Effectiveness of a Respiratory Day Hospital Program to Reduce Admissions for Exacerbation in Patients with Severe COPD: A Prospective, Multicenter Study

ABSTRACT

The respiratory Day Hospital (DH) is a care facility currently operating at various healthcare institutions. It monitors patients with severe chronic obstructive pulmonary disease (COPD) presenting repeated exacerbations with at least two hospital admissions per year. The main aim of the study was to evaluate the effectiveness of the DH program for controlling admissions for COPD exacerbations in this cohort of patients, and to identify clinical factors associated with hospitalizations and mortality. An observational prospective multicenter study was carried out at three hospitals. The sample comprised 150 consecutive patients (median age 70 [65–76] years, FEV₁ 33 [26–43]%, 97% males), included at the DH program. Over a one-year period, variables assessing effectiveness and use of healthcare resources were recorded. Factors associated with hospitalizations and mortality were identified. Patients made a median of 4[2–5] emergency visits due to COPD exacerbations with a median of 1[0–2] hospitalization(s)/year. Most of exacerbations (77%) were evaluated at the DH, but there were fewer hospitalizations from the DH than from the emergency department (21% vs. 81%, p < 0.001). In all, 29% of the patients had at least two admissions; these were the patients with the most severe disease. Age, readmission at 30-days and the presence of respiratory failure were the predictors of mortality. In conclusion, the DH program is an effective model for reducing hospitalizations in this cohort of patients. In all, 29% of the patients required two hospital admissions or more; these patients had more advanced disease and poorer prognosis, and would be most likely to benefit from additional care support.

KEYWORDS:

• Health resources
• hospitalizations
• integrated care
• readmissions

Introduction

Chronic obstructive pulmonary disease (COPD) is the most prevalent chronic respiratory illness and the one that incurs the highest health costs, both in terms of resource consumption and loss of quality of life. Patients who have severe disease and present frequent exacerbations (which in turn are related) consume the most health resources (1–3). It is estimated that COPD exacerbations cause more than 110,000 deaths and over 500,000 hospitalizations per year, with an annual direct economic cost of more than $18 billion (4,5). In addition to this economic burden, the inability to work and the severe limitations on the quality of life are other important restrictions (6). Previous reports have estimated that hospital admission costs represent 40.4% of the total in patients with moderate COPD and 62.6% in patients with severe COPD (7).

The significant economic impact of this condition has led the medical community to seek out new models of care for its control. Treating exacerbations at emergency departments (ED) and short-stay units does not appear to be effective; the high readmission rates reflect poor disease control and, from an economic perspective, increase health costs (8). Other models that have sought to train patients to recognize the signs of exacerbation at an early stage and have provided personalized self-management plans under the tutelage of a “case manager”, especially after hospital discharge, have improved the impact of exacerbations on the individual patients, but have not managed to reduce either the rate of exacerbations or the use of health resources (9–11).

In recent years, various day hospital (DH) care models have been designed for the treatment of a particular group of patients with severe COPD who experience a high frequency of exacerbations and hospital admissions. In most instances, recurrent exacerbations remain quite stable over time and so this subset of patients is assigned a specific phenotype, “the frequent exacerbator phenotype” (12). In the DH program, these patients attend scheduled visits or in case of an emergency they are seen directly by pulmonologists. The program complements the attention provided by the primary care service. Hospitals which have applied this combined model of specialized attention at the DH and primary care have recorded initial improvements at the local level in disease control (i.e., improvements in quality of life, exercise tolerance and mortality) and significant reductions in the number of hospital admissions and emergency service consultations (13–16). However, the impact of these respiratory units is often difficult to assess, since the same center may implement several models of care for the treatment of the same pathology and the ability of these units to control hospital admissions in the long term is currently unknown.

The aim of this prospective, multicenter study was to assess the global effectiveness of the DH respiratory care program to reduce hospitalizations for acute exacerbation of COPD (AECOPD) in a cohort of patients with severe disease and frequent previous admissions. Furthermore, we explored the clinical and functional factors associated with poor outcome (admissions and readmissions) and mortality in this cohort of patients.

Materials and methods

Study and population

We conducted an observational prospective and multicenter study in a cohort of patients with severe COPD attended regularly on a comprehensive care program at the respiratory DHs of three university hospitals in the Barcelona area (patients in follow-up during 2013, n = 379): Hospital Universitari de Bellvitge (HUB), Hospital Parc Tauli (HPT) and Hospital Germans Trias i Pujol (HGTiP). All the patients included in these programs previously presented a high consumption of health resources [average of 2.3 hospital admissions/year (standard deviation: 1.6) and average of length of hospital stay of 19.2 days/year (standard deviation: 14.1)] (17). The specific criteria for inclusion were: diagnosis of COPD (smokers >10 pk/year, ratio of FEV₁ (expiratory volume in one second) to FVC (forced vital capacity) post-bronchodilator test <70% according to GOLD recommendations) and history of two or more severe exacerbations requiring hospitalization in the previous year despite of conventional treatment. The exclusion criteria were: advanced (terminal) disease requiring palliative care, comorbidity associated with poor short-term prognosis (life expectancy below six months), severe or total dependency in a stable situation (Barthel score ≤35), severe cognitive impairment, lack of motivation, multiple comorbidity without predominance of respiratory disease and lack of a caregiver for patients >80 years or with Barthel index <90.

For this study, the first 50 eligible patients presenting a COPD exacerbation at the respiratory DH at each center from January 1, 2013 were consecutively selected.

The study complied with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of our hospital (reference number PR 337/13) and by the local ethics committee of each participating center.

Care model

The DH program at the three centers comprised: (1) scheduled visits (every 3 ± 1 months), (2) unscheduled visits (due to exacerbations) attended by a pulmonologist and a nurse, (3) early control visits after an exacerbation, (4) medical call center, (5) health education (including smoking cessation, annual vaccination, adherence to treatment), and (6) microbiological monitoring of sputum samples. This comprehensive patient-centered care was provided at a DH facility, which has rooms for visits (for both the physician and the nurse) and reclining armchairs for treating exacerbations (between two and four per site).

At the DH, patients presenting COPD exacerbations are urgently attended by the pulmonologist. In addition, diagnostic tests can be performed and intensive medical treatment administered if necessary. After assessing the exacerbation, the patient may be discharged or admitted to the hospital. Comorbidities are also evaluated at each visit and treated if required. The DH is open Monday through Friday, during daytime (8 am to 10 pm). Outside these times, patients requiring attention must go to the conventional hospital ED or primary care. The program nurses provide specific health education (lifestyle changes, healthy habits, education regarding inhaled therapies, home oxygen therapy and self-management of exacerbations). They also perform follow-up controls by telephone and monitor the microbiology results.

Definitions and variables analyzed

Demographic and anthropometric variables were recorded at the time of inclusion, as well as smoking history, comorbidities (Charlson index), symptoms (modified Medical Research Council dyspnea scale), spirometry, COPD treatment and the use of long-term oxygen therapy (LTOT) in each patient.

The number of emergency consultations due to exacerbation during the year of follow-up was recorded prospectively, along with the health facility providing care for the exacerbation (emergency department, DH or primary care). The consultations due to exacerbation were classified as: initial acute exacerbations of COPD (iAECOPD), relapses (R) and treatment failures (TF). Relapse was defined as a worsening of symptoms during the 4 weeks following the end of treatment for the exacerbation, and treatment failure as a worsening of symptoms during the exacerbation itself which required additional treatment (18). Rest of the exacerbations were classified as iAECOPD. Exacerbations that required a change in the usual inhaled medication were considered mild; those requiring treatment with corticosteroids and/or antibiotics were defined as moderate, and those requiring hospitalization as severe (19). In all exacerbations in which the patient expectorated a sputum culture was collected for conventional microbiological analysis.

The^{Table 1} following indicators of the effectiveness of the care model were analyzed during the 1-year follow-up: the number of total emergency consultations, number of admissions and readmissions (defined as new admissions within 30 days) for COPD exacerbations, days of hospital stay and all-causes mortality. In addition, an extended study of mortality was conducted in this cohort of patients at 2 years follow-up.

Patients were classified into three groups according to the number of admissions during follow-up: patients without admissions (group A), those with a single admission (group B) and those with two or more (group C).

We considered that the program of DH was effective in this population if the number of hospital admissions due to an AECOPD during 1-year was ≤1 (A and B groups).

Statistical analysis

Quantitative data were expressed as means ± SD or as medians [interquartile range] depending on whether the distribution was normal according to the Kolmogorov-Smirnov test. The Student t-test or the Mann–Whitney test was used to compare continuous variables where appropriate; in the case of comparisons between more than two categories, analysis of variance (ANOVA) or the Kruskal–Wallis test was used. Categorical variables were described as frequencies and percentages. The qualitative variables were compared using Chi-square test or Fisher's exact test if fewer than five frequencies were expected. Lineal and logistic regression analyses were performed to determine which clinical factors were associated with the effectiveness variables (relapses, admissions, readmissions and mortality) using the forward stepwise method. Survival curves were calculated using the Kaplan–Meier method, and comparisons between them with the log-rank test. A value of p < 0.05 was considered statistically significant. The statistical analysis was performed using the PASW Statistics v18.0 software (SPSS Inc., Chicago, IL).

Results

General characteristics of the population

One hundred and fifty patients (50 from each center) were included. The median age was 70 [65–76] years, and 97% of the patients (n = 146) were male. Patients had a cumulative dose of 50 pack-years [40–75], and 9% were active smokers. Clinical and functional variables are displayed in Table 1, both in total and per center. Sixty-six per cent of patients had a cardiovascular risk factor (other than tobacco consumption) and 68% had at least one established chronic cardiovascular disease (ischemic or valvular heart disease, arrhythmia or peripheral artery disease). The left ventricular ejection fraction (LVEF) was preserved in most patients (nine had LV systolic dysfunction, defined as LVEF <45% as assessed by Doppler echocardiography). Sixteen patients (11%) presented echocardiographic criteria of pulmonary hypertension (systolic pulmonary arterial pressure >40 mmHg). No statistically significant differences were observed between the centers in terms of functional-related characteristics and comorbidities. All patients belonged to the GOLD group D (20).

Table 1. Clinical and functional characteristics of the patients.

	Global n = 150	HUB n = 50	HGTiP n = 50	HPT n = 50	p
Age, years	70 [65–76]	68 [64–76]	71 [66–77]	70 [66.7–77.2]	0.306
Males, n (%)	146 (97)	48 (96)	49 (98)	49 (98)	0.773
Active smokers, n (%)	14 (9)	3 (6)	5 (10)	6 (12)	0.576
AD, packets-year	50 [40–75]	50 [40–80]	47.5 [40–62.5]	60 [40–78.5]	0.355
BMI, kg/m^2	27 [23.5–30.1]	27.7 [24.2–31.7]	25.8 [22.9–29.4]	27.2 [23.2–29.8]	0.170
Charlson index, points	2 [1–3]	2[1–2.3]	2 [1–3]	2 [1–2]	0.274
Arterial hypertension, n (%)	70 (47)	27 (54)	24 (48)	19 (38)	0.269
Diabetes mellitus, n (%)	44 (29)	15 (30)	17 (34)	12 (24)	0.543
Dyslipidemia, n (%)	51 (34)	19 (38)	19 (38)	13 (26)	0.343
Chronic heart failure n (%)	21 (14)	9 (18)	4 (8)	8 (16)	0.313
FEV1 post BD (%)	33 [26–43]	37 [28–48]	31 [25–43]	33 [26–40]	0.088
mMRC dyspnea score, points	2 [2–3]	3 [2–3]	2 [2–3]	2 [2–3]	<0.001^*
LTOT, n (%)	74 (49)	21 (42)	24 (48)	29 (58)	0.271
ICS, n (%)	144 (96)	49 (98)	45 (90)	50 (100)	0.026^*
Roflumilast, n (%)	27 (18)	7 (14)	13 (26)	7 (14)	0.197
Nebulized ATB, n (%)	21 (14)	10 (20)	9 (18)	2 (4)	0.043^*
Azithromycin, n (%)	39 (26)	16 (32)	7 (14)	16 (32)	0.060

Data expressed as medians [IQR] or frequencies.

^* A p value <0.05 was considered statistically significant.

HUB, Hospital Universitari de Bellvitge; HUGTiP, Hospital Universitari Germans Trias i Pujol; HPT, Hospital Parc Taulí; AD, accumulated dose; BMI, body mass index; FEV₁ post BD, maximum expired volume in the first second of forced expiration after bronchodilator test; LTOT, long term oxygen therapy; ICS, inhaled corticosteroids; and ATB, antibiotic.

Exacerbations and use of health resources

Five hundred and seventy-five emergency consultations were made: 386 (67%) for the iAECOPD, 151 (26%) for R and 38 (7%) for TF, which represented a median of 4 [2–5] consultations per patient per year. The median number of iAECOPD per patient during follow-up was 2 [2–3] (99% of them were moderate-severe).

Of the consultations, 77% (n = 444) were performed at the DH (68% for iAECOPD, 25% for R and 7% for TF,), 16% (n = 90) at the ED (59% for iAECOPD, 34% for R and 7% for TF) and 7% (n = 41) at other health facilities, mainly primary care services (73% for iAECOPD, 22% for R and 5% for TF). The distribution of iAECOPD, R and TF at different health care units was similar (p = 0.371).

Of note, in all, 81% (n = 73) of the ED consultations required hospitalization, compared with only 21% (n = 95) of the consultations at the DH (p < 0.001) (Figure 1).

Figure 1. Destination of emergency consultations due to COPD exacerbation according to the health facility used. A higher percentage of consultations attended at emergency department required hospitalization compared with those attended at the day hospital. DH, day hospital; ED, emergency department.

Effectiveness of the care model: Number of hospital admissions

The variables of effectiveness are shown in Table 2. During the year of follow-up, the median number of admissions per patient and year was 1 [0-2] with a median hospital stay of 8 [5-14] days per admission. During follow-up, most patients (71%, n = 107) were not hospitalized or they presented only one admission (groups A and B). This low number of admissions for AECOPD was not associated with the time the patient had been on the DH program, indicating, in some extent, a similar short- and long-term effectiveness of the program.

Table 2. Variables of effectiveness of the Day Hospital model for control of exacerbations of COPD.

	Overall n = 150
Emergency consultations by patient/year, n	4 [2–5]
Total emergency consultations, n	575
Attended at DH, n (%)	444 (77)
Attended at the ED, n (%)	90 (16)
Attended at other services, n (%)	41 (7)
Hospital admissions, n (% EC)	169 (29)
HA generated at DH, n (% of EC at DH)	95 (21)
HA generated at the ED, n (% of EC at ED)	73 (81)
HA from other services, n (% of EC at other services)	1 (2)
Admissions by patient/year	1[0–2]
Patients requiring admission, n (% total patients)	88 (59)
Median stay per HA, days	8 [5–14]
Median stay per year, days	5 [0–17]
Readmissions at 30 days, n (% of admissions)	32 (19)
Patients requiring readmission, n (% total patients)	18 (12)
All-cause mortality at two years of follow-up, n (%)	33 (22)

Data are expressed as frequencies or median [IQR].

EC, emergency consultations; DH, day hospital; ED, emergency department; and HA, hospital admissions.

The remaining 29% of the patients (n = 43) presented at least two admissions (group C). The patients with more admissions had a higher total number of emergency consultations for COPD exacerbations, than those with one admission or none at all (groups A and B) due to the higher number of relapses. Group C patients also presented more symptoms (increased mMRC dyspnea score), greater airflow obstruction (FEV₁) and increased mortality. Comparing the microbiological data, patients in group C had higher rates of isolation of Pseudomonas aeruginosa in sputum culture during an exacerbation, and a higher proportion was receiving continuous nebulized antibiotics (Table 3).

Table 3. Clinical variables by the number of admissions during the year of follow-up.

	Group A (n = 62)	Group B (n = 45)	Group C (n = 43)	p
Age, years	69 [63–77]	71 [67–4.5]	71 [66–9]	0.233
Active smokers, n (%)	7 (11)	5 (11)	2 (5)	0.458
AD, packets-year	44.5 [32.7–60]	55 [40–70]	60 [40–80]	0.009^*
Charlson index	2 [1–3]	1.5 [1–2.8]	2 [1–3]	0.451
BMI, kg/m^2	28.2 [24.3–30.6]	26 [22.9–29]	27.3 [22.9–29.9]	0.154
mMRC dyspnea score, points	2 [2–3]	2 [2–3]	3 [2–3]	0.029^*
FVC postBD, %ref	65 [48–80]	61 [43–70]	60 [49–77]	0.337
FEV1 postBD, %ref	37 [28–48]	30 [25–35]	33 [26–45]	0.012^*
LTOT, n (%)	22 (36)	22 (49)	30 (70)	0.003^*
ICS, n (%)	60 (97)	41 (91)	43 (100)	0.096
Nebulized ATB, n (%)	7 (11)	3 (7)	11 (26)	0.028^*
Azithromycin, n (%)	12 (19)	11 (24)	16 (37)	0.117
Pseudomonas in COPD exacerbation, n (%)	8 (13)	10 (22)	15 (35)	0.028^*
Time of inclusion, months	3.5 [0–26.3]	19 [0–30]	4.5 [0–33.5]	0.569
EC, n	3 [2–4]	3 [2–5]	5 [4–6]	<0.001^*
iAECOPD, n	2 [1.8–3]	3[1–4]	3 [2–4]	0.194
Relapses, n	0 [0–1]	0 [0–2]	1 [0–3]	0.013^*
Median stay per HA, days	–	11 [7–17]	8.5 [5–15]	0.248
Median stay per year, days	–	11 [7–17]	20 [10–39]	0.005^*
Mortality at 2 years, n (%)	4 (7)	12 (27)	17 (40)	<0.001^*

Data expressed as medians [IQR] or frequencies. Group A: no admissions, Group B: 1 admission, Group C: 2 or more admissions.

^* A p value <0.05 was considered statistically significant.

AD, accumulated dose; BMI, body mass index; FVC, forced vital capacity; FEV₁ post BD, maximum expired volume in the first second of forced expiration after bronchodilator test; LTOT, tong term oxygen therapy; ICS, inhaled corticosteroids; ATB, antibiotic; EC, emergency consultations; iAECOPD, initial acute exacerbations of COPD; and HA, hospital admissions.

In the cohort as a whole, the number of relapses and the presence of chronic respiratory failure were predictors of having at least two admissions (OR 1.61, 95% CI 1.2–2.15, and OR 3.42, 95% CI 1.54–7.59, respectively).

There were 32 readmissions at 30 days (19% of the total admissions), generated by 18 patients (12% of the study population). The multivariate analysis showed that higher baseline dyspnea (mMRC score) (OR 4.05, 95% CI 1.52–10.74) and a greater number of relapses (OR 1.71, 95% CI 1.23–2.36) were independently associated risk factors of presenting at least one readmission.

In turn, relapse after initial treatment for exacerbation was also associated with the degree of dyspnea and with having a prior or subsequent isolation of P. aeruginosa in sputum culture (β coeff. 0.606, 95% CI 0.1–1.09, p = 0.005; β coeff. 0.492, 95% CI 0.134–0.765, p = 0.025, respectively).

No clinically relevant differences were found between the centers when analyzing the effectiveness of the DH program.

Mortality

Overall mortality was 10% after one year and 22% after 2 years. The univariate analysis showed that the patients with the highest mortality at 2 years had the most admissions (group C, Table 3), with at least one readmission at 30 days (50% vs 18% in patients without readmissions, p = 0.001) and a longer time of inclusion in the program (33% in patients with >24 months included vs 18% in patients with ≤24 months included, p = 0.035) (Figure 2).

Figure 2. Kaplan–Meier survival curves by: (A) number of admissions, (B) readmissions at 30 days, and (C) time of inclusion in the Day Hospital program. Survival was worse in patients with 2 or more hospital admissions, with at least one readmission at 30 days and with a longer time of inclusion in the program (more than 24 months). Abbreviations: HA, hospital admission.

The multiple logistic regression analysis showed that age, presence of readmissions and chronic respiratory failure were predictors of mortality at 2 years (OR 1.14, 95% CI 1.06–1.22; OR 2.34, 95% CI 1.18–4.63; OR 4.31, 95% CI 1.64–11.28, respectively).

Discussion

This multicenter study shows that the comprehensive care program in operation at DH respiratory care facilities is effective for the management and control of COPD exacerbations in a cohort of patients with severe disease and frequent admissions. The main results of the study were: (1) 71% of the patients with severe COPD with two admissions or more in the previous year due to COPD exacerbations reduced their annual number of admissions to one or zero, and this reduction was maintained over time; (2) despite the restrictions on the opening times of the facilities, most patients consulted the DH when presenting a COPD exacerbation, and had a lower rate of hospitalization than their counterparts who consulted the emergency services for the same reason (21% vs. 81%, p < 0.001); (3) the prognostic factors associated with mortality were age, presence of chronic respiratory failure and hospital readmissions. To our knowledge, this is the first multicenter study to analyze the effectiveness of a DH care program in this cohort of patients.

In COPD management, reducing admissions and readmissions has been a priority objective of healthcare services for many years (1). Indeed, admissions, and above all readmissions, are considered as key parameters of quality of care and in many countries, they are heavily penalized by the health authorities (21,22). On the respiratory DH care program described here, most patients with severe COPD and frequent exacerbations with two or more admissions in the previous year due to COPD exacerbations reduced their annual number of admissions to one or none. Although health education was provided to all patients from the first visit onwards (smoking cessation and adherence to treatment), this was only a part of the support that they received in the program. By providing regular monitoring of the patients and prompt evaluation of exacerbations by the pulmonologist, the DH program aims to minimize the progression of the disease and avoid the need for hospitalization.

Several factors may be involved in the significant reduction in admissions observed in our study. The first is that care at the DH service is almost always provided by the same team, comprising a pulmonologist and a nurse. This means that the health staffs are familiar with the characteristics of each of their patients (i.e., disease severity, baseline situation, and social and family environment) and have greater leeway for taking decisions regarding hospitalization. In addition, the greater flexibility in the monitoring program (with both telephone and face-to-face consultations) means that patients with moderate exacerbations can be re-evaluated within a few days, which helps clinicians to detect early TFs and to apply the appropriate therapeutic modifications. The DH respiratory program also provides two important advantages over “traditional” systems, which increase patients' adherence and enhance the impact of the model. The first is the short waiting time for medical attention at the DH unscheduled visits (conventional emergency units are often saturated specially in winter periods). The second is lower impact on the daily life of patients and their family: chronic patients remain at home longer, thus reducing stress levels and avoiding the medical problems that may be generated by hospital admission.

Despite the restrictions on the opening times at the DH, in the event of a COPD exacerbation, most patients (77%) consulted this facility. Among these patients, the rate of admissions was much lower than in those who consulted the emergency services (above all during weekends). Our study did not evaluate clinical or laboratory parameters of severity in the exacerbations. However, a previous single-center study carried out in a cohort of similar patients (13) found that the level of severity of COPD exacerbations attended at DH facilities and emergency services was similat in terms of gas exchange (PaO₂/FiO₂ ratio and respiratory acidosis). Moreover, this study did not record differences in readmissions and mortality at two months after the exacerbation. These findings suggest that disease severity was not the reason for the differences in the admission rates from the two services.

In the US, in 2012, the Centers for Medicare and Medicaid Services launched the Hospital Readmissions Reduction Program, a program designed to penalize readmissions in its member hospitals. Initially the diseases included were acute myocardial infarction, pneumonia and heart failure, but in 2014 readmissions due to COPD exacerbation were added (21,23). The inclusion of COPD generated a broad debate on whether there is a relationship between readmissions at 30 days and quality of care, and whether there really are proven interventions that significantly reduce readmissions due to exacerbations of the disease (24). Multiple randomized controlled clinical trials have been conducted to evaluate these different interventions (25–29), which focus on self-management of the disease, comprehensive patient care, and shared care arrangements involving specialists, primary care and home care. A systematic review by Prieto-Centurion et al. (30) evaluated a selection of post-discharge interventions carried out in these patients with the aim of preventing readmissions, but the authors were unable to perform a meta-analysis because of the heterogeneity in the design of the various studies and in the presentation of the results. None of the studies evaluated the reduction of readmissions at 30 days, since most were designed to examine outcomes at 6–12 months. Prieto-Centurion et al. concluded that there is insufficient evidence at present to recommend specifically targeted interventions for reducing readmissions at 30 days in COPD exacerbations.

In our study, although we achieved a reduction in admissions, 29% of the patients still presented two or more hospitalizations for COPD exacerbations. The group with two or more hospitalizations presented more dyspnea, poorer lung function, more presence of chronic respiratory failure, higher isolation of P. aeruginosa in COPD exacerbations and, moreover, higher mortality. Patients with readmissions were also included in this group: we found a rate of readmissions due to COPD exacerbations of almost 19% at 30 days, but generated by a very small percentage of patients – only 12%. Population-based studies have shown that the occurrence of every new severe exacerbation requiring hospitalization worsens the course of the disease and increases the risk of a subsequent exacerbation (31). In this regard, our results showed that mortality after two years in this group with readmissions was significantly higher than in the group without readmissions (50% vs. 18% respectively) with a median survival of 17.2 (95% CI 13.5–20.9) months. This analysis (which includes not just clinical severity of the disease, but also parameters of hospital resource use that incur penalties in the Hospital Readmissions Reduction Program) helps identify patients with severe COPD at their “end of life” stage, a situation that is often very difficult to predict in this disease. For this reason, readmission should be considered not as an indicator of poor care, but as a prognostic parameter that can help to adapt the care provided to the needs of patients. In patients at the “end of life” stage, the DH facility is less effective and palliative care units are needed to assist in the management of patients outside the hospital setting.

One of the strengths of this management study is that it was carried out by the clinicians themselves. Furthermore, the patients are highly selected in terms of disease severity and are representative of the population responsible for the greater part of the cost associated with exacerbations.

The main limitation of our study is its observational design; it compares the number of hospital admissions before and after a specific intervention, without a control group, and does not compare two care models. At our centers, it would have been unethical to include a control group of COPD patients and to deny them the support of the DH program. However, our study provides sufficient evidence to carry out an appropriate randomized controlled trial able to reach more consistent conclusions.

Conclusion

In conclusion, the DH respiratory care program proved effective for the control of exacerbations in selected patients with severe COPD and frequent exacerbations, significantly reducing hospital admissions. The effectiveness of this model decreases with the progression of the disease; however, the predictive model including clinical parameters and management helps to identify patients at the end of life who would benefit from palliative care support outside the hospital setting.

Declaration of interest

The author reports no conflicts of interest in this work.

Author contributions

All authors contributed to data analysis, drafting and critically revising the paper and agree to be accountable for all aspects of the work.

Acknowledgments

The authors thank the nursing staff of the DHs for their dedication to the care of these patients.

Funding

Dr. Huertas is the recipient of a grant from the Catalan Respiratory Society (SOCAP, 2014). The study was also supported by an educational grant from Menarini. Menarini had no involvement in study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.