Asthma is a chronic condition [17,18]. The prevalence of asthma is significant; worldwide, the number of people suffering from asthma is approximately 300 million [17]. According to a Cochrane review, one of the most efficient tools that patients can use to manage their asthma is a written AAP [19]. The AAP is a set of instructions prepared with a health care professional that helps recognize signs that asthma is worsening, indicates which medication to use, or provides nonmedication strategies to keep asthma under control [20]. Written AAPs are individualized documents that must be updated (eg, once a year) by a clinician to match the evolution of the individual’s asthma condition [20].

When properly used, written AAPs are associated with fewer visits to the emergency department with an asthma exacerbation, fewer hospital admissions, better lung function, and an overall improvement of asthma symptoms [19]. Although having an up-to-date AAP is highly recommended, written AAPs are widely underused among adults; only 1 in 5 patients actually possess an up-to-date and usable AAP [21,22]. Most initiatives to improve the uptake of the written AAP have been targeted at health care professionals. Few have targeted patients.

This randomized controlled trial (RCT) is designed to test whether a PCHMS, tailored to help adults with asthma, would increase their rate of obtaining or updating a written AAP from a health care professional and whether this would lead to an improvement in asthma control.

Compared to participants allocated to the control group (ie, static online educational page), we hypothesized that those using a PCHMS are

In this parallel 2-arm RCT, participants were stratified by gender and level of asthma severity (intermittent vs persistent), and randomized 1:1 to have immediate access to the PCHMS or to control.

Participant recruitment took place between April and August 2013. All individuals who expressed an interest were assessed with a 5-minute online eligibility questionnaire. Eligible individuals were then invited to complete a 10- to 15-minute prestudy questionnaire. Participants in both arms continued to receive usual care from their health services and were surveyed monthly for asthma symptoms, asthma exacerbation, asthma control, and other competing priorities, and followed up with a 10- to 15-minute poststudy questionnaire between April and June 2014.

Eligible participants were adults (aged 18 years and older) living in Australia diagnosed with asthma, who had at least monthly access to the Internet and email, and had sufficient English language skills.

On completion of the prestudy questionnaire, participants who had been randomly allocated to the control arm were redirected to a static webpage with links to patient websites (eg, the Asthma Foundation, HealthInsite, myDR) that provided educational information on asthma. They were advised that they would be contacted to complete monthly surveys to elicit their asthma status during the study and would receive a follow-up questionnaire on conclusion of the study.

Full details of the Healthy.me Web-based PCHMS are described elsewhere [11,12,23]. During the study, Healthy.me provided participants with evidence-based information about asthma, the importance of a written AAP, and ways of obtaining a plan from a health care professional. Additionally, participants received monthly email reminders about the various interactive features of Healthy.me (eg, forum, poll, PHR).

An expert steering group was formed to tailor educational content for patients with asthma and to customize the interactive features of Healthy.me to deliver this content over the 12 months. Three asthma “journeys” were developed, providing evidence-based material to consumers about the written AAP. A usability study with 10 individuals was conducted and all major usability issues associated with the content and the intervention were addressed before commencing the study.

A review of online interventions found that those built on a theoretical framework demonstrated greater efficacy [24]. The Health Belief Model (HBM) [25], a prominent model of behavioral change, was used to guide the design of the 3 asthma journeys. More details are available in Multimedia Appendix 1.

There are strong theoretical reasons why the PCHMS features drive behavioral change:

All primary, secondary, and ancillary analyses are outlined in the published protocol [23]. Sample size calculation and expected effect size are also documented in the protocol [23]. No major changes from the protocol were introduced during study execution. Statistical significance was defined as a P value of less than .05 (2-sided test). Effect sizes are reported with 95% confidence intervals. Data were analyzed using SPSS version 20.

Recruitment was conducted over a period of 5 months between April and August 2013, during which 485 participants were assessed for eligibility (Figure 1). Recruitment was complete in August 2013 and follow-up conducted between April and June 2014. In all, 330 participants were assessed eligible and randomized (intervention: n=154; control: n=176). No participants with available data were excluded from the analyses. No harm or unintended effects were reported by participants during the study.

Baseline characteristics were similar for all allocated participants, participants lost to follow-up, and remaining participants (Table 1). The majority of participants were female (control: 79.5%, 140/176; intervention: 80.5%, 124/154), in their late thirties / early forties (control: mean 39, SD 13 years; intervention: mean 40, SD 14 years), and were very familiar with the Internet and social networking sites such as Facebook and Twitter. Asthma-related characteristics, such as smoking status, use of asthma medication, and contact with health care professional for asthma in the past 12 months, were similar to rates identified in a national survey of adults with asthma [22]. However, their rate of possessing a written AAP was higher than the national rate (this study: 38.8%, 128/330 vs Australian Centre for Asthma Monitoring: 21.3% [22]). For those with a written AAP at prestudy, 75.8% (97/128) could not recall when they last obtained or updated it.

Analyses of the primary outcome (possession of a written AAP at poststudy) was conducted by intention-to-treat using the data of all 330 allocated participants and the 153 participants who completed the poststudy questionnaire.

We did not apply the intention-to-treat principle to secondary and ancillary outcomes due to the availability of data for analyses. Analyses of secondary outcomes relating to use of the AAP and visits to a health care professional were conducted using the data of 153 participants who completed the poststudy questionnaire. Other study outcomes (ie, asthma exacerbation, asthma control, worsening of asthma, and loss days from work or study) were conducted using the data of 242 participants who completed at least one monthly questionnaire.

Analysis of the primary outcome is outlined in Table 2 (more details in Multimedia Appendix 2). There were no significant differences in the proportion of participants who reported having a written AAP poststudy between the intervention and control groups (all participants: χ² ₁=0.6, P=.43; all participants with imputation method: χ² ₁=0.4, P=.52; remaining participants at study end: χ² ₁=0.9, P=.36).

Binary logistic regression was adjusted for differences in baseline characteristics and potential confounding factors that might influence the primary outcome measure. Only one independent variable made a statistically significant contribution to the regression model: possession of written AAP at prestudy (χ² ₂₂=299.6, P<.001). Allocation to the intervention group (OR 0.43, 95% CI 0.15-1.23) did not contribute a significant effect to the proportion of participants possessing a written AAP at poststudy.

Analyses of secondary outcomes are presented in Table 3 (more details in Multimedia Appendix 2). There were no statistically significant differences in the proportion of participants between the control and intervention groups for any secondary outcome.

Access to the Healthy.me PCHMS did not improve the rate of possession of written AAPs, planned visits to health services for nonemergency asthma management, asthma status, control, or work and study productivity. These results are in stark contrast to earlier trials of the same PCHMS, which showed significant improvements in outcomes associated with consumer behavior change, including influenza vaccination [7] and sexually transmitted infection screening [8]. The negative results also occur in a context where intervention and study designs were in accord with factors typically associated with successful uptake and efficacy of online consumer interventions.

High attrition rates are common in eHealth intervention studies [14], with a recent systematic review revealing that completion of protocol rates for depression sites ranged from 43% to 99% [34]. This study suffered from moderate to high rates of attrition in the intervention (64%) and control (45%) groups. A previous study of Healthy.me did not experience this degree of attrition [7]. However, that study was conducted in 2009, its duration was shorter, and the mean participant age was 26.2 years compared to 40 years among those allocated to the intervention in this RCT [7].

In decision theoretic terms, the expected utility of any eHealth intervention is a product of the utility or benefit of each individual interaction with the system to the user and the number of times the interaction takes place [35]. Systems that engage their users and, as a result, are used frequently are theoretically more likely to deliver benefit. This is reflected in the research evidence, where consumer eHealth systems seem to demonstrate a clear dose-response relationship between use and benefit [35].

In this trial, some participants suggested that they saw little benefit in using the system, either because they or their health professionals saw little value in having an AAP, because asthma management was not a major priority in their life compared to other competing priorities, or that they have already developed their own strategies to manage the condition and needed no further assistance. Perhaps as suggested by some participants, the intervention would be more helpful for those who are newly diagnosed with asthma.

A systematic review of PHRs used for chronic conditions found that unless a system clearly assisted consumers in self-management tasks, they were unlikely to be successful [36]. This benefit might come from tracking important parameters to control an illness, such as blood pressure or glucose levels, or by delivering feedback when changes to management are needed. That review identified diabetes, hypertension, asthma, HIV, fertility management, glaucoma, and hyperlipidemia as having the most evidence for PHR benefit. However, only one study in that review actually included asthma patients and these were grouped together with other patients who had diabetes or hypertension [36]. The only outcome measure was patient activation, and asthma patients represented only 7% of the sample, providing weak evidence of PHR benefit in asthma management outcomes.

Recent systematic reviews concluded that although there is evidence that some digital interventions are associated with positive asthma self-management outcomes [37,38], most interventions do not use behavioral change theory, clinical guidelines, and/or assessment tools to inform their design [37]. A Cochrane review on smartphone apps for asthma concluded there is currently lack of evidence to advise clinical practitioners, policy makers, and the general public on ways to implement these interventions for asthma self-management programs [39]. Relevant to the AAP, a theoretical model has proposed 4 elements that are essential in facilitating the “right” contexts between patients and professionals, but few studies have used all these elements in their implementation [40].

Our own earlier trials of this intervention focused on supporting preventive health tasks. A trial aimed at encouraging influenza vaccination demonstrated a significant doubling in vaccination rates, most likely because the system allowed easy and immediate access to booking a vaccination with a primary care center, for a condition where seasonality and acting in a timely manner is important [7]. Similar benefits were demonstrated when the system was targeted at increasing screening rates among young adults for sexually transmitted infections, where use of an online booking system may have additionally reduced any sense of stigma associated with making a decision to act [8].

Although the lack of uptake of eHealth interventions is a widely known phenomenon [13-15], the literature on negative findings in this field is still scarce. Our study provides a number of lessons:

Study strengths include nationwide recruitment, use of recommended care indicators for outcome measures, and triangulation of participant feedback with quantitative results.

Notable limitations of this study include the gender and age distribution of participants, the attrition rate, and the use of self-reported data. The majority of participants were female in their late thirties / early forties and it is possible this population sample behaved differently than a more representative sample.

Participants had a higher rate of AAP possession than reported in other studies. As a result, as a cohort, they may already be better engaged and confident in their self-management and less likely to benefit from the intervention compared to the population average, reducing the potential effect size. Further, because the outcome measure was focused on having an up-to-date written AAP that was updated by a clinician (eg, once a year), we may have missed some participants as the study duration was only 12 months. Future studies should consider extending the trial period to more than 12 months.

Our primary recruitment strategy is online, which has a number of limitations, such as high rates of attrition. More effective recruitment could potentially result when it is channeled through influencers such as health care providers or with the encouragement of caregivers who help patients to deal with issues every day. However, this is an intervention designed primarily for consumers, to be delivered online, thus it is important that there is a direct channel to recruit consumers who are already online.

Consumers are increasingly turning to the Internet and social media for health advice, yet we still do not fully understand why some online interventions work and others do not. In this study, participant goals were poorly aligned with the clinical goals of the system despite there being clear evidence underpinning these latter clinical goals. It may be that a different approach is required in the domain of asthma management in adults, at least as far as AAPs are concerned, focusing not so much on task support as on belief change. More generally, researchers should not feel discouraged to publish negative findings because in failure many significant lessons can be learned.