Background

To date, no consensus exists on the effects of steroid use on pneumonic chronic obstructive pulmonary disease (COPD) owing to trial design issues in previous trials involving these conditions. Therefore, we aimed to evaluate steroid effectiveness in pneumonic COPD exacerbation patients.

Methods

This multi-centred, retrospective, observational study was conducted across five acute general hospitals in Japan. We analysed the association between parenteral/oral steroid therapy and time to clinical stability in pneumonic COPD exacerbation.

We used a validated algorithm derived from the 10th revision of the International Classification of Diseases and Related Health Problems (ICD-10) to include pneumonic COPD exacerbation patients. We excluded patients with other hypoxia causes (asthma exacerbation, pneumothorax, heart failure) and complicated pneumonia (obstructive pneumonia, empyema), those who required tracheal intubation/vasopressors, and those who were clinically stable on the admission day.

The primary outcome was time to clinical stability. Multiple imputation was used for missing data. Propensity scores within each imputed dataset were calculated using potential confounding factors. The Fine and Gray model was used within each dataset to account for the competing risk of death and hospital discharge without clinical stability, and we combined the results.

Results

Altogether, 1237 patients were included. The pooled estimated subdistribution hazard ratio of time to clinical stability in steroid versus non-steroid users was 0.89 (95% confidence interval, 0.78 to 1.03). However, there were potentially unmeasured confounders, and we could not assess longer-term outcomes.

Conclusions

The current study recommends that steroid therapy should not be used routinely for pneumonic COPD exacerbation.

Study design and participants

This retrospective cohort study was conducted across five acute general hospitals in Japan: Kameda Medical Center, Hyogo Prefectural Amagasaki General Medical Center, Awa Regional Medical Center, Saiseikai Yokohamashi Tobu Hospital, and Ichinomiyanishi Hospital. Kameda Medical Center and Hyogo Prefectural Amagasaki General Medical Center are tertiary care hospitals while Awa Regional Medical Center, Saiseikai Yokohamashi Tobu Hospital, and Ichinomiyanishi Hospital are secondary care hospitals. These hospitals cover a wide medical area in Honshu, the main island in Japan. The study protocol adhered to the Declaration of Helsinki. Additionally, this project was approved by the institutional review board of each hospital, and the need for written informed consent was waived by the institutional review board of Kameda Medical Center because of the study’s retrospective nature (approval number, 19-076). 

We analysed the association between steroid therapy and the clinical important outcomes among patients with pneumonic COPD exacerbation. Each hospital included patients who were treated during the following different time periods because the hospitals had different storage terms for their electronic medical records: Kameda Medical Center, April 1, 2008–March 31, 2019; Hyogo Prefectural Amagasaki General Medical Center, July 1, 2015–August 31, 2019; Awa Regional Medical Center, April 1, 2010–August 31, 2019; Saiseikai Yokohamashi Tobu Hospital, April 1, 2009–August 31, 2019; and Ichinomiyanishi Hospital, June 1, 2014–August 31, 2019. The inclusion criteria were patients who were 40 years or older and who were hospitalized for both pneumonia and COPD exacerbation. We used a modified version of the patient selection algorithm based on the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10), as described in a previous article. To validate the patient selection algorithm, we conducted a validation study at Kameda Medical Center between October 1, 2018 and March 31, 2019 using the chart review results of all the patients who were hospitalized in the Department of Internal Medicine and Pulmonology as a reference standard. We made a clinical diagnosis of pneumonic COPD exacerbation at the time of admission based on previous diagnostic criteria for each disease [6, 7]. For admission, we used either criteria (ⅰ) or (ⅱ) as follows: (ⅰ) admission-precipitating the diagnosis of pneumonia (ICD-10 codes; J12, J13, J14, J15, J16, J18, J69, and P23) with comorbidities present at the time of admission for COPD (ICD-10 codes; J44.1 and J44.9) and (ⅱ) admission-precipitating the diagnosis of COPD exacerbation (ICD-10 code; J44.1) with comorbidities present at the time of admission for pneumonia (ICD-10 code; J12, J13, J14, J15, J18, J69, and P23). By reviewing patient charts, we excluded those who had the following conditions at the time of admission: hypoxia caused by other respiratory diseases (asthma exacerbation, pneumothorax, or heart failure), complicated pneumonia (obstructive pneumonia or empyema), severe cardiopulmonary conditions (patients who needed tracheal intubation or vasopressors on the day of admission), or conditions requiring daily steroid use. We also excluded patients who fulfilled the clinical stability criteria outlined below on the day of admission.

Data extraction

The patients’ medical records provided the following data: age, sex, types of inhalers used for COPD (inhaled corticosteroid, long-acting beta-agonist, and long-acting muscarinic antagonist), use of steroid treatment for pneumonic COPD exacerbation, use of home oxygen therapy, activity of daily living (full assistance or not), blood test (white blood cell counts, eosinophil counts, and urea nitrogen) results on admission, information regarding wheezing lung sounds, and tracheal intubation.

Exposure

Treatments of interest included systemic steroid therapy versus no steroid therapy. Steroid therapy was defined as oral or parenteral administration of steroids after hospitalization regardless of the doses or timings.   

Outcomes

The primary outcome was time to clinical stability. The definition of clinical stability that we used is outlined in a previous RCT. Patients were determined to be clinically stable when they met all the following criteria: temperature ≤ 37.2°C, heart rate ≤ 100 beats/minute, systolic blood pressure ≥ 90 mmHg, arterial oxygen tension ≥ 60 mmHg, or blood oxygen saturation level (SpO2) ≥ 90% in ambient air or with the preadmission oxygen flow. The secondary outcomes were time to hospital discharge, time to in-hospital death, number of clinically diagnosed patients with delirium, number of new insulin users, and number of tracheal intubated patients following day 2 of hospitalization. We selected patients with delirium and new insulin users based on the chart review that included medical records written by doctors, charts filled in by nurses, and patients’ drug prescriptions during hospitalization.

Covariates

The potential confounding factors were age, sex, use of home oxygen therapy, activity of daily living, respiratory rate, altered mental status, heart rate, and blood urea nitrogen. We selected these variables as confounding factors based on previously reported prognostic factors for pneumonia and COPD exacerbation [9, 10]. As patients admitted to the same hospital may have received similar treatments, we considered each hospital as a cluster.

Statistical analysis

The participants’ baseline characteristics were summarized as proportions for categorical variables and as means with standard deviations for continuous variables. We compared the differences between the treatment groups (steroid users vs. non-steroid users).

              We used multiple imputation and propensity score analysis for assessing the time to clinical stability, time to hospital discharge, and time to in-hospital death. Missing data were imputed by 100 complete datasets with multiple imputation by chained equations on the assumption of data missing at random. We used potential confounders and outcome variables for estimating the missing data. The propensity score was calculated for each imputed dataset, and patients who were steroid users were matched to steroid non-users in a 1:1 ratio without replacement based on the propensity score. The nearest neighbour technique was used with a calliper of width equal to 0.2 of pooled standard deviation of the logit of propensity score. Disparities in baseline characteristics between the two groups were assessed with the standardized mean difference. Then, we used the survival analysis within each dataset for comparing the event rates based on the treatment group. Regarding time to clinical stability and time to hospital discharge, we estimated the cumulative incidence of patients who reached each event. We calculated the subdistribution hazard ratio (sHR) using the Fine and Gray model by adjusting for potential confounding factors to account for the competing risk of death and hospital discharge without clinical stability [11]. Regarding time to in-hospital death, we used the Cox proportional hazards model. Finally, we averaged with Rubin’s rule over datasets to estimate the treatment effects

variables' information

id: anonimized patients' ID, age: patients' age on hospitalization, gender: 1=male, 2=female, steroid: steroid use (0=not, 1=steroid use), hospital: categorized hospital name (1=Kameda Medical Center, 2=Awa Regional, 3=Hyogo Prefectural Amagasaki General Medical Center, 4=Saiseikai Yokohamashi Tobu Hospital, 5=Ichinomiyanishi hospital), adl: activity daily living before hospitalization (0: full support, 1: not full support), wheeze: wheezing lung sound on admission (0:not, 1:wheezing lung sound), bun: blood urea nitrogen (mg/dL), rr: respiratory rate (/minute), ams: altered mental status (0=not, 1=altered mental status),  hr: heart rate (/minute), hot: home oxygen therapy use before hospitalization (0=not, 1=users), insulin: new insuline users during hospitalization (0=not, 1=new users), delirium: clinical diagnosis of delirium during hospitalization (0=not, 1=delirium), stability: reaching the clinical stability (0=not, 1=reached clinical stability), time_to_stability: time to clinical stability (days), death: in-hospital death, non-informative_censoring: non-informative censoring (0=not, 1=non-informative censoring), discharge (0=not, 1=discharged patients), intubation: tracheal intubation during hospitalization (0=not, 1=intubated), hospitalization=length of stay (days)