With one notable exception, there were no statistically significant differences in outcomes among groups. Cowie et al16 reported an 11-fold decrease in total emergency room visits for the group using a peak-flow action plan (5 vs 55, P = .02), and also reported a reduction in hospitalizations of a similar magnitude (2 vs 12) that did not reach statistical significance. However, this study suffers from notable flaws that diminish confidence in the results. It is a post-intervention comparison among groups, which does not compare change from baseline, or incorporate baseline values as covariates in the analysis. Moreover baseline utilization data were provided by patient recall and not corroborated by medical records. There was a substantially larger variability in the baseline utilization rates for the peak flow group compared with the control group. This suggests that a subset of very high frequency users may have been over-represented in the peak flow group, and the reduction in emergency room visits may be concentrated in this subset.
Peak-flow meter-based written action plan versus peak flow meter with no written action plan
Two studies18,19 addressed the independent effect of a written action plan when added to peak flow self-monitoring (Table 3). Charlton19 reported no significant group differences for main outcomes, while Ignacio-Garcia18 reported large and statistically significant differences in most of the outcomes, favoring the group that used the written action plan.
The Ignacio-Garcia study, however, suffers from notable flaws suggesting the results may be attributable to bias. The sole participating physician, not blinded to treatment assignment, was highly involved in all phases of patient assessment, monitoring, and treatment. There was evidence of baseline differences between the two groups. A total of 25% of patients were withdrawn after randomization, and an unexplained decline in lung function occurred in the control group. Thus, the potential for selection bias, withdrawal bias, and ascertainment bias limits confidence in the results of this study
Symptom-based written action plan compared with peak flow-based written action plan
In 4 studies,16,17,20,21 reported outcomes were generally equivalent between groups and comparisons were not statistically significant, with one exception (Table 3). The 3-arm study by Cowie et al16 reported a striking reduction in the total number of emergency room visits with a peak flow meter-based written action plan compared with a symptom-based written action plan (5 versus 45, P
Discussion
The objective of this systematic review was to assess the independent effects of 2 specific components commonly included in asthma self-management plans—a written action plan and a peak flow meter. Few studies, however, are designed to permit reviewers to isolate the effects of these components. Moreover, the studies we reviewed did not clearly identify the population expected to benefit from interventions or specify the primary outcomes of interest; nor was the level of clinically meaningful improvement prospectively defined.
Most of the trials we reviewed, including the largest community study of 569 patients, did not demonstrate improved outcomes. The 2 trials that reported statistically significant results favoring a peak flow-based written action plan suffer from notable flaws suggesting the results may be attributable to bias. In the other 7 trials, there was little difference in outcomes between groups. However, these studies had insufficient power to detect group differences or confidently conclude equivalence between groups.
Thus, available evidence is insufficient to demonstrate that asthma outcomes are improved by use of a written asthma action plan, with or without peak flow monitoring. While this body of literature does not establish that these interventions are ineffective, it suggests they will not have a large effect on outcomes when applied to the general asthmatic population. The application of written action plans to all asthmatics indiscriminately may be a wasteful use of resources. This systematic review also questions the validity of written action plans as an indicator of asthma quality of care, or as a means to achieve quality improvement.
This analysis also highlights several obstacles to assessing the effects of disease management interventions. First, while the impact of whole intervention programs can be evaluated in controlled trials, it may be unfeasible to isolate each component of such programs and subject it to a rigorous analysis. Furthermore, as a behavioral intervention, the general principle of engaging patients in self-management may be more important that the specific components of these programs. Finally, regarding the optimization of medications (most obviously initiation of inhaled steroids) the impact of written action plans is likely to be relatively small, particularly on lung function or symptom control.
Future clinical trials should be done selectively, aimed at producing rigorous results that can improve the effectiveness of self-management interventions. Further study is warranted for specific subpopulations, such as those with higher baseline severity of illness or those with high baseline utilization rates. Available data suggest that, if there is benefit to be gained from self-management interventions, it will most likely be seen among these patients. Specific components of self-management that might be tested individually are those that are relatively high-cost, resource intensive, or risky for the patient.