Care Quality for Patients with Chronic Obstructive Pulmonary Disease in the Readmission Penalty Era

Abstract

Rationale

Chronic obstructive pulmonary disease (COPD) is the fifth-leading cause of admissions and third-leading cause of readmissions among U.S. adults. Recent policies instituted financial penalties for excessive COPD readmissions.

Objectives

To evaluate changes in the quality of care for patients hospitalized for COPD after implementation of the Hospital Readmissions Reduction Program (HRRP).

Methods

We conducted a retrospective cohort study of patients older than 40 years of age hospitalized for COPD across 995 U.S. hospitals (Premier Healthcare Database).

Measurements and Main Results

Quality of care before and after HRRP implementation was measured via adherence to recommended inpatient care treatments for acute exacerbations of COPD (recommended care, nonrecommended care, “ideal care” [all recommended and no nonrecommended care]). We included 662,842 pre-HRRP (January 2010–September 2014) and 285,508 post-HRRP (October 2014–December 2018) admissions. Recommended care increased at a rate of 0.16% per month pre-HRRP and 0.01% per month post-HRRP (P < 0.001). Nonrecommended care decreased at a rate of 0.15% per month pre-HRRP and 0.13% per month post-HRRP. Ideal care increased at a rate of 0.24% per month pre-HRRP and 0.11% per month post-HRRP (P < 0.001).

Conclusions

The pre-HRRP trends toward improving care quality for inpatient COPD care slowed after HRRP implementation. This suggests that financial penalties for readmissions did not stimulate higher quality of care for patients hospitalized with COPD. It remains unclear what policies or approaches will be effective to ensure high care quality for patients hospitalized with COPD exacerbations.

At a Glance Commentary

Scientific Knowledge on the Subject

In 2006, a landmark paper on chronic obstructive pulmonary disease (COPD) showed that only one-third of hospitalized patients with COPD exacerbations received ideal care. There has not been a study published since the Centers for Medicare and Medicaid Services instituted the Hospital Readmissions Reduction Program (HRRP) that has sought to reevaluate if care quality for hospitalized patients with COPD has improved after penalty implementation. Our study therefore sought to evaluate whether care quality improved after implementation of the HRRP.

What This Study Adds to the Field

We found modest changes in trends related to the quality of care for patients with COPD after the introduction of financial penalties through the HRRP for excessive rehospitalizations. Importantly, these results suggest that financial penalties for readmissions did not stimulate higher quality of care for patients hospitalized with COPD, and it remains unclear what policies or approaches will be effective to ensure high care quality for patients hospitalized with COPD exacerbations.

More than 16 million people have been diagnosed with chronic obstructive pulmonary disease (COPD) in the United States (1, 2). COPD is the fifth overall leading cause of hospital admissions and the third-leading cause of hospital readmissions in the United States, with a projected economic impact of almost $49 billion in healthcare costs in 2020 (3). These data led the Centers for Medicare and Medicaid Services (CMS) to designate COPD as a Hospital Readmissions Reduction Program (HRRP) condition in October 2014 to improve patient outcomes and curtail excessive expenditures associated with COPD readmissions (4, 5).

Despite the burden of COPD on the U.S. healthcare system, data on care quality received by patients hospitalized with acute exacerbations of chronic obstructive pulmonary disease (AECOPD) are sparse (6). A previous landmark study found that only about one in three patients received “ideal” care when hospitalized with AECOPD (7). Using COPD treatment guidelines (8), the authors concluded that treatment could be improved by increased use of systemic corticosteroid and antibiotic therapy and by decreasing the use of unnecessary and potentially harmful treatments and decreasing variability in treatment between hospitals. More recently, Spece and colleagues performed a cohort study of veterans older than 40 years of age hospitalized for COPD between 2005 and 2011 at five Department of Veterans Affairs hospitals (9). Similar to Lindenauer and colleagues (7), they found that few patients received all of the recommended care items.

To our knowledge, our study is the first to evaluate the quality of care for patients hospitalized with AECOPD across U.S. hospitals after HRRP implementation and the first to specifically evaluate if the trends in quality of care changed in association with the timing of the penalty. The HRRP galvanized many U.S. hospitals to implement interventions to improve transitional care for patients to prevent readmissions and avoid financial penalties. Currently, data on the effectiveness of the HRRP policy to reduce readmissions and/or reduce costs are mixed. Clinical trials focused on strategies to prevent COPD rehospitalizations have produced conflicting results, with all recent studies failing to demonstrate benefit and some reporting harm (10–21).

We sought to determine whether the HRRP was associated with improvements in care quality with respect to the management of patients hospitalized for AECOPD. We hypothesized that quality of care for patients hospitalized with AECOPD across U.S. hospitals would improve between 2015 and 2018 because of the threat of financial penalties related to COPD readmissions under HRRP. Some of the results of these studies have been published previously in the form of an abstract (22).

Methods

Study Design, Setting, and Participants

We conducted a retrospective cohort study using data from the Premier Healthcare Database (PHD) representing more than 1,000 U.S. hospitals and approximately 25% of U.S. inpatient discharges (23). The PHD is a large, U.S. hospital–based, service-level, all-payer database containing data on inpatient discharges, primarily from geographically diverse nonprofit, nongovernmental, and community and teaching hospitals and health systems from rural and urban areas (23). The University of Chicago Institutional Review Board exempted this study from review (IRB20-0248).

Of the more than 1,000 U.S. hospitals in the PHD, 995 had inpatient admissions for patients 18 years of age or older with principal diagnosis codes pertaining to COPD or respiratory failure from 2010 through 2018. The PHD database included patients discharged between January 1, 2010, and December 31, 2018. Patients were included if they were seen in an inpatient facility, were older than 40 years of age, had a principal diagnosis of COPD or a principal diagnosis of respiratory failure paired with a secondary diagnosis of COPD, and were discharged between January 1, 2010, and December 31, 2018. Because administrative data can be imprecise with respect to identifying patients admitted for AECOPD, we conducted a sensitivity analysis with just patients with COPD codes as a primary diagnosis (24, 25). We categorized admissions between January 1, 2010, and September 30, 2014, as pre-HRRP and those admitted between October 1, 2014, and December 31, 2018, as post-HRRP. International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes were used to assess diagnostic information of hospital data from 2010 to 2015 (26). Similarly, data from 2016 to 2018 were assessed using ICD-10-CM codes to assess diagnostic information (26) (see Table E1 in the online supplement). Patients were excluded in the primary analysis if they had a secondary diagnosis of pneumonia, both to be consistent with the methods used in the prior study by Lindenauer and colleagues (7) and because there may be important differences in clinical approaches and/or clinical outcomes for patients with both AECOPD and pneumonia. However, we conducted a sensitivity analysis with a diagnosis of pneumonia to determine if this exclusion impacted the results In addition, patients admitted for pregnancy-related reasons or patients with unavailable age and/or sex information were excluded from our analysis.

Data Elements

In addition to age, sex, and race, we used the Elixhauser comorbidity index to identify the presence of diabetes, depression, deficiency anemias, fluid and electrolyte disorders, hypothyroidism, hypertension, obesity, other neurologic disorders, and solid tumors without metastasis (27, 28). Although our primary objective was to evaluate the quality of care provided to patients hospitalized for AECOPD in relation to the HRRP, we also report in-hospital mortality; length of stay; and disease-specific, pulmonary-specific, and overall readmission rates at 30 days that were obtained from the PHD discharge data. Of note, readmission rates included only readmissions to the same hospital, because each unique patient identifier (e.g., medical record number) in the PHD dataset is limited to each hospital site that shares data with Premier, a limitation of the PHD dataset. Hospital features, including bed size, annual number of AECOPD admissions, teaching status, geographic region, and whether institutions served urban or rural populations, were also noted. We used admission date to determine if patients fell into pre- or post-HRRP cohorts. Adherence to quality measures was assessed by using a combination of pharmacy billing data and charges for other diagnostic and therapeutic services provided during each hospitalization.

Adherence to Guideline Recommendations

The Institute for Healthcare Improvement’s concept of a “bundle,” defined as a “collection of processes needed to effectively care for patients with a particular condition” (29), was used to classify patients as receiving recommended care (e.g., chest radiography, arterial blood gas, supplemental oxygen), nonrecommended care (e.g., sputum examinations, acute spirometry, mucolytic agents), or ideal care (i.e., all recommended care and no nonrecommended care) (Figure 1). Quality of care was evaluated by comparing the treatment of patients hospitalized with AECOPD with the 2017 Global Initiative for Chronic Obstructive Lung Disease guidelines and guidelines established by the American Thoracic Society and European Respiratory Society in 2017 (30, 31). After review of COPD care guidelines from 2010 to 2018, it was determined that there were minimal changes, so the 2017 guidelines were used together as the basis for our analysis. On the basis of recommendations of the American College of Physicians and the American College of Chest Physicians in 2017, we categorized the following diagnostic evaluations and treatments as beneficial (i.e., recommended): chest radiography, arterial blood gas analysis, supplemental oxygen, inhaled anticholinergic bronchodilators, inhaled short-acting β₂-agonists, systemic corticosteroids, antibiotics, and noninvasive positive pressure ventilation. However, because all patients may not need oxygen therapy or noninvasive positive pressure ventilation, we excluded both therapies from our definition of recommended care for our interrupted time series (ITS) analysis. Furthermore, because of the nature of administrative data, the data are not granular enough to determine if these treatments were clinically indicated. Nonrecommended care was defined as receiving any one of the following: sputum examinations, acute spirometry, mucolytic agents, and methylxanthine bronchodilators. These treatment strategies or tests have uncertain benefits or may cause potential harm when administered during admission for COPD.

Figure 1.

Definitions of ideal care, recommended care, and nonrecommended care. Quality of care was evaluated by comparing the treatment of patients with chronic obstructive pulmonary disease with the 2017 Global Initiative for Chronic Obstructive Lung Disease guidelines and guidelines established by the American College of Physicians and the American College of Chest Physicians in 2017.

Statistical Analysis

Patient- and hospital-level data were described with summary statistics using means and standard deviations for continuous variables. Medians and interquartile ranges were used to report interhospital variability in care. To assess the impact of CMS readmission penalties on hospital care provision for patients with AECOPD, we evaluated trends in quality of care over time using an ITS analysis with segmented regression. ITS is one of the strongest quasiexperimental approaches for evaluating the longitudinal effects of interventions (32–35). ITS analysis is a valuable study design for evaluating the effectiveness of population-level health interventions that have been implemented at a clearly defined point in time (34). It is increasingly being used to evaluate the effectiveness of interventions ranging from clinical therapy to national public health legislation (35). To determine the impact of the penalty, we used an interaction term between time and intervention to evaluate the change in slope (rate of change per month) before and after the intervention (e.g., HRRP) (35). The model parameters (βs) represent the baseline intercept (β0), preinterruption slope (β1), change in level at the interruption (β2), and the change in slope (i.e., interaction of time and intervention; β3) (35). In our primary ITS analysis, we adjusted for age, Medicare status, month, and hospital region (e.g., South, West) as potential confounders.

As a sensitivity analysis, we conducted an ITS analysis while excluding the period from October 1, 2013, through September 30, 2014, to see if the effects of adding COPD as a target condition under the HRRP created a significant change in the quality of care for patients with AECOPD after excluding the partial penalty phase. This time period was chosen because prior work found that a small reduction in COPD readmission rates began before the full financial penalty took effect (19, 36). In addition, we conducted another sensitivity analysis with a focus on just Medicare status as the HRRP, because the HRRP is a Medicare-based policy, to ensure that the findings from our main analysis did not differ when evaluating the effect of the penalty on the quality of care for Medicare patients. Last, we ran two additional sensitivity analyses using cohorts with only a primary diagnosis of COPD and those patients who had a secondary diagnosis of pneumonia to ensure that a more precise or more broad cohort selection process did not change our results. All analyses were performed using R (version 3.6.3).

Role of the Funding Source

This study was supported by a seed award from the Center for Health Administration Studies at the Crown Family School of Social Work, Policy, and Practice at the University of Chicago.

Results

In total, our analysis included 662,842 patients pre-HRRP (2010–September 2014) and 285,508 patients post-HRRP (October 2014–2018) from 995 hospitals in the PHD (Figure 2). The mean ages of the pre- and post-HRRP cohorts were 67 (SD, 12) and 67 (SD, 11) years, respectively; 60% and 61% of the pre-HRRP and post-HRRP cohorts were female, respectively (Table 1). The majority of the study sample was White (75% pre-HRRP, 77% post-HRRP). The most commonly observed comorbidities in both cohorts were hypertension (55% pre-HRRP, 53% post-HRRP), fluid and electrolyte disorders (29% pre-HRRP, 31% post-HRRP), and diabetes (28% pre-HRRP, 23% post-HRRP).

Figure 2.

Flow diagram for the Premier Healthcare Database to identify pre- and post-HRRP COPD inpatient admissions. *COPD admission is inpatients older than 40 years of age admitted with an International Classification of Diseases, Ninth or Tenth Revision, Clinical Modification, code with a primary diagnosis of COPD or a primary diagnosis of respiratory failure with a secondary code for COPD. **Exclusions based on patients with incomplete demographic information and patients with a secondary diagnosis of pneumonia. COPD = chronic obstructive pulmonary disease; HRRP = Hospital Readmissions Reduction Program.

Table 1.

Baseline Patient Characteristics Before and After Hospital Readmissions Reduction Program Chronic Obstructive Pulmonary Disease Designation

Characteristic	January 2010–September 2014 (n = 662,842)	October 2014–December 2018 (n = 285,508)
Age, yr, mean (SD)	67 (12)	67 (11)
Sex
Female	394,715 (60)	174,235 (61)
Race
Black	90,042 (14)	43,200 (15)
Other	74,989 (11)	23,178 (8)
White	497,811 (75)	219,610 (77)
Ethnicity
Hispanic status, N	435,358 (66)	221,151 (77)
Hispanic status, U	202,252 (31)	54,959 (19)
Hispanic status, Y	25,232 (4)	9,398 (3)
Length of stay, mean (SD), d	4.5 (5.8)	4.4 (7.6)
In-hospital mortality	14,176 (2)	5,092 (2)
Readmission within 30 d*
COPD 30-d readmission	43,595 (7)	19,773 (7)
Respiratory 30-d readmission	70,706 (11)	34,379 (12)
All-cause 30-d readmission	104,524 (16)	48,813 (17)
Elixhauser comorbidities
Hypertension	363,214 (55)	152,427 (53)
Fluid and electrolyte disorders	192,141 (29)	89,472 (31)
Diabetes	184,436 (28)	66,816 (23)
Obesity	133,503 (20)	68,432 (24)
Depression	133,002 (20)	61,185 (21)
Hypothyroidism	94,939 (14)	44,098 (15)
Peripheral vascular disease	56,358 (9)	26,590 (9)
Other neurologic disorders	46,770 (7)	25,002 (9)
Deficiency anemias	23,686 (4)	11,008 (4)
Solid tumor without metastasis	16,060 (2)	6,686 (2)

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; N = no; U = unavailable; Y = yes.

Data are presented as count (percent) unless otherwise noted. All values besides age were significantly different (P < 0.001) between before and after Hospital Readmissions Reduction Program cohorts.

^*All readmission values were determined for patient readmission within the same hospital.

With respect to the geographic representation of the hospitals in the study, 43% of the hospitals were in the South, 25% were in the Midwest, 18% were in the West, and 14% were in the Northeast. The majority of the hospitals were urban (72% urban vs. 28% rural) and nonteaching (71% nonteaching vs. 29% teaching). The majority (66%) of the hospitals in our cohort had 0–299 beds, with the remaining 34% being hospitals with 300 or more beds (Table E2). There was no substantial change in the characteristics of the hospitals represented in our analysis when comparing pre-HRRP with post-HRRP years.

The mean length of stay was similar pre-HRRP and post-HRRP at 4.5 days (SD, 5.8 d) and 4.4 days (SD, 7.6 d), respectively. Similarly, all-cause 30-day readmission rates were similar pre- and post-HRRP (16% vs. 17%). Overall, few inpatient deaths were observed pre- or post-HRRP (14,176 [2%] vs. 5,092 [2%]) deaths.

Quality of Care

The trend of patients receiving recommended care before HRRP was increasing at a rate of 0.16% per month, whereas the trend after HRRP increased at a rate of 0.01% per month (P < 0.001) (Figure 3). The trend of patients receiving nonrecommended care before HRRP was deceasing at a rate of 0.15% per month, whereas the trend after HRRP decreased at a rate of 0.13% per month (P = 0.3) (Figure 3). Last, the trend of patients receiving ideal care pre-HRRP was increasing at a rate of 0.24% per month, whereas the trend post-HRRP increased at a rate of 0.11% per month (P < 0.001) (Figure 3). In other words, before HRRP, all indicators (recommended, nonrecommended, and ideal care) were improving (increasing, decreasing, and increasing, respectively), whereas after HRRP, the trends all slowed (recommended, nonrecommended, and ideal). The absolute additional change after the implementation of the HRRP penalty was not significant for recommended (P = 0.51), nonrecommended (P = 0.13), or ideal care (P = 0.35). The full results from the ITS analysis are shown in Table E4. Changes in recommended and nonrecommended care treatments in relation to the HRRP rollout are plotted in Figure 4, with annual counts of various treatment types listed in Table E5.

Figure 3.

Interrupted time series analysis for chronic obstructive pulmonary disease quality of care from 2010 through 2018. This figure depicts segmented regression for ideal care, recommended care, and nonrecommended care from January 2010 through September 2014 and October 2014 through December 2018. The dotted vertical line in the graph depicts the month in which the Hospital Readmissions Reduction Program implemented a penalty for excessive readmission of patients with chronic obstructive pulmonary disease.

Figure 4.

Trends in chronic obstructive pulmonary disease (COPD) treatment from 2010 through 2018. Each panel in this figure depicts the trend in administration of a defined element of COPD care from 2010 through 2018. The dashed line in each panel represents the implementation of the Hospital Readmissions Reduction Program penalty for COPD readmissions (October 2014).

Our primary sensitivity analysis conducting ITS analysis excluding data from October 2013 to September 2014 did not demonstrate a significant change in our findings for trends in recommended, nonrecommended, and ideal care (Table E6, Figure E1). Our secondary sensitivity analysis with a focus on just Medicare status did not demonstrate a significant change in our findings for trends in recommended, nonrecommended, and ideal care (Tables E7 and E8, Figure E2). Last, we ran two additional sensitivity analyses using a different patient cohort either with patients with only a primary diagnosis of COPD (Table E9, Figure E3) or including those patients who had a secondary diagnosis of pneumonia (Table E10, Figure E4) to ensure that a more precise or more broad cohort selection process did not change our results. There were no clinically significant changes to recommended, nonrecommended, and ideal care in these different cohorts.

Median hospital performance greatly varied with regard to the quality of care provided to patients hospitalized with AECOPD (Table E11). The distribution of recommended care and ideal care had a smaller standard deviation (e.g., less variability), whereas nonrecommended care had a larger standard deviation, after HRRP. Table E12 reports the changes in the overall distribution of those patients receiving recommended, nonrecommended, and ideal care before and after the introduction of the HRRP penalty.

Discussion

This is the first study, to our knowledge, to use a nationally representative dataset to evaluate the quality of care for hospitalized patients with AECOPD before and after the designation of COPD as a condition under the HRRP. We found that introducing financial penalties for excess rehospitalizations did not meaningfully improve the quality of care for patients admitted to the hospital with an AECOPD. Furthermore, greater variation in care quality was found across hospitals between 2010 and 2018 than was found in an earlier study (7).

There are several potential explanations for why there was not significant improvement in quality of care for patients admitted with COPD, despite changes in federal policy. First, the inpatient quality-of-care measures we evaluated do not describe all care provided for patients with COPD. For instance, post-HRRP implementation, many hospitals implemented additional transition-of-care interventions, such as discharge bundles, follow-up phone calls, and care management programs, to reduce the risk of readmission (5, 12, 15, 21). These intervention metrics, such as smoking cessation, referral to pulmonary rehabilitation, health coaching, and patient education, were not available in the PHD for analysis of quality of care (15, 37–41). The use of hospital- or health system–wide quality improvement programs in COPD care from acute through postacute care settings may have helped patients get more urgent care outside of the hospital (42). These innovative programs were not captured in our analysis. Second, the designation of COPD as a target condition simply may not have been sufficient to improve COPD care practices. Without clear, effective strategies at the initiation of the HRRP penalty, different hospitals employed different tactics but without guidance regarding which tactics would have the greatest impact on improving the quality of care delivered (5, 6, 43, 44). However, it is important to note that an ITS analysis of CMS data suggests that the HRRP was associated with a reduction in the readmission rate for patients with COPD when compared with patients with other medical conditions not designated under the HRRP (19). Because the HRRP penalty did not meaningfully improve the quality of care for patients admitted to the hospital with an AECOPD, it is likely that improvements in quality of care were not responsible for the readmission rate reduction found in the prior analysis (19).

The designation of other conditions (e.g., congestive heart failure [CHF], acute myocardial ischemia, and pneumonia) has similarly mixed data on improvement in mortality or other relevant patient-centered outcomes. A recent analysis, which included Medicare beneficiaries hospitalized with CHF along with other conditions covered by the HRRP from 2008 to 2014, reported that reductions in hospital 30-day readmissions were weakly but statistically significantly correlated with reductions in 30-day mortality rates after discharge (45). However, the other side of the statistical correlation reported in this study is that hospitals with no change in or increasing 30-day readmission rates, and thus facing greater HRRP financial penalties, had increases in 30-day postdischarge mortality for patients with CHF (46). Another study found that although 30-day readmission rates fell after HRRP implementation, mortality rose for patients admitted with COPD, further raising concerns that a metric focused on reducing readmissions not only does not improve quality of care but also may in fact raise the level of harm (18). Another risk of harm associated with the HRRP penalty is risk of increasing health disparities. A study found that hospitals caring for more at-risk and vulnerable patient populations were more likely to be penalized (47). In a similar fashion to studies done to evaluate the impact of the penalty on outcomes in CHF, acute myocardial ischemia, and pneumonia, more studies are needed to further evaluate the relationship between reduced readmissions and care quality for patients hospitalized for AECOPD. Ultimately, we believe that our work demonstrating that there was no meaningful differences in inpatient quality of care for patients admitted with COPD supports the argument that a penalty focusing on reducing readmission rates may not directly impact relevant patient-centered outcome such as the quality of care delivered to patients.

Our retrospective cohort study using administrative data from the PHD allowed examination of the changes in the quality of care provided to patients hospitalized with COPD since the prior data were published (7) and since the HRRP policy was implemented. Large interhospital variation in quality of care was likely attributable to variability in sputum testing and arterial blood gas analysis. Although the lack of adherence to our COPD guidelines and guideline implementation may have contributed to this variation, it is also possible that our use of administrative data was unable to capture the diversity of patient presentations and that comorbidities could have contributed to interhospital variability.

Our study has a few notable limitations. The PHD, though a large, geographically, and structurally diverse sample of U.S. hospitals, may not be fully representative of all U.S. hospitals. In addition, because of the use of an administrative database for our analysis, study methods were limited to the use of ICD-9/10 codes to identify COPD encounters. As a result, it is possible that some patients identified as having COPD may have truly been hospitalized for another respiratory illness, and some patients admitted for COPD may have inadvertently been excluded, given potential coding for other diagnoses that are often comorbid, such as CHF (48–51). In addition, we observed a declining number of patient encounters for COPD between 2010 and 2018, with the greatest reduction occurring after 2015. It is important to note that this period coincides with the change to ICD-10 reporting for patient diagnoses. This indicates that the changes in diagnosis coding practice could have introduced bias if patients receiving more guideline-adherent care were not included in our ICD-10 cohort. In addition, specific therapies such as oxygen may not have been reliably reported for billing purposes, which would limit the validity of administrative data in assessing the use of supplemental oxygen for patients with COPD. These types of clinical nuances and readmission rates to a different hospital were not available in our data. Many of the care transition strategies (e.g., smoking cessation, postdischarge follow-up visits, health coaching) that hospitals might have implemented to reduce COPD readmissions were not captured within the data set (20, 39–41, 52, 53). Hospitals likely focused more on these care transition strategies than on direct inpatient management because inpatient treatments addressed in current guidelines have not been shown to reduce the risk of COPD readmissions. Last, the composition of the hospitals available in the PHD changed over the time period of our study, with less than half of the 955 hospitals reporting data consistently during our study period (2010–2018). This change in hospital reporting could have confounded the results if hospitals that started reporting data later were performing worse.

In summary, we found no meaningful improvement in the quality of care for patients admitted to the hospital with an AECOPD. To date, there are few data to suggest that the HRRP policy led to improvements in overall quality of care for hospitalized patients with COPD.