Do spirometry and regular follow-up improve health outcomes in general practice patients with asthma or COPD? A cluster randomised controlled trial

Most guidelines for asthma1,2 and chronic obstructive pulmonary disease (COPD)3,4 stress the importance of spirometry for diagnosis and evaluation of management options. A review of COPD has also recommended that forced expiratory volume in 1 second (FEV1) after administration of a bronchodilator be measured repeatedly over the course of the disease to define the rate of decline in lung function,5 to help focus both treatment decisions and discussions regarding prognosis. This implies that spirometry should be used on a routine basis, to identify patients with a rapid decline in FEV1 and to improve quality of care.

Few long-term studies have evaluated the benefit of regular spirometry in the management of COPD and asthma in general practice. The Detection, Intervention and Monitoring of COPD and Asthma program found that it was possible to detect COPD and asthma at an early stage.6 However, early initiation of inhaled corticosteroids did not demonstrate any effect on the rate of FEV1 decline.7

Victoria has the lowest rate of Medicare claims of any Australian state for spirometry and complex lung function tests, particularly office-based tests,8 despite having a similar prevalence of asthma9 and COPD10 to other states. This presents a unique opportunity to investigate the role of spirometry in improving the management of chronic respiratory diseases. We therefore conducted a randomised controlled trial (RCT) of spirometry with regular medical review as an intervention for managing asthma and COPD in a general practice setting. Our aim was to determine whether this intervention resulted in improvements in quality of life, increased written asthma action plans, or reductions in respiratory symptoms, days lost from usual activities, emergency presentations or hospital admissions, compared with usual care.

Group allocation was concealed by the study statistician (R W) from other research staff and general practices until after the practices had recruited their first participant.

Of 48 practices that agreed to participate, 31 (with 75 full-time and 49 part-time GPs) proceeded to randomisation and recruited 305 patients (Box 1). The other 17 practices (25 full-time and 28 part-time GPs) dropped out before randomisation, due to the principal GP changing his or her mind, and did not recruit any patients.

This RCT did not find any significant improvement in quality of life associated with spirometry and regular medical review among general practice patients with asthma and/or COPD. There was a brief spike in PCS score among participants who received spirometry alone without reports being sent to their GPs. This might have been the result of their interaction with the respiratory scientist, but more likely represents a chance finding in the context of multiple comparisons. Nor was there any consistent improvement in secondary outcomes, such as reduced respiratory symptoms, asthma attacks, emergency presentations or hospital admissions, or increases in written asthma action plans.

The lack of positive findings might reflect the insensitivity of a generic quality-of-life measure such as the SF-36. We chose this well validated and widely used instrument rather than a disease-specific questionnaire because we did not know in advance what proportions of patients would have asthma or COPD. We could instead have used a combination of instruments, such as the Asthma Quality of Life Questionnaire and the St George’s Respiratory Questionnaire, but this would have increased responder burden. Nonetheless, we have previously been able, using the SF-36, to detect effects of respiratory symptoms such as wheezing on the quality of life of young adults.20 Therefore, we do not think that the SF-36 was too insensitive to detect a clinically important change in quality of life, had one occurred.

Practices that participated in the trial were likely to be those most interested in the management of chronic respiratory diseases and might not be representative of Australian general practices. We originally anticipated that half the participants would be children and eventually recruited 305 adults. However, even with 253 adults completing the trial, we still had greater than 80% power to detect a moderate effect of the intervention on quality of life. As shown in Box 4, the 95% confidence intervals around observed differences between groups in change from baseline all excluded + 4.6 units, the planned effect size on the PCS. Nonetheless, power to detect different changes between groups in health service use was limited because of small numbers of events (asthma attacks, presentations or hospital admissions) during the trial. More events would have occurred with a longer duration of follow-up.

The numbers of participants were slightly unbalanced between groups at baseline, because practices randomly assigned to spirometry alone tended to recruit more patients. Losses to follow-up were greater than expected, with 52 patients (17%) not completing the trial. Loss was greater in Groups A and B and could be due to the burden of spirometry, but this is unlikely to explain the negative results. Further analysis based on multiply imputed SF-36 outcomes for people who were lost to follow-up found similar results to those presented in Box 4 (data not shown). There were some protocol violations — 10 patients allocated to Group A and eight in Group B never had spirometry performed. Conversely, at 12-month follow-up, 12 patients in Group C reported having spirometry outside the trial. Nine of these were from three practices, suggesting some heterogeneity of “usual care” between practices. While the intention-to-treat analysis was inherently conservative, it was also the least prone to bias.21

It is quite likely that the trial was adversely affected by co-interventions. During the period that practices were being recruited and patients followed, the National Asthma Council Australia launched a major initiative to promote spirometry to GPs and practice nurses.22 However, these spirometry training workshops were initially directed towards rural areas and only five were conducted in Divisions of General Practice participating in our trial. Furthermore, we were trialling a different model, in which a trained respiratory scientist performed spirometry in the general practice according to ATS–ERS standards, and a specialist report was provided later by a qualified respiratory physician.

Another explanation for the negative results might be a failure by GPs to act on spirometry reports and change patient management. However, around half the patients in Groups A and B had moderate–severe airflow limitation (data not shown), so it is unlikely that patients not considered to require more intensive treatment because they had mild airflow limitation would account for the negative results. Most general practices have well established processes for responding to abnormal pathology and radiology reports and recalling patients for further consultation and referral to specialists as required.23,24 However, this might not extend to the occasional faxed spirometry report. We are currently investigating what occurred in a sample of the consultations by conducting focus groups with patients and in-depth interviews with practice staff.

Regular medical review is only one component of the management of chronic respiratory diseases. For example, pulmonary rehabilitation has been shown to improve quality of life for patients with COPD.25 Spirometry by itself would not be expected to increase referrals for pulmonary rehabilitation. Unfortunately, we did not collect data on patients’ use of outpatient rehabilitation services during this trial.

There are few other studies with which our trial can be directly compared. An RCT in Italian general practices did not find that office-based spirometry improved the accuracy of diagnosis of asthma or COPD.26 However, the trial randomised by patient (leading to frequent protocol violations), did not include administration of a bronchodilator, and did not report patient outcomes. Conversely, a before-and-after study of spirometry in American family medicine practices found that there were changes in the management of 48% of patients with asthma or COPD with completed tests.27 However, due to the lack of a control group, we cannot be confident that these changes resulted from spirometry.

Recently, negative results have been reported from an RCT of spirometry conducted in South Australian and Tasmanian general practices.28 The trial did not find any improvement in quality of life, days off work or school, asthma exacerbations, or daytime or nocturnal symptoms in either adults or children with asthma whose management was guided by spirometry. However, the trial’s intervention model was different from ours — GPs and practice nurses were given training in spirometry so they could perform it themselves in their practices. The standards and interpretation of spirometry were also less certain compared with our study.

In conclusion, regular 3-monthly spirometry and medical review made little difference to quality of life, respiratory symptoms, asthma attacks, written asthma action plans or health care service use in adult patients with asthma and/or COPD. We would thus have considerable difficulty recommending more widespread use of spirometry in Australian general practices to guide the management of such patients. However, there is still likely to be a role for spirometry as a “gold standard” to establish the diagnosis when patients first present.