The study was conducted and reported according to the Consolidated Standards of Reporting Trials (CONSORT) statement for cRCTs [39]. This study had a 2-arm cRCT design with the HA dispensing practice (henceforth: practice) as the unit of allocation. Cluster randomization was preferred over individual randomization to minimize the risk of contamination. The sample size calculation had indicated that 70 practices should participate in the study. Participating practices were prestratified by level of urbanization (located either in relatively rural or urban areas). Within both strata, a statistician randomized practices to either the intervention or control arm. The randomization sequence was generated by the statistical software R (R Foundation on Statistical Computing) with random permutation in blocks of size 4 with a fixed seed. Thirty-four practices were allocated to the control arm and 36 to the intervention arm.

Participants, HA dispensers (henceforth: dispensers), and researchers were aware of the practices’ and participants’ trial arm allocation. Participants and dispensers could not be blinded due to the nature of the intervention. Researchers could not be blinded because they administered all questionnaires to the participants, including the International Outcome Inventory (IOI), which revealed trial arm allocation. Additionally, researchers actively monitored the uptake of SUPR for the purposes of the process evaluation (submitted), also revealing arm allocation. The study was approved by the Dutch Institutional Review Board of the VU University Medical Center Amsterdam (IRB00002991; FWA number: FWA00017598) and registered [ISRCTN77340339].

Participants were recruited between February and September 2016 by the dispensers or their supporting staff of the participating practices. They informed each potentially eligible client about the study when clients were about to enter an HA trial period. Interested clients received an information package. The package contained information on the general study aims and group allocation (SUPR or control). No details about the SUPR program other than that it is a support program aimed at improving communication was provided at that point to ensure that clients would not ask or seek information about SUPR (thereby preventing contamination of the control group). Clients could sign up for the study via an online webpage and provided their consent there.

Eligible participants were Dutch speaking adults aged 50 years or over who had decided to purchase one or two HAs, had access to a device with internet connection, and were owner of an email account for the total duration of the study. These purchased HAs were the first ones for the first-time HA users and replacement HAs for the experienced users. We excluded participants who received additional care via an audiology clinic because this type of care might interfere with that provided in SUPR. We also excluded participants who purchased HAs to suppress tinnitus because tinnitus management was not included in SUPR.

All outcomes were collected via online questionnaires at baseline (t0: before the HA trial period), immediately postintervention (t1: 6 months after the HA purchase), 6 months postintervention (t2), and 12 months postintervention (t3). The outcomes were measured at all time points except for the ones that did not apply at t0 (ie, HA use—participants had not yet obtained their HAs).

Figure 2 displays the flow of first-time and experienced clients through the study. Between February and September 2016, 1107 (739 first-time and 368 experienced) clients in the intervention arm and 1169 (809 first time and 360 experienced) clients in the control arm were invited to participate in the study. The number of invited clients is an estimation based on the reported numbers of invited clients of the practices complying with the protocol to report this.

In total, 248 clients (across 35 practices in the SUPR arm) and 252 clients (34 practices in the control arm) enrolled for the study and consented to participate. Of these, 180 and 163 clients were included in the ITT analysis, respectively. Loss to follow-up was 2.8% (5/180) in the SUPR arm and 8.0% in the control arm (13/163). Sixty-three clients in the SUPR arm and 161 controls were included in the per protocol analysis.

Table 2 presents the baseline characteristics of the participants. The number of participants in each practice ranged from 1 to 11 (mean 5). Of the total sample, 60% (206/343) were male (mean age 68.1 [SD 8.5] years), and mean pure-tone hearing loss was 43.7 (SD 11.1) decibels Hearing Level. The characteristics for SUPR and control participants were similar, as were the outcomes at baseline (P>.05), indicating that correction in the analyses due to significant group differences was not required.

Using a cRCT design, we evaluated the effectiveness of SUPR, a web-based self-management support program provided as an addition to regular HA fitting to HA users aged 50 years and over in order to improve the self-management of hearing difficulties and HA use. The study showed that SUPR did not lead to more frequent use of communication strategies (primary outcome) compared with care as usual. Nonetheless, SUPR significantly improved clients’ self-efficacy for advanced HA handling and HA satisfaction at 12 months, as well as HA use immediately postintervention. No group differences were observed for any of the remaining secondary outcomes.

It is encouraging to see that SUPR was able to significantly increase HA use immediately postintervention. Although the effect seems small (mean difference of 0.3 on a scale from 1 to 5), the fact that SUPR was able to improve HA use can be considered valuable because a recently published systematic review indicated that there is no evidence of interventions showing any improvements on HA use on the short, medium, or long term [10]. The evidence was judged as limited because the majority of previous studies had nonpowered small sample sizes (limiting the occurrence of statistically significant differences) or were carried out in nonrepresentative samples (ie, military veteran populations). The SUPR study can thus be regarded as an important addition to the existing body of evidence. Moreover, the per protocol analysis revealed a significant positive effect of SUPR on HA use stability in the long term, indicating a more stable HA use pattern (ie, the same number of hours of HA use every day) among SUPR participants than among controls. The absence of a long-term improvement in HA use may imply that without support like SUPR, HA clients may tend to fall back and be left to full self-management. This underlines the importance of having follow-up support after clients have completed SUPR to increase the likelihood of extending the effects to the long term.

It has been argued that more frequent or more stable HA use does not automatically imply more satisfaction with HAs [56] and that HA use (in hours) alone cannot be viewed as an indicator of successful HA use [57]. From this perspective, it is interesting to see that we not only found increased HA use in SUPR recipients, but also significantly greater HA satisfaction than the controls postintervention. This shows that the increase in hours of use indeed coincided with greater HA satisfaction.

This study demonstrated that the SUPR group had a significantly higher score of 5.3 points (95% CI 0.3 to 10.4; scored on a scale from 0 to 100) on the advanced HA handling self-efficacy scale directly postintervention than controls that lasted up to 12 months later. Both groups seemed to perform at ceiling on the basic HA handling self-efficacy scale, suggesting that during appointments dispensers already ensured clients became skilled in basic HA handling (ie, HA and battery insertion and removal, HA cleaning and maintenance). This was also stated by Ferguson et al [32], who found similar ceiling effects. Although the instruction videos mostly focused on basic rather than advanced HA handling skills, watching them might have increased clients’ confidence in their ability to also handle more advanced skills, like troubleshooting or naming the model of a particular HA. We are uncertain about the clinical meaning of a difference of 5.3 points, however. Further research should address this.

Bennett et al [58] found that most problems HA owners experienced were related to HA management including HA use, handling, and ongoing care and that these had the greatest impact on HA success. SUPR improving HA clients’ confidence in their ability to manage their HAs can thus be considered an encouraging finding. Other work by Bennett et al [59] showed that HA management can be classified according to two overarching themes: the device and the person. The results of our study suggest that SUPR is successful in improving HA management concepts related to the device (HA maintenance and repairs, daily HA use, advanced HA knowledge) but not for HA management related to the person (learning to come to terms with HAs, communication strategies, working with a clinician). This is further discussed below.

Previous studies have shown that both individual and group auditory rehabilitation interventions can effectively increase communication strategy use [22,31,33,60] both in the short- and long term. A plausible explanation for the absence of an effect in this study may be related to the video-watching rates as reported in our process evaluation study (submitted). Whereas up to 37% of the participants in the intervention arm had watched the instruction videos, only 7% to 16% had watched the videos on communication strategies and personal adjustment. Not engaging with web-based interventions as was intended is a well-known problem in intervention effectiveness research [61,62]. The per protocol analysis still showed no effects on communication strategy use, but it must be noted that the sensitivity analysis was most likely underpowered (SUPR n=60, care as usual n=151). We intended to perform the per protocol analysis with clients who had clicked on more than two video links, but this resulted in samples too small to allow a meaningful statistical analysis.

An alternative explanation for the absence of a communication strategy use or personal adjustment effect may be related to the setting in which SUPR was provided. Given that the HA dispensing setting is a primary care one, mainly focusing on HA fitting and dispensing, clients may not have expected or been ready for receiving educational videos. Such type of rehabilitation may better fit in specialized (ie secondary or tertiary) hearing health care settings. Kramer et al [33] chose a tertiary setting to provide their home education program, resulting in improved communication strategy skills.

Control group participants reported significantly greater benefit, satisfaction, and quality of life because of their HAs (IOI-HA) than SUPR participants reported for SUPR (IOI-AI). This suggests that HAs were viewed as more impactful than SUPR. In a way, this is not surprising, since HAs can be considered as a basis. They amplify sounds and may thus improve listening ability and thereby quality of life [63]. It is the combination of the two (HAs and additional support) that may be most beneficial [10]. Also, it is important to consider that HAs have a long history (more than a century) of development while the development of (web-based) support programs is still in its infancy. There is much to learn still to further refine communication programs to ensure they fit clients’ needs and have a larger impact.

The URICA questionnaire outcomes conflicted with what was expected both in the direction of the effect and changes over time and were therefore difficult to interpret. We expected the SUPR group to show an increase in the action scores, while in fact the follow-up scores were lower than baseline in both groups. Differences in interpretation of what taking action meant to someone may have occurred between participants but also within participants over time (as the intervention might have influenced what taking action was) causing invalid measurements. This is further discussed in Meijerink et al [40].

This was the first audiological rehabilitation study implementing and evaluating a web-based self-management support program on such a large scale and in a real-life HA dispensing setting. The large sample size, use of a robust RCT design, and outcome assessment at the short, medium, and long term can be considered unique in our research field. Given these strengths, the study tackles most of the limitations mentioned in a recent meta-analysis on intervention studies to improve HA use [10]. Including 70 clusters across the Netherlands and purposefully sampling for spread in degree of rural/urban areas minimized imbalance across treatment groups and increased the generalizability of the findings to the Dutch real-world practice. Nevertheless, there are a number of limitations.

First, while we did reach our targeted sample size for the first-time clients, we did not for the experienced clients due to recruitment difficulties. The limited number of experienced clients may have resulted in nonsignificant differences in effects between first-time and experienced clients. A second limitation is that clients had to provide consent for study participation while knowing their group allocation. This may have affected clients’ willingness to participate in the study and introduced selection bias. Unfortunately, it was not possible to prevent this because randomizing the practices after obtaining clients’ consent would have delayed the intervention period by 7 months until clients had provided their consent. This was deemed unacceptable given the real-life character of the study. A third limitation was that participating clients, dispensers and researchers could not be blinded. This may have introduced bias. Possibly, clients who were aware of receiving SUPR may have responded more favorably compared to controls, while controls being aware of receiving standard care only may have sought alternative treatments (which would have increased the likelihood of contamination). We attempted, however, to prevent controls from seeking access to SUPR by reducing the amount of information given about SUPR content and by offering controls SUPR after study completion. A fourth limitation is that we initially aimed to measure HA use using data logging in order to measure HA use objectively, but this appeared unfeasible. Hence, all outcomes were self-reported, possibly resulting in overreporting of HA use [18,48]. Finally, behavior change was expected at multiple levels, and therefore many outcomes were evaluated. This increased the likelihood of finding statistically significant results by chance. We therefore applied Bonferroni corrections for the primary outcomes and the post hoc analyses but not for the secondary outcomes. It should be noted that there is debate among statisticians as to when multiple outcomes should be corrected for [64,65]. Using a Bonferroni correction for all outcomes is concerned too strict by some as it would increase the chance for false negatives [66].

While the popularity of web-based platforms to complement HA fitting is rising [30-32], high-quality evidence (ie, assuring external validity and power) to show the long-term benefits of eHealth in HA rehabilitation is still lacking [37]. This study is a valuable addition to the existing evidence for such platforms in hearing health care. While SUPR did not significantly enhance the use of communication strategies, this study provides evidence for the effectiveness of SUPR to improve self-efficacy for HA handling and HA satisfaction in the long term and HA use in the short term. Given that the effects were mainly found in the HA handling domain, this study indicates that an intervention offering web-based instructions is a promising addition to the services provided by dispensers. Further research is needed to evaluate if adjustments to SUPR will lead to a higher adherence of clients in following the intervention to improve the (long-term) effectiveness of communication strategy use and other psychosocial outcomes.