Determining the adoption intention and influencing factors of telerehabilitation by people with rehabilitation needs: a cross-sectional survey

Abstract

Background

Telerehabilitation is considered a promising way to address the gap between supply and demand for rehabilitation services. To sure that telerehabilitation will be widely applied in clinical, a comprehensive and thorough understanding of the extent of usage intention for telerehabilitation is needed. This study aimed to identify the use intention and the barriers and facilitators factors to telerehabilitation utilization among people with rehabilitation needs.

Methods

This study is a cross-sectional study. A structured survey questionnaire on the usage intention of telerehabilitation was designed by the research team based on the extended UTAUT theoretical model. This survey recruited a total of 386 participants with rehabilitation needs from May 2022 to July 2023. The partial least squares structural equation model (PLS-SEM) to analyze the data and test the research hypotheses to determine the factors that influence the public’s intention to use telerehabilitation.

Results

A total of 327 valid questionnaires were collected. The model of this study explained 62.0% of the variance in behavioral intention, 39.1% of the variance of perceived trust, and 36.7% of the variance of performance expectations. The results of the structural model analysis indicated that performance expectancy (β = 0.220, P < 0.001), social influence (β = 0.108, P = 0.011), perceived trust (β = 0.265, P < 0.001) and self-efficacy (β = 0.185, P = 0.009) had significant effects on behavioral intention to use telerehabilitation; performance expectancy (β = 0.402, P < 0.001) and effort expectancy (β = 0.208, P = 0.010) had a significant effect on perceived trust; self-efficacy had a significant effect on performance expectancy (β = 0.608, P < 0.001); performance expectancy and effort expectancy had indirect effect on behavioral intention with perceived trust as the mediator, and self-efficacy also had significant indirect influences with performance expectancy as the mediator.

Conclusions

This study applied an extended UTAUT model to investigated the intention of and influencing factors for using telerehabilitation service among individuals with rehabilitation needs. The results indicate that performance expectancy, social influence, perceived trust and self-efficacy are important factors influencing the use intention of telerehabilitation in individuals with rehabilitation needs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27678-6.

Background

It has become a basic international consensus that the need for rehabilitation is rapidly increasing worldwide [1]. The main reasons include the aging and longevity of the population worldwide as well as the ongoing incidence of injury and child developmental conditions. Healthcare and health policy will change to the long-term management of chronic diseases or impairments, implying an unprecedented increasing need for medical rehabilitation services [1, 2]. However, there is a large demand-supply gap of rehabilitation medical services globally especially in developing areas, most of the rehabilitation need has yet to be met [3]. It is estimated that more than 50% of the rehabilitation needs of people from some low- and middle-income countries is not met. An investigation carried in China indicated that more than 75% of the rehabilitation needs for middle-aged and elderly Chinese people were not met [4]. In the COVID-19 pandemic, this gap becomes larger. The World Health Organization launched Rehabilitation 2030: a call for action, which emphasized the increasing unmet rehabilitation service requirements in developing countries and called for action to strengthen rehabilitation to meet new challenges that increasing rehabilitation needs [5, 6].

Telerehabilitation can be utilized to expand rehabilitation medical services and improve the accessibility and quality of current rehabilitation services [7, 8]. Therefore, telerehabilitation is recognized as a potential means of bridging the demand-supply gap for rehabilitation services [9]. Currently, several systematic reviews have demonstrated that telerehabilitation is an effective and clinically safe new method of providing rehabilitation services for people with rehabilitation needs [10, 11]. The COVID-19 pandemic accelerated the sharp increase in China’s mHealth market [12]. Several large medical organizations in China have also developed related mobile application to implement telerehabilitation medical services [13–15]. Meanwhile, China’s rehabilitation service system is undergoing a profound transformation driven by national policies. The Opinions on Accelerating the Development of Rehabilitation Medical Services (2021) [16] explicitly encourage the extension of rehabilitation services to communities and homes through “Internet +” and home-based care models. More recently, the Guideline on Accelerating the Cultivation and Openness of Scenarios to Promote the Large-Scale Application of New Scenarios (2025) [17] has identified remote terminal service systems and rehabilitation medical services as key application scenarios for innovation. These initiatives reflect a strategic shift toward a diversified, technology-enabled rehabilitation service network, aiming to address the growing demand-supply gap and improve accessibility, particularly through telerehabilitation. However, the efficacy and adoption of telerehabilitation greatly depends on the intention and perseverance of the end users, i.e., those with rehabilitation need to use this untraditional kind of medical service [18]. To sure that telerehabilitation will be widely applied in clinical, a comprehensive and thorough understanding of the extent of usage intention for telerehabilitation is needed. In addition, it is important to identify the factors that drive and constrain people with rehabilitation needs to use telerehabilitation [19].

Therefore, the objective of this study was to identify the use intention and the barriers and facilitators factors to telerehabilitation utilization among people with rehabilitation needs. Therefore, advices could be provided for enhancing the adoption of digital rehabilitation interventions to multi-stakeholders.

Theoretical background

With the extraordinarily rapid development of digital technologies, e-health services have been continuously expanded in many aspects. Among them, telemedicine is now one of the most widely used e-health services and could be benefit to any individual with geographic, economic or other difficulty to go in the hospital or clinic. Scholars have proposed several theoretical models for understanding and analyzing the intentions and use of telemedicine, of which the Unified Theory of Acceptance and Use of Technology (UTAUT) is one of the most popular and influential theoretical models to understand the intention to use of new technology [20].

The UTAUT model can explain approximately 70% of the variance in behavioral intention (BI) and is better than other models at explaining the influencing factors on behavioral intention [21, 22]. It includes four core predictors: performance expectancy (PE), effort expectancy (EE), social influence (SI), and facilitating conditions (FC) [21]. As one of the best models to understanding the users’ intention to use the new technology, the UTAUT model has been widely used in the field of telemedicine [23]. However, Zhang et al. (2023) [24] suggest that the variables included in UTAUT fail to fully explain users’ usage intention for telemedicine. Several studies have also indicated that telemedicine acceptance and its determinants are complex, and that the UTAUT model is insufficiently in understanding telemedicine adoption intentions and its determinants, suggesting that specific factors should be extended to improve its explanatory power for telemedicine usage intention [25–27].

The extended UTAUT studies proposed several additional influencing factors for adoption telemedicine, particularly perceived risk (PR) and perceived trust (PT) [27, 28]. while telemedicine can improve the quality of health care and expand the scope of healthcare services, it also triggers security and privacy concerns [29]. Previous research has demonstrated that PR increases the tendency for negative outcomes, which increases resistance to the use of new technologies [30–32]. The PR has long been considered a negative factor in the adoption of new eHealth technologies. In contrast, PT has been considered a driving factor a driving factor of the adoption of new eHealth technologies [33]. Veinot et al. (2013) [34] conducted a design-orientated focus groups to advise on the development of web-based informatics intervention. The result suggested that the majority of participants’ recommendations focused on trust. Trust removes major doubts during user decision-making and plays a key role in driving initial use and future sustained use behaviors of new technologies [35].

In addition, evidence has showed that self-efficacy (SE) is also an important factor in analyzing telemedicine usage intentions. Liu et al. [36] suggest that SE is a user’s self-perception of their abilities that is essentially an individual’s subjective judgement, it can significantly influence the user’s intention to adopt a new technology. People with high SE can focus their attention and motivation on tasks that will help them achieve their goals [37].

Research Hypotheses

Performance expectancy (PE) is defined as the extent to which individual using a new technology will help them improve their job performance [22]. In this study, PE refers to the extent to which individuals perceive telerehabilitation to be useful. Sun et al. [38] analyzed users’ acceptance behavior of telemedicine based on the extended UTAUT model, the results indicated that when performance expectations are higher, telemedicine are more likely to be adopted. In addition, when individuals perceive a new technology as useful and can help them perform their job better, they will trust the new technology because of this favoring impression [39, 40]. Therefore, this study proposes the following hypothesis:

• H1: PE has a positive effect on individuals' behavioral intention to adopt telerehabilitation services.

• H2: PE positively influences individuals’ trust in telerehabilitation.

Effort expectancy (EE) is defined as the degree of easiness when using a new technology [22]. In this study, EE refers to the degree of ease or difficulty for the individuals to use telerehabilitation. There is evidence that users’ adoption intentions of telemedicine will increase if they feel that the telemedicine technology is easy to use [41]. On contrast, if the new technology is thought to be difficult to use, a negative impact would bring on the intention and lead to distrust on the new technology by the users [39, 40]. For the elderly users in particularly, EE is often an important influencing factor for adopting telemedicine [42]. Therefore, this study proposes the following hypothesis:

• H3: EE has a positive effect on individuals' behavioral intention to adopt telerehabilitation services.

• H4: EE positively influences individuals’ trust in telerehabilitation.

Social influence (SI) is defined as the extent to which individual is influenced by the opinions of surrounding groups [22]. In this study, SI refers to the extent to which an individual is influenced by the opinions of surrounding groups (e.g., family members, patients, medical staff, etc.) when using telerehabilitation. Research on telemedicine suggests that SI significantly influences users’ behavioral intention to use telemedicine [43]. In addition, researchers in the field of psychology have found that the choices of the majority would influence an individual’s trust preferences when he/she is confronted with something new [44]. If the new thing is not entirely understanding for example a digital health technology, the recommendations of surrounding groups can increase the user’s trust in the new technology. Therefore, this study proposes the following hypothesis:

• H5: SI has a positive effect on individuals' behavioral intention to adopt telerehabilitation services.

• H6: SI positively influence individuals’ trust in telerehabilitation.

Facilitating conditions (FC) is defined as the degree of technical or organizational support that an individual believes to be able to get for using a new technology [22]. In this study, FC refers to the extent to which individuals have acquire to support in using telerehabilitation by technology, service measures, personal equipment, etc. Zhang et al. [28] demonstrated that FC is a determining factor that influences diabetes patients to use telemedicine. Therefore, this study proposes the following hypothesis:

• H7: FC has a positive effect on individuals' behavioral intention to adopt telerehabilitation services.

Perceived risk (PR) is defined in this study as the perceived potential for negative effects when using telerehabilitation. The use of telerehabilitation may involving potential risks such as leakage of personal privacy, medical security or ethical considerations [45]. PR has long been considered a negative factor in the adoption of new health information technologies [32]. In addition, the relationship between PR and PT has attracted significant attention from researchers. Hong et al. [35] examined the influencing factors of individuals’ trust in online healthcare services, the results found that PR significantly reduced individuals’ trust in online healthcare services. Thus, we hypothesize that:

• H8: PR has a negative effect on individuals' behavioral intention to adopt telerehabilitation services.

• H9: PR negatively influences individuals' trust in telerehabilitation.

Perceived trust (PT) refers to the degree to which users believe that telerehabilitation services are reliable. PT is considered a key factor in human adoption of new technologies [46]. Trust can remove major doubts from the decision-making process of adopting new technologies, and thus plays a critical role in promoting initial and future sustained usage of new technologies [35]. Previous studies suggest that trust is an important attractive factor to participate in information technology services (e.g., telemedicine) [27, 47]. Hong et al. [35] demonstrated that people with high level of PT were more willing for continuous usage of online healthcare services in China. Overall, PT is a prerequisite for positive attitudes to behavioral intentions. Thus, we hypothesize that:

• H10: PT has a positive effect on individuals' behavioral intentions to adopt telerehabilitation services.

Self-efficacy (SE) refers to a person’s belief or confidence in his or her ability to perform a task [48]. Liu et al. (2022) [36] argued that SE is an individual’s self-perception of his or her ability, which is essentially a personal subjective judgement, and that it can significantly influence users’ adoption intentions of new technologies. Several studies have suggested that SE has a significant positive influence on user’s intention to use telemedicine [29, 36]. In addition, there is evidence that the stronger the SE, the stronger the PE in telemedicine [36]. Thus, we hypothesize that:

H11: SE has a positive effect on individuals’ behavioral intentions to adopt telerehabilitation services.

H12: SE positively influence individuals PE for using telerehabilitation.

In summary, this paper constructed a theoretical model of telerehabilitation usage intention based on the UTAUT model, and the specific model theoretical is shown in Fig. 1.

Fig. 1.

Research model

Methods

Questionnaire design

This study was a cross-sectional study to investigate the users’ intentions and factors influencing on using telerehabilitation. Based on the research model and research hypotheses, a research survey questionnaire was designed. The questionnaire had two sections, the first section was the demographic characteristics including age, gender, education, occupation, household income, use experience, etc. The second section was a structured scale on the intention of and influencing factors for using telerehabilitation, which included 29 items with each item measured a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The constructs PE, EE, SI, FC, and PR are measured by 4 items each, whereas PT, SE, and BI are measured by 3 items each. These variables were composed of potential items, all items in this scale were adopted or reworded from scales that have been validated in previous studies.

To ensure the reliability of research survey questionnaire, a pre-survey of the questionnaire was conducted in this study. The subjects of this pre-survey included the healthcare personnel of the Department of Rehabilitation Medicine of the First Affiliated Hospital of Gannan Medical University (including rehabilitation physicians, rehabilitation nurses, and rehabilitation therapists) as well as patients with rehabilitation needs, 47 questionnaires were distributed, 43 valid questionnaires were received. The Cronbach’s alpha (ɑ) of the questionnaire was 0.823, indicating that the reliability of the questionnaire was high. In addition, according to the feedback from the pre-survey, the items that were difficult to understand or ambiguous were modified to ensure the readability of the questionnaire items. See Appendix 1 for the specific variables and question items.

Participants and data collection

The participants of this study were individuals with rehabilitation needs, for which the inclusion criteria were patients aged 14 years or older who had completed at least 1 week of rehabilitation therapy. If a participant was unable to effectively complete the questionnaire due to illness or literacy level, their long-term carers assisted them in filling out the questionnaire.

The investigation was carried out in two cities (Ganzhou and Xi’an) from May 2022 to July 2023. We distributed the paper questionnaires at the Department of Rehabilitation Medicine, the First Affiliated Hospital of Gannan Medical University in Ganzhou from May to November 2022, and in Xi’an at the Department of Rehabilitation, the First Affiliated Hospital of Xi’an Jiaotong University and the Shaanxi Provincial Rehabilitation Hospital from April to July 2023. Before completing the questionnaire, the Participants were informed of the purpose of the study and the anonymity of questionnaire filling at the beginning of the survey, and all signed the written informed consents. The study was approved by the Ethics Committee of the First Affiliated Hospital of Gannan Medical University (Approval No. LLSC-2022042001).

In addition, we defined invalid questionnaires as one of the follows: (1) The completion time was less than three minutes, which was much less than it was (5 min) when testing in the pre-survey; (2) Those with an obvious and illogical answer rule (e.g., 1234512345……12345) in the second section of the questionnaire; (3) Those with unfilled information in the questionnaire. The invalid questionnaires were identified and deleted before data analysis.

Statistical analysis

We performed descriptive statistical analysis using SPSS 25.0 (SPSS Inc., Chicago, IL, USA). Between-group differences in behavioral intention (BI) across demographic subgroups were analyzed using independent t-tests for two-group comparisons, and one-way ANOVA for comparisons among three or more groups. The plausibility of the proposed model was verified by partial least squares structural equation modeling (PLS-SEM) following the established guidelines using Smart-PLS 4.0 (free trial version) [49]. The analysis of PLS has two phases. The first phase was to analyze the reliability and validity of the measurement model. The model reliability was assessed by the Cronbach’s alpha (ɑ) and the combined reliability (CR), the recommended value > 0.7 [50]. The model validity was assessed by the convergent validity and the discriminant validity. The convergent validity was evaluated by the Average Variance Extracted (AVE), the recommended value > 0.5 [51]. The discriminant validity was evaluated using the Fornell–Larcker criterion and cross-loadings [52]. The second phase was to verify and evaluate the structural model to show the relationship between the latent variables.

When using PLS-SEM, the sample size of the questionnaire should be larger than 10 times the maximum number of paths pointing to a particular variable in the structural model. In this study, the maximum number of paths pointing to a variable is 10. Therefore, the minimum sample size required for this study is 100.

Results

Demographic characteristics

A total of 386 questionnaires were distributed in this study, and 327 valid questionnaires were collected after the invalid questionnaires were removed, a validity rate of 84.7%. Among the invalid questionnaires, 10 questionnaires were filled in less than 3 min, 25 questionnaires had obviously logical answer rule, 24 questionnaires were incomplete. Of the 327 survey respondents, 58.1% (190/327) were male, 41.9% (137/327) were female. Only 11.0% (36/327) of the participants had used some form of telerehabilitation before this survey. Table 1 shows the demographic characteristics of participants.

Table 1.

Demographic characteristics

Variables		n	Percentage (%)	Mean (SD)	P
Gender	Male	190	58.1	3.905 (0.630)	0.52
	Female	137	41.9	3.861 (0.587)
Age (years)	≤ 30	43	13.1	4.047 (0.480)	0.010
	31—40	62	19.0	3.995 (0.697)
	41—50	62	19.0	3.962 (0.570)
	51—60	68	20.8	3.848 (0.510)
	> 60	92	28.1	3.717 (0.667)
Educational level	Primary school and below	37	11.3	3.658 (0.636)	< 0.001
	Middle school	78	23.9	3.889 (0.585)
	Senior high\Secondary Schools	91	27.8	3.714 (0.600)
	Undergraduate\junior college	104	31.8	4.096 (0.585)
	Master and above	17	5.2	4.020 (0.493)
Occupation	Party and government agencies\institutional workers	64	19.6	4.089 (0.646)	0.019
	Enterprise staff	62	19.0	3.903 (0.593)
	Self-employed\freelance	66	20.2	3.919 (0.582)
	Physical laborers	51	15.6	3.732 (0.529)
	Students	9	2.8	3.963 (0.512)
	Retired\unemployed\non-working	75	22.9	3.769 (0.648)
Monthly family income/capita (RMB)	< 2000	34	10.4	3.951 (0.575)	0.077
	2000–3000	33	10.1	3.788 (0.492)
	3001–4000	78	23.9	3.885 (0.653)
	4001–5000	82	25.1	3.760 (0.638)
	> 5000	100	30.6	4.003 (0.589)
Impact of health condition on one’s life	No effect	25	7.6	3.973 (0.471)	0.85
	Mild impact	86	26.3	3.907 (0.602)
	Moderate impact	102	31.2	3.876 (0.617)
	Severe impact	114	34.9	3.863 (0.645)
Forms of medical costs	Self-funded	53	16.2	3.994 (0.622)	0.079
	Employee Basic Medical Insurance	116	35.5	3.966 (0.611)
	Basic medical insurance for urban and rural residents	121	37.0	3.782 (0.559)
	Publicly-funded medical care	22	6.7	3.773 (0.759)
	Commercial Insurance	7	2.1	3.667 (0.720)
	Other	8	2.4	4.125 (0.589)
Telerehabilitation experience	Used	36	11.0	4.148 (0.481)	0.006
	No used	291	89.0	3.855 (0.619)

Mean (SD) represents the average of the three BI items per participant; subgroup means are the averages of these individual BI means

Descriptive analysis of variables

Descriptive statistics were analyzed for each variable in the model of this study. The mean score of BI was 3.887 ± 0.612, indicating users are willing to use telerehabilitation when they need it. Based on demographic subgroup comparisons, the following differences in BI were observed. Age had a significant effect on BI (P = 0.010), with older participants showing lower intention to use telerehabilitation. Educational level also significantly influenced BI (P < 0.001); participants with an Undergraduate\junior college or higher reported stronger intention compared to those with lower education levels. Occupation was significantly associated with BI (P = 0.019); party/government/institutional workers had numerically higher BI than physical laborers. Additionally, participants with prior telerehabilitation experience had significantly higher BI than those without experience (p = 0.006). No significant differences were found for gender, monthly family income, impact of health condition, or forms of medical costs.

Measurement model analysis

Table 2 summarized the results of measurement model testing. In this study, Cronbach’s alpha (ɑ) and CR values of all variables were greater than 0.8, and all the AVEs were greater than 0.6, indicating the excellent credibility and convergent validity of each construct.

Table 2.

Measurement model statistics

Construct	M ± SD	Items	VIF	Loadings	ɑ	CR	AVE
Performance expectancy (PE)	3.733 ± 0.632	PE1	2.007	0.810	0.874	0.914	0.726
		PE2	2.377	0.869
		PE3	2.431	0.871
		PE4	2.304	0.855
Effort expectancy (EE)	3.572 ± 0.687	EE1	2.972	0.882	0.895	0.927	0.761
		EE2	3.171	0.893
		EE3	2.123	0.855
		EE4	2.290	0.860
Social influence (SI)	3.609 ± 0.645	SI1	1.638	0.823	0.86	0.904	0.702
		SI2	2.347	0.832
		SI3	2.646	0.853
		SI3	2.048	0.842
Facilitating conditions (FC)	3.657 ± 0.690	FC1	2.192	0.847	0.85	0.899	0.69
		FC2	1.713	0.794
		FC3	2.083	0.822
		FC4	2.022	0.857
Perceived risk (PR)	3.320 ± 0.682	PR1	1.686	0.788	0.842	0.872	0.632
		PR2	1.935	0.706
		PR3	2.334	0.756
		PR4	2.011	0.915
Perceived trust (PT)	3.791 ± 0.633	PT1	1.832	0.840	0.844	0.906	0.762
		PT2	2.238	0.902
		PT3	2.095	0.875
Self-efficacy (SE)	3.706 ± 0.630	SE1	1.653	0.807	0.834	0.901	0.752
		SE2	2.767	0.915
		SE3	2.261	0.876
Behavioral intention (BI)	3.887 ± 0.612	BI1	2.218	0.885	0.883	0.927	0.81
		BI2	2.908	0.918
		BI3	2.556	0.897

The discriminant validity was evaluated using the Fornell–Larcker criterion and the Cross-Loadings. Our results showed that the square root of AVE values of the of each construct greater than the squared correlation coefficient of other constructs, and the factor loadings for each construct are greater than the other variables factor loadings (Tables 3 and 4). Therefore, the measurement model had good discriminant validity.

Table 3.

Discriminant validity of the constructs (square root of the AVE values in bold)

	PE	EE	SI	FC	PR	PT	SE	BI
PE	0.852
EE	0.683	0.872
SI	0.538	0.502	0.838
FC	0.614	0.735	0.509	0.830
PR	0.031	0.021	0.100	0.063	0.795
PT	0.597	0.533	0.416	0.512	-0.056	0.873
SE	0.608	0.721	0.425	0.649	-0.002	0.670	0.867
BI	0.675	0.646	0.515	0.617	0.035	0.665	0.674	0.900

Table 4.

Results of cross-loadings

	PE	EE	SI	FC	PR	PT	SE	BI
PE1	0.810	0.500	0.439	0.442	0.022	0.438	0.436	0.508
PE2	0.869	0.613	0.485	0.596	0.057	0.514	0.541	0.626
PE3	0.871	0.591	0.439	0.521	0.023	0.559	0.565	0.580
PE4	0.855	0.617	0.470	0.521	0.003	0.515	0.518	0.577
EE1	0.594	0.882	0.428	0.605	-0.014	0.473	0.632	0.538
EE2	0.561	0.893	0.421	0.624	0.016	0.452	0.613	0.523
EE3	0.644	0.855	0.454	0.680	0.039	0.495	0.661	0.623
EE4	0.577	0.860	0.445	0.650	0.030	0.434	0.605	0.561
SI1	0.557	0.558	0.823	0.516	0.041	0.410	0.453	0.534
SI2	0.421	0.337	0.832	0.367	0.090	0.282	0.330	0.381
SI3	0.387	0.327	0.853	0.365	0.121	0.310	0.268	0.345
SI4	0.398	0.402	0.842	0.418	0.101	0.361	0.333	0.421
FC1	0.505	0.600	0.443	0.847	0.067	0.410	0.501	0.495
FC2	0.521	0.567	0.346	0.794	0.026	0.468	0.527	0.504
FC3	0.402	0.611	0.400	0.822	0.056	0.379	0.537	0.465
FC4	0.592	0.659	0.492	0.857	0.060	0.439	0.585	0.574
PR1	0.038	0.113	0.109	0.105	0.788	-0.036	0.030	0.027
PR2	0.004	-0.056	0.098	-0.041	0.706	-0.042	-0.037	-0.040
PR3	0.001	0.021	0.130	0.052	0.756	0.017	-0.002	0.037
PR4	0.026	-0.025	0.065	0.040	0.915	-0.060	-0.015	0.045
PT1	0.463	0.405	0.309	0.424	0.045	0.840	0.537	0.546
PT2	0.583	0.514	0.404	0.487	-0.066	0.902	0.588	0.636
PT3	0.508	0.468	0.369	0.424	-0.117	0.875	0.629	0.552
SE1	0.535	0.531	0.326	0.473	-0.016	0.609	0.807	0.508
SE2	0.523	0.689	0.365	0.623	0.008	0.584	0.915	0.596
SE3	0.525	0.649	0.410	0.586	0.000	0.554	0.876	0.643
BI1	0.621	0.622	0.486	0.592	0.059	0.560	0.623	0.885
BI2	0.604	0.560	0.440	0.547	0.005	0.633	0.615	0.918
BI3	0.596	0.561	0.465	0.526	0.031	0.602	0.582	0.897

The values of bold are standardized loading loadings; the other values are cross loadings

The variance inflation factor (VIF) was tested the multicollinearity among each item. All VIF values of this study were smaller than 5, indicating no obvious multicollinearity problem between each item (Table 2).

Structural model analysis

To test the statistically significant of hypothesis in this study, the bootstrapping procedure (5000 resample) method to analysis and examine the measurement model.

The coefficient of determination (R²) is a commonly parameter to evaluate the level of variance explained. The R² value is between 0 and 1, with values closer to 1 indicating that the model has greater explanatory power. The model explained 62.0% (R² = 0.620) of BI to use telerehabilitation, suggesting it has moderate explanatory power.

The validity of the research hypothesis is determined by the significance of the standardized path coefficients (t > 1.92 and P < 0.05). Table 5; Fig. 2 demonstrated the results for the hypothesis testing. The relationship between BI and PE (β = 0.220, P < 0.001), SI (β = 0.108, P = 0.011), PT (β = 0.265, P < 0.001) and SE (β = 0.185, P = 0.009) were significant. Similar results were shown between PT and PE (β = 0.402, P < 0.001), EE (β = 0.208, P = 0.010). In addition, the path coefficient between SE and PE (β = 0.608, P < 0.001) was also statistically significant. Therefore, the hypotheses 1, 2, 4, 5, 10, and 11 were supported, and the hypotheses 3, 6, 7, 8, and 9 were not supported. It is worth noting that we found that PE could directly influence BI or indirectly through PT as the mediator, and the amount of mediating effect (0.106/0.220) was 48.2%; SE could directly influence BI or indirectly through PE as the mediator, and the amount of mediating effect (0.134/0.185) was 72.4%. In addition, although the hypothesized relationship between EE and BI was not supported, the results suggested that the PT played a significant mediating role between EE and BI.

Table 5.

Structural model results

Hypothesis	Path	R 2	β	t value	P value	Result
Direct effects
	BI	0.620
H1	PE -> BI		0.220	3.802	0.000	supported
H3	EE -> BI		0.090	1.346	0.178	Not supported
H5	SI -> BI		0.108	2.533	0.011	supported
H7	FC -> BI		0.104	1.661	0.097	Not supported
H8	PR -> BI		0.024	0.507	0.612	Not supported
H10	PT-> BI		0.265	4.130	0.000	supported
H11	SE -> BI		0.185	2.613	0.009	supported
	PT	0.391
H2	PE -> PT		0.402	5.829	0.000	supported
H4	EE -> PT		0.208	2.565	0.010	supported
H6	SI -> PT		0.104	1.708	0.088	Not supported
H9	PR -> PT		-0.084	1.240	0.215	Not supported
	PE	0.367
H12	SE -> PE		0.608	14.030	0.000	supported
Specific indirect effects
	PE -> PT -> BI		0.106	3.346	0.001
	EE -> PT -> BI		0.055	2.120	0.034
	SE -> PE -> PT -> BI		0.065	3.077	0.002
	SE -> PE -> BI		0.134	3.581	0.000
	PR -> PT -> BI		-0.022	1.200	0.230
	SI -> PT -> BI		0.028	1.523	0.128

Fig. 2.

Structural model analysis results

Discussion

This paper investigated the intention of and influencing factors for using telerehabilitation service among individuals with rehabilitation needs by adopting the extended UTAUT model. The results found that only one in nine of the respondents with rehabilitation needs had the experience of using telerehabilitation, but their intention to use telerehabilitation was high. There are four factors positive directive effects on intention for telerehabilitation: PE, SI, PT, and SE. Among these factors, PE and EE also had significant indirect influences with PT as the mediator, while SE also had significant indirect influences with PE as the mediator.

Of the 327 survey respondents in this study, only 11.0% of the respondents had used some form of telerehabilitation before this survey, indicating a low usage rate of telerehabilitation, which was similar to the findings of Jiang Yingyu et al. (2018) [53] (14.2%) and Nie Li et al. (2021) (11.22%) [54]. This may be because: (1) telerehabilitation is still in its early stage of development with not many relatively mature technical equipment, and a complicated prescribing process; (2) people have low awareness of telemedicine, especially telerehabilitation. Jiang Yingyu et al. (2018) [53] showed that the overall percentage of people mHealth services among middle-aged and elderly people in Beijing was 58.3%, while a study by Nie Li et al. [54] showed that the overall awareness rate of those aware of mHealth services among elderly patients with chronic diseases in Xinxiang was only 24.39%. Overall, the awareness of telemedicine in China is low. In addition, differences in awareness in people from different regions may be related to the level of economic development, healthcare infrastructure and access to information technology in the region.

This paper investigated the relationship between PE, EE, SI, FC, PR, PT, SE and BI. The results showed that PT, PE, SI and SE were direct influential factors on individuals’ use of telerehabilitation. PT had significant positive impacts on BI of using telerehabilitation, suggesting that PT was the primary issue for individuals using telerehabilitation, and that the users were more likely to continue to use telerehabilitation only if they have trust. Meanwhile, our finding showed that PE and EE had significant positive impacts on PT, which was found in other research [39]. Telerehabilitation is a new technology that has not yet been widely used in China and patients and their caregivers lack sufficient understanding about it. Therefore, it is difficult for users to have trust in telerehabilitation directly. If users believe that telerehabilitation can meet their expectations of their actual rehabilitation needs or it is easy to use, however, they would have initial trust in it and thus more willing to using telerehabilitation.

Consistent with the findings of previous studies [28], our study found that PE and SI had a significant positive impact on BI using telerehabilitation. Evidence showed that usefulness is a key facilitator for the use of digital health technologies in clinical practice [45, 55, 56]. PE is the basis for the implementation and development of telerehabilitation, and only if individuals perceive that telerehabilitation can meet their actual rehabilitation needs and benefit their health, their intentions to use telerehabilitation will be strong. In addition, telemedicine is still in the early stages of development in China, with recommendations from family members, patients, and medical staff being the main ways in which individuals become known to telemedicine services [57]. A previous survey on the use of telemedicine by Chinese diabetic patients showed that among half of those using diabetes apps, the reason for them to use these apps were recommendation from other patients or doctors. Li et al. [58] showed that the interaction between older people and their communication groups is one of the most important factors in promoting the use of telemedicine, and that when people around them (e.g., friends, family members, and medical staff) mention or use telemedicine often, older people would be significantly influenced to increase their subjective acceptance of the telemedicine technology, which in turn creates a tendency to try it out. Thus, social groups often play a crucial role in promoting users’ adoption of telemedicine services [59]. Moreover, consistent with the findings of Liu et al. (2022), our results indicate that self-efficacy has a direct positive effect on BI, and can also indirectly affect BI with PE as the mediator. This suggests that the stronger the user’s self-efficacy, the stronger perception of the usefulness of telerehabilitation, and the stronger intention to use telerehabilitation.

It is worth noting that our findings show that EE, FC, and PR have no significant effect on the intention to use telerehabilitation. In terms of EE, some previous studies on intention to use new medical technologies have also found no association between EE and BI, which is consistent with our findings [60, 61]. However, Cimperman et al. [62] and Wang et al. [63] suggested that EE was a determinant factor influencing patients’ use of telemedicine. This difference may be related to the ease of use of the different tele-technologies. Van et al. [64] noted that the effect of EE on using intentions will become less important with time as people continue to use and are familiar with new technological systems. In addition, our results also indicate that EE positively influence individuals trust in telerehabilitation. Overall, EE is not a key factor influencing using intentions, but mastery and proficiency in the use of technological systems associated with telerehabilitation will promote individual’s trust in telerehabilitation, which would further facilitate the individual’s use of telerehabilitation.

Regarding the FC, Zhang et al. [28] and Zhu et al. [59] concluded that FC had positively influence intention to use. The essence of telemedicine is the application of advanced communication technologies in the medical field, so the individuals first need to have the infrastructure and technical support of these advanced telecommunication technologies to access the convenience of telemedicine. However, our finding indicated that the effect of FC on patients’ intention of using telerehabilitation was insignificant, with this result being consistent with the finding of Yuan et al. [65] and Suwannapusit et al. [66]. This non-significant result may be explained by contextual factors. First, it is important to note that most telerehabilitation services currently available in China are primarily led by large public hospitals, often functioning as simple extensions of in-person rehabilitation care. In such settings, users are typically required to follow structured guidance from healthcare providers throughout the tele-rehabilitation process, with limited need for independent navigation of complex platforms. Consequently, users may not yet have developed a strong awareness of the importance of facilitating conditions such as ongoing technical support, training, or platform-specific guidance because these elements are often implicitly managed by the healthcare institution. As China’s rehabilitation medical service system continues its transformation toward community and home-based models, and as telerehabilitation services become more diversified, personalized, and technically complex, users will likely encounter new challenges that require more robust facilitating conditions. Second, the widespread adoption of smart mobile devices (e.g., smartphones, tablets) and the high penetration of internet infrastructure in daily life have contributed to generally good information literacy and basic network access among the population. Moreover, telerehabilitation services in China are currently concentrated in relatively economically developed regions, where digital literacy and infrastructure are typically higher. These contextual factors may have further diminished the perceived relevance of FC in our sample. However, this does not imply that FC is unimportant. On the contrary, in resource-limited settings or among populations with lower digital health literacy, the availability of clear user guidance, technical assistance, and accessible training could be critical determinants of adoption. Therefore, we believe the non-significance of FC in this study should be interpreted as a reflection of the early-stage characteristics of telerehabilitation implementation and the specific profile of our sample, rather than evidence of its irrelevance.

In our present study, PR shows no significant effect on the intention to use telerehabilitation, which was contrary to our hypothesis but consistent with previous telemedicine studies in China [24, 63]. This non-significant finding may be attributed to the specific context of telerehabilitation implementation in China. Currently, most telerehabilitation services are led by large public hospitals, which carry strong credibility and may buffer users’ risk perceptions through institutional trust. Additionally, telerehabilitation in China remains at an early developmental stage, with services often serving as simple extensions of in-person rehabilitation care where users are passively guided by healthcare providers. The limited prior experience among participants (only 11%) further suggests that most respondents lacked direct exposure to potential risks, making concrete risk perceptions less likely to have formed. As telerehabilitation services become more diverse, commercially driven, and technically complex, and as public awareness of healthcare data privacy grows alongside evolving legal regulations, the salience of risk perceptions may increase significantly. Future research should re-examine the role of PR in more mature telerehabilitation contexts and among populations with greater experience and digital health literacy.

Limitations

There are several limitations of this study. Firstly, the subjects of this study were individuals with rehabilitation needs from the Department of Rehabilitation Medicine of a level-tertiary hospital in China. Whereas in China, there are more individuals with rehabilitation needs in lower-level hospitals or community hospitals, and they tend to have less knowledge about telerehabilitation, so our sample is restricted by geography and resources, and may be subject to sample bias. Consequently, the findings may not be generalizable to all populations with rehabilitation needs, particularly those in under-resourced settings. This limitation is particularly salient given the ongoing transformation of China’s rehabilitation service system toward community and home-based models, which will increasingly serve populations with different characteristics and service expectations than those captured in our tertiary hospital sample.

In addition, only 11.0% of the participants in this survey had experience in the use of telerehabilitation, and most of them may not have a well understanding of telerehabilitation, which may affect the accuracy of the results in this case. Given the low proportion of experienced users, our results primarily reflect the intentions of potential users rather than actual adopters. As policy initiatives continue to promote the integration of rehabilitation services across different levels of the healthcare system and as telerehabilitation becomes more widely implemented, users’ experience and perceptions may evolve. Future research should include more diverse samples across different geographic regions, healthcare settings (including community health centers and primary care facilities), and user experience levels to enhance the generalizability of the findings and to capture the evolving dynamics of telerehabilitation adoption as China’s rehabilitation service transformation progresses.

Conclusions

This paper applied the extended UTAUT model to determine the factors influencing the usage intention of telerehabilitation. The results indicate that PE, SI, PT, SE are important factors influencing the individual intention to use telerehabilitation; PE and EE have a positive effect on PT; SE has a significant positive impact on PT and indirectly affects intention to use by PE; furthermore, PE and EE are mediated by PT as a mediator that indirectly effect on BI of using telerehabilitation.

Abbreviations

AVE

Average Variance Extracted

BI

Behavioral intention

CR

Combined reliability

EE

Effort expectancy

FC

Facilitating conditions

PE

Performance expectancy

PLS-SEM

Partial least squares structural equation modeling

PR

Particularly perceived risk

PT

Perceived trust

SE

Self-efficacy

SI

Social influence

UTAUT

Unified Theory of Acceptance and Use of Technology

VIF

The variance inflation factor

Authors’ contributions

Conceptualization: Y-Q W and HC; Methodology: HC, SC, YZ, and Y-Q W; Formal analysis and investigation: SC, YZ and HC; Writing-original draft preparation: Y-Q W, HC, and C-M W; Writing - review and editing: all authors; Funding acquisition: C-M W; Supervision: C-M W.

Funding

This work was supported by the Humanities and Social Sciences Project of Jiangxi Colleges and Universities (Grant No. JC21204).

Data availability

The questionnaire data of this study can be available from the corresponding author email.

Declarations

Ethics approval and consent to participate

This study received approval from the Ethics Committee of the First Affiliated Hospital of Gannan Medical University. Informed consent was obtained from all participants prior to data collection. The purpose and procedures of the study were explained in detail, ensuring participants’ comprehensive understanding of data usage and their involvement. Participation was entirely voluntary, and all methods adhered to relevant guidelines and regulations.

Consent for publication

Written and oral consent was obtained from all individuals involved in this study during the data collection period of January to May 2023.

Competing interests

The authors declare no competing interests.