Factors Influencing the Length of Hospital Stay Among Patients with Chronic Obstructive Pulmonary Disease (COPD) in Macao Population: A Retrospective Study of Inpatient Health Record

Introduction

Chronic obstructive pulmonary disease (COPD) is a major chronic disease characterized by persistent airflow limitation. Currently being the fifth leading cause of death worldwide, COPD is predicted to become the third by 2030.¹ From 1990 to 2015, there was a 44.2% increase of COPD prevalence globally, and the number of patients reached 17.5 million in 2015.² COPD also posed substantial economic and social burden on both patients and healthcare systems^3,4 It has been reported that in China the direct medical costs of COPD ranged from 72 to 3565 USD per capita per year, accounting for 33.33% to 118.09% of local average annual income.⁵ Another study showed that the direct costs of COPD accounted for 56% of the total costs of respiratory diseases (about 6% of the total annual health care budget) in the European Union.⁶ Notably, nearly 35% to 84% of the COPD medical costs were due to acute exacerbations (AECOPD).^7–9 It was found that hospitalized COPD patients were prone to decreased quality of life, high hospital costs, increased mortality, and a long length of hospital stay (LOS).^10,11

LOS is a key indicator of COPD patients medical resource use and hospitalization costs.^12,13 A meta-analysis reported that an increase in LOS was one of the significant risk factors for 30- and 90-day all-cause readmission.¹⁴ LOS may be affected by a range of factors such as age, history of tobacco smoking, Charlson index, cause of exacerbations, and therapies-related variables.^15–17 Key therapies in COPD clinical practice included smoking cessation, vaccinations, pharmacological therapy, pulmonary rehabilitation (PR), oxygen therapy, and non-invasive ventilation (NIV).^18,19 Regarding pharmacological therapy, the most common drug classes used for managing and preventing COPD symptoms and exacerbations were inhaled bronchodilators and/or corticosteroids (ICS), systemic steroids and antibiotics.^18,20 Exploring these LOS-related factors can help inform clinical practice about how to reduce LOS, thereby alleviating the disease burden of COPD patients.

The analysis of characteristics of COPD inpatients and contributing factors to LOS is currently under-reported, especially among Asian populations. Therefore, this study aims to identify the effects of different patient and clinical therapy factors on LOS for COPD-related admissions. By addressing the gaps in COPD-related research in Macao, the study findings is expected to supplement the current demographic and clinical evidence for the study of COPD inpatients in Asia, thereby contributing to a more comprehensive reference for the management of COPD.

Methods

Data Collection

COPD is one of the top-ten causes of death in the past decade in Macao. Statistics showed that 14% local manpower was influenced by a second-hand cigarette in smoking permitted gambling casinos before October 2014, which potentially increased the morbidity and mortality of COPD in Macao.^21,22 We retrospectively examined the COPD inpatient health record dated between 2017 and 2019 in Kiang Wu Hospital, which is one of the 3 major hospitals in Macao with 1928 staffs and 775 hospital beds (47% of the total in Macao).²³

This dataset contained all COPD-related admissions to Kiang Wu Hospital from January 1, 2017 to December 31, 2019. All medical records were collected from the hospital’s electronic medical record system. Three inclusion criteria were applied: (1) patients were admitted with a main diagnosis of COPD (ICD-10 code J44), (2) all records were confirmed by the specialists as admission due to acute exacerbation, and (3) there was no missing data among all the included admission items. This study used data on hospitalization information per patient admission.

The study was approved by the ethical review committee at the Kiang Wu Hospital [Ref. No. 2019–015]. All data were collected retrospectively from the hospital electronic medical record system and fully anonymized in a standardized case report form to ensure data confidentiality. As such, written consent from the patients or their guardians was exempted and this study was carried out in full compliance with the Helsinki Declaration.

Variables and Measurements

The data was divided into 4 categories including demographic data, blood test results, clinical therapies, and LOS. Demographic data included patient’s age, gender, history of tobacco smoking and number of comorbidities. Blood test results included blood eosinophil count, hemoglobin, white blood cells (WBC) and creatinine. Eosinophil count is usually used as the biomarkers to predict and drive treatment choices in inflammatory lung disease like COPD.^24,25 Hemoglobin, the main protein in red blood cells, can be used to evaluate the oxygen-carrying capacity and help diagnose anemia, and is also an independent prognostic predictor of premature mortality and a higher likelihood of hospitalization in COPD patients.^26,27 WBC is the circulating leukocytes I whole blood and is also one of inflammatory biomarkers for the prediction of mortality in COPD.²⁸ Measurement of creatinine is to assess renal function. Afzal, A et al found there was a significant relationship between the admission creatinine values and mortality of COPD.²⁹ All blood tests were carried out at the time of admission to provide references for diagnosis and to assess COPD disease status and inform treatment plan.³⁰ Regarding the clinical therapies for COPD in Macao, data about the usage of recommended therapies including inhaled medications, systemic steroids (prednisolone, dexamethasone, methylprednisolone), antibiotics, oxygen therapy, NIV and PR during hospitalization were extracted.^18,31

Based on the use of inhaled COPD medications of COPD, the patients were assigned to 1 of the 4 groups: Group 1 included the patients who used only one type of inhaled medication, such as long-acting bronchodilators (long-acting beta2 agonist (LABA) only, long-acting muscarinic antagonist (LAMA) only or both), short-acting bronchodilators (short-acting beta-agonist (SABA) only, short-acting muscarinic antagonist (SAMA) only or both) or inhaled ICS only. Group 2 included the patients who received two types of inhaled medications, such as long-acting bronchodilators (LABA, LAMA or both) combined with short-acting bronchodilators (SABA, SAMA or both), long-acting bronchodilators (LABA, LAMA or both) combined with ICS, short-acting bronchodilators (SABA, SAMA or both) combined with ICS. Group 3 were the patients who used the combination of all 3 types of inhaled medications ((SABA, SAMA or both) and (LABA, LAMA or both) and ICS). Group 4 referred to the patients who did not use any inhaled medications during hospitalization. The LOS in this study referred to the number of days between admission and discharge.

Statistical Analysis

All information was analyzed in a descriptive manner by using median (IQR), mean (SD), and proportions. Blood test results were categorized into below normal range, normal, and above normal ranges following the reference ranges for Macao’s blood tests. To investigate the factors influencing LOS, univariate analysis and multivariate were both applied. Unary linear regression, t-tests, and one-way ANOVA were used to explore the potential associations between LOS and independent variables (demographic information, blood tests results, and clinical therapies information). Furthermore, all the significant factors were entered into the final multivariate linear regression model as independents variables. Statistical analyses were performed in SPSS 23.0. (IBM Corporation, Armonk, NY, USA) with a significance level of 0.05.

Results

From 2017 to 2019, a total of 1116 COPD admissions were included in this study. Figure 1 summarised the key steps of data analysis.

Figure 1 Key steps of data analysis in this study.

Descriptive Analysis

Demographic Characteristics and Blood Test results

Demographic information including gender, age, history of tobacco smoking and number of comorbidities was summarized in Table 1. Among the patients, the average age was 79.42 (±10.35) years old and 66.6% of them were male. It was found that 37.5% cases never smoked, and 45.0% had quitted smoking accounting for the majority of the patients. Only 13.9% cases did not have any comorbidities while 32.2% had 3 or more comorbidities. Results of blood tests showed that 24.7% of cases were with blood eosinophils ≥300 cells/μL, 60.0% had hemoglobin below normal range, 94% had a higher level of WBC compared with normal range. For creatinine, 75.6% and 73.7% of male and female cases were within the normal range.

Table 1 Demographic Information and Blood Test Results (n=1116)

Clinical Therapies and LOS

Among all the cases, 596 (53.4%) and 828 (74.2%) used systemic steroids and antibiotics, respectively, while 88.8% received oxygen therapy and 88.5% did not receive NIV. Only 10.8% received PR during hospitalization. Nearly 95% of the cases received inhaled medications during hospitalization, with 22% using only one type, 20.6% using two types, and 52.2% using a combination of all 3 types of inhaled medications. The average LOS was 12.28 (±9.23) days, whereas the median value was 11.0 days. A detailed summary is presented in Table 2.

Table 2 Descriptive Analysis of Clinical Therapies and Length of Hospital Stay

Multivariable Regression Analysis on LOS

LOS was a non-normal dependent variable (Skewness=3.373, Kurtosis=30.601), so log transformation was used (Skewness=−0.400, Kurtosis=0.234). Univariate analysis was firstly applied to identify potential factors. Then selected significant factors were included into the final multiple regression model.

Age (B =0.285 P<0.001), gender (t=6.173, P<0.001), history of tobacco smoking (F=22.145, P<0.001), number of comorbidities (F=6.559, P<0.001), blood eosinophil count (B =−0.071, P=0.024), hemoglobin (B =−0.212, P<0.001), use of systemic steroids (t=−2.807, P=0.038), oxygen therapy (t=−0.179, P<0.001), NIV (t=−7.668, P<0.001), PR (t=−14.955, P<0.001) and inhaled medications (F=17.290, P<0.001) during hospitalization were all associated with LOS. All these 10 factors were selected into the final multivariable regression model.

Table 3 summarized the results of univariate analysis and multivariable regression (F=24.401, p<0.001). Age (B=0.178, 95% CI: 0.535–1.072, P<0.001), gender (B=−1.147, 95% CI: −0.138–0.056, P<0.001), hemoglobin (B=−0.118, 95% CI: −0.629- −0.214, P<0.001), NIV (B=0.067, 95% CI: 0.009–0.124, P=0.023) and PR (B=0.232, 95% CI: 0.181–0.297, P<0.001) had statistically significant associations with LOS. For tobacco smoking, people who still smoked at admission (B=−0.086, 95% CI:-0.124–0.018, P=0.08) or people who quitted smoking prior to admission (B=0.072, 95% CI:0.004–0.087, P=0.030) had significant differences in LOS compared against people who never smoked. There were statistical differences between people who had comorbidities and people who did not have any comorbidity in LOS (B=0.126, 95% CI: 0.034–0.147, P=0.002; B=0.125, 95% CI: 0.031–0.144, P=0.002; B=0.116, 95% CI: 0.028–0.146, P=0.004; B=0.090, 95% CI: 0.021–0.166, P=0.011). Similarly, compared with patients who did not use any inhaled medications, COPD patients who used two or three types of inhaled medications had significant associations with longer LOS (B=0.109, 95% CI: 0.003–0.166, P=0.043, B=0.255, 95% CI: 0.083–0.237, P<0.001).

Table 3 Significant Factors Associated with LOS

Discussion

This is the first retrospective study investigating the influencing factors of LOS among COPD inpatients in Macao. Patients’ demographic information, blood test results, clinical therapies, and LOS were investigated separately. The average LOS was around 12 days. Age, gender, hemoglobin, history of smoking, the presence of comorbidities, and the use of NIV, PR and inhaled medications were found to be associated with an increase in LOS. The characteristics of LOS of COPD inpatients in Macao and the relationships between LOS and contributors are further discussed below.

The mean LOS among COPD patients in Macao (12.28 days) was higher than that in China (9.38 days),¹² North West England (9.8 days),¹⁶ United States (5.9 days),³² and European countries (8.7 days).¹⁷ This may be partly explained by the older age of the patients in this study which found that an increase in log (age) by one-unit resulted in hospital day increase by 1.148 units. The average age of COPD inpatients was 79.42 years old in this study. This is considered high compared to the mean age of 69.9–70.4 years old among hospitalized COPD patients in the US,³³ and 70.7 years old among COPD inpatients in European countries.¹⁷ In mainland China, a national cross-sectional study found that over 48% patients with severe COPD were only 70 years old or above.⁴

Having older COPD inpatients in Macao is not surprising considering the outlook of life expectancy and the aging population. The average life expectancy at birth in Macau in 2016–2019 was 83.8 years, and that for men and women was 80.8 years and 86.7 years, respectively.³⁴ By 2025, the proportion of the elderly population is expected to reach 16.3% and Macau will become an aging society.³⁵ Increasing age is prone to a progressive deterioration in lung function. Older COPD patients generally have a higher risk of acute exacerbations, a higher level of comorbid illness, longer LOS, and higher hospital cost.³⁶ Poorer physical condition and frailty might make it harder for the patients to recover from an exacerbation.^37,38 With the aging population in sight, the prevalence of severe COPD in older individuals resulting in hospitalization and negative impact on LOS will only be increasing.

Three out of 5 COPD inpatients in this study had at least one comorbidity which was similar to previous studies.^39,40 The number of comorbidities was also associated with an increase in LOS reaffirming that comorbidities were important predictors of prolonged hospitalization.^14,41–44 Such association might be contributed by pathophysiologic changes that compromised lung function leading to increased risks for other diseases or complications,^45–47 while shared risk factors such as smoking, age, sex, and inactivity might also add to the complexity of the association between comorbidity and COPD. Collectively, the associations of age and comorbidities with LOS posed great implications to the healthcare provision drawing attention to the importance of an integrative and interdisciplinary approach for older patients.^48,49 Pharmacists should be actively involved to improve adherence to pharmacotherapy and to optimize treatment outcome. Allied health care professionals such as physiotherapists are also important to support recovery and physical training.⁵⁰

Regarding clinical interventions, 11.5% and 10.8% of the patients in this study used NIV and PR during hospitalization. The use prevalence of NIV and PR was consistent with what was reported about the COPD patients in UK (12.1% of COPD patients received NIV during hospitalization)⁵¹ and US (only 1.5% initiated pulmonary rehabilitation within 90 days of their hospital discharge).⁵² Both NIV and PR were associated with a longer LOS in this study. NIV is used to improve the blood oxygen and carbon dioxide exchange rate of the patient’s lungs to prevent endotracheal intubation and reduce mortality in severe cases.^53,54 As such, NIV may be interpreted as an indirect indicator of more severe COPD, which could explain its association with a longer LOS.⁵⁵

Special attention should be paid to the role of PR for COPD patients. Being a comprehensive clinical intervention for AECOPD managements,⁵⁶ PR might benefit clinical outcome for the majority of the patients reducing dyspnea scale, LOS, risk of readmission and annual hospital day.^57,58 The findings about PR being associated with a longer LOS in this study may be explained by the PR data being limited to each hospitalization. At present, various methods and therapies of PR have been used in COPD patients.⁵⁹ Details of the PR design (components, frequency and duration) and the long-term benefits of PR for inpatients with COPD warrants further investigation in the future.

Regarding inhaled medications, 94.8% of the patients used them and 78.2% used a combination that mostly involved a ICS and either a short-acting or a long-acting beta-agonist. While the benefits of inhaled medication in reducing bronchospasm and inflammation,²⁰ relieving ongoing breathing problems,¹⁸ and improving the overall management of COPD^18,60 have been well established, the use of inhaled medications was found to be associated with a longer LOS in this study. Such results should be interpreted carefully as the need for multiple pharmacotherapy during hospitalization might at best indicate the severity of COPD. To further evaluate the association between inhaled medications and LOS, future studies should focus on the medication use and adherence prior to admission.⁶¹

At last but not least, the negative impact of cigarette smoking to the LOS and the overall COPD management should be emphasized.^62,63 In this study, only 17.5% patients smoked at admission which was a relatively low ratio compared to previous findings,^4,64 and 45% patients already quitted smoking which was a higher ratio compared to another study.⁶⁵ While current smokers and former smokers were found to have a longer LOS,^66–68 non-smokers in this study were found to have longer LOS when compared with current smokers in this study. Upon an in-depth investigation, the average age of the smoking population in this study was lower (73.4 vs 79.8 years old) and the smokers in this study might have a milder condition and so a shorter LOS. Nevertheless, smoking cessation at an early stage of COPD has been shown to improve prognosis largely⁶⁹ and any efforts to prevent cigarette smoking and support smoking cessation should be reinforced continuously.

There were some limitations in this study. Firstly, due to the limitation of data, the severity of the disease, the dosage and duration of medications used by the patients during hospitalization and the cost factor was not involved in this research, which may collectively affect LOS. Furthermore, future studies may supplement the current findings by taking into account other factors that might indicate the COPD disease stages such as intensive care unit admission data, detailed information about NIV, the laboratory measurement data, oxygen saturation and the use of supplemental oxygenation data, etc. Pulmonary function test indicators such as baseline spirometry, FEV1/FVC ratio, were not considered in this study due to a limited number of tested patients, which prevented a full glimpse of the patients’ overall lung status. Despite the data limitation, this current study has provided valuable analytical results about the influencing factors of LOS among COPD patients in Macao which may serves a foundation to support future research on reducing disease burden and improving COPD management.

Conclusion

COPD patients treated with NIV, PR, 2 or 3 types of inhaled medications during hospitalization had longer LOS in Macao, which may be associated with more severe acute exacerbations before admission. Patients with COPD should avoid risk factors that cause acute exacerbations, such as smoking, more comorbidities, and low hemoglobin level. Early smoking cessation, use of inhaled medications and timely clinical treatments of other diseases such as anemia are effective interventions for COPD patients to reduce LOS and long-term disease burden.