Comparative Effectiveness of In-Person and Virtual Picture-Naming Treatment for Poststroke Anomia

Abstract

Purpose:

In light of COVID-19, telepractice for speech therapy has been increasingly adopted. Telepractice promotes accessibility to therapy services for those in rural environments, lowers the frequency of missed appointments, and reduces the costs of rehabilitation. The efficacy of telepractice has been scarcely explored in the aphasia literature. Preliminary research has demonstrated comparable results of telepractice and in-person therapy for people with aphasia, but the current scope of research is insufficient to guide clinical practice. The present study examined whether the virtual administration of a picture-naming therapy paradigm was as effective as in-person administration.

Method:

The treatment effects of two similar clinical trials, one completed in-person (n = 13) and one completed virtually (n = 13), are compared. Participants were adults with chronic (> 6 months) poststroke aphasia. Both clinical trials administered Cued Picture-Naming Therapy 4 days a week for 2 weeks (eight treatment sessions). Treatment outcomes were analyzed using Tau-U effect sizes and Mann–Whitney U tests.

Results:

Weighted Tau-U averages showed an advantage of telepractice over in-person treatment in the acquisition effects of trained words, with participants demonstrating a very large effect (0.84, p < .01) following telepractice and a large effect (0.75, p < .01) following in-person treatment. Both telepractice and in-person rehabilitation demonstrated significant treatment effects and were not significantly different from each other per Mann–Whitney U independent-samples t tests.

Conclusions:

The present study demonstrated that telepractice of a picture-naming paradigm is as effective as in-person treatment administration. This justifies the use of telepractice to overcome accessibility and cost barriers to speech therapy administration and justifies taking patient preference into account. Future research should explore the efficacy of telepractice for treatments that promote greater generalizability to functional communication.

Supplemental Material:

https://doi.org/10.23641/asha.27641031

Approximately 30% of poststroke adults experience aphasia, or a disturbance in receptive and/or expressive communication (Grönberg et al., 2022). People with aphasia may experience various communication deficits, including impairments in comprehension, verbal expression, reading, and writing (Brady et al., 2012). A hallmark deficit of poststroke aphasia is anomia, or the inability to retrieve the appropriate word during communication (Goodglass, 1993). The consistent inability to retrieve a correct word needed for conversation may result in frequent communication breakdowns and, thus, reduced social participation (Dalemans et al., 2010; H. Lee et al., 2015). Research has shown that therapy for anomia not only has a direct impact on word retrieval (Diedrichs et al., 2023; Kendall et al., 2019; Schuchard et al., 2020; Wisenburn & Mahoney, 2009) but can also have an impact on a patient's life participation and quality of life (Best et al., 2008). Thus, therapies tailored to address word retrieval deficits in people with aphasia are well established (Raymer & Kohen, 2006; Raymer & Roitsch, 2023; Sze et al., 2021).

In light of the COVID-19 pandemic, there was a rapid shift toward telepractice, or the delivery of therapies using remote internet technology, to address language deficits in people with aphasia (Aggarwal et al., 2021; American Speech-Language-Hearing Association, n.d.; Fong et al., 2021). There is now a critical need to determine whether telepractice is as effective as routine, in-person aphasia treatment administration. Generally, telepractice presents key benefits for nonclinical and clinical populations, including people with aphasia. First, telepractice reduces the risks of spreading infectious illnesses, such as influenza or COVID-19, by maintaining appropriate social distancing measures (Monaghesh & Hajizadeh, 2020). Second, the implementation of telepractice promotes greater accessibility to rehabilitation services for people in rural environments. Individuals living in rural environments experience limited access to speech therapy services and an absence of clinicians with expertise in aphasia rehabilitation (Hall et al., 2013). Third, telepractice reduces the barrier of transportation to receive services as telepractice can occur in an individual's home environment. Less than 15% of poststroke adults demonstrate complete motor recovery following injury (Hendricks et al., 2002) and only 30%–66% of poststroke adults return to driving (Aufman et al., 2013; Blonski et al., 2014). Many people with aphasia rely on care partners for transportation. The reliance on transportation is costly in terms of time and money and directly increases the care partner's burden (Tindall & Huebner, 2009). Moreover, travel costs can prohibit adults from attending scheduled appointments (Silver et al., 2012; Syed et al., 2013). The frequency of missed appointments is critical as missed appointments not only impede adults with aphasia from achieving adequate language treatment dosage for optimal language recovery but can also negatively impact clinical productivity, cost of care, and resource planning (Bhogal et al., 2003; Kheirkhah et al., 2016). Importantly, a study by Covert et al. (2018) found that participants enrolled in telerehabilitation for intensive Lee Silverman Voice Treatment completed significantly more scheduled appointments than participants enrolled in the in-person administration, suggesting that telerehabilitation may reduce the frequency of missed appointments. Lastly, telepractice may reduce rehabilitation costs by an estimated 30% (Wenke et al., 2014). The cost reduction may be due to reduced transportation costs, facility fees, and loss of work time (Hoffmann et al., 2010; Kairy et al., 2009; McGrath et al., 2008; Tindall & Huebner, 2009). It has been reported that approximately 2.2–2.5 million people with aphasia in the United States alone would receive treatment if the accessibility to rehabilitation services improved (Simmons-Mackie, 2018). Thus, telepractice may present the opportunity to promote access to speech therapy services to a greater number of individuals across the nation and reduce negative outcomes of poststroke aphasia.

The effectiveness of speech therapy delivered via telepractice for people with aphasia has been documented scarcely in the literature. As of 2021, only five published studies were eligible for inclusion in a systematic review exploring the effectiveness of telepractice for poststroke aphasia (Cacciante et al., 2021). However, preliminary studies have reported that telepractice outcomes are comparable to in-person rehabilitation outcomes in people with aphasia across constructs, including word finding, functional communication, auditory comprehension, and generalization (Agostini et al., 2014; Øra et al., 2020; Woolf et al., 2016; Zhou et al., 2018). Regarding word finding, Agostini et al. (2014) suggested that telepractice outcomes for anomia in a sample of five poststroke adults with chronic aphasia were both feasible and comparable to in-person treatment. More recently, telepractice for anomia in bilinguals with poststroke aphasia has produced equivalent rehabilitative outcomes to in-person treatment and high treatment reliability (Peñaloza et al., 2021), though the study was also limited by a small sample size (N = 16). Beyond treatment effectiveness, many people with aphasia have reported higher satisfaction with telepractice in comparison to routine, in-person treatment (Carr et al., 2022; Jacobs et al., 2021).

Whereas preliminary studies of telepractice effects and patient reports speak to the promise of telepractice for aphasia, the results and brevity of the literature remain insufficient to definitively inform clinical practice. To extend the existing, yet limited, body of literature exploring telepractice for people with aphasia, the present study aims to examine the effectiveness of telepractice in comparison to the routine, in-person rehabilitation of word retrieval deficits in poststroke aphasia. In line with preliminary investigations of telepractice and picture naming (Agostini et al., 2014; Peñaloza et al., 2021), we posit that the treatment effects following the telepractice administration of a picture-naming paradigm will be comparable to the outcomes of an in-person administration of the same treatment paradigm.

Method

The present study is a retrospective analysis comparing treatment outcomes from two aphasia clinical trials. The first clinical trial, completed in 2019, was completed entirely in-person, whereas the second clinical trial was administered virtually and is ongoing. Both studies received approval from The Ohio State University's Institutional Review Board (IRB Protocol #: 2019H0314) before recruitment and all participants signed informed consents prior to data collection.

Participants

Thirteen adults with chronic aphasia (i.e., greater than 6 months poststroke) completed each clinical trial for a combined sample of 26 unique participants included in this study. Participants were eligible for inclusion if they (a) had a single left-hemisphere stroke, without right-hemisphere involvement, at least 6 months before participation; (b) were native English speakers; (c) were between the ages of 18–85 years; (d) had a raw score from 4 to 44 on the Boston Naming Test (BNT; Kaplan et al., 1983); and (e) had no severe depression per the Beck Depression Inventory (Beck et al., 1961; in-person trial) or Quality of Life in Neurological Disorders Depression Short Form (Cella et al., 2012; telepractice trial). The presence of aphasia was determined by the Western Aphasia Battery–Revised (WAB-R; an Aphasia Quotient of less than 94.7; Kertesz, 1982) for the in-person clinical trial and by the Comprehensive Aphasia Test (CAT; modality mean T-score of 70 or below and Comprehension of Spoken Language of 35 or above; Swinburn et al., 2004) for the telepractice clinical trial. Participants were required to demonstrate relatively preserved auditory comprehension abilities to ensure adequate comprehension of all tasks and therapy instructions. Auditory comprehension was determined by performance on auditory comprehension sections of the WAB-R (within 2 SDs of the norming sample mean on the Auditory Verbal Comprehension Score) or the CAT (within 1.5 SDs of the norming sample mean on the Comprehension of Spoken Language Section, excluding Comprehension of Spoken Paragraphs). Participants with severe apraxia of speech (AOS) were excluded based on a brief examination of AOS as described by Harnish et al. (2018). AOS severity was rated by two certified speech-language pathologists who were otherwise unaffiliated with the study. If the two speech-language pathologists disagreed on the severity of AOS, a third was consulted.

Picture-Naming Probes

Stimulus lists for therapy were individually curated for each participant. Sixty target words that the participant was unable to name accurately on two separate occasions were chosen from a corpus of 500 items. When a participant was unable to accurately name fewer than 60 words on these two occasions, words that the participant was unable to accurately name once were included as possible target words. The 60 selected items were divided across three-word lists with 20 items in each list. Stimulus items in each list were matched for word frequency, syllable length, and living versus nonliving status (e.g., an animal vs. an inanimate object). Stimulus lists were matched within participant stimulus lists but were not matched between participants on the aforementioned linguistic variables. List 1 consisted of the 20 words targeted in treatment and probed frequently, before each treatment session. List 2 words were also probed frequently, but not trained, as a control for practice effects. List 3 words were probed seldomly (i.e., 2 times throughout the duration of study participation) and not trained, to examine response generalization effects, or the generalization of therapy gains from trained items to untrained items.

Treatment outcomes were measured using picture-naming probes. List 1 (frequently probed, trained) and List 2 (frequently probed, untrained) were probed across all baseline sessions, before each treatment session (except Treatment Session 1), and posttreatment. Picture probes were not administered before the first treatment session because they were implemented to measure improvement from the previous session's treatment administration. List 3 was probed only once at baseline and once at posttreatment. For the in-person clinical trial, additional baseline sessions were added until there was no ascending or descending trend in List 1 or List 2 performance across the three sessions before initiating treatment. For the telepractice trial, participants only completed three baseline sessions due to the decision to use Tau-U for individual effect size analysis, which corrects for baseline trends (see the Data Analysis section).

Reliability for the picture-naming probes was calculated by subtracting the difference in ratings (between the original score and rescore) from the total number of words across all probes and then dividing by the total number of words. Of the 17% of sessions assessed for the in-person trial, intrarater reliability was 97.8% and interrater reliability was 97.2%. Of the 18% of sessions assessed for the virtual trial, intrarater reliability was 98.8% and interrater reliability was 98.5%.

Treatment

The present study administered a computerized Cued Picture-Naming Therapy (CPNT; Harnish et al., 2014; D. Kendall et al., 2014), which has been found to produce positive, robust effects on trained words (Diedrichs et al., 2023; Harnish & Lundine, 2015; Kendall et al., 2014). Participants in the in-person trial completed treatment 4 times a week for 2 weeks (eight sessions), whereas participants in the telepractice trial completed treatment 4 times a week for 4 weeks (16 sessions). To keep treatment length consistent across both trials for comparison of outcomes, only the treatment effects after the first 2 weeks of treatment (eight sessions) for the telepractice trial were examined.

During each treatment session, participants named each word from List 1 across eight consecutive presentations with a set sequence of cueing from the computer program and administrator. Each target word was presented as a black-and-white image. The sequence included (a) independent naming, (b) orthographic cueing (i.e., the written word beneath the picture), (c) repeating, (d) independent naming after a short delay (i.e., approximately 3 s), (e) semantic cueing (i.e., three descriptive statements about the target were read aloud by the administrator), (f) phonological cueing (i.e., the first sound and letter were provided by the administrator), (g) repeating, and (h) independent naming after a short delay. If a participant inaccurately named the item following the cue, the treatment administrator provided a model of the target word and asked the participant to repeat it. If the participant still produced the target word incorrectly after the model, the treatment moved on to the next cue. Figure 1 illustrates the therapy protocol.

Figure 1.

Cued Picture Naming Therapy Protocol. Images are numbered in order of appearance during treatment. For the orthographic cue, participants read the word on the screen aloud. For the repetition cue (Images 3 and 7), the administrator verbalized the target word and asked the participant to repeat it. For the 3-s delay cue (Images 4 and 8), the participant viewed a blank white screen for 3 s before the image appeared and then they named the image independently. For the semantic cue (Image 5), three descriptive statements (e.g., “worn outside” for “shoe”) were read aloud by the clinician and then the participant was asked to name the item. For the phonemic cue (Image 6), the administrator named the letter and sound that the target word starts with (e.g., “starts with the letter ‘s’ … ‘sh’ for ‘shoe’”).

Fidelity for the treatment protocol was calculated as a percentage of accurate pictures, cues, or models provided out of total opportunities. In the 28% of in-person sessions assessed for treatment fidelity, fidelity was 93.8%. In the 25% of virtual sessions assessed for treatment fidelity, fidelity was 98.06%.

Procedure

All sessions were audio- and video-recorded for fidelity and reliability purposes. Both in-person and telepractice treatment were administered on Dell Latitude laptop computers utilizing E-prime Professional (Psychology Tools; www.pstnet.com) software. E-prime Version 2.0 was used for the in-person trial, and Version 3.0 was used for the virtual trial. The in-person trial utilized 14- or 15.6-in. Dell Latitude E6540 laptops or 15.6-in. Dell Latitude 5590 laptop computers, and the telepractice trial used 15.6-in. Dell Latitude 5500 laptop computers. The in-person treatment administration was completed face-to-face in a private room, and the telepractice was administered using Microsoft Teams in a remote setting. Participants in the telepractice trial were provided a Dell Latitude laptop computer to complete treatment in their home environment. When the participant received the laptop computer for the research study, the research administrator provided written and verbal education on how to access and use Microsoft Teams. The written instructions were created by certified speech-language pathologists with picture support and simple verbiage to support comprehension of user instructions. Research administrators observed each participant access a trial link to ensure comprehension and implementation of access instructions. Treatment administrators recorded accuracy at the time of treatment, and the recordings were used to verify scoring. All treatment administrators were either certified speech-language pathologists or trained research assistants.

Data Analysis

Treatment effect sizes were analyzed using Tau-U (Parker et al., 2011), an approach to analyzing single-case experimental research derived from Mann–Whitney U and Kendall rank correlation. Tau-U determines effect sizes as a nonoverlap between a baseline and intervention phase, accounting for the treatment phase trend while correcting for the baseline trend. Specifically, an undesirable baseline trend is accounted for by adding trend data to the nonoverlapping data in the case of a negative trend and subtracting trend data in the case of a positive trend. Tau-U has shown promise as an approach to quantitative analysis of small-sample aphasia treatment studies (J. B. Lee & Cherney, 2018). For descriptive purposes, Tau-U effect sizes were described as having no effect (≤ 0.20), small effect (< 0.20), moderate effect (0.20 to < 0.60), large effect (0.60 to < 0.80), or very large effect (≥ 0.80; Vannest & Ninci, 2015).

To determine if there was a statistically significant difference between treatment groups on treatment acquisition, the Mann–Whitney U independent-samples test across Tau-U effect sizes was conducted. The Mann–Whitney U independent-samples test is a rank-based, nonparametric test that can determine whether there are differences between two groups on a continuous variable. To further analyze group treatment effects, the weighted average, or the average of all participants' Tau-U treatment effects weighted by the number of baseline sessions, of Lists 1 and 2 for each clinical trial was calculated and compared. The presence of response generalization, or the generalization of therapy gains from trained items (i.e., List 1) to untrained items (i.e., List 2), was examined by reviewing the performance on the seldom-probed list of words (List 3). Stimulus generalization, or the ability to produce a learned response in a new stimulus environment, was not examined within the present study (Harnish et al., 2018).

Results

Participants

For the in-person clinical trial, there were two women and 11 men with a mean age of 60.5 years (SD = 14.4, range: 36–78). All participants had 12 or more years of education (M = 15.23, SD = 2.59, range: 12–20). Across the participants, the mean time poststroke was 51 months (SD = 56.5, range: 6–178). The mean WAB Aphasia Quotient was 62.6 (SD = 15.2, range: 39–93.5), and the mean BNT score was 19.38 (SD = 14.33, range: 5–44). The average AOS rating was 2.05 (SD = 1.76, range: 0–5) on a Likert scale of 0 (no AOS) to 7 (profound AOS) across participants. See Table 1 for complete demographics and linguistic profiles of the in-person trial participants.

Table 1.

In-person participant demographics and linguistic profiles.

ID	Age	Gender	Education	Months post-CVA	AOS	BNT	WAB-AQ	WAB classification
I01	72	M	20	178	5.0	14	60.0	Broca
I02	36	M	17	33	1.0	33	71.8	Conduction
I03	62	M	12	14	4.3	5	57.4	Broca
I04	36	M	18	47	2.0	19	54.8	Broca
I05	40	F	14	12	1.0	9	63.8	Anomic
I06	67	M	14	21	1.5	38	76.5	Conduction
I07	63	M	17	154	0.0	41	93.5	Anomic
I08	70	M	12	6	4.3	6	48.1	Broca
I09	78	F	12	74	1.0	10	39.0	Broca
I10	69	M	16	27	0.5	6	54.1	Conduction
I11	72	M	13	81	4.0	14	45.0	Broca
I12	67	M	17	11	0.0	44	78.2	Anomic
I13	55	M	16	9	2.0	13	71.6	Conduction

Note. CVA = cerebrovascular accident; AOS = apraxia of speech, reported as the average of two to three raters on a scale of 0 (absent) to 7 (profound); BNT = Boston Naming Test (Kaplan et al., 1983); WAB = Western Aphasia Battery; AQ = Aphasia Quotient (Kertesz, 1982); M = male; F = female.

For the telepractice clinical trial, there were six women and seven men with a mean age of 60.07 years (SD = 9.30, range: 51–78). All participants had 12 or more years of education (M = 14.46, SD = 2.02, range: 12–16). Across the participants, the mean time poststroke was 44 months (SD = 42.67, range: 6–130). The average CAT modality mean score was 50.23 (SD = 6.29, range: 39.5–65.7), and the mean BNT score was 25.46 (SD = 13.6, range: 4–44). The average AOS rating was 1.38 (SD = 1.22, range: 0–5) out of 7 across participants. See Table 2 for complete demographics and linguistic profiles of the telepractice trial participants.

Table 2.

Virtual participant demographics and linguistic profiles.

ID	Age	Gender	Education a	Months post-CVA	AOS	BNT	CAT modality b T score	CAT receptive T score	CAT expressive T score
V01	55	M	12	33.8	0.5	27	49.5	48.5	49.6
V02	53	F	12	13.7	3.5	10	45.8	46.5	44.6
V03	56	F	16	116.5	0.5	12	46.0	43.0	48.6
V04	60	M	16	13.9	1.0	4	46.3	48.0	43.3
V05	73	F	16	6.4	0.0	36	57.0	57.5	57.6
V06	54	F	16	11.5	1.0	13	51.2	49.0	52.3
V07	78	M	16	57.8	1.0	41	65.7	64.5	66.7
V08	52	F	12	92.1	4.0	36	39.5	47.0	31.7
V09	51	M	16	44.2	2.0	15	50.7	56.0	46.7
V10	75	M	16	130.6	0.0	34	49.7	59.0	60.0
V11	55	M	12	11.6	2.0	20	48.7	49.5	50.7
V12	63	F	16	21.2	1.5	39	54.3	50.5	56.3
V13	56	M	12	19.2	1.0	44	48.8	45.0	51.5

Note. CVA = cerebrovascular accident; AOS = apraxia of speech, reported as the average of two to three raters on a scale of 0 (absent) to 7 (profound); BNT = Boston Naming Test (Kaplan et al., 1983); CAT = Comprehensive Aphasia Test (Swinburn et al., 2004); M = male; F = female.

^aParticipants were asked if they had greater than or equal to 16 years of education and did not provide an exact number should they have greater than 16 years of education. A value of 16 was used for all of these participants and the mean years of education may be underestimated.

^bThe modality mean T score is the average of the following subtests: comprehension of spoken language (excluding paragraph comprehension), comprehension of written language, repetition, naming, reading, and writing.

The Mann–Whitney U independent-samples test was conducted to compare treatment session time and demographics of the two clinical trials. Dosage was kept constant across participants to control for the number of exposures/teaching episodes. Aphasia severity was variable across participants, so the length of treatment sessions varied depending on the reaction time for each naming attempt. The length of each treatment session was recorded throughout the study. Across participants, time in treatment did not significantly differ between the in-person (M = 23 min; range: 11–57 min) and telepractice (M = 23 min; range: 13–46 min) clinical trials. There was a significant difference between the aphasia severity of the in-person and telepractice trials, such that participants in the telepractice trial presented with less severe language severity as a group in comparison to the in-person treatment group. No other clinical or linguistic variables showed a significant difference between the in-person and telepractice clinical trials. See Table 3 for Mann–Whitney U test statistics and significance values.

Table 3.

Comparison between the clinical trials.

Variable	Test statistics	p value
Age	−0.591	.579
Months post-CVA	−0.026	1.00
Education	−1.127	.287
Aphasia severitya	13	.001*
BNT	1.104	.287
AOS	−0.962	.362
Treatment session time	65.0	.336

Note. Mann–Whitney U independent-samples t test was used to compare clinical and linguistic variables of the in-person and telepractice trials. CVA = cerebrovascular accident; BNT = Boston Naming Test; AOS = apraxia of speech; WAB = Western Aphasia Battery; AQ = Aphasia Quotient; CAT = Comprehensive Aphasia Test.

^aTo compare the WAB-AQ and CAT modality mean, WAB quotients were converted to T scores in order to be compared in the Mann–Whitney U test.

^* p < .01.

Individual Treatment Effects

Individual effect sizes were calculated for each participant to investigate whether virtual treatment sessions were as effective for treatment acquisition as in-person treatment sessions. Acquisition effects, or the baseline performance compared with performance during and immediately following treatment, were calculated using Tau-U (Parker et al., 2011) via a web-based application (the Tau-U calculator; singlecaseresearch.org; Vannest et al., 2016). Ten of the 13 participants in the in-person clinical trial responded to the aphasia treatment with statistically significant large or very large effect sizes (see Table 4). Two participants demonstrated nonsignificant moderate (I01) and very large (I13) effect sizes. One participant (I03) demonstrated no effect on trained words. Ten of the 13 participants in the virtual clinical trial responded to the aphasia treatment with significant very large treatment effect sizes (see Table 5). Two participants demonstrated nonsignificant large (V04, V09) or small (V06) effect sizes. See Figure 2 for an example of a participant's performance (i.e., V01) across the protocol. Additional individual participant plots can be found in Supplemental Material S1, Figures S1–S25.

Table 4.

In-person participant Tau-U effect sizes comparing treatment acquisition.

Participants	Trained items	Untrained items
I01	0.42 (moderate)	0.17 (small)
I02	0.85 (very large)*	−0.38 (no effect)
I03	−0.38 (no effect)	0.21 (moderate)
I04	0.79 (large)*	0.79 (large)
I05	1.00 (very large)*	0.75 (large)
I06	1.00 (very large)*	0.79 (large)
I07	0.83 (very large)*	−0.54 (no effect)
I08	1.00 (very large)**	0.75 (large)*
I09	0.84 (very large)*	0.5 (moderate)
I10	1.00 (very large)*	0.17 (small)
I11	0.75 (large)*	0.63 (large)
I12	0.94 (very large)*	0.72 (large)
I13	0.67 (very large)	0.58 (moderate)

Note. Tau-U values reported are for trained and untrained items that were frequently probed (Lists 1 and 2, respectively). Per Vannest & Ninci (2015), Tau-U effect sizes of < 0.20 were considered small; 0.20 to < 0.60, moderate; 0.60 to < 0.80, large; and ≥ 0.80, very large.

^* p < .05.

^** p < .01.

Table 5.

Virtual participant Tau-U effect sizes comparing treatment acquisition.

Participants	Trained items	Untrained items
V01	0.88 (very large)*	−0.04 (no effect)
V02	0.92 (very large)*	−0.13 (no effect)
V03	0.95 (very large)*	0.58 (moderate)
V04	0.75 (large)	0.54 (moderate)
V05	1.04 (very large)*	0.04 (small)
V06	0.13 (small)	0.38 (moderate)
V07	0.92 (very large)*	0.83 (very large)*
V08	0.83 (very large)*	.086 (very large)*
V09	0.79 (large)	−.29 (no effect)
V10	1.08 (very large)**	0.92 (very large)*
V11	1.13 (very large)**	0.25 (moderate)
V12	1.04 (very large)*	0.88 (very large)*
V13	0.91 (very large)*	1.00 (very large)*

Note. Tau-U values reported are for trained and untrained items that were frequently probed (Lists 1 and 2, respectively). Per Vannest & Ninci (2015), Tau-U effect sizes of < 0.20 were considered small; 0.20 to < 0.60, moderate; 0.60 to < 0.80, large; and ≥ 0.80, very large.

^* p < .05.

^** p < .01.

Figure 2.

Example of V01's performance across the treatment protocol. All treatment lists are out of 20 total words. Probe accuracy performance is presented as a raw score of the total number of words accurately retrieved. V01 completed three baseline sessions, eight treatment sessions (seven of which included picture probes for Lists 1 and 2), and one posttreatment session. Only seven picture probes were collected during treatment (i.e., Sessions 4–10) because picture probes do not occur in the first treatment session. V01, the participant presented in this figure, demonstrated a large improvement on List 1 words, no effect on List 2 words, and a small improvement on three words.

For the in-person clinical trial, one of the 13 participants (8%) demonstrated a significant large treatment effect on the untrained, frequently probed word list when comparing baseline performance with performance during and immediately following treatment. For the telepractice clinical trial, five of the 13 participants (38%) demonstrated significant very large treatment effects on the untrained, frequently probed word list. Raw scores for the seldomly probed, untrained word lists are presented in Table 6. Seven out of 13 participants in the telepractice clinical trial (54%) demonstrated small improvements from pre- to posttreatment. Nine out of 13 participants in the in-person clinical trial (69%) demonstrated small improvements from pre- to posttreatment.

Table 6.

Raw scores of seldomly probed, untrained items (N = 20).

In-person	Pretreatment	Posttreatment	Virtual	Pretreatment	Posttreatment
I01	5	5	V01	3	4
I02	9	12	V02	3	6
I03	2	4	V03	0	0
I04	3	2	V04	2	0
I05	3	4	V05	11	12
I06	5	10	V06	6	4
I07	14	11	V07	17	12
I08	3	2	V08	4	4
I09	4	5	V09	6	5
I10	2	3	V10	2	9
I11	5	9	V11	3	3
I12	11	14	V12	3	9
I13	3	5	V13	7	11

Note. List 3 raw scores are out of 20 total.

Group Treatment Effects

A Mann–Whitney U test was conducted to determine whether there were statistically significant differences in treatment acquisition scores between the in-person and telepractice trials. Median treatment acquisition scores for in-person and telepractice trials were not statistically significantly different on the trained words (List 1; U = 101, z = 0.847, p = .418) or the untrained, frequently probed words (List 2; U = 84.5, z = 0.00, p = 1.00).

When comparing the Tau-U weighted averages, or the average of all participants weighted by the number of baseline sessions, the telepractice trial presented with an advantage over the in-person trial when examining acquisition effects of trained words (i.e., List 1), evidenced by a very large effect following telepractice administration (0.84, p < .01) and a large effect following in-person administration (0.75, p < .01). When examining acquisition effects of the untrained, frequently probed words (i.e., List 2), both the in-person trial (0.39, p < .01) and telepractice trial (0.44, p < .01) demonstrated a moderate effect following rehabilitation.

Discussion

The present study aimed to determine whether a telepractice administration of a picture-naming treatment paradigm for people with aphasia would demonstrate effectiveness similar to that of an in-person administration of the same picture-naming treatment. Consistent with previous research (Agostini et al., 2014; Peñaloza et al., 2021), Mann–Whitney U tests revealed that the Tau-U treatment effect sizes for both telepractice and in-person administrations of treatment for anomia produced significant and comparable treatment gains. Examination of group weighted averages of Tau-U effect sizes showed that the average effect size for telepractice administration of the CPNT protocol was larger than that for the in-person administration (very large vs. large effect, respectively). Although Mann–Whitney U tests did not identify this as a statistically significant difference, it is possible that with greater power afforded by a larger sample size, a statistical difference could emerge. However, the telepractice clinical trial group presented with significantly less severe aphasia, which may have promoted a slight advantage in the treatment outcomes per the weighted Tau-U values. Nonetheless, the results of the present study support the use of telepractice to deliver anomia treatment as it was found to be as effective as in-person administration.

Treatment effects on the untrained, frequently probed words (List 2) were moderate in both in-person and telepractice trials, and the improvements between clinical trials were not statistically significantly different from one another. A greater proportion of participants from the in-person trial demonstrated improvements from pre- to posttreatment on the untrained, seldomly probed list (List 3), suggestive of response generalization. It is important to note that the magnitude of improvement remained small for these participants, as with the virtual participants who showed improvement. Moreover, a similar number of participants declined in performance on the untrained, seldomly probed words (List 3) in the in-person (three participants) and virtual (four participants) groups. Therefore, it is unlikely that there were true differences in response generalization effects between the two groups. Future studies should continue to control for practice effects and further explore factors contributing to response and stimulus generalization effects for picture-naming treatment paradigms. While aphasia treatment literature consists of predominantly single-case design or small group studies (Fridriksson & Hillis, 2021), conducting virtual clinical trials offers the opportunity to recruit participants remotely and acquire larger samples to draw firmer conclusions about the effects of telepractice for anomia, as well as factors that contribute to generalization.

Beyond the treatment administration mode, other personal and external characteristics may have influenced response to anomia treatment. The aphasia literature has acknowledged the notable heterogeneity that is largely unexplained in rehabilitative effects at the individual level. Despite our attempt at controlling for individual variability via strict inclusion criteria, heterogeneity across participants in both trials is still present. One explanation of individual variability in treatment outcomes may be the presence and variability of AOS, a disorder of motor planning, across participants in both clinical trials. For example, I03 presented with one of the most severe AOS ratings and also demonstrated no effect on the trained treatment protocol. Similarly, I01 presented with the most severe AOS ratings and only presented with a moderate treatment effect on trained words. Thus, it is possible that the severity of AOS, despite accounting for the exclusion of people with severe AOS, could impact treatment response. Similarly, though no significant differences were found between groups, the in-person treatment group had more participants who demonstrated higher AOS ratings as a group. It is possible that the in-person sessions are more effective for individuals with severe apraxia given the comparable treatment outcomes. Future research may need to further explore the severity of apraxia when deciding between in-person and telepractice interventions.

Previous research has supported that most individuals with communication disorders are satisfied with telepractice modalities for speech-language rehabilitation (Guglani et al., 2023; Jacobs et al., 2021; Tousignant et al., 2018). Importantly, 94% of adults with mild traumatic brain injury reported a preference for telepractice over in-person rehabilitation (Brockway et al., 2016). With this knowledge and the findings of the present study, aphasia rehabilitation may be better suited to take client preference into account when determining a treatment regimen. Some patients may find therapy to be less stressful when completing it in a familiar home environment in comparison to a controlled research laboratory or clinical setting. Stress is a normal physiological and perceptual response to novel, unfamiliar, or challenging situations (Anisman & Merali, 1999; Schneiderman et al., 2005). However, acute stress, such as stress that may occur before or during the completion of speech therapy, has been found to impede memory and executive functions that are critical for learning (Buchanan et al., 2006; Schwabe & Wolf, 2010; Starcke et al., 2016). People with aphasia report a preference for participating and communicating in familiar environments (Dalemans et al., 2010; Howe et al., 2008), and familiar environments are more favorable for learning success (Shadiev et al., 2023). Thus, for some participants with aphasia, completing therapy in an unfamiliar, controlled laboratory environment may cause increased stress, discomfort, or performance anxiety, which could negatively impact cognitive functions needed for learning in comparison to participants with aphasia completing therapy in a familiar environment. The present study did not examine the stress levels of participants before or after treatment, but this may be one factor that could influence individual patient preference for telepractice versus in-person therapy. Future research should seek to explore the differences in stress levels of virtual versus in-person treatment paradigms to further explore the benefits of one form of administration over the other.

Given the advantage of telepractice in terms of accessibility and, in some cases, patient preference for telepractice over in-person therapy, it is promising that telepractice administration can offer the same benefit to a person's aphasia rehabilitation as in-person treatment. Not only does telepractice promote increased access to rehabilitation in rural and urban environments, but it also provides an additional outlet to achieve optimal treatment dosage. This is critical for aphasia rehabilitation as there are previous indications that an increased dose yields greater improvements in trained language functions (Carr et al., 2022; Cavanaugh et al., 2021; Cherney, 2012). Moreover, allowing participants to select a treatment delivery modality (i.e., in-person vs. telepractice) may promote motivation and, thus, compliance with treatment protocols (Biel et al., 2022).

An important aspect to consider when recommending telepractice for individuals is cognitive factors that may promote or reduce satisfaction and/or effectiveness with telepractice regimens. Prior research has supported that verbal short-term memory (Dignam et al., 2017), nonverbal working memory (Harnish & Lundine, 2015), visuospatial skills (Votruba et al., 2013), and selective attention (Lambon Ralph et al., 2010) may be predictive factors of aphasia treatment response. These cognitive abilities may have differential implications on treatment response to telepractice rehabilitation as all the above studies solely examined in-person treatment paradigms. Additionally, individuals with more severe aphasias and/or cognitive deficits may have a more difficult time completing telepractice without direct clinician help due to difficulty in utilizing technology independently (Menger et al., 2020; Rosenberg et al., 2009). This may result in a preference for in-person rehabilitation. Future research should seek to explore patient-specific factors, such as cognitive abilities, that may impact response to telepractice.

Our study presents critical implications for the accessibility of speech-language rehabilitation services for individuals with aphasia. Access barriers, such as living in rural environments and reliance on care partners for transportation, may inhibit individuals from receiving speech-language therapy. Research has found that the more a person with aphasia participates in speech-language therapy, the more positive the communication and language outcomes (Enderby & Sutton, 2020). Our research indicates that individuals can demonstrate comparable treatment outcomes following telepractice as compared to completing the treatment in person at an outpatient facility for picture-naming treatment paradigms. However, comparable treatment outcomes for other language domains and treatment targets may vary. Clinical practice should begin to better leverage the comparable effectiveness of telepractice to promote accessibility and health equity for individuals with aphasia.

Limitations

This study was limited by a small sample size. Replication of this work in a larger sample is warranted. Additionally, the present study solely examined picture naming as a therapeutic outcome. Thus, the results of the study cannot be extended or generalized to other language domains, such as reading, writing, comprehension, or functional communication. Future research should seek to explore the comparative effectiveness of telepractice on various language aspects for a more holistic understanding of telepractice efficacy.

Aphasia rehabilitation is most meaningful when generalization of trained items, or the ability to take a learned skill and utilize it in a novel, untrained context, is acquired (Mayer et al., 2024). Whereas picture naming is a common treatment regimen for people with aphasia, response and stimulus generalization following picture-naming protocols are often poor (Boyle & Coelho, 1995; Lambon Ralph et al., 2010; Wisenburn & Mahoney, 2009b). Previous research by our group has reported evidence for stimulus generalization for some individuals following CPNT (i.e., the picture-naming therapy utilized in this study). However, the study concluded that there is a relationship between working memory and stimulus generalization, such that all participants did not demonstrate generalization following therapy (Harnish et al., 2018). Thus, while the present study offers evidence for the effectiveness of telepractice for anomia, the effectiveness of telepractice when targeting generalization to untrained contexts is not clear. Given the importance of generalization following aphasia rehabilitation, future studies should explore the effectiveness of telepractice speech-language rehabilitation for therapies that better target response and stimulus generalization. It may be possible that the generalization of language skills to daily life would be more effective in a home environment as opposed to a laboratory setting (Macoir et al., 2017).

Conclusions

The present study presents evidence that further examines the efficacy of telepractice for anomia in people with aphasia. The results show that the effectiveness of a telepractice picture-naming paradigm is similar to that of routine, in-person treatment administration. The results of the study present important clinical implications. Should telepractice continue to demonstrate treatment outcomes at least as effective as in-person therapies, speech-language pathologists can be confident that it will provide evidence-based benefits to patients with aphasia while improving accessibility and reducing the costs of treatment. These findings also support taking patient preference into account if they have the option of both telepractice and in-person speech therapy. Future research should continue to examine the efficacy of virtual treatment for anomia in studies with larger sample sizes and across treatments that address other aspects of language beyond naming, such as functional communication. Future work should also more thoroughly examine whether there are differences in treatment generalization for telepractice versus in-person therapy.

Data Availability Statement

The data sets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Funding Statement

Research reported in this publication was supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under Award Number R01DC017711 (S.M.H.). The first author (C.C.J.) received support during the article review from the National Institute on Deafness and Other Communication Disorders of the National Institute of Health under Award Number 1F31DC022142. The second author (V.A.D.) received support during article preparation from the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR) under Award 90ARHF0007. The NIDILRR is a center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The third author (D.S.B.) was supported by National Institute on Deafness and Other Communication Disorders Grant T32DC014435. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIDILRR, ACL, HHS, or National Institutes of Health, and you should not assume endorsement by the Federal Government.