Acceptance of a patient portal: Analysis of an Italian regional public system

Abstract

Background

Patient portals allow patients to access their health information and interact with their healthcare system. While their use is still limited, this article explores the behavioral intention to use a patient portal implemented by a public regional authority in Italy. The authors also investigate the role of sociodemographic moderators - age, gender, education, and occupation - on the intention to use the portal.

Methods

While most of the literature on patient portals is focused on small sets of respondents or is verticalized on specific diseases, this paper describes the results of a survey sent to 34,256 users registered on the patient portal. Of these, 15,102 users answered the questionnaire. The survey explored the acceptance of the patient portal through the extended unified theory of acceptance and use of technology model (UTAUT2). Descriptive and SEM analyses were also conducted.

Results

The model has good explanatory power for the behavioral intention to use a patient portal. One construct of the UTAUT2 model showed insignificant effects on the intention to use. The results indicate that the impact of the constructs affecting the intention to use the patient portal is significantly moderated by individuals’ sociodemographic characteristics.

Conclusions

The analysis results confirm a good acceptance of patient portals provided by public authorities. This supports the idea that public institutions can also develop innovative tools. The results confirm the desirability of these tools among citizens. The results have important policy implications for public health investments.

Keywords

patient portal technology acceptance UTAUT2 health digitalization public health e-health

Background

Digital transformation is changing the way healthcare services are provided. This move has a growing impact on citizens’ behaviors and the healthcare system. Digital technologies are changing how citizens access health data, stimulating more integrated and less time-consuming interactions between users and healthcare providers.¹ Among the many digital innovations in healthcare over the last decade, patient portals have arisen as a new digital communication channel between actors of a healthcare system.^2,3 While the first patient portals were applications allowing patients to access health information electronically, they evolved into complex systems encompassing a whole set of online services that radically innovate many processes between citizens and healthcare service providers.⁴

Patient portals have grown fast, and the market potential of telemedicine has been demonstrated to be strong and is expected to grow at a compound annual growth rate of 14% in the last few years.^5,6 However, their use is subject to new challenges. The literature shows that many sociodemographic characteristics may facilitate or hinder the acceptance of digital platforms.^7,8 Characteristics such as education and gender might affect the intention to use a patient portal.^9,10 Image, identity, and reputation may also play a role in accepting digital platforms.¹¹ Ethnicity, age, income, and residence were also found to affect portal usage.¹² Other non-demographic variables affecting portal adoption are security concerns about unauthorized access¹³ and the system’s usability.¹⁴

The current literature has some limitations. Most studies have concentrated on vertical aspects, such as particular actors or diseases. For example, some studies focus on apps designed for specific patient groups (e.g., those targeting underrepresented depressed patients) or apps aimed at professionals (nurses, doctors, and other staff).^15–18

While patient portals are increasingly promoted as tools for enhancing access, transparency, and efficiency in public healthcare systems, few studies have examined the factors influencing their adoption at scale, particularly when these systems are made universally available to an entire population.¹⁹ This gap in the literature limits our ability to generalize findings beyond pilot projects or narrowly defined user groups. As recent research has emphasized, further investigation is needed to understand how patient portals function as public services, how they are received across diverse population segments, and what factors influence their sustained use.^2,20 The present study contributes to this area by offering empirical evidence on the differential acceptance of a region-wide public patient portal, with implications for designing, governance, and evaluating large-scale eHealth initiatives. Its findings are thus relevant not only to scholars interested in technology acceptance and digital health but also to practitioners and policymakers engaged in implementing inclusive, population-level digital services.

The key research question of this research is as follows: What sociodemographic moderators influence the acceptance of the use of a public patient portal? The extended version of the unified theory of acceptance and use of technology (UTAUT2) model was used on a theoretical basis. The research question was conducted with users of a portal implemented in a regional public health system.^20,21

This paper is organized as follows. The next subsection introduces the institutional setting of the study. Section 3 introduces the theoretical background and research hypotheses. Section 4 describes the methods. Section 5 describes the results of the analysis. Sections 6 and 7 discuss the results and conclude this work.

Institutional setting

In Italy, public health governance is entrusted to local institutions (regions and provinces). Since 2008, the local government of a province in Northern Italy has fostered a series of experiments that gradually became a patient portal.²² This process also anticipated national laws regarding digital health infrastructures according to national and international standards. It originally started as a personal health record system and evolved into a more complex system. It is possible to identify the following three phases.

Personal health record phase

The original system was initially designed around 2010 to integrate data produced by healthcare institutions (e.g., medical reports and laboratory analyses) and data that only the patient knows, comprising so-called observations of daily living (e.g., weight, blood pressure, symptoms, and over-the-counter medications), the family’s clinical history, and medical data not stored in the database of the local health authority (e.g., paper-based reports and reports produced by unconnected healthcare facilities).

Patient portal phase

Since 2014, the system has increased its functions to become a patient portal. The novelty of the portal was that it allowed one to consult the (simplest) results of laboratory tests within 4–5 business hours. During this phase, the number of users of the portal was about 61,409.

Multiplatform phase

Thanks also to the findings of this survey, the portal was re-engineered into a new multiplatform system with new services. Among others, the added services are: payments, digital administrative procedures such as the change of the general practitioner, one’s medical record with allergies and vaccines.²²

This new system comprises an app in line with recent developments and the growing interest in e-health.⁸ In late 2024, over 200,000 users were registered on the service out of a total population of about 500,000 in the region.

Theoretical background and hypotheses development

The literature on the factors influencing people’s acceptance and use of technology employs many theoretical models. Among others, UTAUT is one of the most recognized models for studying the acceptance of technologies by users.^3,23

The Unified Theory of Acceptance and Use of Technology (UTAUT) is a model that explains how factors like performance expectancy, effort expectancy, social influence, and facilitating conditions influence individuals' intention to adopt and use technology.

UTAUT2 is an extension of the original UTAUT model, incorporating three additional constructs: hedonic motivation, price value, and habit, to better account for consumer technology adoption in personal settings. While UTAUT primarily focuses on organizational contexts, UTAUT2 adapts the model to more accurately reflect the factors influencing individual users’ decisions to accept and use technology in everyday life. The extended version, UTAUT2, has shown stronger predictive power in explaining technology usage behavior.^24,25

Within this literature, acceptance of technology has been studied as the behavioral intention to use various innovative technology products and services.²³ Behavioral intention is used as a predictor of actual technology usage behavior. In other words, it reflects how likely an individual is to adopt or continue using a particular technology based on their attitudes, beliefs, and perceptions.

Following the studies that adopted this literature in the healthcare sector,^26,27 we analyzed the acceptance of a patient portal by adopting UTAUT2. UTAUT2 predicts behavioral intention through seven constructs: performance expectancy, effort expectancy, social influence, facilitating conditions, use habit, price value, and hedonic motivation.²³

For this work, we did not consider all the constructs of the original UTAUT2 model (Figure 1). Price value was not considered because the users do not pay for the service. Habit is not significant, as the patient portal is used sporadically rather than daily. On the other hand, we added a construct concerning security perception, which is connected to the perception of the privacy and confidentiality of data managed by an electronic system. Previous studies have defined perceived security as the perception of the privacy and confidentiality of personal data (Dünnebeil et al., 2012.^28,29; This element was introduced given the growing perception of the criticality of data security and in light of European regulations introduced in 2016, specifically the General Data Protection Regulation.

Figure 1.

The UTAUT2 model used in this work.

In the following sections, the authors describe the different variables and the associated hypotheses of the model.

Performance expectancy refers to the degree to which citizens who use technology perceive a benefit from using it (e.g., downloading medical reports). Usefulness is measured based on time savings, convenience, and ease. The hypothesis is that citizens perceive the benefits of using the patient portal for specific healthcare-related activities.

Performance expectancy has a significant and positive relationship with behavioral intention.

Effort expectancy concerns the degree of ease associated with using the system, which is strongly connected to the degree to which a person believes that using the patient portal would require less effort than traditional channels. The hypothesis is that the patient portal will ease the performance of certain healthcare-related activities.

Effort expectancy has a significant and positive relationship with behavioral intention.

Social influence is the degree to which an individual perceives it important that others believe he/she should use the patient portal. The hypothesis is that the influence of significant others may play a role in encouraging the use of the patient portal for health-related activities.

Social influence has a significant and positive relationship with behavioral intention.

Facilitating conditions are related to the degree to which an individual believes that an organization or technical infrastructure exists to support the use of the system. The hypothesis is that citizens will be more inclined to use the patient portal when facilitation conditions, such as the presence of a contact center, are in place.

Facilitating conditions significantly and positively influence behavioral intention.

Hedonic motivation is linked to the motivation to do something due to internal satisfaction. In this case, it is hypothesized that hedonic motivation is related to pleasantness when using the patient portal.

Hedonic motivation significantly and positively affects behavioral intention.

Perceived security, although not present in the original model, proved to be an important factor affecting the acceptance of a system that manages personal data. Security and privacy are two elements that have gained growing attention over the last few decades and are in line with a growing digital environment. The hypothesis is that security concerns may hinder patient portal use.

Perceived risk significantly and positively affects behavioral intention.

Moderators

In the UTAUT model, a moderator is a variable that changes the strength of the relationship between an independent and a dependent construct. Age, gender, and experience are the original moderators used in the UTAUT2 model.²³ While for age and gender, the authors used the traditional categories, experience was dropped as a moderator. The questionnaire was sent to specific users, namely those who accessed the system at least three times a year. This aligns with recent literature, which indicates that the threshold of 3–5 accesses per year represents the median usage of electronic health record systems.³⁰ The authors thus considered the sample of respondents to be already experienced.

On the other hand, two more moderators were added: occupation and education.^31,32 These were investigated to account for disparities between citizens and their knowledge/job background. The literature shows that acceptance of health technologies increases with the level of education and the type of occupation (Or & Karsh, 2009.^33–35 Figure 1 shows the six predictors of the model and the moderators used.

Methods

To verify the hypotheses formulated with the theoretical model presented above, a questionnaire was constructed and administered by computer-assisted web interviewing (CAWI), a data collection methodology based on completing a web questionnaire provided through a link, panel, or website. The questionnaire consisted of two sections. The first section addressed the UTAUT2 model. This was investigated by 16 questions in the form of statements to which the respondents answered through a four-point Likert scale, with the anchors being “strongly disagree” and “strongly agree.” The four-point scale was chosen to eliminate neutrality and encourage participants to express more defined opinions.³⁶ Table 1 illustrates the questions asked for each individual construct. The second section aimed to gather data about the sociodemographic profiles of the respondents.

Table 1.

UTAUT2 variables and questions.

Variable	Questions
Performance expectancy (PE)	PE1. The patient portal is useful in my everyday life PE2. The patient portal system helps me to manage my health PE3. The patient portal system allows me to save time
Effort expectancy (EE)	EE1. Learning how to use the patient portal is easy for me EE2. After the first accesses, the patient portal is easy to use
Social influence (SI)	SI1. My relatives/friends think that it is useful for me to use the patient portal SI2: The medical staff is in favor of using the patient portal
Facilitating conditions (FC)	FC1. I Have the technological resources to use the patient portal FC2. I Have the knowledge necessary to use the patient portal FC3. I Can get the necessary help when I have difficulties using the patient portal
Hedonic motivation (HM)	HM1. Using the patient portal is enjoyable HM2: In terms of content, organization, and graphics, the patient portal is overall satisfactory
Perceived security (PS)	PS1. I Believe the patient portal protects the confidentiality of my personal data PS2: I Am concerned about the possibility of making errors when entering data in the patient portal
Behavioral intention (BI)	BI1. I Intend to continue using the patient portal in the future BI2. I Would like to continue using the patient portal more and more

At the time of the investigation, 61,409 users were registered on the patient portal. Out of a total of 500,000 citizens in the entire region, this is a significant number, especially considering that adults can also have a delegation of use for minors and non-self-sufficient individuals. Among the 61,409 individuals, the questionnaire was sent by email to those people who accessed the system at least three times since 2014. The questionnaire was submitted in the fourth quarter of 2016.

A total of 34,256 users out of 61,409 (55.78% of registered users) were invited to answer the questionnaire. Of these, 17,703 users (51.7%) viewed the web questionnaire and 15,102 users (44%) completed the questionnaire (85% of those who clicked on the link).

The UTAUT2 model (Figure 1) was empirically validated by estimating a structural equation model (SEM), as it is the proper methodology for assessing causal relationships among latent constructs.⁸ Essentially, SEM can be estimated using the covariance-based method (CB-SEM), which is the appropriate approach if the aim is to confirm theoretical propositions, or the partial least squares method, which is more appropriate for prediction and theory development.³⁷ According to the objective of this research, we opted for CB-SEM. The evaluation of a CB-SEM model is conducted in two distinct stages: (1) the outer model (or measurement model) evaluation and (2) the inner model (or structural model) evaluation.

In the outer model evaluation, the focus is on the relationship between the latent constructs of the UTAUT2 framework—such as performance expectancy, effort expectancy, and social influence—and the observable survey items (also referred to as measurement variables) used to measure them. This step assesses whether the measurement variables reliably and validly represent their corresponding latent constructs, ensuring that the model’s indicators accurately capture the theoretical concepts.

The inner model evaluation involves examining the relationships between the latent constructs that correspond to the theoretical paths in the model. This stage tests the structural relationships proposed in the UTAUT2 framework and evaluates the research hypotheses. For instance, in UTAUT2, the inner model assesses how performance expectancy, effort expectancy, and other constructs influence behavioral intention and actual technology use.

By systematically evaluating both the measurement and structural components, CB-SEM can provide the empirical evidence needed to validate the theoretical relationships and overall framework of UTAUT2.

Results

Descriptive statistics

We compared the population using the patient portal with the demographic and structural population assisted by the regional health system. The data shows a clear division of the users by gender, with women slightly overrepresented with respect to the population (Figure 2). Regarding occupation, the white-collar middle class is over-represented in using the Patient Portal, particularly at the expense of the working class (Figure 3). Figure 4 compares users’ ages with the reference population. Young individuals are underrepresented in this study compared to other groups.

Figure 2.

Men and women users’ distribution compared with the reference population.

Figure 3.

Users’ occupation distribution compared to the reference population (%).

Figure 4.

Users’ age compared with the reference population (%).

CB-SEM analysis

Goodness-of-fit

Table 2 reports some of the main goodness-of-fit indices. Their values establish that the SEM assessing the UTAUT2 model fits the data sufficiently well in absolute terms.

Table 2.

Goodness-of-fit indices.

Model	$χ^{2}$	RMSEA	CFI	TLI	SRMR	AIC	BIC
UTAUT2	8028.890^***	0.080	0.916	0.878	0.061	411,756.552	412,160.549
Alternative model 1	14221.395^***	0.095	0.850	0.827	0.142	417,907.057	418,150.980
Alternative model 2	25368.365^***	0.127	0.732	0.691	0.074	429,054.028	429,297.950

^***p-value < 0.001.

^**p-value < 0.01.

^*p-value < 0.05

Outer model

The evaluation of the outer model in the SEM framework is fundamental, as it constitutes the logical basis for defining latent constructs. Table 3 reports the main results of the measurement model: factor loadings (along with their standard errors), standardized estimates, p-values, and coefficients of determination (R2). In the context of the outer model, R2 indicates the proportion of variability explained by the latent construct for a given observed item. All constructs have at least one observed item with a standardized loading greater than 4, thus indicating that the outer model is valid as a whole. However, three items (PE3, SI2, and PS1) load poorly on their factors; hence, they have been removed from the outer model to improve the reliability of performance expectancy, social influence, and perceived security constructs.

Table 3.

Estimates of loadings of the outer model.

Latent variable	Measurement variable	Loading	Standardized estimate	$R^{2}$
PE	PE1	1.000 (0.000)	0.502	0.524
	PE2	1.087^*** (0.014)	0.545	0.549
	PE3	0.697^*** (0.010)	0.350	0.407
EE	EE1	1.000 (0.000)	0.674	0.714
EE	EE2	0.968^*** (0.009)	0.653	0.793
SI	SI1	1.000 (0.000)	0.513	0.403
SI	SI2	0.449^*** (0.012)	0.230	0.139
FC	FC1	1.000 (0.000)	0.510	0.710
	FC2	1.069^*** (0.011)	0.545	0.676
	FC3	0.839^*** (0.013)	0.428	0.281
HM	HM1	1.000 (0.000)	0.584	0.638
HM	HM2	0.763^*** (0.009)	0.445	0.510
PS	PS1	1.000 (0.000)	0.097	0.013
PS	PS2	6.430^*** (0.919)	0.621	0.946
BI	BI1	1.000 (0.000)	0.435	0.746
BI	BI2	1.003^*** (0.013)	0.436	0.494

p-value < 0.001.

p-value < 0.01.

p-value < 0.05.

Inner model

The structural part of the SEM model allows us to assess the system of associations between the latent constructs and, in particular, how performance expectancy, effort expectancy, facilitating conditions, hedonic motivation, perceived security, and social influence affect behavioral intention. Table 4 reports the main results: path coefficient estimates (along with their standard errors), standardized estimates, p-values, and coefficient of determination (R²).

Table 4.

Estimates of the structural model.

Endogenous variable	Exogenous variable	Estimate	Standardized estimate
Behavioral intention	Performance expectancy	0.169^*** (0.010)	0.208
	Effort expectancy	0.004 (0.010)	0.006
	Social influence	0.016^** (0.005)	0.029
	Facilitating conditions	0.338^** (0.010)	0.395
	Hedonic motivation	0.193^** (0.017)	0.259
	Perceived security	0.041^** (0.006)	0.060
	$R^{2}$	0.597

^***p-value < 0.001.

^**p-value < 0.01.

^*p-value < 0.05.

The inner model explains 59.7% of the variance in the intention to use the patient portal (behavioral intention).

As summarized in Table 5, all the constructs except effort expectancy had a statistically significant effect on the intention to use the patient portal. The positive and significant coefficients suggest that higher levels of performance expectancy, social influence, facilitating conditions, hedonic motivation, and perceived security are associated with a stronger intention to use the portal.

Table 5.

UTAUT2 validation model.

Structural model	Significant	Hypothesis confirmed
H1: PE → (+) BI	OK	YES
H2: EE → (+) BI	NO	--
H3: SI → (+) BI	OK	YES
H4: FC → (+) BI	OK	YES
H5: HM → (+) BI	OK	YES
H6: PS → (+) BI	OK	YES

Performance expectancy, facilitating conditions, and hedonistic motivation are the most relevant constructs of the behavioral intention to use the patient portal. As reported in Table 1, performance expectancy is represented by the perceived usefulness of the patient portal (PE1), by the perceived improvements in health management (PE2), and by the perceived time savings offered by the portal (PE3). The relevant role of performance expectancy in the prediction of behavioral intention is in line with results from several other studies related to different digital services.^23,32

According to Table 1, facilitating conditions are linked to the availability of the technological resources necessary to use the patient portal (FC1), the knowledge necessary to use it (FC2), and the availability of help/support structures (FC3). In the present study, a significant influence of facilitating conditions on behavioral intention emerges. Similar to other studies, this suggests that respondents consider resources and support to be essential for this kind of service.³⁸

According to Table 1, hedonic motivation (HM) is conceptualized as the internal satisfaction and enjoyment derived from using the portal, captured through the respondents’ perceptions of the portal’s pleasantness and overall design quality (HM1 and HM2). In our case, the results indicate that hedonic motivation has a significant and positive relationship with Behavioral Intention, suggesting that users find the experience of interacting with the portal not only functional but also gratifying.

Regarding perceived security, represented by sentences PS1 and PS2 (Table 1), the results lead us to think that citizens are positively interested in the system’s functionality and that they do not perceive problems with the system’s security. Social influence (Beta = 0.016) is a low-impact indicator of behavioral intention. The portal is built on the idea of personal use, through which people find answers to their need for timely and updated clinical data.

In predicting behavioral intention, no importance appears to be devoted to effort expectancy. Effort expectancy reflects the perceived ease of use in the two different components: ease of learning how to use the system at the beginning and ease of use after the first period of use. In other studies, effort expectancy is important in promoting acceptance, especially in the early stages of service introduction.²³ The lack of effort in the present study can be ascribed to the characteristics of the CAWI sample, which is composed of people skilled in technology use and who utilize Internet services in their daily lives.

Role of moderators

We also investigated the role of four moderators in the UTAUT2 model—gender, age, occupation, and education — by including these variables in the outer model, along with their interactions with the exogenous latent constructs (Table 6). Concerning age, the questionnaire distinguished eight categories of age from the category “less than 18 years old” to the category “more than 76 years old.” We stratified the sample into three age categories referring, respectively, to young, middle age, and elderly people: “less than 36 years old” (millennial and newer generations), “36–55 years old” (generation X), and “more than 55 years old” (boomers).

Table 6.

Effects of the moderators on the relationships between each exogenous construct and Behavioral Intention.

Moderator		PE	EE	SI	FC	HM	PS
Gender	Female	−0.042	-	-	-	-	-
Gender	Male	0.170	−0.310	-	-	-	-
Age	Millennial	−0.042	-	0.653	-	−0.035	0.579
	Generation X	−0.042	0.617	0.265	0.217	−0.035	0.160
	Boomer	0.133	0.475	0.197	0.273	0.241	0.030
Education	Low. Secondary	−0.042	-	0.653	-	−0.035	0.579
	Upp. Secondary	−0.042	−0.356	0.438	0.420	−0.035	0.368
	Tertiary	0.170	−0.336	0.385	0.270	0.257	0.355
Occupation	Unemployed	-	-	0.653	-	-	-
Occupation	Employed	-	0.102	0.456	-	-	-

Concerning the occupation moderator, the questionnaire distinguished six categories: employed, unemployed or laid off, retired, homemaker, invalid, and student. For the purpose of the analysis, due to the particularity of the sample using the portal, with only 64% of respondents employed, the categories were grouped into two macro-categories: employed and non-employed.

Finally, the education moderator was distinguished into four categories: category 1 “primary school or no title,” category 2 “lower secondary school or professional title,” category 3 “high school,” and category 4 “degree or post degree.” The previous categories one and two were grouped together due to the lower number of people responding to category 1 (less than 500).

The results of the estimation of the model, including the interactions between moderators and exogenous constructs, are shown in Table A1 in the Appendix. Table 6 summarizes the significant effects of moderators on the relationship between behavioral intention and each of the other constructs according to the significant estimated parameters of the interaction terms. In particular, it reports the standardized path coefficient for the different categories of the moderator variables, computed by summing the estimated directs of the exogenous constructs with the coefficients of the corresponding interaction terms characterized by a value greater than 0.1 and a p-value < 0.05. These results provide evidence that the intensity of the determinants of the intention to use the platform is not homogenous among all patients. In particular, gender, age, education, and occupation significantly shape the determinants of portal acceptance. Males show a stronger positive association between Performance Expectancy (PE) and Behavioral intention (BI) than females but are also more negatively affected by Effort Expectancy (EE), suggesting heightened sensitivity to both perceived usefulness and usability barriers. Age also plays a crucial role: younger users (Millennials) are more responsive to Social Influence (SI) and Perceived Security (PS), while older users (Boomers) respond more uniformly to all constructs, especially PE, EE, Facilitating Conditions (FC), and Hedonic Motivation (HM). Educational background further moderates these effects. More educated users are influenced primarily by PE and HM, whereas users with lower levels of education are more reliant on SI and PS. Employment status also differentiates user responses: unemployed individuals are more strongly influenced by SI, indicating the importance of social endorsement, whereas employed users show a stronger response to EE. These results confirm that the intensity and direction of key determinants of behavioral intention vary across population subgroups. Consequently, policies and communication strategies aimed at increasing the adoption of digital health tools should be tailored to specific user profiles, accounting for generational, educational, and occupational differences to ensure inclusive and effective engagement with patient portals.

Discussion

The analysis shows that all constructs except effort expectancy have a significant relationship with behavioral intention. The non-significance of effort expectancy in this study compared to previous research^10,39 may be due to the characteristics of the sample, which is predominantly composed of digitally literate individuals.

The main contributors to behavioral intention are performance expectancy, facilitating conditions, and hedonic motivation. These findings highlight a predominantly utilitarian approach to system adoption (Schomakers et al., 2022), reinforcing the idea that users engage with the patient portal as a functional tool rather than as an optional convenience. Given that health systems are generally viewed as essential public services, our findings suggest that users recognize these constructs as key drivers of engagement. The portal is an innovative public service aligned with healthcare needs.⁴⁰

Moderating variables such as gender, age, education, and occupation offer a more nuanced understanding of digital health adoption. Among these, social influence and perceived security emerged as consistent and significant factors across demographic subgroups, underscoring their relevance in shaping behavioral intention. The data suggest that these two constructs play a particularly important role in reinforcing the legitimacy and trustworthiness of digital platforms, which is critical in healthcare settings. In this light, perceived effort acts as a deterrent for male users, potentially linked to lower usability expectations or reduced perceived benefits.

The impact of age on acceptance varies between generational groups. Millennials show a strong relationship between social influence and perceived security over behavioral intention. This suggests that their effect on behavioral intention is driven by social endorsement and security considerations. Given their familiarity with digital technologies, they perceive the system as a seamless extension of existing online tools, with minimal concerns about trust and security. Generation X primarily influences effort expectancy, with social influence and facilitating conditions playing a secondary role. This suggests that this group interacts with usability and accessibility.⁴¹ A well-designed interface and user-friendly features are critical drivers of adoption. Boomers are associated with effort expectancy, while facilitating conditions and hedonic motivation have a lower influence. This suggests that while they still rely on traditional healthcare interactions, they also recognize the benefits of digital access. The role of hedonic motivation implies an alternative yet convenient method of performing routine healthcare tasks. Education as a moderator shows that more educated users interact mainly with social influence, facilitating conditions, hedonic motivation, and perceived security constructs over behavioral intention. This group seems to be an optimal target for adoption of patient portals. However, potential friction in effort expectancy needs to be addressed to maximize engagement through an efficient user interface. The correlation between hedonic motivation and tertiary education reinforces the notion that for this group, the portal is a practical tool and an enhanced, convenient experience for managing healthcare services. Employment status exhibits a minimal direct influence on behavioral intention. However, social influence affects unemployed users at a higher level, suggesting that peer recommendations and suggestions from healthcare professionals may play a role in their adoption decisions. At the same time, the good interaction between employed users and social influence indicates that workplace-driven incentives or promotional efforts could be an effective strategy for increasing adoption among the people of this group.

Conclusions

This study provides empirically grounded insights into the behavioral intentions of users engaging with a large-scale public patient portal. As such, the portal represents a critical node in the ongoing digital transformation of publicly funded healthcare systems.

Applying the UTAUT model, the analysis confirms that performance expectancy, social influence, and facilitating conditions are statistically significant predictors of users’ behavioral intention to adopt the platform. Notably, effort expectancy—widely acknowledged in the literature as a determinant of technology adoption (e.g.,^10,39—was found to be non-significant in this context. This divergence may be attributed to the user sample’s digital competence, who perceive the system as sufficiently intuitive. This finding contributes to a growing body of research suggesting that as digital literacy increases, the relative weight of usability-related constructs may diminish, shifting attention toward motivational and social drivers.

Moderator - particularly gender, age, and education - enables a more nuanced interpretation of adoption dynamics. The findings indicate that older adults with higher educational attainment exhibit a stronger intention to use the system, highlighting the portal’s potential to empower users traditionally viewed as late adopters. Furthermore, gender appears to moderate the relationship between effort expectancy and behavioral intention: for male users, perceived effort represents a more substantial barrier, potentially indicating differentiated expectations or engagement patterns. These insights underscore the need for gender- and age-sensitive approaches in interface design and digital health communication strategies.

The study thus reinforces the value of integrating sociodemographic moderators into technology acceptance models, particularly in the context of universalist digital health interventions. It also points to the importance of accounting for heterogeneity in user capabilities and expectations, particularly as digital health infrastructures scale and mature. In line with the recommendations of Ravangard et al.,⁴² we advocate for future longitudinal studies that trace the evolution of user engagement across different stages of technological deployment and diverse population subgroups.⁴³

The results further underscore the importance of adopting differentiated strategies to foster digital health adoption. Policy-makers and designers should tailor interventions:

• Highlighting usefulness and satisfaction for educated and employed users;

• Emphasizing security and social endorsement for younger or less-educated populations;

• Providing assistance and training for older users who require support on multiple dimensions.

The findings also emphasize that a “one-size-fits-all” approach risks ignoring key nuances determining user engagement in public digital health services. Tailored communication, usability design, and support infrastructure should reflect the socio-demographic subgroup-specific sensitivities to maximize uptake and sustained use of patient portals.

In conclusion, this research contributes to an emerging understanding of patient portals as complex socio-technical systems whose success hinges on technical affordances and their capacity to accommodate differentiated user needs and expectations. Especially within publicly funded systems, attention must be paid to the interplay between user characteristics, perceived system value, and contextual factors. The moderation effects suggest that effective digital health strategies must combine robust system design with inclusive outreach, adaptive governance, and targeted user engagement mechanisms to ensure equitable and sustainable adoption.

Limitations and future work

This study has some limitations. First, the study was conducted in a small and culturally homogeneous territory. The results may differ in more complex areas. More research is needed on the moderators' role to better understand the interactions between various predictors and acceptance in systems delivered by different players. Some limitations result from the self-selection of users from the population. Profiling users in different contexts and portals can help to observe specific interactions in the behavioral intention to use the system. Another limitation derives from the fact that this survey predates the pandemic and that the remote use of health services has generally improved and intensified, even if there are signs of a return to traditional ways of communicating with healthcare facilities due to the resilience of organizations.

While this quantitative work gives insights for the development of digital tools in the healthcare sector, further research is needed to understand better the point of view of different stakeholders of such tools.⁴⁴

Footnotes

Acknowledgment

The authors wish to thank the important collaboration of the university that provided access to the server of the survey software and supervised the technical process of sending the questionnaire.

ORCID iDs

Alberto Zanutto

Diego Ponte

Diego Giuliani

Ethical approval

All methods were carried out in accordance with relevant guidelines and regulations. Ethics approval was obtained from the local health authority.

Consent to participate

Informed consent was obtained from all the respondents.

Author contributions

AZ developed the original research idea, prepared the survey online, disseminated it, and collected the data. DP wrote Sections 1 and 2. DG and MMD performed the analysis of the data. CB-SEM was conducted using the Lavaan package in R 4.2.2. All the authors interpreted and discussed the results. DG, MMD, and DP wrote Section 3. All authors wrote Sections 4 and 5. All the authors reviewed and approved the submitted version.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Datasets and output files of the data analysis are available upon request from the corresponding author.

Appendix

Table A1.

Estimates of structural model with moderators and interaction effects.

Endogenous variable	Exogeneous variable	Estimate	Standardized estimate
BI	PE	−0.035^** (0.012)	−0.042
	EE	−0.018 (0.017)	−0.029
	SI	0.354^*** (0.021)	0.653
	FC	−0.044 (0.029)	−0.048
	HM	−0.027^*** (0.007)	−0.035
	PS	0.398^*** (0.025)	0.579
	SexM	0.014^* (0.006)	0.033
	AgeGenX	−0.043^*** (0.010)	−0.097
	AgeBoom	−0.055^*** (0.011)	−0.126
	Edu2	0.060^*** (0.007)	0.136
	Edu3	−0.017^* (0.009)	−0.038
	EmplYes	−0.045^*** (0.008)	−0.102
	intSexMPE	0.102^* (0.044)	0.212
	intSexMEE	−0.122^** (0.040)	−0.281
	intSexMSI	−0.033^* (0.016)	−0.080
	intSexMFC	0.046 (0.062)	0.079
	intSexMHM	0.042 (0.024)	0.099
	intSexMPS	−0.032 (0.017)	−0.067
	intAgeGenXPE	0.000 (0.039)	0.000
	intAgeGenXEE	0.267^*** (0.044)	0.617
	intAgeGenXSI	−0.152^*** (0.022)	−0.388
	intAgeGenXFC	0.124^* (0.056)	0.217
	intAgeGenXHM	0.025 (0.018)	0.059
	intAgeGenXPS	−0.197^*** (0.024)	−0.419
	intAgeBoomPE	0.072^* (0.035)	0.133
	intAgeBoomEE	0.217^*** (0.039)	0.475
	intAgeBoomSI	−0.213^*** (0.022)	−0.456
	intAgeBoomFC	0.156^*** (0.045)	0.273
	intAgeBoomHM	0.127^*** (0.026)	0.276
	intAgeBoomPS	−0.274^*** (0.025)	−0.549
	intEdu2PE	0.086 (0.045)	0.177
	intEdu2EE	−0.164^** (0.053)	−0.356
	intEdu2SI	−0.091^*** (0.020)	−0.215
	intEdu2FC	0.253^** (0.080)	0.420
	intEdu2HM	0.029 (0.027)	0.064
	intEdu2PS	−0.103^*** (0.019)	−0.211
	intEdu3PE	0.114^** (0.043)	0.212
	intEdu3EE	−0.174^*** (0.051)	−0.336
	intEdu3SI	−0.125^*** (0.020)	−0.268
	intEdu3FC	0.197^** (0.070)	0.270
	intEdu3HM	0.141^*** (0.034)	0.292
	intEdu3PS	−0.129^*** (0.021)	−0.224
	intEmplYesPE	0.037^* (0.015)	0.079
	intEmplYesEE	0.045^* (0.018)	0.102
	intEmplYesSI	−0.078^*** (0.018)	−0.197
	intEmplYesFC	0.054^* (0.023)	0.091
	intEmplYesHM	0.023^* (0.011)	0.053
	intEmplYesPS	−0.047^* (0.020)	−0.100
	R²	0.543

***p-value < 0.001.

**p-value < 0.01.

*p-value < 0.05

References

Bossen

Pine

Cabitza

, et al. Data work in healthcare: an Introduction. Health Inf J 2019; 25(3): 465–474.

Dendere

Slade

Burton-Jones

, et al. Patient portals facilitating engagement with inpatient electronic medical records: a systematic review. J Med Internet Res 2019; 21(4): e12779.

Carini

Villani

Pezzullo

, et al. The impact of digital patient portals on health outcomes, system efficiency, and patient attitudes: updated systematic literature review. J Med Internet Res 2021; 23(9): e26189.

Davidson

Østerlund

Flaherty

. Drift and shift in the organizing vision career for personal health records: an investigation of innovation discourse dynamics. Inf Organ 2015; 25(4): 191–221.

Eu Commission . Market study on telemedicine. October 2018. https://health.ec.europa.eu/system/files/2019-08/2018_provision_marketstudy_telemedicine_en_0.pdf

Leventhal

. Patient portal market expected to reach $900M by 2017. Septentrion 2013; 27: 2013. https://www.hcinnovationgroup.com/clinical-it/electronic-health-record-electronic-medical-record-ehr-emr/news/13022058/patient-portal-market-expected-to-reach-900m-by-2017

Leb

Magnin

Prokosch

, et al. Patient portals: objectives, acceptance, and effects on health outcome-A scoping review of reviews. Stud Health Technol Inf 2021; 283: 194–201.

Palos-Sanchez

Saura

Rios Martin

MÁ

, et al. Toward a better understanding of the intention to use mHealth apps: exploratory study. JMIR Mhealth Uhealth 2021; 9(9): e27021.

Graetz

Gordon

Fung

, et al. The digital divide and patient portals: internet access explained differences in patient portal use for secure messaging by age, race, and income. Med Care 2016; 54(8): 772–779.

10.

Harle

Listhaus

Covarrubias

, et al. Overcoming barriers to implementing patient-reported outcomes in an electronic health record: a case report. J Am Med Inf Assoc 2016; 23(1): 74–79.

11.

Saura

Palos-Sanchez

Velicia-Martin

. What drives volunteers to accept a digital platform that supports NGO projects? Front Psychol 2020; 11: 429.

12.

Reiners

Sturm

Bouw

, et al. Sociodemographic factors influencing the use of eHealth in people with chronic diseases. Int J Environ Res Publ Health 2019; 16(4): 645.

13.

Clarke

Lyden

, et al. Sociodemographic differences and factors affecting patient portal utilization. J Racial Ethn Health Dispar 2021; 8: 879–891.

14.

Khairat

Coleman

Newlin

, et al. A mixed-methods evaluation framework for electronic health records usability studies. J Biomed Inf 2019; 94: 103175.

15.

Matthews

Savoy

Paranjape

, et al. Acceptability of health information exchange and patient portal use in depression care among underrepresented patients. J Gen Intern Med 2022; 37(15): 3947–3955.

16.

Ben Arfi

Ben Nasr

Khvatova

, et al. Understanding acceptance of eHealthcare by IoT natives and IoT immigrants: an integrated model of UTAUT, perceived risk, and financial cost. Technol Forecast Soc Change 2021; 163: 120437.

17.

Pan

Gao

. Determinants of the behavioral intention to use a mobile nursing application by nurses in China. BMC Health Serv Res 2021; 21(1): 228.

18.

Rouidi

Elouadi

Hamdoune

, et al. TAM-UTAUT and the acceptance of remote healthcare technologies by healthcare professionals: a systematic review. Inform Med Unlocked 2022; 32: 101008.

19.

Yan

Kwan

Thumboo

, et al. Characteristics and health care utilization of different segments of a multiethnic Asian population in Singapore. JAMA Netw Open 2019; 2(9): e1910878.

20.

Antonio

Petrovskaya

Lau

. The state of evidence in patient portals: umbrella review. J Med Internet Res 2020; 22(11): e23851. DOI: 10.2196/23851.

21.

Alturkistani

Qavi

Anyanwu

, et al. Patient portal functionalities and patient outcomes among patients with diabetes: systematic review. J Med Internet Res 2020; 22(9): e18976.

22.

Eccher

Gios

Zanutto

, et al. TreC platform. An integrated and evolving care model for patients’ empowerment and data repository. J Biomed Inf 2020; 102: 103359.

23.

Venkatesh

Thong

. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Q 2012; 36: 157–178.

24.

Rondan-Cataluña

Arenas-Gaitán

Ramírez-Correa

. A comparison of the different versions of popular technology acceptance models: a non-linear perspective. Kybernetes 2015; 44(5): 788–805.

25.

Williams

Rana

Dwivedi

. The unified theory of acceptance and use of technology (UTAUT): a literature review. J Enterprise Inf Manag 2015; 28(3): 443–488.

26.

Rahimi

Nadri

Lotfnezhad Afshar

, et al. A systematic review of the technology acceptance model in health informatics. Appl Clin Inf 2018; 9(03): 604–634.

27.

Kruse

Argueta

Lopez

, et al. Patient and provider attitudes toward the use of patient portals for the management of chronic disease: a systematic review. J Med Internet Res 2015; 17(2): e40.

28.

Kim

Bell

, et al. Patient perspectives about decisions to share medical data and biospecimens for research. JAMA Netw Open 2019; 2(8): e199550.

29.

Dünnebeil

Sunyaev

Blohm

Leimeister

, et al. Determinants of physicians’ technology acceptance for e-health in ambulatory care. International journal of medical informatics. 2012; 81(11): 746–760.

30.

Strawley

Richwine

. Individuals access and use of patient portals and smartphone health apps, 2022. ASTP Health IT Data Brief, 2023.

31.

Sapio

Turk

Cornacchia

, et al. Building scenarios of digital television adoption: a pilot study. Technol Anal Strat Manag 2010; 22(1): 43–63.

32.

Niehaves

Plattfaut

. Internet adoption by the elderly: employing IS technology acceptance theories for understanding the age-related digital divide. Eur J Inf Syst 2014; 23(6): 708–726.

33.

Sutarto

Ansori

Widyanti

, et al. Driving the future: using TAM and UTAUT to understand public acceptance of fully autonomous vehicles. 2023 10th International Conference on ICT for Smart Society (ICISS). IEEE, 2023, pp. 1–8.

34.

Chopdar

. Adoption of Covid-19 contact tracing app by extending UTAUT theory: perceived disease threat as moderator. Health Policy Technol 2022; 11(3): 100651.

35.

Tavares

Oliveira

. Electronic health record portal adoption: a cross country analysis. BMC Med Inf Decis Making 2016; 16(1): 2–16.

36.

Guy

Norvell

. The neutral point on a Likert scale. J Psychol 1977; 95(2): 199–204.

37.

Hair Jr

Matthews

, et al. PLS-SEM or CB-SEM: updated guidelines on which method to use. International Journal of Multivariate Data Analysis 2017; 1(2): 107–123.

38.

Tieu

Schillinger

Sarkar

, et al.

Online patient websites for electronic health record access among vulnerable populations: portals to nowhere?

J Am Med Inf Assoc 2017; 24(e1): e47–e54.

39.

Tao

Intention to use and actual use of electronic information resources: further exploring Technology Acceptance Model (TAM). In AMIA Annu Symp Proc 2009; 2009: 629–633.

40.

Huang

Whitehead

Kuper

. Competing discourses, contested roles: Electronic health records in medical education. Medical Education. 2024; 58(12): 1490–1501.

41.

Calvo-Porral

Pesqueira-Sanchez

. Generational differences in technology behaviour: comparing millennials and Generation X. Kidorui 2020; 49(11): 2755–2772.

42.

Ravangard

Kazemi

Abbasali

, et al. Development of the UTAUT2 model to measure the acceptance of medical laboratory portals by patients in Shiraz. Electron Physician 2017; 9(2): 3862–3869.

43.

Ayangunna

(2023). Factors associated with patient portal utilization, preventive services utilization, and health promoting behaviors among adults in the United States.

44.

Zanutto

‘Two clicks and I’m in!’Patients as co-actors in managing health data through a personal health record infrastructure. Health Informatics Journal. 2019; 25(2): 389-400.