Association of Opioid and Benzodiazepine Use with Adverse Respiratory Events in Older Adults with Chronic Obstructive Pulmonary Disease

Abstract

Rationale: Older adults with chronic obstructive pulmonary disease (COPD) are at substantially increased risk for medication-related adverse events. Two frequently prescribed classes of drugs that pose a particular risk to this patient group are opioids and benzodiazepines. Research on this topic has yielded conflicting findings.

Objectives: The purpose of this study was to examine, among older adults with COPD, whether: 1) independent or concurrent use of opioid and benzodiazepine medications was associated with hospitalizations for respiratory events, and 2) this association was exacerbated by the presence of obstructive sleep apnea (OSA).

Methods: We conducted a case–control study of Medicare beneficiaries aged ≥66 years, who were diagnosed with COPD in 2013, using the 5% national Medicare database. Cases (n = 3,232) were defined as patients hospitalized for a primary COPD-related respiratory diagnosis in 2014 and were matched with up to two control subjects (n = 6,247) on index date, age, sex, socioeconomic status, comorbidity, presence of OSA, COPD medication, and COPD complexity.

Results: In comparison to the referent (no opioid or benzodiazepine use), opioid use alone (adjusted odds ratio [aOR], 1.73; 95% confidence interval [CI], 1.52–1.97), benzodiazepine use alone (aOR, 1.42; 95% CI, 1.21–1.66), and concurrent opioid/ benzodiazepine use (aOR, 2.32; 95% CI, 1.94–2.77) in the 30 days before the event/index date were all associated with an increased risk of hospitalization for a respiratory condition. Risk of hospitalization was higher with concurrent opioid and benzodiazepine use when compared with use of either medication alone. There was no statistically significant interaction between OSA and either of the drugs, alone or in combination. However, the adverse respiratory effects of concurrent opioid and benzodiazepine use were increased in patients with a high degree of COPD complexity. All of the above findings persisted using exposure windows that extended to 60 and 90 days before the event/index date.

Conclusions: Among older adults with COPD, use of opioid and benzodiazepine medications alone or in combination were associated with increased adverse respiratory events. The adverse effects of these medications were not exacerbated in patients with COPD–OSA overlap syndrome. However, the adverse impact of dual opioid and benzodiazepine was greater in patients with high-complexity COPD.

Chronic obstructive pulmonary disease (COPD) increases with age (1, 2) and is associated with multiple comorbidities (1, 3–5) and polypharmacy (3, 5). Older adults with COPD are therefore at substantially increased risk for medication-related adverse events (5). Two frequently prescribed classes of drugs that pose a particular risk to this patient group are opioids (6–9) and benzodiazepines (10, 11).

Dyspnea is a significant cause of impaired activity and poor quality of life in patients with advanced COPD (12, 13). Oral opioids are recommended as a potential treatment for refractory dyspnea in patients with advanced disease (12, 14, 15). Likewise, benzodiazepines are often prescribed to patients with COPD for refractory dyspnea (16, 17), as well as for anxiety (11, 18) and insomnia (18). A complex bidirectional relationship may occur in such patients, as anxiety may be both a cause and a consequence of dyspnea (19). Although current practice guidelines recommend against routine use of opioids or benzodiazepines to improve dyspnea and quality of life in patients with COPD (12, 15), studies assessing the adverse respiratory effects of these drugs in patients with COPD—both randomized controlled trials (16, 20) and observational studies (8–10, 14, 21–24)—have yielded conflicting results. Specifically, among patients with COPD, some studies have reported adverse respiratory effects of opioids (8, 9), benzodiazepines (10, 21), and the two drugs combined (10), and other studies have reported no such effects (16, 20, 23). Although limited by small sample sizes, some recent studies have even shown modest improvement of dyspnea with low doses of morphine (20) and with benzodiazepines (16, 23).

Obstructive sleep apnea (OSA) is prevalent in patients with COPD (25–27) but is often undiagnosed and untreated (25, 26, 28). COPD–OSA overlap syndrome refers to the coexistence of these two conditions and is associated with a severe clinical course (29, 30). Given the adverse impact of opioids and benzodiazepines on upper airway muscle tone, central chemosensitivity, and arousability (31–35), patients with overlap syndrome may be at particularly high risk of adverse respiratory events when prescribed these medications. The current literature on this topic, however, is conflicting (31–38).

To date, no large-scale or nationally representative study has examined the independent and concomitant effects of opioids and benzodiazepines on patients with COPD alone and COPD–OSA overlap syndrome. Given the high prevalence of these conditions and the frequent use of these medications in older adults, understanding these associations holds broad clinical and public health significance. We therefore sought to examine these issues in a large and nationally representative sample of Medicare beneficiaries.

Methods

Data Source

This case–control study used administrative health data from the 5% national Medicare database. The Medicare database contains eligibility, inpatient and outpatient claims, and medications from retail pharmacies through a beneficiary’s Part D benefit. For each medication, the database contains medication name, date of fill, formulation (e.g., oral, transdermal, injectable), dose, quantity, and days of supply. This study was approved by the institutional review board of the University of Texas Medical Branch at Galveston (project approval number 16-0068). All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Review Board of University of Texas Medical Branch and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Study Cohort

We identified 33,240 Medicare beneficiaries aged ≥66 years who were diagnosed with COPD in 2013 and had complete enrollment data for 2013. Patients who met any of the following criteria were classified as having COPD: 1) at least two outpatient or consultation visits (Evaluation and Management codes 99201–99205, 99211–99215), each with a diagnosis of COPD at least 30 days apart within 12 months; or 2) one acute care hospitalization with a primary discharge diagnosis of COPD on the basis of the following International Classification of Diseases, Ninth Revision (ICD-9) codes: 491.x (chronic bronchitis), 492.x (emphysema), or 496 (chronic airway obstruction); or 3) hospitalization for respiratory failure (ICD-9 codes 518.81, 518.82, 518.84) listed as the primary diagnosis at hospital discharge and COPD listed as the secondary diagnosis at hospital discharge.

Cases

Within the 2013 COPD cohort, we identified 3,232 beneficiaries who were hospitalized for a respiratory condition in 2014 on the basis of having an ICD-9 code for respiratory disease (ICD-9 460.xx–519.xx) in the primary diagnosis position of their hospital discharge diagnosis. The date of admission served as the event date for all analyses.

Control Subjects

We identified 6,247 control subjects who were not hospitalized for a respiratory condition in 2014 using the criteria described above. We used the following stepwise algorithm, broadening the matching criterion at each stage. First, we matched 95.5% of cases with control subjects on event/index month, age at diagnosis/index date (±5 yr), sex, comorbidity (0, 1, ≥2), OSA, COPD medication, COPD complexity, and socioeconomic status (SES). Second, we matched 2.3% of the cases with control subjects on all of the above criteria except SES. Third, we matched 0.8% of the cases with control subjects on all of the second-step criteria except COPD complexity. Fourth, we matched 1.5% of the cases with control subjects on all of the third-step criteria except comorbidity. Finally, we matched the last 0.03% of control subjects with cases on only age (±5 yr) and sex. Overall, the matching process resulted in 96.7% of cases having two matches and 3.3% having only one match.

Exposure

Patients whose most recent prescription period for an opioid alone, a benzodiazepine alone, or both an opioid and a benzodiazepine within the 30-day period before the event/index date were defined as exposed in each of the aforementioned categories. We included all doses and formulations of the above pharmacotherapies in our analyses. We also conducted sensitivity analyses in which we examined exposure windows that extended to 60 and 90 days before the event/index date.

Covariates

Geographic region was divided into four census bureau regions. The Elixhauser comorbidity index, except for COPD, was identified from both outpatient and inpatients claims to generate the number of comorbidities for beneficiaries as an index score (0, 1, 2, ≥ 3) (39). Sociodemographic characteristics including age at hospitalization/index and race and ethnicity (white, black, or other) were obtained from Medicare enrollment files. Low SES was identified by either a patient’s eligibility for state buy-in coverage provided by the Medicare program or a patient’s receipt of a low-income subsidy for the Medicare Part D program. Presence of OSA was defined as having at least one outpatient visit with an encounter diagnosis of OSA (ICD-9 codes 780.51, 780.53, 780.57, 786.03, 327.20, 327.23) (27, 40). Receipt of positive airway pressure (PAP) therapy was assessed using Healthcare Common Procedure Coding System (HCPCS) codes (E0470, E0471, E0601, E0561, E0562) in the Medicare durable equipment files.

Receipt of COPD medication was defined as having received at least one prescription for any of the following: long-acting β-agonists, long-acting muscarinic agonists, inhaled corticosteroids, or fixed-dose long-acting β-agonists/inhaled corticosteroid combinations. For each year, the proportion of days covered (PDC), defined as the proportion of days that a patient had COPD maintenance medication available in the year before respiratory hospitalization, was calculated. For prescriptions extending beyond the respiratory hospitalization, the days covered were truncated at the date of hospitalization. Beneficiaries were classified as having no COPD medication, COPD medication with <50% PDC, and COPD medication with ≥50% PDC.

COPD Complexity

A classification of COPD complexity, on the basis of administrative claims data, was created to serve as an indicator of COPD severity. Comorbid respiratory conditions and medical procedures during the 12 months before the hospitalization/index date served as a basis for assigning patients to one of three COPD complexity levels (high, moderate, or low), as described in previous studies (41). High-complexity conditions included cor pulmonale, tuberculosis, or malignant neoplasm; moderate-complexity conditions included pneumonia, cyanosis, bronchoscopy, or dependence on supplemental oxygen. All patients with COPD who did not have high- or moderate-complexity conditions were classified as low complexity.

Statistical Analysis

We used conditional logistic regression models to estimate the risk of respiratory hospitalization associated with opioid exposure, benzodiazepine exposure, and dual opioid–benzodiazepine exposure expressed as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). All multivariable analyses were adjusted for age, race, region, SES, comorbidity, OSA, and COPD complexity. We also conducted analyses stratified on age, sex, and COPD complexity. All analyses were performed using Statistical Analysis System version 9.4 (SAS Institute).

Results

Table 1 shows the baseline characteristics of the study sample for cases versus control subjects. The mean age was 77 years, and 60% were women. The majority of subjects in the sample were non-Hispanic white and had three or more other comorbidities. The standardized mean differences show that, with the exception of region, all of the study variables were distributed relatively evenly across cases and control subjects. Table 2 shows that, in the 30 days before the event/index date, opioid use alone (aOR, 1.73; 95% CI, 1.52–1.97), benzodiazepine use alone (aOR, 1.42; 95% CI, 1.21–1.66), and concurrent opioid and benzodiazepine use (aOR, 2.32; 95% CI, 1.94–2.77) were all associated with an increased risk of hospitalization for a respiratory event. Likewise, in the 60 days before the event/index date, opioid use alone (aOR, 1.66; 95% CI, 1.47–1.88), benzodiazepine use alone (aOR, 1.44; 95% CI, 1.24–1.67), and concurrent opioid and benzodiazepine use (aOR, 2.27; 95% CI, 1.93–2.67) were all associated with an increased risk of hospitalization for a respiratory event. Finally, in the 90 days before the event/index date, opioid use alone (aOR, 1.58; 95% CI, 1.40–1.78), benzodiazepine use alone (aOR, 1.40; 95% CI, 1.20–1.63), and concurrent opioid and benzodiazepine use (aOR, 2.21; 95% CI, 1.88–2.59) were associated with an increased risk of hospitalization for a respiratory event. In each of these analyses, concurrent opioid and benzodiazepine use remained statistically significantly elevated when compared with either of the benzodiazepine-alone or opioid-alone referent groups. In each of the above outcome categories, there was no statistically significant interaction between OSA and either of the drugs, alone or in combination. This finding persisted even after adjusting for PAP. Finally, our analyses stratified by COPD complexity (Figure 1) showed that the adverse respiratory impact of concurrent opioid and benzodiazepine was greater in patients with high COPD complexity than in patients with low or moderate COPD complexity.

Table 1.

Baseline characteristics for cases and matched control subjects

	Matched Control Subjects n (%)	Cases n (%)	Standardized Mean Difference (%)
All	6,247 (100)	3,232 (100)	—
Sex			0.7
Female	6,740 (59.8)	1,921 (59.4)
Male	2,517 (40.2)	1,311 (40.6)
Age, mean (SD)	77.0 (6.8)	76.6 (6.9)	−6.4
Race group			6.5
Black	352 (5.6)	224 (6.9)
Other	197 (3.2)	122 (3.8)
White	5,708 (91.2)	2,886 (89.3)
Region			22.7
Midwest	1,565 (25.0)	802 (24.8)
Northeast	1,619 (25.9)	556 (17.2)
South	2,498 (39.9)	1,486 (46.0)
West	575 (9.2)	388 (12.0)
Low SES			6.0
No	4,570 (73.0)	2,274 (70.4)
Yes	1,687 (27.0)	958 (29.6)
Comorbidity			5.5
0–2	845 (13.5)	404 (12.5)
3–4	1,775 (28.4)	880 (27.2)
5–6	1809 (28.9)	928 (28.7)
≥7	1,828 (29.2)	1,020 (31.6)
COPD complexity			−5.4
Low/moderate	5,225 (83.5)	2,633 (81.5)
High	1,032 (16.5)	599 (18.5)
COPD medication*			1.6
None	414 (6.6)	206 (6.4)
<50%	2,673 (42.7)	1,404 (43.4)
≥50%	3,170 (50.7)	1,622 (50.2)
OSA			2.0
No	5,422 (86.6)	2,779 (86.0)
Yes	835 (13.4)	453 (14.0)
PAP therapy	257 (30.8)	135 (29.6)
No PAP therapy	578 (69.2)	319 (70.4)

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; OSA = obstructive sleep apnea; PAP = positive airway pressure; SD = standard deviation; SES = socioeconomic status.

^*Proportion of days covered before the index date.

Table 2.

Association between opioid/benzodiazepine exposure and respiratory hospitalization

Drug Exposure	Control Subjects (n = 6,247) No. (%)	Cases (n = 3,232) No. (%)	Unadjusted OR (95% CI)	Adjusted* OR (95% CI)	Drug–OSA Interaction P Value
30-d exposure†	4,431 (70.8)		1.00 (Referent)	1.00 (Referent)	0.86
None	584 (9.3)	1,908 (59.0)	1.73 (1.54–1.94)	1.73 (1.52–1.97)
Opioid only	875 (14.0)	640 (19.8)	1.40 (1.21–1.62)	1.42 (1.21–1.660
Benzodiazepine only	367 (5.9)	342 (10.6)	2.34 (1.98–2.76)	2.32 (1.94–2.77)
Both		342 (10.6)
60-d exposure‡	4,092 (65.4)				0.28
None	640 (10.2)	1,729 (53.5)	1.00 (Referent)	1.00 (Referent)
Opioid only	1,050 (16.8)	713 (22.1)	1.63 (1.45–1.82)	1.66 (1.47–1.88)
Benzodiazepine only	475 (7.6)	364 (11.3)	1.39 (1.20–1.60)	1.44 (1.24–1.67)
Both		429 (13.2)	2.27 (1.96–2.64)	2.27 (1.93–2.67)
90-d exposure§	3,924 (62.7)		1.00 (Referent)	1.00 (Referent)	0.25
None	655 (10.5)	1,663 (51.5)	1.54 (1.38–1.72)	1.58 (1.40–1.78)
Opioid only	1,158 (18.5)	748 (23.1)	1.35 (1.17–1.56)	1.40 (1.20–1.63)
Benzodiazepine only	520 (8.3)	365 (11.3)	2.20 (1.90–2.54)	2.21 (1.88–2.59)
Both		459 (14.1)

Definition of abbreviations: CI = confidence interval; OR = odds ratio; OSA = obstructive sleep apnea.

^*Multivariable analyses were adjusted for race and region.

^†Patients whose most recent prescription period overlapped by at least 1 day within the 30-day period before the event/index date were defined as exposed.

^‡Patients whose most recent prescription period overlapped by at least 1 day within the 60-day period before the event/index date were defined as exposed.

^§Patients whose most recent prescription period overlapped by at least 1 day within the 90-day period before the event/index date were defined as exposed.

Figure 1.

Association between opioid/benzodiazepine use and adverse respiratory events, by chronic obstructive pulmonary disease (COPD) complexity.

Discussion

In this case–control study of 9,479 older adults with COPD, we found that opioid and benzodiazepine use—both alone and in combination—were associated with an increased risk of respiratory hospitalization. The adverse effects of these medications were not exacerbated in patients with COPD–OSA overlap syndrome. However, the adverse impact of dual opioid and benzodiazepine was greater in patients with high-complexity COPD. These findings persisted across three time windows—30, 60, and 90 days—and after adjusting for multiple covariates. To our knowledge, this is the first nationally representative study to examine the independent and concomitant effects of opioid and benzodiazepines on patients with COPD alone and COPD–OSA overlap syndrome.

Overall, our study results were consistent with some studies but in conflict with others. For example, our finding that opioids and benzodiazepine, alone or in combination, were associated with an increased risk of respiratory events was consistent with two large Canadian-based cohort studies (6, 11) but inconsistent with a large Swedish-based cohort study (10). Vozoris and colleagues investigated administrative healthcare data for cohort of older patients with COPD in Ontario. They found that up to two-third of patients had new prescription of opioids (6), and nearly one-third were new users of benzodiazepines (11). Compared with nonusers, new users had more frequent adverse respiratory outcomes with both opioids (8) and benzodiazepines (21) and a significantly higher rate of hospitalization and pneumonia-related, cardiac-related, and all-cause mortality in new opioid users (8, 9). In their nationwide Swedish prospective longitudinal cohort study, Ekström and colleagues (10) reported that lower doses of opioids (≤30 mg morphine/d) improved dyspnea in patients with COPD on long-term oxygen therapy without a significant increase in hospitalizations and mortality. However, benzodiazepines and higher doses of opioids, especially when used concurrently, were associated with higher mortality. Finally, Donovan and colleagues (23), in a large cohort of patients with COPD and post-traumatic stress disorder, observed an increased risk of adverse respiratory outcomes after short-term benzodiazepine use compared with long-term use. It is unclear why prior research on this topic is so divergent. It is possible that these conflicting findings are, to some degree, attributable to variations in medication dose, duration of use, severity of COPD, comorbidities, and missing data.

We observed no statistically significant interaction between OSA and either opioids or benzodiazepines, alone or in combination, with adverse respiratory events. These findings persisted after adjustment for PAP therapy. In view of the previous research showing an adverse impact of opioids and benzodiazepines on upper airway muscle tone, central chemosensitivity, and arousability, patients with overlap syndrome have been believed to be at particularly high risk of adverse respiratory events when prescribed these medications (31, 32, 34, 35). The recent literature on this topic, however, is conflicting. In their randomized controlled trial of 60 men with severe OSA, Roswell and colleagues (36) reported that moderate doses of morphine did not worsen OSA. Likewise, in a study of 10 patients with OSA before versus after receiving morphine, Wang and colleagues (37) observed no increase in adverse respiratory symptoms. In a randomized controlled trial of 20 men with OSA, Wang and colleagues (38) reported that benzodiazepine therapy did not adversely affect respiratory function but that outcomes varied substantially according to OSA phenotype. In attempting to reconcile these disparate findings across studies, it is important to consider that an OSA–drug interaction effect is concentrated among patients with severe OSA. Our inability to examine OSA severity may have limited our analysis. Future investigations with an ability to precisely examine OSA severity may be particularly helpful in assessing this important interaction effect.

Limitations

The results of our study may have been influenced by several limitations. First, the medical conditions under study were assessed using diagnostic coding by clinicians, rather than clinical laboratory testing such as spirometry for COPD or polysomnography for OSA, which can lead to inaccuracies (42–45). Second, it is possible that our observed association between the two medications under study and adverse respiratory events was attributable, at least in part, to confounding by indication. That is, patients with more severe COPD had a higher degree of pain and insomnia. Our adjustment for COPD complexity, however, should have mitigated this potential confounding effect. Third, the Medicare insurance pharmacy plan used by the study participants did not cover over-the-counter medications (e.g., sleep medications) or opioids and benzodiazepines obtained from friends and from illicit sources. Fourth, reliance on claims data precluded assessment of a number of potential confounding factors, such as diet, alcohol, and use of herbal supplements, especially sedating herbs like valerian root and kava. Fifth, our data provide information on the date the prescription was filled but not the date it was picked up or actually used by the patient. It is possible, therefore, that some of the drug exposure periods used in this study underestimated the true medication exposure period. Our sensitivity analyses assessing 30- and 60-day exposure windows, however, help to address this issue. Sixth, our database lacked information on several important risk factor variables, such as smoking status and diet. Seventh, we did not have information on PAP compliance.

Despite these limitations, this study has a number of strengths, including a large representative sample, matching based on sociodemographic and disease risk factors, simultaneous adjustment for potentially confounding medical conditions and medications, and assessment of multiple exposure windows.

In patients with advanced COPD with dyspnea refractory to conventional therapy, nonpharmacological interventions, including pulmonary rehabilitation, chest vibration, pursed-lip breathing, relaxation therapy, and behavioral techniques, should be optimized before consideration of pharmacotherapy with opioids (1, 12). Use of pulmonary rehabilitation continues to be low (46), despite evidence that it reduces symptom burden and healthcare utilization (12, 15, 47). One way to reduce opioid- and benzodiazepine-related toxicity is to encourage patient participation in pulmonary rehabilitation. For management of chronic insomnia—in addition to treatment of COPD-related symptoms like cough and dyspnea to reduce frequent arousals—current data support efficacy of cognitive behavioral therapy, relaxation training, and stimulus control, interventions that may provide sustained benefit over benzodiazepines without concern of side effects (48).

In conclusion, patients with COPD and their physicians should judiciously assess the risks and benefits of opioids and benzodiazepines, alone and in combination, and preferentially recommend nonopioid and nonbenzodiazepine approaches for pain, sleep, and anxiety management in patients with COPD.