classified based on more restrictive criteria (19). It is therefore important to investigate the impact of different ACOS definitions on outcomes and disease burden.

Objectives

This study aimed to assess whether differences in co-morbidities, markers of disease burden, and medication treatment patterns exist among various definitions of asthma/COPD overlap (based on the International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] diagnostic codes) in large US administrative healthcare claims databases.

Methods

Data source and study population

A cohort of adults (18+ years) with a diagnosis code of COPD (ICD-9- CM: 490.xx–492.xx or 494.xx–496.xx) or chronic obstructive asthma (ICD-9-CM: 493.2x) was identified between 2008 and 2011 from the Truven Health Analytics MarketScan Commercial and Medicare Supplemental databases. The commercial database contains data from commercially insured individuals aged less than 65 years, and the Medicare supplement provides claims for individuals aged 65 years and over who receive coverage through Medicare. The database contains information on inpatient and outpatient treatment, as well as dispensed prescriptions based on ICD-9-CM diagnostic codes, Current Procedural Terminology (CPT) 4th edition and Healthcare Common Procedure Coding System (HCPCS) procedure codes.

At least one COPD diagnosis (ICD-9-CM codes: 490.xx–492.xx, or 494.xx–496.xx) or chronic obstructive asthma (493.2x) code was required during the study period. The date of the first COPD or obstructive asthma diagnosis between January 1, 2008, and December 31, 2011, was defined as the index date. Patients were required to have 24 months of continuous insurance enrollment in the 24 months surrounding the index date (Figure 1). Patients with a diagnostic code for a medical condition incompatible with a COPD diagnosis at any time in their history were excluded. These included conditions related to lung or bronchial developmental anomalies, degenerative processes (cystic fibrosis, pulmonary fibrosis), bronchiectasis, pulmonary resection, or other significant respiratory disorders (but not including cancer) that could interfere with a clinical COPD diagnosis or substantially change the natural history of COPD.

Figure 1. Study design.

Overlap classification

Patients with asthma/COPD overlap were defined as 1) having either a chronic obstructive asthma diagnostic code or a COPD code and 2) a non-obstructive asthma diagnostic code in the 24-month window around the index date (a maximum of 12 months prior to and 12 months post-index date). Non-obstructive asthma ICD-9-CM codes included 493, 493.0, 493.1, 493.8, and 493.9. Patients with no asthma code and no chronic obstructive asthma code within the 24-month window were considered to have COPD alone. Patients with asthma/COPD overlap were further stratified into mutually exclusive groups based on the codes used to define overlap, a) chronic obstructive asthma only, with no COPD diagnosis (≥1 chronic obstructive asthma code), b) COPD and chronic obstructive asthma, or c) COPD and ≥1 non-obstructive asthma code. Groups a through c were subsequently combined to form the overall asthma/COPD overlap group.

Co-morbidities and disease burden indicators

Co-morbidities and disease burden indicators were defined using ICD-9-CM codes, CPT codes, and prescription records in the 24-month window around the index date. Markers for disease burden included inpatient hospital visits for any reason, oxygen therapy, short-acting bronchodilator (SABD) use, shortness of breath (ICD-9 code: 786.05), oral corticosteroid (OCS) prescriptions, COPD-related emergency department (ED) visits, and COPD-related hospitalizations with intensive care unit stays. Oxygen therapy was defined as a medical code for oxygen therapy or a procedural code for oxygen. SABD use was defined as a prescription for a short-acting beta2-agonist (SABA), short-acting muscarinic antagonist (SAMA), or fixed combinations of SABA/cromoglycate or SABA/SAMA.

Respiratory medications

A separate analysis was performed to examine respiratory medication use in the 6-month window around index date. Respiratory medications included SABD, long-acting bronchodilators (LABA), long-acting muscarinic antagonists (LAMA), and inhaled corticosteroids (ICS), as well as fixed-dose combinations of ICS and LABA. Respiratory medications were categorized into mutually exclusive groups that were determined in a hierarchical manner beginning with triple therapy, then dual therapies, followed by monotherapies. The entire 6-month window around the index date was used to determine the medication category. Patients with no prescription fills for any of the included therapies in the 6-month window were also identified. Monotherapies were defined as a prescription of SABD, LABA, LAMA, or ICS that never overlapped with any of the other respiratory therapies in the 6-month window. “SABD only” was defined as having only SABD prescription(s) in the 6-month window. Dual therapies included combinations of LABA/LAMA, LAMA/ICS, and LABA/ICS where prescriptions of each treatment overlapped the other for at least one day, or a fixed combination of an ICS and LABA (in one device) was dispensed. Triple therapies were defined as a fixed combination of ICS/LABA that overlapped by at least one day with a LAMA or dispensing of a LABA, ICS and LAMA that overlapped by at least one day for all three prescriptions.

Data analysis

Demographics (age and gender) were evaluated on the index date. The number and proportion of patients with COPD only and asthma/COPD overlap were determined along with the proportion of patients in each overlap classification category (i.e., chronic obstructive asthma only, COPD and chronic obstructive asthma, and COPD and ≥1 asthma). Co-morbidities and indicators of disease burden in the 24-month window around the index date and respiratory medications in the 6-month window around the index date were described for patients with COPD only and asthma/COPD overlap, as well as each of the sub-categories of overlap. Chi-square tests were used to compare the COPD and asthma/COPD overlap (total) groups and to compare the asthma/COPD overlap groups (chronic obstructive asthma only, COPD and chronic obstructive asthma, and COPD and ≥1 asthma code). Analyses were conducted using SAS version 9.4. A post hoc sensitivity analysis stratifying co-morbidities and disease burden indicators by age (18–49; 50–64; 65+ years) was conducted to understand variation by age group.

Results

Study population

The cohort contained 1,488,613 adults with COPD or obstructive asthma; 316,987 classified as diagnosis asthma/COPD overlap (including obstructive asthma), and 1,171,626 had COPD alone. The asthma/COPD overlap cohort was stratified into obstructive asthma only (14.5%, n = 45,988), COPD and chronic obstructive asthma (15.3%, n = 48,489), and COPD and ≥1 asthma (non-obstructed) code (70.2%, n = 222,510) (Table 1).

Table 1. Demographics and co-morbidities in a cohort of patients with COPD and asthma/COPD overlap stratified by overlap diagnoses: United States, 2008-2011.

						Asthma/COPD overlap
	COPD only (n = 1,171,626)		Asthma/COPD overlap, total (n = 316,987)			Chronic obstructive asthma only (n = 45,988)		Chronic obstructive asthma + COPD (n = 48,489)		COPD + ≥1 asthma (n = 222,510)
	N	%	N	%	COPD only vs. asthma/COPD overlap, total	N	%	N	%	N	%	Asthma/COPD overlap groups
Gender: Female	631,529	53.9	206,119	65.0	p < 0.0001^a	29,438	64.0	30,470	62.8	146,211	65.7	p < 0.0001^a
Age at diagnosis					p < 0.0001^b							p < 0.0001^b
18–49 years	405,626	34.6	109,941	34.7	—	17,976	39.1	7,535	15.5	84,430	37.9	—
50–64 years	464,790	39.7	135,348	42.7	—	19,708	42.9	21,962	45.3	93,678	42.1	—
65+ years	301,210	25.7	71,698	22.6	—	8,304	18.1	18,992	39.2	44,402	20.0	—
Co-morbidities
Anxiety	127,677	10.9	47,734	15.1	p < 0.0001	5,993	13.0	7,565	15.6	34,176	15.4	p < 0.0001
Arrhythmia	197,177	16.8	61,177	19.3	p < 0.0001	7,073	15.4	15,237	31.4	38,867	17.5	p < 0.0001
Cancer	110,288	9.4	28,543	9.0	p < 0.0001	3,388	7.4	6,518	13.4	18,637	8.4	p < 0.0001
Cataracts	120,035	10.3	35,403	11.2	p < 0.0001	4,403	9.6	7,502	15.5	23,498	10.6	p < 0.0001
Cerebrovascular disease	130,910	11.2	34,790	11.0	p = 0.0017	3,892	8.5	8,969	18.5	21,929	9.9	p < 0.0001
Congestive heart failure	110,611	9.4	35,146	11.1	p < 0.0001	3,174	6.9	11,475	23.7	20,497	9.2	p < 0.0001
Depression	143,129	12.2	52,741	16.6	p < 0.0001	6,533	14.2	8,815	18.2	37,393	16.8	p < 0.0001
Diabetes	225,820	19.3	71,853	22.7	p < 0.0001	9,260	20.1	14,975	30.9	47,618	21.4	p < 0.0001
Diabetes with complications	55,874	4.8	18,053	5.7	p < 0.0001	2,321	5.1	4,356	9.0	11,376	5.1	p < 0.0001
Emphysema	78,034	6.7	19,629	6.2	p < 0.0001	.	.	7,351	15.2	12,278	5.5	p < 0.0001
LRTI excluding pneumonia	841,743	71.8	222,151	70.1	p < 0.0001	13,255	28.8	34,085	70.3	174,811	78.6	p < 0.0001
Peripheral vascular disease	122,906	10.5	31,331	9.9	p < 0.0001	3,157	6.9	8,480	17.5	19,694	8.9	p < 0.0001
Pneumonia	127,143	10.9	53,779	17.0	p < 0.0001	4,633	10.1	14,580	30.1	34,566	15.5	p < 0.0001

COPD, chronic obstructive pulmonary disease; LRTI, lower respiratory tract infection.

^a p-value for gender was calculated for the whole category, rather than for the individual subgroups.

^b p-value for age at diagnosis was calculated for the whole category, rather than for the individual age subgroups.

The majority of patients were female (54% of COPD alone, 65% of asthma/COPD overlap) with a similar age distribution between patients with COPD alone and asthma/COPD overlap (Table 1). Within the asthma/COPD overlap overlap groups, ages were similar across strata, except the group coded with COPD and chronic obstructive asthma diagnoses. These patients were more likely to be 65+ years (39%) compared with the other asthma/COPD groups (18–20%) and COPD (26%). The distribution of patients aged 50–64 years was similar among all the asthma/COPD overlap groups and slightly higher than the COPD only group. The proportion of patients aged 18–49 years was similar among all the asthma/COPD groups and COPD (35–39%) except for the group coded as obstructive asthma/COPD in which the proportion was half (16%) that of the other groups.

Co-morbidities and disease burden indicators

Patients with asthma/COPD overlap had higher anxiety (15% vs. 11%), depression (17% vs. 12%), diabetes (23% vs. 19%), and pneumonia (17% vs. 11%) compared with patients with COPD alone; but other co-morbidities were similar between groups (Table 1). Patients with asthma/COPD overlap additionally had higher rates of an inpatient hospital visit (33% vs. 25%), shortness of breath (2+ codes: 20% vs. 10%), oxygen therapy (9% vs. 6%), SABD use (63% vs. 28%), and OCS dispensing (2+: 30% vs. 12%) compared with COPD only (Table 2).

Table 2. Disease burden indicators in a cohort of patients with COPD and asthma/COPD overlap stratified by overlap diagnoses: United States, 2008–2011.

						Asthma/COPD overlap groups
	COPD only (n = 1,171,626)		Asthma/COPD overlap, total (n = 316,987)			Chronic obstructive asthma only (n = 45,988)		Chronic obstructive asthma + COPD (n = 48,489)		COPD + ≥1 asthma (n = 222,510)
	N	%	N	%	COPD only vs. asthma/COPD overlap, total	N	%	N	%	N	%	Asthma/COPD overlap
Inpatient hospital visit	297,767	25.4	105,582	33.3	p < 0.0001	12,640	27.5	29,482	60.8	63,460	28.5	p < 0.0001
Shortness of breath: Mean (SD)	2.3 (2.7)		2.6 (3.2)		—	2.2 (2.3)		3.4 (4.5)				—
Shortness of breath:					p < 0.0001^a							p < 0.0001^a
0	925,236	79.0	199,230	62.9	—	33,417	72.7	20,502	42.3	145,311	65.3	—
1	130,793	11.2	54,578	17.2	—	6,612	14.4	9,640	19.9	38,326	17.2	—
2+	115,597	9.9	63,179	19.9	—	5,959	13.0	18,347	37.8	38,873	17.5	—
Oxygen therapy	68,802	5.9	29,022	9.2	p < 0.0001	1,735	3.8	10,709	22.1	16,578	7.5	p < 0.0001
SABD	325,941	27.8	200,413	63.2	p < 0.0001	26,339	57.3	33,735	69.6	140,339	63.1	p < 0.0001
OCS count:					p < 0.0001^b							p < 0.0001^b
OCS count: 0	837,816	71.5	157,477	49.7	—	26,330	57.3	20,357	42.0	110,790	49.8	—
OCS count: 1	193,143	16.5	63,626	20.1	—	8,638	18.8	8,846	18.2	46,142	20.7	—
OCS count: 2+	140,667	12.0	95,884	30.3	—	11,020	24.0	19,286	39.8	65,578	29.5	—
ED visit	506,882	43.3	169,552	53.5	p < 0.0001	20,053	43.6	32,069	66.1	117,430	52.8	p < 0.0001

COPD, chronic obstructive pulmonary disease; ED, emergency department; OCS, oral corticosteroid; SABD, short-acting bronchodilator; SD, standard deviation.

^a p-value for shortness of breath was calculated for the whole variable, rather than the individual counts.

^b p-value for OCS count was calculated for the whole variable, rather than the individual counts.

Within the asthma/COPD overlap group, patients coded with COPD and chronic obstructive asthma tended to have greater co-morbid conditions and disease burden indicators compared with other asthma/COPD classification groups, as well as patients with only COPD, with the exception of lower respiratory tract infections and anxiety (Table 3).

Table 3. Disease burden indicators in a cohort of patients with COPD and asthma/COPD overlap stratified by overlap diagnoses and age: United States, 2008–2011.

							Asthma/COPD overlap
		COPD only (n = 1,171,626)		Asthma/COPD overlap (n = 316,987)			Chronic obstructive asthma only (n = 45,988)		Chronic obstructive asthma + COPD (n = 48,489)		COPD + ≥1 asthma (n = 222,510)
	Age category years	N	%	N	%	COPD only vs. asthma/COPD overlap, total	N	%	N	%	N	%	Asthma/COPD overlap groups
Oxygen	18–49	6,287	0.5	4,239	1.3	p < 0.0001	481	1.1	906	1.9	2,852	1.3	p < 0.0001
	50–64	24,757	2.1	12,675	4.0	p < 0.0001	841	1.8	4,669	9.6	7,165	3.2	p < 0.0001
	65+	37,758	3.2	12,108	3.8	p < 0.0001	413	0.9	5,134	10.6	6,561	3.0	p < 0.0001
SABD	18–49	99,057	8.5	69,278	21.9	p < 0.0001	10,772	23.4	5,022	10.4	53,484	24.0	p < 0.0001
	50–64	127,545	10.9	83,126	26.2	p < 0.0001	11,023	24.0	14,518	30.0	57,585	25.9	p < 0.0001
	65+	99,339	8.5	48,009	15.2	p < 0.0001	4,544	9.9	14,195	29.3	29,270	13.2	p < 0.0001
Inpatient hospital visit	18–49	54,935	4.7	24,690	7.8	p < 0.0001	3,673	8.0	3,626	7.5	17,391	7.8	p < 0.0001
	50–64	103,746	8.9	43,737	13.8	p < 0.0001	5,575	12.1	12,684	26.2	25,478	11.5	p < 0.0001
	65+	139,086	11.9	37,155	11.7	p < 0.0001	3,392	7.4	13,172	27.2	20,591	9.3	p < 0.0001
OCS count 1	18–49	69,269	5.9	23,313	7.4	p < 0.0001^a	3,513	7.6	1,314	2.7	18,486	8.3	p < 0.0001^a
	50–64	76,650	6.5	25,705	8.1	p < 0.0001^a	3,555	7.7	3,783	7.8	18,367	8.3	p < 0.0001^a
	65+	47,224	4.0	14,608	4.6	p < 0.0001^a	1,570	3.4	3,749	7.7	9,289	4.2	p < 0.0001^a
OCS count 2+	18–49	40,700	3.5	31,282	9.9	p < 0.0001^a	4,326	9.4	3,017	6.2	23,939	10.8	p < 0.0001^a
	50–64	57,415	4.9	41,538	13.1	p < 0.0001^a	4,919	10.7	8,455	17.4	28,164	12.7	p < 0.0001^a
	65+	42,552	3.6	23,064	7.3	p < 0.0001^a	1,775	3.9	7,814	16.1	13,475	6.1	p < 0.0001^a
Shortness of breath 1	18–49	29,892	2.6	17,134	5.4	p < 0.0001^b	2,409	5.2	1,509	3.1	13,216	5.9	p < 0.0001^b
	50–64	52,941	4.5	23,815	7.5	p < 0.0001^b	2,906	6.3	4,393	9.1	16,516	7.4	p < 0.0001^b
	65+	47,960	4.1	13,629	4.3	p < 0.0001^b	1,297	2.8	3,738	7.7	8,594	3.9	p < 0.0001^b
Shortness of breath 2+	18–49	13,988	1.2	14,019	4.4	p < 0.0001^b	1,953	4.3	2,412	5.0	9,654	4.3	p < 0.0001^b
	50–64	43,000	3.7	27,602	8.7	p < 0.0001^b	2,652	5.8	8,035	16.6	16,915	7.6	p < 0.0001^b
	65+	58,609	5.0	21,558	6.8	p < 0.0001^b	1,354	2.9	7,900	16.3	12,304	5.5	p < 0.0001^b
ED visit	18–49	164,754	14.1	59,074	18.6	p < 0.0001	7,900	17.2	5,150	10.6	46,024	20.7	p < 0.0001
	50–64	180,932	15.4	67,578	21.3	p < 0.0001	8,219	17.9	13,770	28.4	45,589	20.5	p < 0.0001
	65+	161,196	13.8	42,900	13.5	p < 0.0001	3,934	8.6	13,149	27.1	25,817	11.6	p < 0.0001
Congestive heart failure	18–49	6,177	0.5	3,133	1.0	p < 0.0001	398	0.9	648	1.3	2,087	0.9	p < 0.0001
	50–64	32,359	2.8	13,007	4.1	p < 0.0001	1,326	2.9	3,993	8.2	7,688	3.5	p < 0.0001
	65+	72,075	6.2	19,006	6.0	p < 0.0001	1,450	3.2	6,834	14.1	10,722	4.8	p < 0.0001
LRTI	18–49	361,062	30.8	86,242	27.2	p < 0.0001	5,472	11.9	5,962	12.3	74,808	33.6	p < 0.0001
	50–64	325,127	27.8	93,208	29.4	p < 0.0001	5,858	12.7	15,614	32.2	71,736	32.2	p < 0.0001
	65+	155,554	13.3	42,701	13.5	p < 0.0001	1,925	4.2	12,509	25.8	28,267	12.7	p < 0.0001
Pneumonia+	18–49	24,173	2.1	13,309	4.2	p < 0.0001	1,479	3.2	1,916	3.95	9,914	4.5	p < 0.0001
	50–64	45,329	3.9	22,399	7.1	p < 0.0001	2,015	4.4	6,042	12.5	14,342	6.5	p < 0.0001
	65+	57,641	4.9	18,071	5.7	p < 0.0001	1,139	2.5	6,622	13.7	10,310	4.6	p < 0.0001

COPD, chronic obstructive pulmonary disease; ED, emergency department; LRTI, left respiratory tract infection; OCS, oral corticosteroid; SABD, short-acting bronchodilator.

^a p-value for OCS count was calculated for the whole variable, rather than the individual counts.

^b p-value for shortness of breath was calculated for the whole variable, rather than the individual counts.

Respiratory medications

The majority (74%) of patients with COPD alone had zero pharmacy dispensing for a respiratory medication in the 6-month window around the index date compared with 42% of patients with asthma/COPD overlap (Table 4). Within the asthma/COPD overlap groups, the proportion of patients not dispensed one of the respiratory treatments ranged from 34% to 45%. The most frequently dispensed treatments for patients with COPD alone and patients with asthma/COPD overlap were SABD only (15% COPD alone, 22% overlap) and ICS/LABA (5% COPD alone, 22% overlap). Among patients with COPD alone who filled a prescription for a respiratory medication, 59% received SABD only in the 6-month window and 6% received triple therapy. Among patients with asthma/COPD overlap who filled a prescription, 37%, 3%, 37%, and 6% were dispensed SABD only, a LAMA, an ICS/LABA, and triple therapy, respectively.

Table 4. Respiratory medications prescribed in the 6-month window around the index date, stratified by overlap classification.

							Asthma/COPD overlap
	COPD only (n = 1,171,626)		Asthma/COPD overlap total (n = 316,987)			Chronic obstructive asthma only (n = 45,988)		COPD + chronic obstructive asthma (n = 48,489)		COPD + ≥1 asthma (n = 222,510)
Therapy of interest	N	%	N	%	COPD only vs. asthma/COPD overlap, total	N	%	N	%	N	%	Asthma/COPD overlap groups
SABD	180,535	15.4	69,056	21.8	p < 0.0001	8,323	18.1	8,651	17.8	52,082	23.4	p < 0.0001
LABA	1,743	0.2	1,322	0.4	p < 0.0001	179	0.4	322	0.7	821	0.4	p < 0.0001
LAMA	24,427	2.1	6,590	2.1	p = 0.8357	459	1.0	2,196	4.5	3,935	1.8	p < 0.0001
ICS	21,443	1.8	24,893	7.9	p < 0.0001	4,124	9.0	3,353	6.9	17,416	7.8	p < 0.0001
LABA/LAMA	1,077	0.1	384	0.1	p < 0.0001	17	0.0	171	0.4	196	0.1	p < 0.0001
ICS/LABA	54,552	4.7	69,678	22.0	p < 0.0001	11,527	25.1	12,119	25.0	46,032	20.7	p < 0.0001
ICS/LAMA	2,238	0.2	1,646	0.5	p < 0.0001	155	0.3	623	1.3	868	0.4	p < 0.0001
ICS/LABA/LAMA	17,723	1.5	11,314	3.6	p < 0.0001	735	1.6	4,670	9.6	5,909	2.7	p < 0.0001
No treatment	867,888	74.1	132,104	41.7	p < 0.0001	20,469	44.5	16,384	33.8	95,251	42.8	p < 0.0001

COPD, chronic obstructive pulmonary disease; ICS, inhaled corticosteroid; LABA, long-acting beta2-agonist; LAMA, long-acting muscarinic antagonist; SABD, short-acting bronchodilator.

For patients with asthma/COPD overlap coded as COPD and chronic obstructive asthma, or chronic obstructive asthma only, ICS/LABA was the most common medication (25%), while for those classified by COPD and ≥1 asthma code, SABD only (23%) was more common than ICS/LABA (21%). Triple therapy was more likely to be used by treated patients coded as COPD and chronic obstructive asthma (10%) than other asthma/COPD groups (2–3%).

Sensitivity analysis

The age distribution of patients varied by the definition of asthma/COPD overlap, and the group coded with COPD and chronic obstructive asthma had almost double the proportion of people over the age of 65 years compared with the other three overlap definitions (Table 1). Sensitivity analyses were conducted to better understand the impact of this age distribution on the observed differences in disease burden and co-morbidities associated with this COPD and chronic obstructive asthma group compared with other asthma/COPD overlap definitions.

Among patients over the age of 65 years, the COPD and chronic obstructive asthma overlap group had the highest proportion of co-morbid congestive heart failure (CHF), lower respiratory tract infection, and pneumonia compared with all other overlap definitions (14.1%, 25.8%, and 13.7%, respectively). This group also had the highest proportion of disease burden indicators among patients over the age of 65 years including oxygen therapy (10.6%), SABD (29.3%), inpatient hospital visit (27.2%), OCS use (1 code 7.7% and 2+ codes 16.1%), shortness of breath (1 code 7.7% and 2+ codes 16.3%), and ED visits (27. 1%). Increased frequencies of these disease burden indicators were also observed among patients with chronic obstructive asthma and COPD in the 50–64 age strata compared with other asthma/COPD overlap definitions, where the proportion of patients in the age strata was similar across all definitions of asthma/COPD overlap. The same trend was observed regarding comorbid conditions; the COPD and chronic obstructive asthma overlap group often had twice the prevalence of co-morbid conditions and disease burden. CHF (8.2%), pneumonia (12.5%), oxygen therapy (9.6%), inpatient hospital visit (26.2%), OCS count 2+ (17.4%), shortness of breath (2+ codes,16.6%), and ED visits (28.4%) were all higher than those expected among the COPD and chronic obstructive asthma overlap group based on age distribution.

Discussion

Prior literature has shown that patients with asthma/COPD overlap have worse outcomes including exacerbations, quality of life, perceived health status, hospitalization rates, and symptoms compared with persons with COPD alone (10–12). Our results confirm these observations and consistently demonstrated higher disease burden indicators in patients with asthma/COPD overlap, including more frequent inpatient hospital stay, shortness of breath codes, oxygen therapy, use of OCS, and ED visits as compared with COPD alone. Interestingly, this trend was observed regardless of the definition of asthma/COPD overlap applied; an exception was the marker of oxygen use, which was coded more frequently for patients with chronic obstructive asthma only. Among the patients with asthma/COPD overlap, patients with obstructive asthma and COPD had greater disease burden and co-morbidities; this association may be related to the older age of that group, or the combination of codes may represent an asthma/COPD overlap population with more severe symptoms and overall disease burden. The data seem to indicate that among patients who received a code for chronic obstructive asthma, there was a spectrum of severity; those who received a code for only chronic obstructive asthma had less severe disease burden than those with a concomitant COPD code. It is plausible that patients coded with chronic obstructive asthma only may have evidence of obstruction, but their clinician may not feel that they meet the criteria for COPD yet. This group of patients was also the youngest, which may indicate that younger patients may not be receiving a COPD diagnosis. In contrast, the group coded with chronic obstructive asthma and COPD had the highest disease burden and was older than the other asthma/COPD groups.

Since the chronic obstructive asthma and COPD overlap group had a higher proportion of older patients than other overlap categories and the patients with COPD only, it is expected that this group would have greater co-morbidities, medication use and disease burden indicators as compared with the others. However, examination by age showed that these frequencies not only remained highest among the older asthma and COPD overlap groups, but also were higher than those expected based on the distribution of age. For example, based on the distribution of patients over the age of 65 years and the prevalence of co-morbidities of overlap groups, we would expect the patients with chronic obstructive asthma and COPD overlap to have approximately twice the prevalence of co-morbidities because the proportion of patients over the age of 65 years is twice as large in this overlap definition compared with other overlap definitions.

The distribution of age also did not fully account for the increased frequency of co-morbidities, medications, and disease indicators among the chronic obstructive asthma and COPD overlap group. Although the chronic obstructive asthma and COPD overlap group are older than the other overlap groups, the larger than expected disease burden, co-morbidities, and medication use suggest that the combination of chronic obstructive asthma and COPD codes represents a population with more severe symptoms and overall disease burden.

Published studies are conflicting with regard to the prevalence of co-morbidities in patients with asthma/COPD compared with patients who have COPD. One study using information from a self-reported questionnaire showed that patients with asthma/COPD overlap in the United States experienced a higher age-adjusted prevalence of a variety of co-morbidities including diabetes, heart disease, stroke, arthritis, mental distress, and high blood pressure, compared with patients with COPD or asthma alone (20). A recent retrospective cohort study found that allergic rhinitis, anxiety, gastroesophageal reflux disease, and osteoporosis were more frequently observed in patients with ACOS compared with patients with COPD. However, chronic kidney disease and ischemic heart disease were observed less frequently in patients with ACOS than those with COPD. The study also demonstrated that overall, cardiovascular diseases showed the strongest association with hospitalization risk in patients with ACOS (21). In addition, another study using a US Medicaid database demonstrated that patients with asthma/COPD overlap had similar rates of coexisting diseases as measured by the Charlson Comorbidity Index (CCI), and the mean CCI for patients with COPD and ACOS was 2.8 and 2.7, respectively (12). The current study results showed slightly higher percentages of diabetes, CHF, arrhythmia, and anxiety among patients with asthma/COPD overlap compared with patients who had COPD alone. The conflicting results may be due to the nature of data collection about co-morbidities across the studies or the definition of asthma/COPD overlap that was used in the studies.

Little is known about treatment in patients with asthma/COPD overlap compared with COPD alone. Patients in the asthma/COPD groups were more likely to be treated with an ICS-containing medication, which is more consistent with the asthma treatment paradigm than the COPD paradigm where ICS is only indicated in those at higher risk for exacerbations. Physicians may be treating this asthma/COPD group with ICS due to the asthmatic component of their disease and/or to prevent exacerbations. Additionally, the obstructive asthma and COPD overlap group had the most treatment with prescription fills for triple therapy as compared with the other overlap groups, which is consistent with the indications that this group was treated as having more severe disease from our analysis of co-morbidities and disease burden.

Interestingly, a much greater proportion of patients (74.1%) with COPD only had no pharmacy-dispensed prescriptions of respiratory medication during the 6-month period around the index date, compared with patients with asthma/COPD overlap (41.7%). This may be a reflection of the previously discussed increased disease burden of patients with asthma/COPD overlap compared with COPD alone. As a result, patients with COPD only may have been less likely to be prescribed the medication, compared with those with asthma/COPD overlap. However, other factors should also be considered. As patients in the asthma/COPD overlap group had likely been diagnosed with asthma prior to the study period, these patients may have been more likely to be receiving treatment at the time of the index date. Additionally, it is likely that patients in the asthma/COPD group had more visits to healthcare practices due to the nature of having multiple diagnoses, which potentially increases the likelihood of receiving a prescription. Finally, the study excluded patients with a diagnostic code for a medical condition incompatible with a COPD diagnosis at any time in their history. This may have resulted in some patients diagnosed with COPD only in the index period who received respiratory treatment being excluded from the analysis.

With improved information on asthma/COPD overlap, further prospective studies can be designed to explore the most appropriate targets and pathways to optimize treatment paradigms for patients with asthma/COPD overlap and ultimately improve outcomes. Proper treatment and prevention of asthma/COPD overlap requires well-defined, universal diagnostic criteria, as well as comprehensive therapeutic intervention paradigms.

There are limitations to the current study. The database captures diagnoses, procedures, and dispensed prescriptions in a US commercially insured population, and the Medicare database contains records for patients aged 65+ years, but generalizability of the descriptive findings outside the US commercially insured population and outside the population with Medicare may be limited as practices may be different in this population compared with other US populations and the rest of the world.

Coding of asthma, COPD, and chronic obstructive asthma may differ across insurance plans and/or be related to physician practices. As cohorts were defined by diagnoses for asthma or chronic obstructive asthma during a 24-month window around COPD or chronic obstructive asthma diagnosis, codes outside this window were not considered, and a patient needed to have received a diagnosis code, so it is possible that patients may have been misclassified. Objective clinical testing such as spirometry would aid in classification of these patients; however, values for this test are not available from the MarketScan data for many patients. Drug exposure was based on dispensed prescriptions, and it is possible that medications were not claimed through insurance and therefore not recorded in the data source. For example, medications administered in hospital were not available. Additionally, patients with dispensed prescriptions also may not have used the medication or used it less frequently than prescribed, as well as, potentially being misclassified regarding treatment group if they received medication prescription medications dispensed outside of the 6-month window.

Conclusions

Patients with medical codes for asthma/COPD overlap have higher disease burden indicators and ICS/LABA use as compared with COPD alone. This trend was consistent for all definitions of asthma/COPD overlap. Patients coded with obstructive asthma and COPD were older and had evidence of higher disease burden as compared with the other overlap definitions. Independent of older age, the chronic obstructive asthma and COPD overlap population may represent a more severe form of asthma/COPD overlap. More clinical insight and detailed phenotyping are needed to determine the reasons for coding variation observed in asthma/COPD overlap with implications for further research to better understand the natural history and address unmet needs across the diverse group of patients with chronic obstructive lung disease with asthma features.

Acknowledgments

Natasha Thomas, PhD of Fishawack Indicia Ltd, provided editorial support for this manuscript in the form of formatting the manuscript to journal requirements. The authors wish to acknowledge the following individuals for their input and critical review during the development of this manuscript: Kaitlin Kelly-Reif and Greta Bushnell.

Funding

This work was supported by GSK, study number ODA2451 (previously ODA2263).

Declaration of interest

KEW, SSL, DRH, and KJD are employees of GSK, the study sponsor. SSL, DRH, and KJD own stocks/shares in GSK.