This observational prospective study was approved by the local ethics committee of Sechenov University (IRB No.10-23 dated April 27, 2023) and conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Written informed consent was obtained from each patient.

The study included 72 patients hospitalized for acute exacerbation of COPD. The diagnosis of COPD was established based on clinical signs and symptoms, validated through the patient’s medical history, and lung function tests per the Global Initiative for Chronic Lung Disease (GOLD) guidelines [2]. All participants had received DPI therapy as maintenance treatment prior to admission. Exclusion criteria encompassed current diagnoses of asthma, interstitial lung diseases, lung cancer, decompensated heart failure, and pneumonia, as well as cognitive impairments.

The analysis encompassed demographic and clinical characteristics such as age, gender, duration of COPD diagnosis, body mass index, smoking history, and clinical symptoms evaluated using the Breathlessness, Cough, and Sputum Scale (BCSS) and the COPD Assessment Test (CAT). It also included the assessment of shortness of breath using the modified Medical Research Council (mMRC) dyspnea scale, the frequency of COPD exacerbations and hospitalizations in the past year, and the monthly need for short-acting bronchodilators during the previous year. The analysis also incorporated the type of inhaler used for maintenance therapy, duration of inhaler usage, outpatient inhalation technique training, and the number of different inhalers used over the previous year. Compliance was assessed with the 12-item Test of Adherence to Inhalers (TAI-12) [23,24].

Pulmonary function tests (spirometry, body plethysmography, diffusion capacity of the lungs) were performed routinely in all patients in accordance with American Thoracic Society (ATS)-European Respiratory Society (ERS) guidelines. Results were reported as absolute values and percentages of predicted values [25]. A COSMED Q-box device (COSMED, Rome, Italy) was utilized for assessing lung function. The parameters analyzed included inspiratory capacity (IC), residual volume (RV), intrathoracic gas volume (ITGV), total lung capacity (TLC), RV/TLC ratio, and diffusing capacity for carbon monoxide (DL_CO). Maximal inspiratory pressure (MIP) and maximal expiratory pressure (MEP) were measured following a modified method by Black and Hyatt. Measurements for MIP were taken from RV, whereas measurements for MEP were taken near TLC.

PIF was evaluated twice with the In-Check DIAL G16: upon the patient’s admission and on the day prior to discharge. The In-Check DIAL features an adjustable dial that simulates the internal resistances of DPIs [26]. It measures inspiratory flow rates from 15 to 120 L/min and is accurate within 10% or 10 L/min, according to the manufacturer’s leaflet [27]. PIF assessments with the In-Check DIAL G16 included (1) day-to-day typical PIF and PIF following explanation of inhalation technique at the resistance of the patient’s inhaler; (2) maximum PIF at medium-low (R2) internal resistance; and (3) maximum PIF at high (R5) internal resistance. For typical PIF measurements, patients were instructed to inhale using the In-Check DIAL G16 as they normally would with their DPI. Subsequently, patients were advised to exhale fully before inhaling forcefully and quickly, with re-measurements of PIF at the inhaler resistance, R2, and R5. Each measurement (post-explanation) was performed in triplicate, and the highest result from the three trials was recorded in the analysis. PIF was categorized as optimal (≥60 L/min) or suboptimal (<60 L/min) for the overall patient sample [16,26,27]. To examine differences in PIF across various inhaler devices, threshold PIF values were set at 50 L/min for the Breezhaler (Novartis Pharma, Basel, Switzerland; low resistance, R1), 60 L/min for the Ellipta (GlaxoSmithKline, London, UK; R2), and 45 L/min for the Turbuhaler (AstraZeneca, Södertälje, Sweden; medium resistance, R3) [28].

Clinical parameters (BCSS scale, mMRC, and CAT questionnaire), spirometry, and PIF measurement at resistance levels of a patient’s inhaler, R2, and R5, were routinely assessed at discharge.

Discrete variables were presented as frequencies, while continuous variables were presented as medians with interquartile ranges. The characteristics of patients at admission and before discharge were compared using the Wilcoxon-matched paired signed rank test (Wilcoxon signed rank test). Comparisons of categorical variables between groups were conducted using Fisher’s exact test or Pearson’s chi-square test; for paired nominal data, McNemar’s test was employed. Stepwise multiple logistic regression was conducted to identify independent predictors of sPIF at admission. Odds ratio (OR) and 95% confidence interval (CI) were computed. Receiver operating curve (ROC) analysis was utilized to evaluate the sensitivity (Sp) and specificity (Se) of sPIF, predicted by risk factors, and to determine the optimal prognostic cut-off values. These cut-off values were calculated using Youden’s index, which was determined only for pairs of Se and Sp values where both exceeded 0.5. A p-value of ≤0.05 was deemed statistically significant. Statistical analyses were performed using IBM SPSS Statistics software, version 26 (IBM Co., Armonk, NY, USA). ROC analyses were conducted with both GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA) and the R programming language using the pROC package (R Foundation for Statistical Computing, Vienna, Austria).

The study involved 72 hospitalized patients with an acute exacerbation of COPD (median age, 68.5 years [IQR, 63.0 to 72.8]; 82% were male). The average smoking history was recorded as 40.0 pack-years (IQR, 22.1 to 50.0). Baseline characteristics of the patients are detailed in Table 1.

The prevalent clinical symptom was dyspnea on exertion (mMRC 3 points [IQR, 3 to 4]). There were 2 (IQR, 1 to 3) COPD exacerbations in the previous year. The number of hospitalizations due to COPD exacerbation during the previous year was 1.0 (IQR, 0.7 to 2.0). The monthly requirement for short-acting beta-2 agonists was 60.0 doses (IQR, 0 to 120.0). During the study period, one death was reported (1.4% of the patients).

The most commonly used DPI was Ellipta (R2) by 33 patients (46%), followed by Breezhaler (R1) by 25 patients (35%), and Turbuhaler (medium-to-high resistance, R4) by 14 patients (19%). The average duration of maintenance therapy with DPIs was 1.5 years (IQR, 0.2 to 3.0). Thirty-nine patients (54%) had received training on the inhaler technique 0.5 years (IQR, 0.0 to 4.0) prior to entering the study. In the previous year, 40 patients (56%) used one type of inhaler, 27 (37%) used two types, and five (7%) used three types; 23 patients (37%) were switched to another inhaler device. The inhaler technique was evaluated using the TAI test; the average score was 50 points (IQR, 45 to 50), indicating good adherence to inhalation therapy.

PIF was measured at a resistance specific to the patient’s inhaler before and immediately after instruction on the proper inhaler technique. Initially, sPIF (<60 L/min) was observed in 38 patients (52.7%). Following instruction on the proper inhaler technique, sPIF (<60 L/min) was still observed in 14 patients (19.4%) (p<0.0001). PIF (at the patient’s device resistance) increased significantly from 55.0 L/min (IQR, 45.0 to 70.0) before instruction to 71.5 L/min (IQR, 63.3 to 85.0) afterward (p<0.0001) (Figure 1). Interestingly, initial PIF before instruction was not associated with prior training in ambulatory settings, switching to another inhaler device during the previous year, the number of inhalers used by the patient, or the duration of the disease.

Moreover, we analyzed patients who achieved sPIF with different DPIs based on the sPIF thresholds provided by the manufacturer. Initially, the sPIF frequencies among users of different inhaler devices were similar: 40% in the Breezhaler group, 35.7% in the Turbuhaler group, and 45.5 % in the Ellipta group. After instruction on the proper inhaler technique, statistically significant differences in sPIF relative to baseline were observed solely in the Ellipta group (45.5% vs. 9.1%, p=0.0001).

During the treatment period in the hospital, there was a significant improvement in dyspnea (mMRC: 3 [IQR, 2 to 4] vs. 3 [IQR, 3 to 4], p<0.0001) and clinical symptoms (CAT test: 20 [IQR, 13 to 26] vs. 25 [IQR, 18 to 31], p<0.0001; BCSS: 3 [IQR, 2 to 5] vs. 5 [IQR, 3 to 7], p<0.0001) (Table 2). Despite the clinical improvements, we observed no significant change in PIF when testing the patient’s device resistance (71.5 [IQR, 63.3 to 85.0] initially vs. 74.5 [IQR, 62.0 to 90.0] at discharge, p=0.239). At discharge, sPIF (<60 L/min) was recorded in 15 patients (20.8%) compared to 14 patients (19.4%) at baseline. Only PIF at R5 showed a statistically significant difference (44.5 [IQR, 35.0 to 50.0] vs. 45.0 [IQR, 38.0 to 54.0], p=0.016) (Figure 2). No significant differences were noted in the proportion of patients in each DPI group reaching the threshold sPIF level at discharge compared to the post-training sPIF at admission: 20% vs. 8% in the Breezhaler group, 14.3% vs. 14.3% in the Turbuhaler group, and 9.1% vs. 15.2% in the Ellipta group.

PIF, measured at admission after explaining the proper inhaler technique and adjusted for the resistance of the patient’s device, correlated with demographic factors (body weight: r=0.26, p=0.026; age: r=-0.32, p=0.005), clinical measures (mMRC: r=-0.29, p=0.011; CAT test: r=-0.32, p=0.006), and functional parameters (forced vital capacity [FVC, L]: r=0.55, p<0.0001; FVC % predicted [pred.]: r=0.44, p<0.0001; forced expiratory volume in 1 second [FEV₁, L]: r=0.44, p<0.0001; FEV₁ % pred.: r=0.35, p=0.002; FEV₁/FVC: r=0.24, p=0.043; IC [L]: r=0.64, p<0.0001; IC % pred.: r=0.52, p=0.003; RV/TLC: r=-0.60, p=0.001; DL_CO % pred.: r=0.64, p<0.0001; ITGV % pred.: r=-0.49, p=0.015; ITGV, L: r=-0.44, p=0.031; and MIP, % pred.: r=0.67, p=0.048).

Regression analysis revealed that the following factors significantly predicted optimal PIF after explaining the proper inhaler technique at admission: age (OR, 0.91; 95% CI, 0.85 to 0.99; p=0.023); FVC, L (OR, 3.33; 95% CI, 1.42 to 7.81; p=0.006); FVC, % pred. (OR, 1.04; 95% CI, 1.00 to 1.07; p=0.032); FEV₁, L (OR, 4.85; 95% CI, 1.29 to 18.29; p=0.020); FEV₁, % pred. (OR, 1.04; 95% CI, 1.00 to 1.08; p=0.038); RV % pred. (OR, 0.98; 95% CI, 0.96 to 1.00; p=0.047); and RV/TLC (OR, 0.89; 95% CI, 0.80 to 0.99; p=0.040). Multiple regression analysis demonstrated that age (OR, 0.89; 95% CI, 0.82 to 0.98; p=0.02) and FEV₁, % pred. (OR, 1.08; 95% CI, 1.02 to 1.15; p=0.009) were predictors of optimal PIF at admission (Table 3).

The ROC analysis demonstrated that independent predictors of sPIF at admission were age >70 years (Se, 64%; Sp, 57%; area under the curve [AUC], 0.677; 95% CI, 0.498 to 0.857; p=0.040); FVC <73% pred. (Se, 77%; Sp, 63%; AUC, 0.702; 95% CI, 0.566 to 0.838; p=0.024); FEV₁ <35 % pred. (Se, 62%; Sp, 69%; AUC, 0.694; 95% CI, 0.544 to 0.844; p=0.03); RV >194% pred. (Se, 75.0%; Sp, 77.3%; AUC, 0.830; 95% CI, 0.576 to 1.000; p=0.04); RV/TLC >70% (Se, 75%; Sp, 96%; AUC, 0.859; 95% CI, 0.665 to 1.000; p=0.024); DL_CO <36% pred. (Se, 75 %; Sp, 86%; AUC, 0.841; 95% CI, 0.621 to 1.000; p=0.033) (Figure 3).

Leving et al. [16] recorded comparable results in a large cohort of patients diagnosed with COPD (n=1,389), showing that among 402 patients with sPIF, 219 (16%) could achieve the optimal PIF but failed to do so routinely, and 183 (13%) were physically incapable of generating sufficient inspiratory flow [16]. Numerous studies have validated the efficacy of this training approach in patients with COPD, utilizing devices that apply resistance during inspiration corresponding to that of specific DPIs.

According to our data, a relatively high frequency of sPIF was observed in patients with exacerbations of COPD, both at admission and discharge from the hospital 20.8% and 19.4% respectively. It is crucial to consider the differences in the frequency of suboptimal flow based on the resistance of the inhaler used (R1 [Breezhaler], 45.5%; R2 [Ellipta], 28.6%; and R3 [Turbuhaler], 87.5%). As reported by Duarte et al. [27], the prevalence of sPIF was about the same, with 61 out of 303 patients (20.1%) having COPD. Similar rates (19%) were reported in other studies involving patients with COPD [29]. COPD exacerbations significantly impact the progression of clinical symptoms and functional disorders, as well as the prognosis of the disease. The risk of PIF dropping below optimal values increases significantly during an exacerbation [30].

However, data on the prevalence of suboptimal flow in patients with COPD exacerbations are limited. In a study by Harb et al. [31], the prevalence of sPIF among 180 patients with a COPD exacerbation was 44.4%, and the distribution of suboptimal flow based on resistance level was similar to our data. Clark et al. [22] reported that approximately half of the patients (56.9%) had sPIF (<60 L/min) for R2, and 14.7% had low PIF (<30 L/min) for R5 during their hospitalization. Mahler et al. [32] discovered a prevalence of sPIF among COPD exacerbation patients as 44.6%: 61.0% at low to medium high resistance and 17.2% at high resistance. Meng et al. [19] observed that a higher proportion of patients experienced insufficient peak inspiratory flow rate during exacerbations compared to the stable phase (61.7% vs. 43.5%, p<0.001). Loh et al. [18] demonstrated that 52% of patients with COPD had sPIF, more commonly in those aged over 65 years.

Sharma et al. [12] showed that approximately one-third of patients had a PIF <60 L/min at discharge following hospitalization for a COPD exacerbation. Broeders et al. [33] found that PIF on day 1 of exacerbation was significantly lower than on day 5 and day 50 of follow-up, though no significant differences were observed between PIF values on day 5 and day 50.

In our study, we observed an improvement in clinical symptoms but no significant change in pulmonary function and parameters of PIF except for PIF at high resistance level (R5). This likely resulted from the lower baseline PIF at this R5 level at admission. These findings underline the importance of careful consideration when selecting baseline therapy for COPD following an exacerbation, particularly with DPIs. Assessing PIF at discharge can assist in determining if a DPI is necessary and in individualizing the choice of inhaler.

Although the availability of the In-Check DIAL device may be limited in real practice, it is crucial to identify predictors of sPIF using routine clinical and functional examinations. There remains some controversy regarding the associations between various demographic, clinical, and functional parameters and PIF in patients with COPD.

The main demographic factors identified as predictors of decreased PIF are age and female sex [12,29]. According to a recently published systematic review of 17 papers evaluating the association between age and PIF, 12 (71%) found that increasing age was associated with decreasing PIF, while nine out of 14 (64%) identified a positive correlation between female gender and low PIF [4].

Selecting the optimal delivery device for elderly patients is challenging due to potential coordination issues, cognitive changes, and fine motor impairments. DPIs are often the preferred inhalers for these patients because they are easier to operate and do not require coordination with inspiration. Sharma et al. [12] demonstrated that the likelihood of lower PIF values, which are suboptimal for DPIs use, increases with age (OR, 1.052; p=0.0058). Additionally, the reduction of PIF with age is linked to a decline in inspiratory muscle strength, and the impact of kyphoscoliosis and comorbid conditions has been explored [18]. Concurrently, several studies found no significant effect of age on PIF [29,31].

The literature presents conflicting data on the effects of BMI, height, and weight on PIF scores [4], with our study identifying only a weak correlation between PIF and patient weight. Duarte et al. [27] also observed that patients with sPIF exhibited lower FEV₁, TLC, and IC. Prime et al. [34] reported a direct correlation between PIF at the Ellipta inhaler and FEV₁ (r=0.69, p<0.0001) as well as DL_CO (r=0.71, p<0.0001). Harb et al. [31] noted that COPD patients with suboptimal flow had significantly lower FEV₁, peak expiratory flow, and forced expiratory flow at 25%, 50%, 75%, and 25%-75% of FVC. Mahler et al.’s [32] study using regression analysis found a significant effect of FVC and IC on PIF, although no correlation between FEV₁ and PIF was detected.

Although published data on the correlation between FVC and PIF are scarce, most suggest an inverse correlation between these two variables [4,21,27,35]. In one study, age (OR, 1.072; 95% CI, 1.019 to 1.128; p=0.007) and FVC (OR, 0.961; 95% CI, 0.933 to 0.989; p=0.006) emerged as significant factors in a multivariate analysis [36]. Moon et al. [28] identified a significant correlation between sPIF and factors such as age (OR, 1.06; 95% CI, 1.02 to 1.09) and FVC, % pred. (OR, 0.97; 95% CI, 0.95 to 0.99) in COPD patients using DPIs.

Despite contradictory evidence in the literature, Leving et al. [4], in a systematic review, demonstrated that two-thirds of the studies (64%) reported a positive correlation between PIF and FEV₁ (9/14 papers). The inconsistencies concerning the relationship between PIF and key spirometry parameters, including FEV₁, may indicate that, along with bronchial obstruction, other pathogenetic factors such as respiratory muscle weakness, air trapping, and pulmonary hyperinflation also influence PIF [37].

Several studies support an association between diminished PIF and markers of lung hyperinflation, specifically reduced IC [4,18,38]. For instance, Loh et al. [18] observed that among various lung function parameters analyzed for correlation with PIF, only IC (r=0.21; p=0.042) exhibited a correlation with PIF, while no associations with FVC or FEV₁ were detected.

It is well established that reductions in MIP and MEP can occur in patients with COPD, due to a variety of factors including malnutrition, muscular atrophy, steroid induced myopathy, and pulmonary hyperinflation leading to increased RV and diminished blood flow to the respiratory muscles. Concurrently, the generation of inspiratory flow is contingent on thoracic geometry and inspiratory muscle force [39]. When conducting a comprehensive functional examination of patients with COPD, Terzano and Oriolo [38] identified a correlation between PIF and MIP. Other studies also support the link between PIF and respiratory muscle strength [33].

In our study, we noted a correlation between PIF and parameters of bronchial obstruction (FEV₁, FVC), pulmonary hyperinflation (IC, ITGV, RV, RV/TLC), and inspiratory muscle weakness (MIP). These elements reflect the principal pathophysiological drivers of reduced PIF in COPD patients. ROC analysis highlighted the significance of age >70 years, FVC <73% pred., FEV₁ <35% pred., RV >194% pred., RV/TLC >70%, and DL_CO <36% pred. as predictors of sPIF. Multivariate regression analysis demonstrated that patient’s age (OR, 0.89) and FEV₁ (OR, 1.08) contributed significantly. These results can be employed to identify patients with COPD who are at high risk of suboptimal DPI use without necessitating additional tests beyond routine clinical assessment.

Our study had several limitations. Firstly, it was a single-center study, which could introduce bias in patient selection. Our patient population was not limited to those with severe exacerbations of COPD. Another study limitation is our choice of sPIF measurement of less than 60 L/min. While a PIF greater than 30 L/min is generally regarded as the minimal inspiratory flow needed to produce some clinical benefit, the total and fine particle doses from a DPI are optimized when the PIF exceeds 60 L/min, constituting this threshold as optimal for most DPIs [18,40]. Another limitation was that we did not fully consider the impact of comorbid conditions on PIF, although numerous studies suggest potential effects from conditions such as cardiovascular disease, anemia, etc. [12].

In conclusion, sPIF values are prevalent in patients with COPD, particularly following an exacerbation, potentially reducing the effectiveness of standard therapy with DPIs. It is crucial to recognize that suboptimal inspiratory flow may result from incorrect inhalation techniques, but proper instruction and training by a physician using devices such as the In-Check DIAL can significantly enhance PIF scores.

The ability to generate the optimal PIF is essential for effective drug delivery with DPI. PIF measurements could help clinicians select an appropriate inhaler for each patient and, if DPI is chosen, PIF could assist a patient in improving their inhalation technique and providing more effective treatment. It is advisable to recheck PIF if there is a change in the patient’s clinical status, such as when a physician plans to prescribe DPI to a patient who has recovered from a COPD exacerbation. This approach ensures personalized inhalation therapy for COPD patients.

PIF reduction in COPD is influenced by various pathophysiological mechanisms, including bronchial obstruction, lung hyperinflation, and inspiratory muscle weakness. This correlation is supported by the relationship between these functional parameters and PIF. Age and FEV₁ are the most important predictors of a sPIF value. Considering that most COPD patients are elderly and have compromised lung function, it is important to consider these factors when selecting optimal delivery devices for baseline therapy and deciding whether to use less flow-dependent delivery devices.