Systematic review of the effects of patient errors using inhaled delivery systems on clinical outcomes in COPD

Introduction

Patients with COPD frequently make errors when using devices to deliver inhaled medications.1 This problem was first recognised soon after hand-held devices were introduced in the 1960s2 and despite numerous efforts, there has been little change in proportion making errors since then.1 Common errors with pressurised metered dose inhalers (pMDIs) and slow mist inhalers relate to coordination, poor inspiratory manoeuvre and lack of breath hold,1 3 4 while incorrect preparation, failure to exhale and perform a forceful, quick and deep inhalation and lack of breath hold were the most common critical errors with dry powder inhalers (DPI).1 5 It is generally, and reasonably, assumed that if patients are not using their inhaled delivery system correctly, they will not achieve optimal benefit from their medication, and if they make a critical error, by definition6 they are unlikely to benefit at all.

Previous systematic reviews have examined the relationship between errors in inhaler technique and outcomes in patients with both asthma and chronic obstructive pulmonary disease (COPD).6 7 Much of the evidence included was about the effects in patients with asthma and there was limited evidence specific to COPD, mainly in meeting abstracts and a few peer-reviewed publications. The relative importance of specific errors has not been considered. As more evidence has now emerged, we have performed a systematic review (PROSPERO, CRD42023393120) in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.8 The objectives of this report are twofold: first to consider the impact of inhalation errors on clinical outcomes in COPD, including the importance of specific errors; and second to provide recommendations for future studies.

Methods

The PubMed, EMBASE, Cochrane and Google Scholar databases were searched according to PRISMA guidelines8 for articles published between 1 January 1990 and 1 September 2023 that met the following inclusion criteria: the study design was a cohort, case–control study or randomised controlled trial, included patients with COPD, included an objective assessment of errors using inhaler and included data on at least one clinical outcome of interest: forced expiratory volume in 1 s (FEV1), dyspnoea, health status, exacerbations. Additional studies were sought by searching the reference lists of the selected articles. Articles that were reviews, editorials or meeting abstracts were excluded, as were studies not reporting the results separately for patients with COPD. Studies were not restricted by language or geography.

PubMed, EMBASE and Cochrane were searched using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) search strategy for identifying studies including patients with COPD9 together with iterative combinations of the following keywords including wildcards for plurals and spelling variants were used during the search process: error, mistake, status, control, critical error, inhaler, outcome, exacerbation (see online supplemental material for search strategies).

Google and Google Scholar were searched using the following key terms, in different combinations: ‘copd’ inhale* (exacerbat* OR outcome* OR control) (error* OR mistake*).

Publication titles and abstracts were then screened, and articles that did not provide information on device errors and outcomes of COPD were excluded. Data were extracted systematically using a predefined extraction template (online supplemental table 1).

Results

A total of 1312 publications met our initial screening criteria (figure 1). An additional seven publications were identified through a manual search. After removing duplicate studies (n=168), 1144 abstracts were examined to assess relevance whether they met the eligibility criteria. 1091 abstracts were eliminated, leaving 55 full-text articles for review in detail. Among these articles, 36 did not contain information about associations between errors and outcomes or were review articles, resulting in a total of 19 publications that were included in this review5 11–28 (table 1).

Request permission

Seven prospective cohort studies of relevance were identified,11 13 16 19 21 24 26 two studies randomised patients to receive inhaler training or not, three also examined the effects of inhaler training but were not randomised and one monitored patients’ inhaler use for 3 months.16 We considered that no conclusions could be drawn from the studies with follow-up of 1 month or less.21 26 None of the intervention studies examined the outcomes in patients whose technique had improved compared with patients who continued to make errors and follow-up was often incomplete. Only the study by Cushen et al16 included more than 70 patients and was considered of good quality (table 2).

10 retrospective case–controlled studies, reported in 12 publications, compared the characteristics of patients who made and who did not make errors using their inhalers (table 1).5 12 14 15 17 18 20 22 23 25 27 28 These included larger numbers of patients than the prospective studies. Three publications23 27 28 reported different outcomes from the same study (The PIFotal COPD study). In all of these studies it is impossible to determine cause and effect, and only the PIFotal COPD study accounted for the effect of adherence.

In many studies, the exacerbation rates were based on patient recall rather than being derived from medical records, and it was not clear how patients were selected for inclusion. Only the PIFotal COPD study23 27 28 was considered of good quality (table 2).

Effect of errors on lung function

Figure 2 summarises the findings of the studies with regard to the outcomes. Maricoto et al showed that FEV₁ improved significantly after training in inhaler use (mean change=145.7 mL; 95% CI: 11.7 to 279.8; p=0.0350).13 One cross-sectional study showed that patients making no errors had a significantly better mean FEV₁ compared with those making errors (p=0.04)17 but another showed no correlation20 (table 1).

Request permission

Figure 2

Summary of study results. Each dot represents a study. Moderate exacerbations—requiring antibiotics or corticosteroids but not hospitalisation; severe exacerbations—requiring hospitalisation. *One publication from the PIFotal study reported no significant association but another found an association between three specific errors and more severe exacerbations. FEV₁, forced expiratory volume in 1 s.

Effect of errors on symptoms

A significant improvement in the mean modified Medical Research Council (mMRC) breathlessness scale scores after training in inhaler use with no change in the control group11 was reported by Göriş et al but this was not confirmed other studies.13 19 Cross-sectional studies have found no significant relationship between mMRC and making errors5 12 20 25 (table 1) (figure 2).

Effect of errors on health status

Göriş et al showed that health status improved after training with no change in the control group11 but other studies have shown no effect12 19 21 24 (figure 2).

The PIFotal study found that the errors in ‘breathing in’, ‘holding breath’ and ‘breathing out calmly after inhalation’ using a DPI were significantly associated with poorer Clinical COPD Questionnaire (CCQ) and COPD Assessment Test (CAT) outcomes (respective estimate of the difference (β) in the absolute score and 95% CIs CCQ β 0.16 CI (0.05 to 0.27), CAT β 0.97 CI (0.18 to 1.77); CCQ β 0.14 CI (0.01 to 0.28), CAT β 1.01 CI (0.16 to 2.02); CCQ β 0.27 CI (0.02 to 0.52); CAT β 2.62 CI (0.73 to 4.50)).28 Other cross-sectional studies show no relationship between errors and health status12 20 25 (table 1, figure 2).

Effect of errors on exacerbations and healthcare resources utilisation

The study by Cushen et al, which used electronic monitoring of adherence and inhalation technique, showed that patients who were adherent but made critical errors experienced more moderate exacerbations (ie, treated with antibiotics or corticosteroids but not hospitalisation) during the 12 months follow-up period than those who were adherent and did not make errors (4.62 (SD 4.69)/person/year vs 4.26 (SD 4.46)) but the difference was not statistically significant.16 Göriş et al reported a significant reduction in moderate exacerbations but no significant difference in severe exacerbations (ie, requiring hospitalisation) after training,11 but the duration of follow-up was only 3 months, and another study with longer follow-up showed no change in moderate or severe exacerbations19 (table 1, figure 2).

The initial publication of the PIFotal study reported that none of the errors made using a DPI were significantly associated with moderate or severe exacerbations.23 However, a subsequent publication claimed that patients with errors in ‘preparation’, ‘hold inhaler in correct position during inhalation’ or ‘breathe in’ had on average significantly more severe exacerbations than patients without these errors (respective rate ratios and 95% CIs: 2.83, 1.30 to 6.16; 1.94, 1.05 to 3.55; 1.85, >1.00 to 3.42).28 Other cross-sectional studies showed an inconsistent relationship between error rates and moderate or severe exacerbations or healthcare utilisation5 14 15 18 20 22 25 (table 1, figure 2).

Before adjusting for adherence, Leving et al27 found that errors in the specific inhalation steps ‘breathe in’ and ‘hold breath’, as well as the overall total number of errors made, were associated with higher COPD-related healthcare costs. However, when adjusted for adherence, no significant associations were observed between the number of errors and specific cost components.

Discussion

Recommendations by professional organisations, such as GOLD, Canadian Thoracic Society and Spanish Society of Pulmonology and Thoracic Surgery, on the management of COPD, have generally emphasised the need to provide education and training on inhaler device technique to patients and the need to assess a patient’s technique before modifying therapy.29–31 But in some guidelines, this consideration is not discussed,32 and in practice, it is frequently overlooked.33 A study of patients in nine countries (Brazil, Canada, France, Germany, Italy, Japan, the Netherlands, the UK and the USA) found that 29% had not had their inhaler technique checked by a healthcare professional within the last 2 years.34

Although it seems obvious that patients who do not use a device correctly will not get the full benefit of the drugs they contain, this systematic review found only weak and inconsistent evidence that making fewer mistakes using inhalers is associated with better clinical outcomes, including lung function, symptoms, health status and exacerbation rates in COPD. The prospective cohort studies did not assess changes in clinical outcomes in patients whose error rates were reduced, nor did they account for adherence.11 13 19 24 26 In the cross-sectional studies, it is impossible to separate cause and effect. For example, patients with poor health status or lung function may make more errors as they may be more breathless and unable to perform inhalation manoeuvres correctly or poor inhaler technique may have affected the benefits of therapy and led to poor health status and lung function.

Differences in the relative importance of specific errors may also confound the interpretation of the data. Critical errors have largely been defined empirically, on the assumption that they will affect the efficacy of the inhaled therapy. For some, such as failing to remove a protective cap from the inhaler, the assumption seems valid, but the clinical significance of others, such as not sitting or standing straight with head tilted, are more open to question. In some instances, there is evidence which shows that incorrect use impacts the effectiveness of the inhaled therapy. For example, the bioavailability of salbutamol in the lungs is greater after slow inhalation (10 l/min) from a pMDI compared with fast inhalation (50 l/min).35 In other cases, there is evidence that casts doubt on their importance. For example, studies using the clinical bronchodilator response to inhaled salbutamol as an endpoint found that there was no significant difference in the bronchodilatation induced by a 10 s as compared with a 4 s or no breath hold.36 37

Most studies identified in our systematic review considered errors collectively and only the PIFotal COPD study reported the relationship between specific mistakes using a DPI and outcomes.23 27 28 The study found that only some errors were related to outcomes but there were inconsistencies. Mistakes in the technique used to breathe in were associated with significantly worse health status, higher secondary healthcare costs and higher total COPD-related healthcare costs.23 27 28 They were also associated with an increased rate of severe, but not moderate, exacerbations, suggesting that the finding may be confounded and more likely to be due to patients with more severe diseases having problems using inhalers correctly. There was no statistically significant relationship between other errors and worse outcomes.23 27 28 Of note, failure to ‘remove protective cap’, was not associated with significantly worse health status or increased exacerbation rates,28 calling into question the relevance of the overall findings of the study and the definition of critical errors.

One of the main strengths of our review is that it examined the relationship between inhaler technique and outcomes in patients with COPD and did not include patients with asthma. It also included recent studies not considered in previous reviews. There are limitations to the review. The main ones relate to the quality of the studies included. There was heterogeneity in the study designs and short follow-up in most of the prospective cohort studies. Exacerbation rates were mostly based on patient recall and it was not clear how patients were selected for inclusion. Most studies failed to take adherence into account and lacked sufficient detail to enable quantitative synthesis. A narrative review was the only form of synthesis possible, but an individual participant data meta-analysis would have offered the best synthesis had the data been available.

Conclusions

The lack of evidence found in this narrative review does not mean that teaching patients how to use their inhalers correctly and regularly checking their technique is unimportant. Better prospective studies with longer follow-up examining clinical outcomes in patients who make fewer errors after training are needed. Ideally, such studies should include objective monitoring of inhalation technique and adherence using digital inhalers, as well as examining the effect of specific errors.