The rate of hypercapnic respiratory failure in a pulmonary function test laboratory database

Extract

Chronic hypercapnic respiratory failure (CHRF) may occur in advanced stages of COPD. The prevalence of CHRF due to any causes in the general population ranges from 0.08% in in 45- to 54-year-old individuals to 1.7% in those aged >85 years [1]. In patients with COPD, the prevalence of CHRF is not well studied. The aim of this study was to estimate the rate of CHRF in individuals with various spirometric categories and various degrees of lung function impairment.

Shareable abstract

Hypercapnia rates are in the range 3.6–12% among those with abnormal spirometry and FEV₁ ≥80% pred, and 53–58% among those with FEV₁ <35% pred. Both airflow obstruction and preserved ratio impaired spirometry are associated with higher risk of CHRF. https://bit.ly/3H8DlfM

To the Editor:

Chronic hypercapnic respiratory failure (CHRF) may occur in advanced stages of COPD. The prevalence of CHRF due to any causes in the general population ranges from 0.08% in in 45- to 54-year-old individuals to 1.7% in those aged >85 years [1]. In patients with COPD, the prevalence of CHRF is not well studied. The aim of this study was to estimate the rate of CHRF in individuals with various spirometric categories and various degrees of lung function impairment.

We retrospectively analysed arterial blood gases (ABGs) of individuals aged >18 years performed for any reason at University of Iowa Hospitals & Clinics pulmonary function test (PFT) labs (Iowa City, IA, USA) between 1 February 2001 and 31 December 2023. Our study protocol was approved by the University of Iowa institutional review board (identifier 202105133). Our inclusion criteria were 1) outpatient ABGs for any clinical reason, and 2) spirometry within 1 year of the ABG. Since our goal was to examine the rate of CHRF in various spirometric categories and various degrees of lung function impairment, we focused only those ABGs accompanied by spirometry testing. If there were several ABGs from one individual, we included only the most recent. If there were more than one set of spirometry data, we included the one closest to the ABG collection date. Demographic data, body mass index (BMI) and smoking history were extracted from the ABG encounter. All data extraction was performed by an independent team (Research Information Technologist team, Institute for Clinical and Translational Science, University of Iowa) using Epic electronical medical records (Epic, Verona, WI, USA). Spirometry was performed according to the PFT lab protocol, which follows the current American Thoracic Society PFT guidelines. Forced expiratory volume in 1 s (FEV₁) and forced vital capacity (FVC) % predicted values were calculated by the PFT lab software based on predicted values per American Thoracic Society PFT guidelines. We excluded records with missing values in pH, partial pressure of arterial carbon dioxide (P_aCO2) and partial pressure of oxygen in the arterial blood (P_aO2). In addition, we excluded records with no pre-bronchodilator FEV₁, FVC and FEV₁/FVC, and those with P_aO2 <40 mmHg, as they were assumed to be venous samples.

Spirometries were classified into three categories: airflow limitation when FEV₁/FVC was <0.7; preserved ratio impaired spirometry (PRISm) when FEV₁/FVC was ≥0.7 and FEV₁ and/or FVC % pred <80%; and normal. We compared characteristics between individuals with and without hypercapnia using t-tests for continuous variables and Chi-squared tests for categorical variables. We compared the rates of CHRF between spirometric categories, and FEV₁ % categories using the Chi-squared test. We created a multivariable logistic regression model with spirometric category as the main independent variable and P_aCO2 ≥45 mmHg as the dependent variable adjusted for age, BMI and smoking history.

Out of 6204 records with ABG testing, 1591 had missing ABG values and 1779 had missing values in pre-bronchodilator spirometry and were excluded. Of the remaining 2834 individuals, 11 had P_aO2 <40 mmHg and were also excluded. Out of 2823 individuals in the analysis, 14.0% (n=394) had P_aCO2 >45 mmHg and 4.9% (n=137) had P_aCO2 >52 mmHg. The median P_aCO2 was 49 mmHg (interquartile interval (IQI) 47–54 mmHg) in those with P_aCO2 >45 mmHg and 37.50 mmHg (IQI 34.00–40.60 mmHg) in those with P_aCO2 ≤45 mmHg.

The group with P_aCO2 >45 mmHg had older age (median 61 years, IQI 53–68 years versus 58 years, 49–66 years; p<0.001), fewer males (58% versus 59%; p<0.001), more ever-smoking history (61.2% versus 54.9%; p<0.001), similar numbers of non-Hispanic white people (90.4% versus 90.7%; p=0.91), more people with BMI >35 kg·m⁻² (32% versus 23.6%; p<0.001), greater pre-bronchodilator FEV₁ (median 45.0% pred, IQI 29.3–65.0% pred versus 77%, 63–91% pred; p<0.001), more airflow limitation (59.4% versus 18.5%) and PRISm (32.2% versus 25.4%; p<0.001), lower pH (median 7.39, IQI 7.37–7.41 versus 7.42, 7.40–7.44; p<0.001) and P_aO2 (median 69 mmHg, IQI 59–83 mmHg versus 83.9 mmHg, 74–95.1 mmHg; p<0.001) relative to the groups with P_aCO2 ≤45 mmHg.

Among individuals with airflow limitation, 18.5% (n=234) had P_aCO2 >45 mmHg and 7.4% (n=93) had P_aCO2 >52 mmHg, while among individuals with PRISm, 17.1% (n=127) had P_aCO2 >45 mmHg and 5.9% (n=44) had P_aCO2 >52 mmHg. Among individuals with airflow limitation, the rate of hypercapnia was 3.6% (10 out of 281), 9.1% (53 out of 586), 26.3% (49 out of 186) and 57.6% (122 out of 212) in those with FEV₁ ≥80% pred, 50–80% pred, 35–50% pred and <35% pred, respectively (p<0.001). Among individuals with PRISm, we observed similar findings (figure 1).

FIGURE 1.

The rates of hypercapnia (partial pressure of arterial carbon dioxide (P_aCO2) >45 mmHg) by spirometric category. Percentage of individuals with P_aCO2 >52 mmHg is also shown. a) Airflow limitation; b) preserved ratio impaired spirometry (PRISm). Chi-squared test was applied for comparison of rates of hypercapnia between forced expiratory volume in 1 s (FEV₁) % predicted groups.

After adjusting for age, BMI and smoking history, both airflow limitation (OR 4.48, 95% CI 3.06–6.75) and PRISm (OR 4.38, 95% CI 2.95–6.69) were associated with P_aCO2 >45 mmHg.

In this real-life sample of individuals referred for an ABG to a single academic centre, 14% had P_aCO2 >45 mmHg and 4.9% had P_aCO2 >52 mmHg. ∼20% of the those with abnormal spirometry had P_aCO2 >45 mmHg and the rate of hypercapnia increased with decreasing FEV₁ % pred. After adjusting for age, smoking history and BMI, both airflow limitation and PRISm are associated with hypercapnia relative to normal spirometry.

The pathophysiological mechanisms of CHRF in chronic lung diseases are complex [2, 3]. It appears that an imbalance in load capability within the respiratory system leads to reduced respiratory drive. A recent secondary analysis of the National Emphysema Treatment Trial, which included only patients with COPD severe airflow limitation and hyperinflation due to COPD, revealed that reduced minute ventilation is associated with CHRF [4]. We found that PRISm is associated with CHRF. Individuals with PRISm may have an increased respiratory load, and thus they have a reduced respiratory drive. Another explanation is that PRISm is more often associated with other comorbidities such as obesity hypoventilation syndrome [5]. Nevertheless, in our analysis we adjusted for BMI, and the association of PRISm and CHRF persisted.

Identifying CHRF is important because CHRF is associated with high risk of hospitalisation and mortality [6–9]. The rates of CHRF in chronic lung diseases are under-studied. Half of the patients admitted to the hospital with hypercapnia due to COPD have persistent hypercapnia (CHRF) after discharge [10, 11]. A German study showed that 25% of patients in outpatient clinics with COPD and severe lung function impairment have P_aCO2 >45 mmHg [12]. In this report, we extended the literature by showing that in a United States sample, CHRF rates can range from 3.6% to 58% depending on severity of lung function impairment.

Apart from its retrospective nature, our study is limited by the fact that it was conducted in a single academic centre with a predominantly white population. We do not have accurate data regarding the clinical reasons for which the ABGs were performed. Furthermore, there is a likelihood of selection bias because we only included individuals who underwent both ABG and spirometry testing for various clinical reasons. This does not undermine the significance of our findings, including the observation that a considerable proportion of individuals with mild to moderate lung function impairment have CHRF, and PRISm is associated with CHRF. Identifying CHRF may help reducing the underutilisation of home noninvasive ventilation [9, 13, 14].

In conclusion, CHRF ranges from 3.6% to 12% among those with abnormal spirometry and FEV₁ ≥80% pred, and between 53% and 58% among those with FEV₁ <35% pred. Both airflow limitation and PRISm are associated with higher risk of CHRF.