Frequent hospital readmissions for acute exacerbation of COPD and their associated factors
Abstract
Objective: The factors that determine frequent hospital readmissions for acute exacerbations of COPD (AECOPD) are poorly understood. The aim of this study was to ascertain rates of re-hospitalizations for AECOPD patients and evaluate factors associated with frequent readmissions for acute exacerbations.
Methods: We conducted a cross-sectional survey of 186 patients with moderate to severe COPD with one or more admissions for acute exacerbations to two large general hospitals. Frequency of previous readmissions for AECOPD in the past year, and clinical characteristics, including depression and spirometry were ascertained in the stable state both before discharge and at 1-month post discharge.
Results: Among them, 67% had one or more previous readmission, 46% had two or more, 9% had 10–20 readmissions in the 1-year period prior to current admission. There was a high prevalence of current or ex-heavy smokers, underweight patients, depression and consumption of psychotropic drugs, and low prevalence of caregiver support, pulmonary rehabilitation and influenza and pneumcoccal vaccination. Univariate analysis showed that male sex, duration >5 years, FEV1 < 50% predicted, use of psychotropic drugs, receipt of pulmonary rehabilitation and vaccination were significantly associated with frequent past readmissions. Multivariate analysis revealed that disease duration >5 years (odds ratio (OR) = 2.32; 95% confidence interval (CI): 1.09–4.92), FEV1 < 50% predicted (OR = 2.60; 95% CI: 1.18–5.74), use of psychotropic drugs (OR = 13.47; 95% CI: 1.48–122.92) and vaccination status (OR = 3.27; 95% CI: 1.12–9.57) were independently associated with frequent readmissions for AECOPD.
Conclusion: Frequent past readmission for AECOPD was associated with disease severity and psychosocial distress and increased use of vaccinations.
INTRODUCTION
COPD remains a poorly understood and inadequately managed problem. The outcomes of medical treatment and care remain poor, with particularly high rates of hospital readmissions for acute exacerbations of COPD (AECOPD). Over half of COPD patients who are hospitalized for acute exacerbations are readmitted at least once in the ensuing 6 months.1–3 Hospitalization for AECOPD accounts for a large part of the high healthcare expenditure for COPD, estimated to be 40% of the total direct cost of medical care for COPD in the USA.2
The factors that determine acute exacerbations and hospitalization in COPD patients are poorly understood. Factors that have been studied as predictors of mortality and other outcomes include FEV1,4–10 blood gases,5,6,9,10 comorbidity,8–10 chronic mucus hypersecretion (CMH),8 muscle weakness,5,11 poor nutritional status, low BMI,6,8–11 socioeconomic status and support,9 number of previous physician visits or hospital admissions,8–10 influenza vaccination,9,10 pneumococcal vaccination,9,10 pulmonary rehabilitation,9,10 inhaled corticosteroids and long-term oxygen therapy.9,10 However, these studies have produced inconsistent results with respect to the influence of FEV1,4–10 BMI6,8,9,11 and smoking.6,9 Most investigators have studied a limited number of specific risk factors in a single study, some of which did not fully take into account potential confounding among variables. Other potential risk or protective factors such as psychological well-being, patient adherence to care and social support have not been reported or rarely so.
The aims of this study were to ascertain in a cross-sectional sample survey in Singapore of hospitalized COPD patients the rates of hospital readmissions for acute exacerbations, the frequency of modifiable and non-modifiable risk factors; and to evaluate the risk factors associated with recurrent hospital admissions for AECOPD.
MATERIALS AND METHODS
Patient population
A cross-sectional survey was conducted in the Respiratory Medicine departments of two large general hospitals which admitted over 80% of COPD patients in Singapore. During a 1-year period, the hospital admission database was used to identify eligible COPD patients with a primary discharge diagnosis of acute exacerbation of COPD. The Global Initiative for Chronic Obstructive Lung Diseases (GOLD) criteria12 were used to define COPD: FEV1/FVC (post bronchodilatation) < 70% and with or without chronic cough and sputum production. The inclusion criteria were: (i) patients with a principal diagnosis of AECOPD; (ii) age over 50 years; (iii) current smokers or ex-smokers with a history of smoking equivalent to at least 20 pack-years; and (iv) surviving patients with stable COPD status on discharge. Both COPD patients with first-episode exacerbation and repeat admissions were included. The exclusion criteria were patients with active pulmonary tuberculosis, pulmonary fibrosis, pneumothorax, lung cancer, acute heart failure, acute myocardial infarction, stroke sequelae, severe renal failure, moderate to severe Parkinson's disease, dementia, hearing impairment, or those who were mentally unable to answer a questionnaire, or were resident outside of Singapore.
Among the 274 patients who satisfied the above eligibility criteria for the study, 71 patients refused to participate in the study, or had language difficulties, or were not able to perform spirometry satisfactorily. Another 17 patients were excluded because they had thyroid cancer (two), Down's syndrome (one), or were moribund (14). The non-participating patients included seven patients with psychiatric disorders. The participation rate was 68% (186/274).
Interviews and clinical assessment were conducted while the patients were in a stable clinical condition at two points in time: (i) during their hospital stay near to the date of discharge; and (ii) at 1 month after discharge (146 COPD patients).
Measures at initial interview
Sociodemographic characteristics included sex, age, ethnicity, housing type and marital status. Clinical characteristics included smoking status, duration of COPD, comorbid medical conditions, CMH, FEV1, oxygen saturation by pulse oximetry (SpO2), BMI, degree of dyspnoea and consumption of psychotropic drugs (hypnotics, tranquilizers, sedatives, anxiolytics, mood stabilizers or antidepressants) in the past year. Patient care characteristics included history of influenza or pneumococcal vaccination, participation in a pulmonary rehabilitation programme in the past year and the patient's reported level of adherence to treatment.
The recorded number of previous readmissions in the past year for acute exacerbations beside the current admission was used to categorize two groups of patients with frequent readmissions (two or more previous admissions in the past year) and non-frequent readmissions (previous readmission or <2 in the past year).
Smoking status (current or ex-smoker), the duration of smoking (years) and the amount of tobacco consumption (cigarettes/day) was used to define two categories of ex-smokers: (i) ex-heavy smoker (≥ 30 years and ≥ 20 cigarettes/day); and (ii) ex-light smoker.
Comorbid diseases were coded according to the International Classification of Diseases, 10th Revision. They included asthma, coronary artery disease, chronic congestive heart failure, diabetes mellitus, renal failure, hypertension, psychiatric illness, other diseases such as gall-stone, renal calculus, benign prostate hyperplasia and cataract, as recorded in the patient problem list maintained in the medical charts.
The degree of dyspnoea was assessed by a graded scale from 0 to 5: (0) not breathless at all; (1) breathlessness on heavy exercise, for example, climbing two or three floors; (2) breathlessness on moderate exertion, for example, climbing one floor or walking quickly; (3) breathlessness on mild exertion, for example, walking at normal speed; (4) breathlessness on minimal exertion, for example, slow walking; and (5) breathlessness on limited exertion, for example, shower, bathing, washing. (New York Classification).
Chronic mucus hypersecretion (CMH) was considered present when cough and sputum had lasted at least 3 months for more than 1 year, using questions from the British Medical Research Council respiratory questionnaire.
Housing type was used as an indicator of socioeconomic status and was categorized into two ordinal categories: (i) public housing 1-, 2- and 3-room apartment; and (ii) public housing 4-, 5-room apartment, private apartment, semidetached house and terrace house.
Low BMI was used to categorize patients in two nutritional status groups: underweight (<20 kg/m2) and normal or overweight (≥ 20 kg/m2).
SpO2 was measured by a microoximeter. We obtained measurements for 183 patients of resting SpO2 while the patients were breathing room air or oxygen because some patients were under continuous oxygen therapy.
Pulmonary rehabilitation consisted of a multidisciplinary rehabilitation programme of 6 weeks of supervised exercise training, education, physiotherapy and nutritional intervention. Patient treatment with rehabilitation was based upon whether the patient had completed a rehabilitation care programme in the past year.
Patient's level of treatment adherence was measured using a scale of three questions relating to the proportion of the times when the patient was able to take his medications as prescribed, attend follow-up and day-rehabilitation visits as required and undertake rehabilitative exercises as prescribed.
Recording of spirometric values was made on an electronic microspirometer (Vitalograph, Birmingham, UK). We recorded spirometric data both on the day of hospital discharge and at 1 month post discharge, and selected the better spirometric value for analysis. Spirometry was performed in accordance with the American Thoracic Society (ATS) criteria for standardization and procedures with the patient in the seated position.13 The values of FEV1 and FVC were based upon the best of four readings. FEV1 was expressed as percentage of predicted from formulas of local normative values for Chinese, Malay and Indian men and women.14 Acceptable spirometric data were obtained for 156 patients.
Measures at 1 month post discharge
On re-interviewing the patients at 1 month after discharge, height, weight, repeat spirometry, psychological distress and family support levels were measured.
Psychological distress: The Hospital Anxiety and Depression Scale (HAD) was used to assess psychological distress.15 The HAD Scale is a measurement tool specifically designed for hospital use, with good internal consistency and is a reliable and valid psychological measure for use in medically ill populations, including critical care patients. This brief 14-item self-report questionnaire had two seven-item subscales for anxiety and depression, with each item scored from 0 to 3 (maximum score of 21 on each subscale). A score of 8 or more on either subscale is suggestive of psychological distress.
Family support level was measured by using a scale of six questions that we developed for the study. The questionnaire was answered by a family member who was the usual caregiver of the patient, and (i) elicited the degree of the caregiver's understanding of the patient's illness and treatment; (ii) elicited the degree of understanding of his/her tasks in relation to the patient's illness (0 = none of it, 1 = some of it, 2 = most of it, 4 = all of it), the proportion of times the caregiver; (iii) supervised the patient taking medicine; (iv) supervised the patient's exercising at home; (v) took the patient for medical follow up and treatment; and (vi) encouraged him/her to regain full recovery (0 = None of the time, 1 = some of the time 2 = most of the time 4 = all of the time).
Complete data were collected for 186 patients in hospital, and for 146 patients at 1 month after discharge from hospital. Data for 18 subjects on BMI and 30 subjects on FEV1 were not collected because of physical ill health and disability.
Statistical analysis
The prevalence of ‘frequent’ and ‘non-frequent’ readmissions and risk factors were expressed as numbers and percentages. Univariate analyses of risk factors that were associated with frequent readmissions were performed as for a case-control study, using the chi-square statistic and estimates of the odds ratio (OR) with their 95% confidence intervals (CI). Multivariate analysis to evaluate the independent risk factors for frequent readmissions for AECOPD was performed using logistic regression for dichotomous dependent variable. Both forward stepwise selection and full saturated models were developed and used to evaluate estimates of ORs. Given that they were consistent and robust from both models, the results from the parsimonious stepwise selection model are presented in this report. All statistical analyses were performed using the Statistical Package for Social Sciences (spss) version 11.0 for Windows (SPSS, Chicago, IL, USA).
RESULTS
During the 1-year period prior to the current admission of the 186 COPD patients, 125 (67.2%) reported at least one previous hospital readmission for AECOPD, 85 patients (45.7%) reported two or more previous hospital readmissions. Among the 85 frequent readmission patients, 16 experienced more than 10 readmissions and the highest number of readmissions was 20.
The 186 subjects aged 50–95 years comprised 70 (38%) subjects aged above 75 years old; 83% were men; 81% were Chinese, 9% were Malays and 10% were Indians; 32% were divorced, widowed or unmarried; 46.2% lived in small (1- to 3-room) public apartments (Table 1).
| Variable | Frequent readmissions n (%) |
Non-frequent readmissions n (%) |
Significance test P-value |
Odds ratio (95% CI) |
|---|---|---|---|---|
| Total | 85 (100) | 101 (100) | ND | ND |
| Sex | ||||
| Male | 76 (89.4) | 79 (78.2) | 0.049 | 2.35 (1.02–5.43) |
| Female | 9 (10.6) | 22 (21.8) | ND | 1.00 |
| Age (years) | ||||
| ≥75 | 28 (32.9) | 42 (41.6) | 0.288 | 0.69 (0.38–1.26) |
| <75 | 57 (67.1) | 59 (58.4) | ND | 1.00 |
| Ethnicity | ||||
| Chinese | 67 (78.8) | 84 (83.2) | ND | 1.00 |
| Malay | 10 (11.8) | 6 (5.9) | 0.194 | 2.09 (0.72–6.04) |
| Indians and others | 8 (9.4) | 11 (10.9) | 1.000 | 0.91 (0.35–2.40) |
| Housing type | ||||
| Public 1-, 2-, 3-room | 41 (48.2) | 45 (44.6) | 0.659 | 1.16 (0.65–2.07) |
| Public 4-, 5-room, executive, private | 44 (51.8) | 56 (55.4) | ND | 1.00 |
| Marital status | ||||
| Single/divorce/widowed | 23 (27.1) | 37 (36.6) | 0.208 | 0.64 (0.34–1.20) |
| Married | 62 (72.9) | 64 (63.4) | ND | 1.00 |
- CI, confidence interval; ND, no data.
Men were about two times more likely to develop frequent readmissions than women (OR = 2.35, 95% CI: 1.02–5.43). Malays were also more likely to have frequent readmissions (OR = 2.09, 95% CI: 0.72–6.04), although statistically insignificant at 5% level. Other sociodemographic variables (age, socioeconomic and marital status) were not significantly associated with increased readmissions for AECOPD.
The clinical characteristics of COPD patients with frequent and non-frequent readmissions are summarized in Table 2. All the subjects were current smokers or ex-smokers with a history of smoking equivalent to at least 20 pack-years. Among them, 61% (114) subjects were current or ex-heavy smokers; 53% (89) subjects were underweight (BMI < 20 kg/m2), 74% (138) had at least one comorbid disease, 39% (73) had two or more comorbid diseases and 42% (62) were depressed.
| Variable | Frequent readmissions n (%) |
Non-Frequent readmissions n (%) |
Significance test P-value |
Odds ratio (95% CI) |
|---|---|---|---|---|
| Total | 85 (100) | 101 (100) | ND | ND |
| Smoking status | ||||
| Current or ex-heavy smoker† | 52 (61.2) | 62 (61.4) | 1.000 | 0.99 (0.55–1.79) |
| Ex-light smoker | 33 (38.8) | 39 (38.6) | ND | 1.00 |
| BMI | ||||
| ≥20 | 33 (38.8) | 46 (45.5) | ND | 1.00 |
| <20 | 43 (50.6) | 46 (45.5) | 0.439 | 1.30 (0.71–2.40) |
| Unknown | 9 (10.6) | 9 (8.9) | 0.525 | 1.39 (0.50–3.89) |
| Duration of COPD (years) | ||||
| >5 | 52 (61.2) | 39 (38.6) | 0.003 | 2.51 (1.39–4.53) |
| ≤5 | 33 (38.8) | 62 (61.4) | ND | 1.00 |
| No. comorbidity | ||||
| ≥2 | 37 (43.5) | 36 (35.6) | 0.465 | 1.32 (0.64–2.75) |
| =1 | 27 (31.8) | 38 (37.6) | 0.849 | 0.91 (0.43–1.94) |
| 0 | 21 (24.7) | 27 (26.7) | ND | 1.00 |
| Chronic mucus hypersecretion | ||||
| Yes | 42 (49.4) | 38 (38) | 0.137 | 1.59 (0.89–2.87) |
| No | 43 (50.6) | 62 (62) | ND | 1.00 |
| FEV1% | ||||
| ≥50 | 23 (27.1) | 42 (41.6) | ND | 1.00 |
| <50 | 46 (54.1) | 45 (44.6) | 0.073 | 1.87 (0.97–3.59) |
| Unknown | 16 (18.8) | 14 (13.9) | 0.119 | 2.09 (0.87–5.03) |
| Pulse oximetry (SpO2)‡ | ||||
| ≤95 | 45 (53.6) | 52 (52.5) | 1.000 | 1.04 (0.58–1.87) |
| >95 | 39 (46.4) | 47 (47.5) | ND | 1.00 |
| Degree of dyspnea | ||||
| ≥3 | 58 (68.2) | 55 (54.5) | 0.070 | 1.80 (0.98–3.28) |
| <3 | 27 (31.8) | 46 (45.5) | ND | 1.00 |
| Anxiety‡ | ||||
| Yes (HAD ≥ 8) | 8 (12.5) | 7 (8.5) | 0.584 | 1.53 (0.52–4.47) |
| No (HAD < 8) | 56 (87.5) | 75 (91.5) | ND | 1.00 |
| Depressed‡ | ||||
| Yes (HAD ≥ 8) | 32 (50) | 30 (36.6) | 0.129 | 1.73 (0.89–3.37) |
| No (HAD < 8) | 32 (50) | 52 (63.4) | ND | 1.00 |
| Psychotropic drugs | ||||
| Yes | 11 (12.9) | 1 (1) | 0.001 | 14.87 (1.88–117.69) |
| No | 74 (87.1) | 100 (99) | ND | 1.00 |
- † (≥30 years and ≥20 sticks/day).
- ‡ Numbers do not total up because of missing data.
- CI, confidence interval; HAD, Hospital Anxiety and Depression Scale; ND, no data.
The results of univariate analysis showed neither smoking, comorbidities nor BMI were associated with increased hospital readmissions. On the other hand, patients with FEV1 < 50% predicted (OR = 1.87, 95% CI: 0.97–3.59), severe dyspnoea (OR = 1.80, 95% CI: 0.98–3.28) and depression (OR = 1.73, 95% CI: 0.89–3.37) were more likely to be frequently readmitted to hospital, although statistically insignificant at P < 0.05. Patients with long duration of COPD (>5 years) were approximately two times more likely to have frequent readmissions than those with short duration (OR = 2.51, 95% CI: 1.39–4.53). Patients with reported consumption of psychotropic drugs were much more likely to have frequent readmissions than those without consumption of psychotropic drugs (OR = 14.9, 95% CI: 1.88–117.69).
The patient care characteristics of COPD patients with frequent and non-frequent readmissions are summarized in Table 3. The large majority (88%) of subjects did not receive influenza or pneumococcal vaccination in the past year. Only 13% (25) subjects completed pulmonary rehabilitation. The majority of patients (85%) reported fair to good patient compliance and 38% subjects had fair to good caregiver support although 35% did not have a caregiver.
| Variable | Frequent readmissions n (%) |
Non-frequent readmissions n (%) |
Significance test P-value |
Odds ratio (95% CI) |
|---|---|---|---|---|
| Vaccination | ||||
| No | 69 (81.2) | 94 (93.1) | ND | 1.00 |
| Yes | 16 (18.8) | 7 (6.9) | 0.024 | 3.11 (1.22–7.98) |
| Influenza vaccination | ||||
| No | 73 | 99 | ND | 1.00 |
| Yes | 12 | 2 | 0.002 | 8.14 (1.77–37.47) |
| Pneumococcal vaccination | ||||
| No | 74 | 95 | ND | 1.00 |
| Yes | 11 | 6 | 0.127 | 2.35 (0.83–6.66) |
| Pulmonary rehabilitation | ||||
| No | 68 (80) | 93 (92.1) | ND | 1.00 |
| Yes | 17 (20) | 8 (7.9) | 0.018 | 2.91 (1.19–7.12) |
| Patient compliance | ||||
| Poor (score: 5–9) | 15 (17.6) | 12 (11.9) | 0.300 | 1.59 (0.70–3.61) |
| Fair to good (score: 3, 4) | 70 (82.4) | 89 (88.1) | ND | 1.00 |
| Family support | ||||
| Fair to good (score ≥ 12) | 32 (37.6) | 39 (38.6) | ND | 1.00 |
| Poor (0 < score < 12) | 6 (7.1) | 4 (4.0) | 0.503 | 1.82 (0.48–7.04) |
| No caregiver (score = 0) | 26 (30.6) | 39 (38.6) | 0.605 | 0.81 (0.41–1.61) |
| Unknown | 21 (24.7) | 19 (18.8) | 0.553 | 1.35 (0.62–2.93) |
- CI, confidence interval; ND, no data.
The results of univariate analysis showed that neither poor patient compliance, nor poor family support was significantly associated with increased hospital readmissions. On the other hand, patients who received influenza and/or pneumococcal vaccination were three times more likely to have frequent readmissions than those who received none of these vaccinations (OR = 3.11, 95% CI: 1.22–7.98). The respective ORs for influenza vaccination and pneumococcal vaccination were OR = 8.14, 95% CI: 1.77–37.47) and OR = 2.35, 95% CI: 0.83–6.66. Patients who were on pulmonary rehabilitation were almost three times more likely to have had frequent readmissions than those without pulmonary rehabilitation (OR = 2.91, 95% CI: 1.19–7.12).
The results of multivariable logistic regression analysis of factors associated with frequent hospitalization for acute exacerbation are shown in Table 4. We included 19 predictive variables as categorical or dichotomous variables in a forward stepwise multiple selection logistic regression analysis: sex, age, ethnicity, housing type, marital status, smoking status, BMI, duration of COPD, number of comorbidities, FEV1%, degree of dyspnoea, CMH, anxiety, depression, consumption of psychotropic drugs, receipt of vaccination and pulmonary rehabilitation programme, patient compliance and family support. Only four variables emerged as being significantly associated with frequent hospital readmissions: long duration of COPD (>5 years) (OR = 2.32; 95% CI: 1.09–4.92), FEV1 < 50% predicted (OR = 2.60; 95% CI: 1.18–5.74), consumption of psychotropic drugs (OR = 13.47; 95% CI: 1.48–122.92) and receipt of vaccination (OR = 3.27; 95% CI: 1.12–9.57).
| Significant independent predictors | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| Duration of COPD > 5 years | 2.32 | 1.09–4.92 | 0.029 |
| FEV1% < 50% | 2.60 | 1.18–5.74 | 0.018 |
| Consumption of psychotropic drugs | 13.47 | 1.48–122.92 | 0.021 |
| Vaccination | 3.27 | 1.12–9.57 | 0.030 |
- CI, confidence interval.
DISCUSSION
Our study results should be interpreted and compared with previous studies with consideration of the differences in study design, type of patients, selection criteria and clinical measurements. The cross-sectional study design with retrospective recall of data is relevant. We defined frequent hospital readmissions for acute exacerbation in our study subjects as having had at least two previous readmissions beside the current admission, compared with those who had at least only one previous readmission (i.e. two admissions or less in a year). Other studies have compared current one time admission for exacerbation versus admissions of patients with stable COPD.9,11 Such differences in case definition may account for varying sensitivities in detecting a given strength of associations. Our study population also tended to exclude the most severely affected patients who were unwilling or unable to participate.
Not surprisingly, we found a high rate of hospital readmissions for acute exacerbations which is in line with previous studies. In large unselected populations of COPD patients, half of those patients who are hospitalized are expected to be readmitted at least once in the ensuing 6 months1–3 with a majority (86%) of readmissions occurring within the first 3 months after hospital discharge.1 The high prevalence of known and putative risk factors in our study is also in support of a similar finding in a previous representative sample of COPD patients in the EFRAM study.16 In particular, there was a high prevalence of current or ex-heavy smokers, malnutrition, depression and consumption of psychotropic drugs, and low prevalence of caregiver support, pulmonary rehabilitation and vaccination. The evidence for recommending pneumococcal vaccination and pulmonary rehabilitation is not firm, judging by inconsistently observed effects on exacerbations in various studies.17–19 However, the basis for recommending influenza vaccination is much less in doubt, as evidenced by the substantial reduction of hospitalizations in elderly subjects with chronic lung disease.20,21
Among the clinical disease and severity variables in this study, we found that FEV1 and past duration of COPD were both independently and significantly associated with frequent hospital readmissions. FEV1 has been found in some studies7–10 to be a predictor of hospital admissions for acute exacerbations, but not in others.4–6 A possible reason for the negative findings in two of the latter studies was the small numbers of patients studied who were also selected on a narrow range and low levels of FEV1.5,6 A positive relationship of FEV1 and risk of hospitalization for AECOPD is more credible given the consistent findings of a strong impact of FEV1 in predicting mortality22–24 and poor overall clinical outcomes that included death, readmission, intubation and intensification of drug therapy.25 No previous study has reported in analysis of COPD disease duration as a potential variable for inclusion in prediction models for increased risk of frequent readmissions.
The degree of dyspnoea was also observed in the univariate analysis to be strongly associated with frequent readmissions with borderline significance. The degree of dyspnoea, in lieu of FEV1 is reported to be associated with increased risk of relapse of acute exacerbations following ambulatory treatment for an index acute exacerbation.8 We did not find a demonstrable relationship for oxygen saturation and readmission in this study. A lack of association with hospitalization has also been reported for oximetry,5 but a preponderance of studies have demonstrated a positive association.6,9,10 In the current study, oximetry data were collected under different conditions (room air or with oxygen therapy), which could possibly explain the observed lack of association. Taken altogether, these findings provide firm support for the important role of clinical disease severity as a factor that increases the risk of acute exacerbations and hospitalization.
The development of psychological sequelae in conjunction with physical illness is recognized to have an impact on subsequent outcomes of care.26 No previous studies have investigated the association of psychological distress with frequent hospital admissions for AECOPD. Only one previous study has investigated ‘sedative’ consumption (as a ‘lifestyle’ variable, indicating psychological disturbance) with no positive association found.9 In the current study, a 1.5-time increased odds of association of depression with frequent readmissions was observed although the association was not statistically significant. The exclusion of seven patients with psychiatric disorders from the study was also likely to have led to a biased underestimation of the association. However, the consumption of psychotropic drugs among those who had depression or anxiety was significantly associated with frequent hospital readmissions in this patient population. Psychotropic drug use may be considered a surrogate marker for more severe clinical depression, and lends support to the relationship between depression and frequent hospital readmission for AECOPD. This relationship may be two-way and reciprocal. Further prospective cohort and interventional studies should be carried out to address this important aspect of care.
Given the retrospective nature of the study, the significant association of frequent readmissions with vaccination should readily be interpreted to mean that patients who had frequent past readmissions for AECOPD were more likely to be given influenza vaccination. However, the frequency of vaccination remains very low. In previous case-control and prospective studies of COPD patients admitted for acute exacerbation, influenza vaccination paradoxically has been found to be associated with an increased risk of hospital admissions for AECOPD.9,10 It is conceivable that COPD patients who had experienced frequent exacerbations resulting from viral infections were more willing to be vaccinated. Vaccination may also be more likely to be given by physicians to those with more severe disease and frequent readmissions. Such a ‘confounding by indication’ occurs when the hospital readmission represents a perceived high risk or poor prognosis that results in an indication for treatment.27
A number of putative predisposing factors were not confirmed in this study to be independently associated with frequent readmission. We found neither age nor sex to be associated with frequent hospital admissions, consistent with other studies.4,6,9 Male sex was significantly associated with more frequent readmissions for AECOPD in univariate analysis; it was also associated with more severe disease (lower FEV1 and longer duration) and hence was not retained in the final model.
There was a high prevalence of comorbid diseases in our patients, but the presence of any number of comorbidities was not significantly associated with frequent readmissions. The presence of comorbidity has been previously identified as a risk factor increasing the risk of hospital admissions in ambulatory patients with mild to moderate COPD in the EOLO Study.8 The data from the current study are similar to two other studies which also studied patients with moderate to severe COPD and indicated that comorbidities with COPD had no significant association with increased hospitalization for acute exacerbation.6,9 The selection criteria in the current study also excluded patients with the most severe comorbidity.
Frequent readmissions were not associated with current smoking. The role of smoking has been studied in only two previous studies.6,9 The first found no significant impact of smoking on the risk of hospitalization;6 the second found paradoxically that current smoking, compared with ex-smoking, was associated with a reduced risk of COPD admission.9 There was also no significant association with CMH. One previous study of ambulatory COPD patients in primary care setting has shown this factor to be associated with more frequent exacerbations of symptoms but not admission, suggesting this may be more related to mild COPD.
A significant association between a low BMI and increased readmissions for acute exacerbation was not evident. Two previous studies of small groups have demonstrated that low BMI6,11 is related to increased risk of readmission, although other studies, with a more unselected group of patients have not demonstrated this association.8,9 Another possible explanation for a lack of positive association with BMI is that anthropomorphic measures of nutritional status in non-Caucasian adults, such as BMI < 20 kg/m2, may not be an adequate indicator of poor nutritional status in a non-Caucasian population. However, the relation between BMI and frequent hospitalizations for AECOPD using different cut-offs such as <18 kg/m2 and <15 kg/m2 also failed to show any association.
Long disease duration, low FEV1, consumption of psychotropic drugs and vaccination were independently associated with frequent hospital readmission for acute exacerbation in COPD patients. The low prevalence of caregiver support, pulmonary rehabilitation and vaccination and the high prevalence of current or ex-heavy smokers, underweight, depression and consumption of psychotropic drugs, indicate that this disease remains poorly understood and an inadequately managed health problem.
ACKNOWLEDGEMENTS
This research was supported by grants from National Medical Research Council (NMRC) and Agency for Science and Technology Research (A-STAR) Biomedical Research Council (BMRC).
