Efficacy of Communication Bridge‐2 for primary progressive aphasia: A randomized controlled trial of communication intervention

Abstract

INTRODUCTION

Primary progressive aphasia (PPA), a language‐based neurodegenerative dementia, negatively impacts communication and quality of life. Previous non‐pharmacologic interventions show promise but lack efficacy trials. Here, outcomes are provided from Communication Bridge‐2 (CB2), a speech‐language randomized controlled trial (RCT) for PPA.

METHODS

CB2 is the first Phase 2, Stage II, parallel‐group RCT delivered via video chat with global enrollment. Ninety‐five dyads were randomized into one of two speech‐language intervention arms. Primary outcomes included communication confidence and participation measures. Marginal linear models assessed efficacy across ≈12 months.

RESULTS

Ninety‐five dyads were randomized from four countries. Experimental arm superiority in communication‐participation measurement of goal attainment was demonstrated (66.7% vs 49.1%, respectively, p = 0.006), and corroborated by post‐study interviews.

DISCUSSION

Outcomes demonstrate the feasibility and initial efficacy of a person‐centered telemedicine intervention for maximizing communication participation for mild‐to‐moderate PPA, providing a pathway for developing and implementing clinically meaningful interventions for Alzheimer's disease and related dementias.

Highlights

• Primary progressive aphasia (PPA) negatively impacts communication participation.

• Communication Bridge‐2 (CB2) is a telemedicine‐delivered randomized controlled trial (RCT).

• Global recruitment of 95 PPA participant dyads into an RCT with low dropout.

• First international superiority trial for PPA using video chat shows efficacy.

• The study provides a model for rigorous non‐pharmacologic trials for Alzheimer's disease/Alzheimer's disease and related dementias (AD/ADRD).

1. BACKGROUND

Primary progressive aphasia (PPA) is a clinical neurogenerative dementia syndrome on the Alzheimer's disease (AD) and related dementias (ADRD) continuum. PPA is characterized by the insidious onset of language and communication impairments (aphasia), with relative sparing of other cognitive and behavioral abilities. ¹ , ² AD neuropathologic change or a form of frontotemporal lobar degeneration are the primary neurodegenerative diseases associated with PPA. ¹ , ² , ³ Progress has been made in the development and approval of disease‐modifying therapies (DMTs) for AD. Persons with PPA were rarely represented in these trials, with primary outcomes tending to be weighted toward memory rather than language symptoms. DMTs for other frontotemporal lobar degeneration, including non‐AD tauopathies and TAR DNA‐binding protein (TDP‐43), remain elusive but active areas of study, aided by emerging in vivo biomarkers and trial designs. ⁴ , ⁵

For individuals living with PPA, activities of daily living and quality of life are negatively impacted by the progressive loss of communication abilities. ⁶ , ⁷ , ⁸ , ⁹ Non‐pharmacological interventions for speech‐language and communication difficulties are the primary management option, with the number of intervention studies expanding in the last 10 years. ¹⁰ However, speech‐language and psychosocial interventions have lacked rigorous randomized controlled trials (RCTs), which limits the identification and subsequent implementation of evidence‐based care for persons with PPA. ¹⁰ , ¹¹

The Communication Bridge (CB) Research Program has been developing and testing interventions for persons with PPA and their care partners using participant‐informed designs and outcomes. The strategically and rigorously designed studies are guided by the National Institutes of Health (NIH) Stage Model and the Readiness Assessment of Pragmatic Trials (RAPT) model. ¹² , ¹³ The Communication Bridge 1 (CB1) pilot trial established the feasibility of delivering an intervention on a global scale using telehealth delivery and demonstrated gains in functional communication outcomes and maintenance 6 months post‐baseline. ¹⁴ , ¹⁵ This report describes the efficacy outcomes from the Communication Bridge‐2 (CB2) trial, a superiority RCT of speech‐language intervention for persons with PPA and their care partners. The primary outcomes are focused on measures of communication participation and confidence, which are closely linked to clinical meaningfulness in everyday life for participants and their families.

2. METHODS

2.1. Intervention, randomization, and study population

The CB2 clinical trial protocol and baseline cohort descriptions were published previously, including the rationale, clinical trial design, intervention, and assessment measures. ¹⁶ Briefly, CB2 is an international, single enrollment site, Phase 2, Stage 2, randomized, parallel‐group, active control, non‐pharmacologic clinical trial delivered over video chat (i.e., telehealth service delivery model) to individuals with PPA and their communication partners ¹⁷ (NCT03371706). The trial was designed to test whether the CB intervention is superior to the control arm intervention for improving (1) participation in everyday communication activities as measured by the Communicative Participation Item Bank (CPIB) ¹⁸ and personally relevant individualized communication participation goals using Goal Attainment Scaling (GAS) ¹⁹ , ²⁰ ; and (2) self‐reported communication confidence as measured by the Communication Confidence Rating Scale for Aphasia (CCRSA). ²¹ , ²² The CPIB and CCRSA are validated participant‐reported outcome measures that inquire about participation and confidence in everyday communication activities, respectively. GAS is a method for setting goals and measuring progress by assigning a standardized outcome scale across participants, and is suitable for use in neurodegenerative populations. ⁵ , ²³ Thus, the primary outcomes are centered on understanding the meaningful and clinically relevant impacts of the interventions.

Ninety‐five adult dyads, consisting of individuals with a clinical diagnosis of mild‐to‐moderate PPA and their communication partners, were enrolled (see refs. [16, 17] for detailed inclusion/exclusion criteria and enrollment data). Briefly, the clinical diagnosis of PPA was made by neurologists and supported by the available medical records. ¹ , ² , ²⁴ A communication partner was defined as an informal caregiver (typically a family member or friend) who knew the participant with PPA for more than 12 months, had close and regular contact, and provided emotional, communication, or activities of daily living support to the participant with PPA. Participants were required to self‐report English as their primary language, have adequate hearing and vision for communicating with others, and have the ability to read functional materials (newspapers, medication lists). In addition, participants were required to pass study‐specific technology and speech‐language therapy readiness screens outlined in the protocol. ¹⁷

The trial intentionally enrolled participants from each of three recognized research subtypes of PPA: logopenic (PPA‐L), agrammatic (PPA‐G), and semantic (PPA‐S), which are differentiated by relative strengths and impairments in word‐finding, grammar, and semantics. ¹⁶ Participants with PPA were assigned to the PPA‐L, PPA‐G, or PPA‐S groups aligned with previously published criteria, ¹ , ² , ²⁴ using screening and baseline assessment data as well as medical records provided by the participant.

The protocol publication describes the randomization plan in detail. ¹⁷ Briefly, a permuted block randomization schedule was stratified by two trial interventionists and subtypes (semantic, agrammatic, logopenic) with a 3:2 ratio of experimental to control.

RESEARCH IN CONTEXT

1. Systematic review: The authors reviewed the literature using traditional sources (e.g., PubMed). Previous studies suggest that non‐pharmacological interventions have the potential to positively impact communication and quality of life for persons with primary progressive aphasia (PPA) and their care partners. However, there has been a lack of efficacy trials. To help fill the efficacy gap, we provide outcome data for the Communication Bridge‐2 (CB2) randomized controlled trial (RCT).

2. Interpretation: Trial outcomes demonstrate feasibility and superiority for a communication‐participation‐based intervention relative to an active control. The results support the use of a person‐centered telemedicine intervention for maximizing communication participation for mild‐to‐moderate PPA.

3. Future directions: The CB2 trial provides methodologic and practical relevance for planning future large‐scale efficacy, effectiveness, and pragmatic trials across the Alzheimer's disease and related dementias (ADRD) spectrum.

Participant dyads were recruited through direct clinician referral, outreach events, mailings, flyers, specialized support groups, postings on relevant websites (e.g., Association for Frontotemporal Degeneration [AFTD] and clinicaltrials.gov), self‐referral, and earned media coverage. The interventions were delivered by speech‐language pathologists (SLPs) with clinical master's degrees and a Certificate of Clinical Competence issued by the American Speech‐Language‐Hearing Association (ASHA).

The protocols for the control and experimental arm interventions are described in the CB2 protocol ¹⁷ and summarized in Table 1. Briefly, the experimental arm intervention is a multi‐component intervention aligned with a participation‐focused aphasia intervention framework. ¹⁷ , ²⁵ The experimental arm intervention utilizes personalized training stimuli and educational materials, as well as a dyadic approach to care. Intervention sessions were goal driven and personalized to the dyad based on the GAS goals developed at the beginning of the intervention and included communication strategy education and training. ¹⁷ Script training and word‐retrieval training were optional in‐session components based on participant preference and whether the prescribed intervention approach was aligned with the participants’ GAS goals. The control arm intervention is most closely aligned with an impairment‐focused framework and utilized fixed training stimuli. The communication partner is a supporting agent and not a direct recipient of the intervention in the control arm. Impairment‐focused behavioral interventions for aphasia commonly target underlying disrupted cognitive processes (e.g., lexical activation) through structured, high‐repetition, intensive, and therapeutic activities. Such interventions propose to improve communication function by rehabilitating the underlying disrupted cognitive process and promoting optimal neural reorganization in the context of injury. Impairment‐focused interventions were the most common approach reported in a recent systematic review of PPA interventions, albeit with much smaller sample sizes (i.e., >80% of the studies had <5 participants). ¹⁰ The control arm intervention sessions focused on addressing underlying language impairments/processes and functional limitations through two components: (1) word‐retrieval training using a semantic–phonological cueing hierarchy ²⁶ and (2) script training, an intervention that capitalizes upon speech automaticity to generate larger units of scripted/scaffolded language exchanges in functional contexts. ²⁷ , ²⁸

TABLE 1.

Key intervention components for each arm.

Protocol component	Experimental arm	(Active) control arm
Theoretical premise	Dyadic, multi‐component, framed in participation‐based models of care	Impairment‐based, aligned with psycholinguistic framework
Intervention stimuli	Personalized	Fixed set
Communication partner role	Direct recipient	Supportive agent
Intervention session activities	Tailored to the participant based on their communication participation goals	Fixed; primarily impairment based
Asynchronous web application exercises	Four exercises; personalized content	Four exercises, fixed content
Dose	Equivalent; activities quantified at each session
Disease education	Provided in session and through the web application
Fidelity	Assessed in three ways: documentation, procedural, theoretical

An overview  of the visit schedule is highlighted in Figure 1A. Both arms receive two blocks of intervention (15 weekly sessions, lasting ≈60 min each) followed by pre‐ and post‐assessments and a final evaluation ≈12 months post‐enrollment. Thus, in addition to assessing superiority, this trial will provide the largest dataset of within‐arm outcomes for each intervention approach in PPA.

FIGURE 1.

Trial overview. (A) Overview of intervention and evaluation schedule. Participants completed two intervention blocks over ≈12 months, with evaluations (Evals) at baseline, after each intervention block, and at 12 months post‐enrollment. Sessions occurred approximately weekly during the blocks (teal lines) and were ≈60 min in duration. (B) Trial enrollment and completion flow chart.

The synchronous sessions were complemented by using the CB web application for both arms. The web application provided reminders/notes from the trial interventionists, educational videos/materials, calendar features for connecting to sessions, a self‐paced asynchronously delivered home exercise program, and a motivational tracking system for self‐monitoring completion of the web application activities. Participants in both arms were encouraged to complete asynchronous home practice exercises on the CB web application 30 min per day, 5 days per week. The four exercises included (1) Picture Cards, a naming (word‐retrieval) exercise; (2) Pronunciation Cards, a word pronunciation exercise; (3) a Word to Picture Matching exercise; and (4) a Script Practice exercise.

2.2. Ethics approval

Informed consent was obtained from all participants and their communication partners. The Northwestern University and University of Chicago institutional review boards (IRBs) approved the trial. Consent procedures followed local IRB committee standards. Enrollment began in May 2018 and ended in April 2022. The last participant visit was in March 2023.

2.3. Statistical considerations

Planned statistical analyses were originally described in the protocol publication and on clinicaltrials.gov (NCT03371706). ¹⁷ In the original power calculation, the target sample size was 90 participant dyads (n = 36 control, 54 experimental). To account for a 5% discontinuation rate over the follow‐up period, the sample size used in the power calculation was 85 participant dyads, with 34 in the control arm and 51 in the experimental arm. There was 80% power at a two‐tailed alpha = .05 to detect an effect size of 0.628 standard deviation (SD). The power calculation was re‐evaluated as part of the interim analysis. The final sample size was 95 participant dyads, 39 in the control arm and 56 in the experimental arm. Key outcomes are summarized here. Data were analyzed according to the intent‐to‐treat principle, where participants were analyzed by arm of randomization, and all randomized participants were accounted for in the analysis.

The primary outcomes included scores from the CCRSA, CPIB, and GAS. CCRSA and CPIB are patient‐reported outcome measures of daily activities (e.g., talking to people you know/do not know). Participants in both arms develop personalized GAS goals that address communication participation through collaboration between participants and their interventionist. All GAS goals are centered on communication‐related activities addressable by speech‐language intervention. Four GAS goals were developed for each participant dyad: three goals targeted communication from the perspective of the person with PPA (primary outcome) and one goal targeted communication from the communication partner's perspective. For each GAS goal, the current level of function/performance is anchored at 0, with +3 and −3 being the highest and lowest possible ratings. GAS goals were set before randomization to minimize bias between arms. GAS ratings were obtained from an interview with the dyad conducted by a blinded, non‐treating clinician. Baseline effect sizes were calculated using a standardized mean difference or Cramer's V.

Analyses examine efficacy (i.e., trial superiority) after each intervention Block and at a 12‐month evaluation post‐enrollment to provide insight into gains and potential maintenance of the intervention over time. In the absence of intervention, functional and language decline is anticipated over a year for PPA, since it is a neurodegenerative condition. ⁷ , ²⁹ , ³⁰ , ³¹ , ³² Thus, improvements, no change from baseline, or slower than expected change could be considered positive intervention outcomes. Intervention efficacy was analyzed using a marginal linear model, considering repeated measures for means of CCRSA and CPIB using SAS (9.4, SAS System for Windows, Copyright 2023, SAS Institute Inc., Cary, North Carolina, USA). The model included all evaluation points. The primary outcomes of CCRSA and CPIB were not adjusted for baseline levels or for participant subtype, since there were no between‐arm differences at baseline, and subtype did not show significant differences in these outcomes when included in the marginal model. Analysis for GAS relied on a three‐category variable (declining: −1 to −3; no change: 0; improving: 1–3), using the cumulative logit link, in which case groups are compared using proportional odds at each evaluation. Because there are three goals per person, a generalized estimating equation model for ordinal data was used, considering clustering of three goals within‐person. ³³ , ³⁴ An overall two‐sided alpha of .05 was used for all analyses.

Secondary outcomes included word retrieval and script performance as well as post‐study interview data. Word and script performance was expected to show within‐arm gains for each group, not to assess trial superiority. Word and script analysis examined the change in the percent correct within each group relative to baseline at each evaluation. In within‐arm comparisons, these outcomes were adjusted for subtype, since subtype showed significant differences in these outcomes when included in the marginal model. Four post‐study interview questions were used to assess usability, acceptability, and self‐reported improvement/uptake. Participants rated each question on a seven‐point scale with a clinician blinded to the study arm. The scale included anchors at 1 (decreased/worse off), 4 (no change), and 7 (improved). Questions included: 1. “When I use what I learned in therapy, communication in daily activities is:” anchors included “1‐is much more difficult” to “7‐is much easier”; 2. “Has your emotional well‐being changed since the beginning of the study?” anchors included “1‐my emotional well‐being is worse” to “7‐my emotional well‐being is much better”; 3. “I found this speech therapy:” anchors included “1‐not helpful” to “7‐very helpful”; and 4. “Did the therapy program meet your expectations?” anchors included “1‐did not meet expectations” to “7‐exceeded expectations.” Analysis of post‐study interview data relied on a three‐category variable (improving: 5–7, no change: 4, and decreasing: 1–3) using the cumulative logit link, where groups are compared using proportional odds at each evaluation. Other outcomes and the measures used to characterize the cohort were described previously and are not the focus of this efficacy article but may be useful in future exploratory analyses to identify factors contributing to intervention response. ¹⁷

3. RESULTS

3.1. Enrollment, sociodemographic, and clinical characteristics

A total of 339 participants were screened (see Figure 1B); of these, 95 participant dyads (28%) met criteria and provided informed consent. The CB2 baseline publication provides additional information about the screen failures. ¹⁶ , ¹⁷ Participant dyads were randomized into the trial from four countries (the United States (USA), Canada, New Zealand, and the United Kingdom) and one U.S. territory.

Demographic and clinical characteristics for the dyads are provided in greater detail in the baseline report ¹⁶ and summarized here (see Table 2). Mild‐to‐moderate participants with PPA included 49 male and 46 female participants, with an average age of 67.1 years and education of 16.4 years. Symptom duration was 3.7 years across arms and on average considered mild as measured by the Western Aphasia Battery‐Revised Aphasia Quotient and the Activities of Daily Living Questionnaire scores at baseline (mean ± SD WAB‐AQ: 81.4 ± 8.4; ADLQ: 14.9 ± 10.0). ³⁵ Participants represented the three research subtypes of PPA: agrammatic, logopenic, and semantic. ¹ , ²⁴ Four participants indicated knowledge of a language other than English (Friulian, Hindi, Lithuanian, and Portuguese). Seventy‐eight percent of the participants living with PPA were retired (experimental: 82%; control: 72%). Most communication partners were spouses or long‐term partners (n = 84, 88%), with more being female (n = 58, 61%), and most dyads lived at the same address (n = 85 dyads, 89.5%). Forty‐two percent of the communication partners were retired (experimental: 43%; control: 41%).

TABLE 2.

Baseline demographic and clinical characteristics of participants with a diagnosis of PPA and their communication partners who were randomized to trial arm.

Participants with PPA	Experimental n = 56	Control n = 39	Effect size
Age at enrollment, years	66.9 +/− 7.2, [53–81]	67.4 +/−7.8, [52–82]	−0.07
Self‐reported sex, male‐to‐female	25:31	24:15	0.17
Self‐reported race/ethnicity:			0.18
White, Non‐Hispanic/Latino	56	37
White, Hispanic/Latino	0	1
Asian, Non‐Hispanic/Latino	0	1
Education, years	16.2 +/− 2.5, [12–21]	16.7 +/− 2.3, [12–21]	−0.22
Symptom duration, years	3.8 +/− 1.8 [0.4–8.4]	3.6 +/− 1.8, [1.3–8.6]	0.14
Prominent PPA subtype	Agrammatic: 15 Logopenic: 26 Semantic: 15	Agrammatic: 11 Logopenic: 17 Semantic: 11

Co‐enrolled communication partners	Experimental n = 56	Control n = 39	Effect size
Age at enrollment, years	65.3 +/− 8.7, [40–90]	63.9 +/− 12.0, [26–78]	0.14
Self‐reported sex, male‐to‐female	27:29	10:29	−0.23
Self‐reported race/ethnicity:			0.22
White, Non‐Hispanic/Latino	56	36
White, Hispanic/Latino	0	1
Asian, Non‐Hispanic/Latino	0	2
Education, years	16.6 +/− 2.9, [12–26]	15.7 +/− 2.4, [10–21]	0.34

Note: Frequency or mean ± SD, [range] are reported.

Abbreviations: PPA, primary progressive aphasia.

3.2. Primary outcomes

All GAS goals were anchored at level zero at baseline, representing the starting function of the goal. For analysis, ratings were collapsed into three categories for increased interpretability of study, including improvement: +1 to +3, no change: 0, and decline: −1 to −3. The experimental arm showed superiority for GAS at the post‐Block 1 evaluation, with the experimental arm goals showing gains in 66.7% of the goals compared to 49.1% of the control arm goals (odds ratio [OR] 2.06, 95% confidence interval [CI]: 1.23–3.47; p = 0.006, Figure 2A). At subsequent evaluations, GAS goals remained above 60% for the experimental arm (range 60.1%–62.0%) and ranged from 47% to 56.1% for the control arm, but these group differences were not statistically significant. For all time points and in both arms, the mean GAS score was positive and significantly different from zero within each arm, indicating significant movement toward goal attainment for both arms relative to baseline.

FIGURE 2.

Primary outcomes. (A) GAS goal outcomes showed superiority for the experimental arm at post‐Block 1. (B) The experimental arm showed significant within‐group gains in the CPIB communication participation measure after Block 1, whereas the control arm did not. (C) The CCRSA communication confidence measure showed low responsivity for both arms across the trial. *Indicates statistical significance p < .05. CCRSA, Communication Confidence Rating Scale for Aphasia; CPIB, Communicative Participation Item Bank; GAS, Goal Attainment Scaling.

The between‐arm comparisons failed to meet significance for the CPIB measure of communication participation. However, there were within‐arm differences in responsivity (Figure 2B). The experimental arm showed significant within‐arm gains at post‐intervention Block 1 (i.e., B1; mean increase: 1.4, CI: 0.2–2.6, p = 0.021) and a non‐significant increase at post‐intervention Block 2 (mean increase: 0.5, CI: −0.5–1.5, p = 0.32). In contrast, the control arm showed a non‐significant increase following post‐Block 1 (mean increase: 0.7, CI: −0.5–1.8, p = 0.26), followed by a steady decline at subsequent evaluations. The control arm showed a significant within‐group decline from baseline over the 12‐month intervention, whereas the experimental arm did not (control mean decrease: −2.1, CI: −3.9 to −0.3, p = 0.026; experimental mean decrease: −0.5, CI: −2.3 to−1.3, p = 0.57).

The CCRSA measure of communication confidence did not show between‐arm superiority for the experimental arm (Figure 2C). The experimental arm showed non‐significant increases in mean performance following each intervention block (mean increase post‐Block 1: 1.1, CI: −1.3–3.4, mean increase at post‐Block 2: 0.1, CI: −0.9–1.2). The control arm showed a non‐significant mean increase of 1.5, CI: −0.6–3.7, at post‐Block 1, followed by a steady decline at post‐Block 2 (mean decrease: −1.5, CI: −3.1–0.2).

3.3. Other outcomes

Other outcomes included word and script performance, with an a priori focus on within‐arm performance rather than between‐arm comparisons. Word performance was scored as a percentage correct. Scripts were scored by trained, independent outcome assessors using a multi‐dimensional scoring system developed for the trial that considers articulation, lexical retrieval, information content, grammar, and syntax structure accuracy as well as difficulty. Script performance is scored as a percentage correct of the maximum difficulty score assigned to the training script. ³⁶ There were significant within‐arm increases in word performance (experimental arm mean % increase post‐Block 1: 23.4%, CI: 18.5–28.3; control arm mean % increase post‐Block 1: 38.3%, CI: 32.8–43.8). There were also significant within‐arm increases in script performance (experimental arm mean % increase post‐Block 1: 35.3%, CI: 28.1–42.6; control arm mean % increase post‐Block 1: 36.3%, CI: 26.5–46.1) from baseline to post‐Block 1. Gains in word and script performance were maintained across all follow‐up evaluations.

Post‐study interviews were conducted with the dyad and a research clinician blinded to the intervention arm. These interviews assessed the dyad‐perceived impact of the intervention on everyday life and provided data regarding the acceptability of the intervention, aligned with the RAPT model. Reports were positive for both arms; however, responses favored the experimental arm for positive changes in communication in daily activities (OR: 2.9, CI: 0.9–9.5, p = 0.073), positive changes in emotional well‐being (OR: 1.4, CI: 0.56–3.53, p = 0.47), overall helpfulness of the speech therapy intervention (OR: 4.4, CI: 0.8–23.0, p = 0.083), and reached between group significance in meeting overall expectations of the intervention (OR: 5.4, CI: 1.7–17.1, p = 0.005; Figure 3).

FIGURE 3.

Post‐study interviews favor the experimental arm. Post‐study interview data favored the experimental arm for positive changes in communication in daily activities, emotional well‐being, overall helpfulness of the speech therapy intervention, and meeting overall intervention expectations. *Indicates statistical significance P < .05. The communication in daily activities anchors included “is much more difficult” to “is much easier”; emotional well‐being anchors included “my emotional well‐being is worse” to “my emotional well‐being is much better”; speech‐language intervention anchors included “not helpful” to “very helpful”; overall expectations anchors included “did not meet expectations” to “exceeded expectations.”

3.4. Safety and fidelity

Discontinuations were low at 6% (n = 6 participant dyads). Five discontinuations were related to personal reasons, and one was due to restrictions from the coronavirus disease 2019 (COVID‐19) pandemic (i.e., the communication partner and person with PPA did not reside in the same residence and could not participate in joint sessions). There were 15 adverse events (AEs) across 12 participants; 80% (12/15) were non‐serious and unrelated, and three were possibly related, categorized as psychosocial events. There were 13 total severe adverse events (SAEs) across 11 participants, all non‐related to the intervention, with the majority being medical events (12/13). Twenty‐two unique participants experienced an AE, an SAE, or both.

Fidelity was assessed from three vantage points (documentation, procedural, and theoretical) by blinded non‐treating study team members and remained high throughout the intervention. The protocol paper describes the fidelity protocol in greater detail. ¹⁷ Briefly, fidelity assessment was completed for 100% of study sessions for the first five participants of each interventionist in each study arm (10 participants/clinician) using audio and video recordings of sessions. After this, fidelity was rated on a randomly selected sample of 25% of study sessions per arm. Documentation and procedural fidelity scores were binned into three compliance categories (full compliance >95%; partial compliance 75%–95%; non‐compliance <75%). Fidelity assessment revealed that 92.4% and 93.1% of the sessions reviewed met full compliance for documentation and procedural fidelity, respectively. Theoretical compliance was rated on a scale where a score of 0 represents an equal representation of participation and impairment approaches, positive scores are reflective of participation approaches, and negative scores reflect impairment approaches (maximum and minimum scores are 25 and −25, respectively). Theoretical fidelity assessments suggested sessions for each arm aligned with their theoretical premise, with the experimental arm receiving a rating of 20.3 ± 6.8, whereas the control arm was at −17.3 ± 9.8. These polarized scores suggest that sessions were with the representative construct for each arm.

4. DISCUSSION

The CB2 trial is the first global RCT showing efficacy for speech‐language intervention for individuals with mild‐to‐moderate PPA and their communication partners delivered via video chat. Post‐study interview data are consistent with the efficacy analysis, showing superiority of the experimental arm. Thus, the experimental arm data show an important combination of statistical superiority and clinical meaningfulness for a complex clinical syndrome that impacts the individual living with the diagnosis, their family, their relationships, and larger psychosocial life circles. With 95 participant dyads and the use of an active control arm, the CB2 trial represents the most extensive study of PPA interventions to date. Both intervention arms were delivered with high fidelity from documentation, procedural, and theoretical perspectives. Both interventions were well tolerated and safe.

A recent systematic review showed that most previous non‐pharmacologic intervention studies for individuals with PPA had samples of five or fewer participants (88.2%), the majority were single‐subject designs (84.2%), and none were RCTs. ¹⁰ Prior studies align with Stage 0 and Stage I according to the NIH Stage Model for Behavioral Intervention Development. ³⁷ In addition, previous non‐pharmacologic intervention studies have tended to focus on impairment‐based outcomes. ¹⁰ In contrast, the CB2 trial design is consistent with a Stage 2 behavioral trial, with primary outcomes relevant to daily life and aligned with U.S. Food and Drug Administration (FDA) requirements for functional and meaningful outcomes. ³⁸ The remote delivery of the intervention provides vital opportunities for wide uptake and improved access to care. ¹² The inclusion of all subtypes of PPA extends generalization; however, there are still unmet needs for those beyond mild‐to‐moderate impairment and for those with multi‐domain cognitive impairment with prominent aphasia.

The progressive nature of PPA necessitates ongoing care that adapts to the disease‐associated changes, but coverage for ongoing intervention can be challenging. Studies with additional intervention blocks or those targeting individuals with more severe deficits will be critical for identifying optimal care practices throughout the disease course. In addition, the optimal number of intervention sessions (i.e., dose) remains an open empirical question. In the CB2 trial, the interventions consisted of two blocks: Block 1 had 10 weekly sessions, whereas Block 2 had 5. This model aligns with common reimbursement structures in the US. Superiority was achieved for the experimental arm after Block 1 but did not reach statistical significance at later evaluation points. Post‐Block 2 evaluations suggested a trend toward increased responsivity in the experimental arm. An extended Block 2 with additional sessions may have yielded stronger benefits, although this requires further empirical study. Post‐study interview data supported this possibility, as the most common participant request (n = 23) was for more sessions.

Although the experimental arm showed superiority in the primary outcome, it is notable that both arms showed significant within‐arm improvements from baseline and maintenance at 12 months in word and script performance. This is particularly striking when compared to observational studies, which report significant declines in naming ability on the Boston Naming Test over the same period for individuals with PPA. ³²

The word and script performance outcomes underscore two relevant points. First, speech‐language interventions can enhance and maintain word retrieval and automaticity/fluency in PPA over ≈12 months. Second, the experimental arm data suggest that with instruction and guidance from an SLP, asynchronous practice may be sufficient for achieving significant gains, allowing valuable synchronous session time to be allocated to other therapeutic activities, education, and counseling.

Despite enrollment across multiple countries, racial diversity was low; this challenge is aligned with the known issues of low PPA awareness in diverse communities and challenges in equitable access for obtaining a PPA diagnosis. ³⁹ Addressing disparities in diagnosis and care remains a key priority. Telehealth assessments and interventions offer a unique solution for overcoming geographic barriers associated with care access. For this trial, it allowed for enrollment across four countries and 25 U.S. states. Expanding the delivery of the intervention to languages other than English, with appropriate cultural adaptations, represents another significant opportunity.

Part of the enrollment and delivery of intervention for this trial occurred during the COVID‐19 pandemic. The video chat delivery intervention allowed for the continuation of the RCT when many other interventions required discontinuation or major adaptations to pandemic‐related restrictions.

In summary, this global, telemedicine, non‐pharmacological RCT demonstrates the efficacy of a speech‐language intervention focused on communication participation. The RCT provides the most robust evidence to date in favor of non‐pharmacologic interventions for positively impacting the life‐altering communication changes in PPA. The trial results elevate the level of evidence for non‐pharmacologic interventions for PPA and support the use of communication‐based interventions for PPA. The positive primary outcome and participant feedback in the experimental arm intervention support the viability of moving toward pragmatic trials and clinical integration. The NIH Stage Model and RAPT model provide important guidance for achieving these goals. ¹² , ¹³ , ³⁷ Cost analysis, potential refinements to determine the optimal dose, and identifying features of responders versus non‐responders are among the key next steps to maximize translational potential. The efficacy evidence is especially important given the paucity of effective pharmacologic options. The opportunity to maximize communication participation, quality of life, and independence has important functional, emotional, and economic impacts on the individual, families, and society. The CB2 trial provides methodologic and practical relevance for planning future large‐scale efficacy, effectiveness, and pragmatic trials across the ADRD spectrum.

CONFLICT OF INTEREST STATEMENT

The authors report grant funding (E.R., A.R.), consultant fees (A.R.), speaker honoraria (E.R.), support for attending meetings/travel (E.R.), and leadership roles on a committee, unpaid (E.R.). There are no additional disclosures from the authors. Author disclosures are available in the supporting information.

CONSENT STATEMENTS

The authors have reviewed and approved the submission of this manuscript. Informed consent was obtained from all human participants prior to trial enrollment.

Supporting information

Supporting Information

Rogalski E, Bona M, Esparza M, et al. Efficacy of Communication Bridge‐2 for primary progressive aphasia: A randomized controlled trial of communication intervention. Alzheimer's Dement. 2025;21:e70088. 10.1002/alz.70088