Skype or Skip? Causes and Consequences of Intimate Self-Disclosure in Computer-Mediated Doctor-Patient Communication

ABSTRACT

Advances in computer-mediated communication have created both opportunities and challenges for online doctor-patient communication. Based on the privacy calculus and objective self-awareness theory, we examined the causes and consequences of intimate self-disclosure in video consultations. We performed a 2 (verbal intimacy) × 2 (physical intimacy) online experiment among a large representative sample of the Dutch population (N = 2,251). Structural equation modeling confirmed expected relationships between benefits, privacy concerns, communication barriers, trust, and self-disclosure. Generally, when people were more willing to self-disclose, they reported more positive (e.g., relief) and less negative (i.e., stress) emotions. However, when it involved revealing physically intimate information (e.g., showing an intimate vs. non-intimate body part), people were less likely to self-disclose and reported more negative emotions as a result. Our findings suggest that people weigh benefits, costs, and trust in their decision to self-disclose and consider the potential negative intrapersonal effects of intimate self-disclosure.

The emergence of computer-mediated communication (CMC) in health provides patients and doctors with new ways to communicate, for example via online video conferencing (Tates, Antheunis, Kanters, Nieboer, & Gerritse, 2017). Video-mediated communication (VMC) will replace and has replaced – for example during the Covid-19 crisis – some of the face-to-face consultations (Greenhalgh, Wherton, Shaw, & Morrison, 2020). There might be opportunities to effectively use VMC in certain doctor-patient consultations. VMC provides several benefits for patients, as it can save time and travel costs, it can enhance self-disclosure, and it can give more rapid access to health professionals (e.g., Elbert et al., 2014; Knight et al., 2016). On the other hand, patients may also express barriers having such consultations, as sharing personal, often intimate information online could elicit concerns about privacy (i.e., individuals not being able to determine when, how, and to what extent information about them is communicated to others; Westin, 1967). The benefits and barriers can be different from face-to-face communication and other modes of communication, such as telephone and e-mail, which makes VMC relevant to explore further.

For VMC to be effective for both patients and providers, patients should be willing to self-disclose to a doctor in a digital environment. Self-disclosure, the act of revealing personal information about oneself (Jourard, 1964), is imperative in doctor-patient communication: it is a crucial prerequisite for information exchange, shared decision-making, and – most importantly – for doctors to make an adequate diagnosis (e.g., Tates et al., 2017). Disclosing personal information in the context of doctor-patient communication often involves discussing sensitive issues, which can disrupt a person’s emotional state both positively and negatively, as it can elicit anxiety but also relief (Afifi & Guerrero, 2000; Farber, Berano, & Capobianco, 2004).

As a result, people might make a cost-benefit tradeoff in their decision to disclose personal information when engaging in VMC with a doctor. The privacy calculus offers a theoretical explanation as to why people disclose information. It posits that the more benefits people expect, such as receiving social support, the more willing they are to self-disclose, whereas the more costs they perceive, such as jeopardizing one’s privacy, the less willing they are to self-disclose (Laufer & Wolfe, 1977). Although previous research has provided empirical support for the privacy calculus in a wide variety of online contexts (e.g., social network sites; Dienlin & Metzger, 2016), it has not yet been applied to the context of VMC. Hence, our first aim is to replicate earlier findings and extend the privacy calculus in understanding how the tradeoff of benefits and costs affects self-disclosure in video consultations.

How people precisely weigh benefits and costs of VMC may depend on the type of information under consideration. For example, patients may be asked to share verbal information that can range from basic (e.g., I have a headache) to very intimate information (e.g., I have a sexually transmitted disease), and could differently influence self-disclosure. Moreover, in medical consultations, patients typically need to share physical information – for example by showing parts of their body – which may also impact self-disclosure. This can range from revealing non-intimate (e.g., hand or arm) to intimate body parts for which patients have to undress (e.g., leg or breast). The level of intimacy of such information may influence how patients weigh benefits against costs, but we currently lack knowledge on how intimacy precisely affects this tradeoff. Therefore, our second aim is to understand how verbal and physical intimacy influence the cost-benefit tradeoff people make in their decision to self-disclose.

Our third aim is to examine the impact of self-disclosure on people’s emotional state. Self-disclosure is instrumental to effective doctor-patient communication, but it can both positively and negatively impact emotional states (Lepore & Smyth, 2002). For example, self-disclosure can improve a person’s emotional state by reducing stress and increasing feelings of relief (Tam, Hewstone, Harwood, Voci, & Kenworthy, 2006), but it can also elicit negative emotions, such as feeling vulnerable (Farber et al., 2004). Objective self-awareness theory (Duval & Wicklund, 1972) states that an increased direction toward the self because of self-disclosure might result in an unpleasant feeling. Self-disclosure in medical settings is no exception, especially since it often not only involves verbal, but also physical disclosure. How and when disclosing verbally and physically intimate information fosters positive or negative emotions has been largely unexplored, but such insights could nuance our understanding of when VMC is an effective tool for medical encounters. This research therefore contributes to further understanding intrapersonal effects of self-disclosure.

Computer-Mediated Doctor-Patient Communication

In an increasingly digitizing society, the opportunities for computer-mediated health communication seem endless: from hospital websites with disease-related information to wearables monitoring chronic patients 24/7. Although remote access to healthcare is not new (e.g., the term telemedicine was coined in the 1970s to describe any medical activity involving an element of distance, Bashshur, 1995), video consultations have only gained minimal attention in medical care. Mostly because the introduction of video consultations is complex in terms of organizational change of processes as well as in terms of guaranteeing medical quality, privacy, and safety (Greenhalgh et al., 2020).

However, in the upcoming years, it is expected that video consultations will replace some of the traditional face-to-face consultations. In fact, the current Covid-19 crisis enabled an instant launch of video consultations in many countries, as face-to-face consultations were not always allowed (Greenhalgh et al., 2020). For example, at a surgical outpatient clinic in the Netherlands, the use of video consultations steeply increased from zero before March 2020 to almost 600 per week during the first wave of the Covid-19 pandemic (Barsom et al., 2021). Although CMC in doctor-patient communication is still not commonplace, VMC may offer several affordances for patients, such as accessibility, perceived anonymity, and availability. Accordingly, there is evidence that patients would like to have video consultations with their doctors next to the regular face-to-face consultations (e.g., Knight et al., 2016). For example, in a study among Dutch patients, 50% indicated that they would like to have a video consultation (Barsom, Rauwerdink, Heldoorn, & Schijven, 2018).

The potential of CMC in doctor-patient encounters has led scholars to study motivations, uses, and effects of computer-mediated consultations. Most of these studies have focused on unimodal communication platforms, such as e-mail or web messaging (e.g., Lee & Zuercher, 2017). For example, research from the early 2000s has shown that both doctors and patients indicated a need for and willingness to use e-mail to communicate with their patients or doctors (e.g., Katz, Nissan, & Moyer, 2004; Moyer, Stern, Dobias, Cox, & Katz, 2002). However, despite their willingness, the actual use of e-mail for doctor-patient interactions was rather low (Santana et al., 2010; Ye, Rust, Fry-Johnson, & Strothers, 2010). Most of the studied platforms are text-based, asynchronous, eliminate time and space restrictions, and allow the patient to communicate with their doctor at their convenience without being afraid of disturbing the doctor. Moreover, the persistence of text-based CMC allows for adequate documentation, which can be convenient for both patient and doctor (Treem & Leonardi, 2013).

Compared to unimodal communication, VMC affords synchronicity and multimodality, which allow for a real consultation but without the burden of travel. Particularly for more personal or complex health issues, or situations in which the doctor needs to see part of the patient’s body, VMC might be more suitable than asynchronous, text-based modalities. The few studies that have been done on VMC have shown its potential by comparing the effects of face-to-face and video-mediated consultations. To illustrate, Tates et al. (2017) found no differences between video and face-to-face consultations with respect to satisfaction, information exchange, relationship building, and shared-decision making. Another study showed there is even a bit more other-attentiveness (i.e., focus on the patient) in video consultations than in face-to-face settings (Stommel et al., 2019). When implemented in clinical care, VMC might be effectively used for certain doctor-patient interactions.

Self-Disclosure in Video-Mediated Communication

Another potential benefit of CMC is that it can enhance self-disclosure, which is essential in medical consultations. In online environments, people might feel less restrained to disclose personal information as there are less cues available and, hence, people feel more anonymous. Ample research has indeed shown that people disclose more about themselves online, via text-based CMC, than face-to-face (e.g., Antheunis, Schouten, Valkenburg, & Peter, 2012; Joinson, 2001; Tidwell & Walther, 2002). Yet, the relationship between visual anonymity and self-disclosure has been challenged. In fact, Croes, Antheunis, Schouten, and Krahmer (2020) found that having eye contact was crucial in people’s decision to disclose intimate information in a video-mediated setting, suggesting there are more factors that contribute to online self-disclosure than affording anonymity.

The privacy calculus can be used to explain self-disclosure behaviors. It posits that people make tradeoffs between potential costs and benefits when making decisions about self-disclosure (Culnan & Armstrong, 1999; Laufer & Wolfe, 1977), where it is expected that as long as the perceived benefits exceed the risks, self-disclosure will occur. In CMC, the benefits often refer to the (social) gratifications of engaging in online self-disclosure and the costs are often operationalized as the privacy concerns associated with online self-disclosure (e.g., Dienlin & Metzger, 2016; Trepte, Scharkow, & Dienlin, 2020). Such privacy concerns associated with online self-disclosure could range from worrying about unwanted surveillance to fear of potential misuse of data by third parties (Bol et al., 2018).

In general, the most important reason for people to reveal personal information are the expected benefits of doing so (e.g., Omarzu, 2000). In the context of video consultations, people may – for instance – expect video consultations to be time and cost efficient (e.g., Almathami, Win, & Vlahu-Gjorgievska, 2020; Elbert et al., 2014). However, online self-disclosure via video consultations may also increase privacy concerns, which refers to individuals’ subjective views of fairness regarding when, how, and to what extent information about them is communicated to others (Malhotra, Kim, & Agarwal, 2004), particularly given the sensitivity of information disclosed during video consultations. Following the privacy calculus, we hypothesize:

H1: The more benefits people perceive, the more willing they are to self-disclose during a video consultation.

H2: The more concerned people are about their privacy, the less willing they are to self-disclose during a video consultation.

Apart from privacy costs, self-disclosure via video consultations might also specifically depend on perceived communication barriers. Moving doctor-patient communication from the consultation room to an online setting causes several changes in communication dynamics. For instance, the fact that the communication does not happen in the same room does not only create a geographical but also a relational distance between doctor and patient. Although VMC has drastically improved over the last decades, it is still limited in terms of facilitating eye contact and expressing and reading non-verbal cues and emotions (Croes et al., 2020). Furthermore, patients and doctors may experience technological disturbances, such as those related to a poor internet connection (Almathami et al., 2020). Such limitations might make it especially hard to perform adequate bodily inspections of patients. Anticipation of such barriers may reduce the willingness to self-disclose during video consultations. We therefore expect:

H3: The more communication barriers people perceive, the less willing they are to self-disclose during a video consultation.

Another important factor in self-disclosure is trust. As doctor-patient communication is oftentimes involved with disclosing sensitive information, having trust in one’s doctor is essential to establish a good doctor-patient relationship (Hillen, de Haes, & Smets, 2011). When such communication happens in a digital environment, another dimension of trust is added to the equation: trust in technology. Such trust depends on specific technology design characteristics, including the extent to which technology safeguards privacy, is reliable, and how valid and accurate the technology is (Rupp, Michaelis, McConnell, & Smither, 2016). Trust is a crucial predictor in the acceptance of technology for online transactions (Pavlou, 2003) and health technologies such as wearables (Rupp et al., 2016). Only when people put trust in how one’s personal data is being handled, people are willing to self-disclose (Bol et al., 2018). This may not be different for CMC in medical settings: one needs to trust the technology that handles their personal health data when engaging in an online video consultation. We therefore hypothesize:

H4: The more trust people have in video-consultation technology, the more willing they are to self-disclose during a video consultation.

Intimacy and Self-Disclosure

Although the privacy calculus may help to explain why people are willing (or not) to self-disclose during video consultations, it does not provide justification for the conditions under which people are likely to self-disclose. Theories that consider the context in which (rational) decision making takes place are, among others, Petronio’s communication privacy management theory (CPM; Petronio, 2002) and Nissenbaum’s theory of privacy as contextual integrity (Nissenbaum, 2010). Both theories argue that privacy behaviors are context-dependent, such that self-disclosure is shaped by contextual norms and expectations at stake in a given context. In the context of health, revealing health information about oneself might be seen as appropriate and expected. Yet, within this context, people create rules about what information they disclose and what to keep private. These rules often depend on a variety of factors.

The level of intimacy of a disclosure could be a factor that affects the willingness to self-disclose. Self-disclosure is considered intimate when it is either emotionally intense or contains potentially negative or (socially) embarrassing information (Altman & Taylor, 1973). In her disclosure decision model (DDM), Omarzu (2000) proposes that the perceived subjective risk of a disclosure plays a significant role in people’s willingness to disclose intimate information. In the context of doctor-patient communication, some disclosures can be considered more risky than others. Self-disclosure in interpersonal communication has been typically defined as verbally revealing personal information to another person (Omarzu, 2000), but medical interactions also often include physically revealing information, which may both vary in their level of intimacy. For example, sharing descriptive information about a certain health situation (e.g., explaining how you hurt your arm while falling off a bike) is considered less intimate than expressing one’s concerns regarding a certain health situation (e.g., being afraid to have cancer; Altman & Taylor, 1973). Regarding physical disclosures, showing a wrist might be considered less embarrassing than showing a hip.

Although self-disclosure intimacy has been commonly studied by asking people how willing they are to disclose certain types of information or by having coders rate the level of intimacy of people’s disclosures (Omarzu, 2000), it has been less commonly investigated (also in doctor-patient communication) how a request for intimate information influences people’s willingness to self-disclose. Moreover, the potential differential impact of verbal versus physical intimate disclosures has not yet been studied. We therefore pose the following exploratory research question (RQ):

RQ1: To what extent do (a) verbal and (b) physical intimacy influence perceived benefits, privacy concerns, communication barriers, trust, and self-disclosure?

Effects of Intimacy and Self-Disclosure on Emotional States

This study further aims to examine the impact of disclosing intimate information during video consultations on people’s emotional states. Intrapersonal consequences of self-disclosure are important to consider, as these are not solely positive, and can improve or harm patients’ psychosocial well-being. Ample research has been conducted on the impact of patients’ self-disclosure on their emotional state, mainly in the field of psychotherapy (e.g., Farber et al., 2004; Kahn, Achter, & Shambaugh, 2001). For example, patients experienced both positive and negative emotions following self-disclosure of very intimate information in face-to-face psychotherapy (Farber et al., 2004). Regarding positive emotions, relief might be experienced after disclosing personal information, particularly when the information elicits emotions, like fear or worries. This is often referred to as catharsis: “the release tension or emotion that can be achieved through verbal disclosure of that emotion.” (Omarzu, 2000, p. 175)

However, as posited by objective self-awareness theory (Duval & Wicklund, 1972), the experience of self-disclosure can also be unpleasant, as self-disclosure directs attention toward the self. Increased self-awareness affects our emotional state, as it leads a person to focus on the disparity between the real and ideal self. VMC particularly stimulates self-focus, as it often affords to not only see the interaction partner, but also oneself (e.g., Joinson, 2001). In case of intimate disclosures, a negative state following self-disclosure can occur, as many of people’s perceived worries or weaknesses can be considered intimate (Archer, Hormuth, & Berg, 1982).

To explore to what degree intimate self-disclosures lead to experiencing positive and/or negative emotions, the following two RQs will be investigated:

RQ2: To what extent does self-disclosure lead to an increase or decrease of experiencing positive and negative emotional states?

RQ3: To what extent do (a) verbal and (b) physical intimacy lead to an increase or decrease of experiencing positive and negative emotional states?

Materials and Methods

Participants

Data were collected in CentERdata’s LISSPANEL between June and August 2019. A priori, we conducted a power analysis using G*Power to calculate the minimum required sample size needed to detect small effect sizes (β = .10), with an alpha set at .05 and an expected power of 80–95%, which showed that a sample size of N = 1,558 to 2,578 was needed to test our proposed model. CentERdata invited a representative sample of the Dutch population aged 16 years or older (N = 2,804). A proportion of 80% fully completed the online questionnaire, resulting in a final sample size of N = 2,251. A post hoc power analysis showed that we reached 92% statistical power.

Participants were on average 53.85 years old (SD = 17.86, range = 16–102) and 53.2% were female. About 28.5% of the participants completed a lower level of education (i.e., primary education, prevocational secondary education), 33.0% completed a middle level of education (i.e., senior general secondary education, pre-university education, senior secondary vocational education), and 38.2% completed a higher level of education (i.e., higher vocational education, university education; for classification, see Statistics Netherlands, n.d.).

Design and Procedure

To test our proposed model, we conducted a scenario-based experiment. In a 2 × 2 between-subjects design, we varied four vignettes in terms of verbal and physical intimacy. A vignette presents a hypothetical situation, to which study participants respond thereby revealing their perceptions, values, social norms, or impressions of events. Scenario-based experiments are well suited for research that seeks to understand how and why individuals, when dealing with complex issues, make decisions (Rungtusanatham, Wallin, & Eckerd, 2011).

Participants were randomly assigned to one of four vignettes. In all vignettes, participants were asked to imagine themselves having a video consultation with a doctor about wound care, during which they were asked to reveal verbal and physical information. In the verbally intimate conditions, participants were asked to share intimate information about their wound (i.e., expressing concerns about a potential surgery) versus non-intimate information (i.e., describing the wound care) in the verbally non-intimate conditions. In the physically intimate conditions, participants were asked to show their wound located on an intimate body part (i.e., their hip) versus a non-intimate body part (i.e., their wrist) in the physically non-intimate conditions. The four vignettes can be found in Appendix A of the online supplemental material (OSM).¹

After reading one of four vignettes, participants answered questions related to perceived benefits, privacy concerns, communication barriers, trust, willingness to self-disclose, and emotional states. All participants provided informed consent before participation, and the study was approved by the ethical review board of the Tilburg School of Humanities and Digital Sciences (reference number: REDC # 2019/35).

Measurements

All variables were assessed on 7-point Likert scales, ranging from 1 = strongly disagree to 7 = strongly agree, unless indicated otherwise. Confirmatory factor analyses (CFAs) were conducted to test the dimensionality and factor validity of all variables separately, as well as all variables included in one measurement model. Based on common fit criteria (e.g., Kline, 2016), our measurement model showed good fit. The factor validity statistics of all variables included in the measurement model can be found in Table 1. All measurement items can be found in Appendix B of the OSM. Figure 1 shows the zero-order correlations and scatterplots of the bivariate relationships between variables, as well as density plots of the variables. Means were calculated by averaging the predicted values for the indicators of the latent variables.

Table 1. Factor validity statistics of all latent variables included in the model (N = 2,251).

	χ^2	df	p-value	CFI	TLI	RMSEA	SRMR	alpha	omega	AVE
Benefits	511.61	20	<.001	.946	.924	.104	.043	.93	.93	.64
Privacy Concerns	68.16	14	<.001	.996	.994	.041	.012	.96	.96	.79
Communication Barriers	16.72	5	.005	.997	.993	.032	.014	.85	.85	.54
Trust	1.36	2	.506	1.000	1.000	.000	.003	.91	.91	.72
Self-Disclosure	573.87	62	<.001	.985	.981	.061	.025	.97	.98	.86
Emotional States	175.32	13	<.001	.979	.966	.074	.044	.89	.93	.70
Overall	3795.80	872	<.001	.956	.962	.039	.040	.86	.96	.74

Self-disclosure is estimated as three-dimensional construct, with lifestyle health-related and medical-related as second-order constructs, and scenario-based self-disclosure and general self-disclosure as first-order constructs. Emotional states is estimated as two-dimensional construct, with positive and negative emotions as two distinct, yet related dimensions. Alpha refers to internal consistency (Cronbach’s alpha); Omega to composite reliability (Raykov’s omega). Abbreviations: CFI = Comparative fit index; TLI = Tucker-Lewis index; RMSEA = Root mean square error of approximation; SRMR = Standardized root mean square residual; AVE = Average variance extracted.

Figure 1. Zero-order correlations are presented in the above diagonal. Density plots are presented in the diagonal. Scatterplots are presented below the diagonal. Statistics are calculated based on the averaged predicted values of the indicators of the latent variables in the model.

Perceived benefits

Perceived benefits reflected the advantages people expected from self-disclosing during a video consultation. Based on a pretest (among a convenience sample of N = 100, see Appendix C of the OSM) and technology acceptance literature (Davis, 1989), we identified eight items that all started with “Sharing personal information in a video consultation:” Example items are “ … would be useful to me” and “ … would save me time.”

Privacy concerns

Privacy concerns assessed how strongly people worry about their privacy when engaging in a video consultation, and were measured with seven items. Five items were adopted from existing scales (e.g., “I am concerned that my personal information can be misused by others when I have a video consultation” Baek & Morimoto, 2012; Bol et al., 2018), and two items were formulated based on the pretest (e.g., “I am afraid that personal information I share via a video consultation is used for other purposes”).

Communication barriers

Whether people expected obstacles that prevent them from having a proper medical encounter with a doctor was referred to as communication barriers. Five items were formulated based on the pretest, and assessed why people find it difficult to share personal information via a video consultation. All items started with “I find it difficult to share personal information via a video consultation, because …,” and example items were: “ … physical contact is impossible” and “ … personal communication is not possible.”

Trust

Trust refers to the belief that the online video software described in the scenario is reliable and confidentially handles one’s personal information. Trust was measured using four items adopted from Chinomona (2013), such as “When I use online video software, I trust that personal information I share is safely stored” and “I believe online video software is reliable.”

Self-disclosure

Self-disclosure assessed participants’ willingness to reveal personal information during video consultations. It was measured with two dimensions: (1) self-disclosure of information requested by the doctor in the scenario (3 items) and (2) self-disclosure of health-related information during a video consultation in the future (10 items), all measured on a scale from 1 = very unlikely to 7 = very likely. CFA suggested suboptimal fit of self-disclosure as a two-dimensional construct. An exploratory factor analysis (EFA) proposed a three-factor structure, with (1) three items about self-disclosure of information related to the scenario (e.g., the wound, concerns about possible follow-up surgery), (2) five items about disclosure of lifestyle health-related types of information (e.g., bodily condition and fitness, nutrition and diet), and (3) five items about disclosure of medical-related types of information (e.g., diagnosis, medical test results). We therefore considered self-disclosure as a three-dimensional construct: (a) scenario-based, (b) lifestyle health-related, and (c) medical-related self-disclosure, with lifestyle health-related and medical-related as second-order constructs, and scenario-based self-disclosure and general self-disclosure as first-order constructs.

Emotional states

Participants were asked to imagine having a video consultation with a doctor in the future and rate their positive and negative emotional states as a result of sharing their personal information. Items were based on the positive affect negative affect scale (PANAS; Watson, Clark, & Tellegen, 1988), and included feelings such as relief, safety, anxiousness, stress (Farber et al., 2004). All items were rated on a 7-point Likert scale, ranging from 1 = not at all to 7 = very much. We considered (a) positive and (b) negative emotional states as separate, yet related constructs in our analyses.

Manipulation check

Perceived verbal and physical intimacy assessed to what extent participants perceived the information and the body part they were asked to reveal in the video consultation scenario as 1 = not intimate/7 = intimate, 1 = not sensitive/7 = sensitive, 1 = not delicate/7 = delicate, and 1 = not personal/7 = personal.

Control variables

We assessed the extent to which participants thought the scenario was likely to occur and its credibility with one item each. Furthermore, power usage was measured to control for participants’ general motivation, expertise, efficacy, and behavior with regard to information technologies. It was measured with eight items, such as “A little bit of intuition is all that is needed to figure out how to use any new technology” and “Using any technological device comes easy to me” (Sundar, Bellur, Oh, Jia, & Kim, 2016). We also assessed participants’ age, gender, education, and subjective health.

Data Analysis

The analyses were conducted with R (Version 3.6.1), using packages such as lavaan (Version 0.6–3; Rosseel, Jorgensen, & Rockwood, 2020) and ggplot2 (Version 3.2.0; Wickham et al., 2020). Structural equation modeling (SEM) was used to test the proposed theoretical model. To test the privacy calculus (H1-H4), assess the impact of intimacy on perceived benefits, privacy concerns, communication barriers, trust, and self-disclosure (RQ1), explore how self-disclosure and emotional states are associated (RQ2), and test the effects of intimacy on emotional states (RQ3), we ran four separate models. The fourth and final model included all paths between intimacy, benefits, privacy concerns, communication barriers, trust, self-disclosure, and emotional states. The results of this final model are displayed in Figure 2. The fit of this full model was good: χ² (980) = 3522.04, p < .001, CFI = .971, TLI = .958, RMSEA = .035, SRMR = .026. We controlled for demographics (i.e., age, gender, education level), subjective health, perceived credibility and likelihood of occurrence of the scenario, and power usage, because these variables are expected to be associated with self-disclosure (Bol et al., 2018). Controlling for these variables changed the significance level of two associations in the model, which are reported in the results.

Figure 2. Overview of all results of the full model test. Estimates reflect the completely standardized solution (i.e., both latent and observed variables are standardized). Paths with more than one estimate indicate the associations between constructs that have more than one dimension (i.e., for self-disclosure and emotional states). Additional analysis showed no mediation effect between the physical intimacy and negative emotional states via self-disclosure (b = 0.02, se = .01, 95% CI [−0.03, 0.04], p = .082, β = .01). Abbreviations: S = Scenario-based disclosure; G = General health-related disclosure; P = Positive emotions; N = Negative emotions.

*** p < .001. ** p < .01. * p < .05.

Results

Manipulation Checks

We checked whether participants perceived our manipulation of physical and verbal intimacy as intended. Participants who were exposed to the physical intimacy conditions rated showing their wound during a video consultation as more intimate (M = 4.22, SD = 1.44) than those in the physical non-intimate conditions (M = 3.10, SD = 1.37), t(2249) = −16.42, p < .001, d = −.61 (which represents a medium effect size). However, participants in the verbal intimacy conditions did not rate the information they were asked to disclose as significantly more intimate (M = 3.87, SD = 1.42) than those in the verbal non-intimacy conditions (M = 3.78, SD = 1.48), t(2249) = −1.47, p = .142, d = −.14 (which represents a negligible effect size).

Privacy Calculus Effects

We interpreted the results of the first model, which included the associations between perceived benefits, privacy concerns, communication barriers, trust, and self-disclosure, to test the privacy calculus (H1-H4). With regard to H1, we found that participants who perceived more benefits were more likely to disclose information asked for in the scenario (b = 0.32, se = .04, 95% CI [0.25, 0.40], p < .001, β = .27), as well as other personal health-related information (b = 0.26, se = .04, 95% CI [0.19, 0.34], p < .001, β = .22), supporting H1. We also found support for H2. When people were more concerned about their privacy, they were less likely to disclose information asked for in the scenario (b = −0.09, se = .03, 95% CI [−0.15, −0.03], p = .004, β = −.10), as well as general health-related information (b = −0.33, se = .03, 95% CI [−0.39, −0.27], p < .001, β = −.35). We found that when participants expected communication barriers, they were less likely to disclose the information asked for in the scenario (b = −0.08, se = .04, 95% CI [−0.16, −0.01], p = .028, β = −.07). However, this effect was not found for disclosure of general health-related information (b = −0.04, se = .04, 95% CI [−0.10, 0.03], p = .323, β = −.03),² providing partial support for H3. We found support for H4, showing that people who put more trust in videoconferencing technology were more willing to self-disclose information asked for in the scenario (b = 0.16, se = .04, 95% CI [0.08, 0.25], p < .001, β = .15), as well as other personal health-related information (b = 0.20, se = .04, 95% CI [0.12, 0.28], p < .001, β = .18).

Effects of Verbal and Physical Intimacy on the Privacy Calculus

The results of the second model were used to assess the impact of (a) verbal and (b) physical intimacy on perceived benefits, privacy concerns, communication barriers, trust, and self-disclosure (RQ1). With regard to RQ1a, we found a small effect of verbal intimacy on communication barriers. Disclosing verbally intimate information (i.e., expressing concerns about the health situation), compared to disclosing verbally non-intimate information (i.e., describing the health situation), was related to expecting less communication barriers, such as Internet connection problems and less personal communication (b = −0.18, se = .08, 95% CI [−0.03, −0.16], p = .022, β = −.08). However, it did not affect benefits (b = 0.03, se = .07, 95% CI [−0.11, 0.18], p = .631, β = .02), privacy concerns (b = −0.08, se = .09, 95% CI [−0.26, 0.10], p = .389, β = −.03), trust (b = 0.12, se = .08, 95% CI [−0.04, 0.27], p = .131, β = .05), and self-disclosure (i.e., self-disclosure of information asked for in the scenario: b = 0.10, se = .09, 95% CI [−0.08, 0.27], p = .269, β = .04; self-disclosure of general health-related information: b = 0.10, se = .09, 95% CI [−0.08, 0.27], p = .273, β = .04).

With regard to physical intimacy (RQ1b), we found that it partially influenced self-disclosure: being asked to share physically intimate information (i.e., showing the hip) during video consultations led to a lower willingness to share such information compared to being asked to share physically non-intimate information (i.e., showing the wrist) (b = −0.24, se = .09, 95% CI [−0.42, −0.06], p = .009, β = −.09). It did not affect the willingness to disclose other, more general health-related information (b = 0.02, se = .09, 95% CI [−0.16, 0.20], p = .824, β = .01). It also did not affect perceived benefits (b = −0.08, se = .07, 95% CI [−0.22, 0.06], p = .270, β = −.04), privacy concerns (b = 0.09, se = .09, 95% CI [−0.09, 0.26], p = .351, β = .03), communication barriers (b = 0.05, se = .08, 95% CI [−0.10, 0.20], p = .536, β = .02), and trust (b = −0.07, se = .08, 95% CI [−0.23, 0.09], p = .375, β = −.03). There were no interaction effects between verbal and physical intimacy on benefits, privacy concerns, communication barriers, trust, and self-disclosure.

Effects of Intimate Self-Disclosure on Emotional States

With respect to RQ2, the third model examined the relationships between self-disclosure and emotional states and showed significant associations. Disclosing information asked for in the scenario as well as other personal health-related information was related to experiencing more positive emotions, such as relief and support (respectively, b = 0.08, se = .03, 95% CI [0.02, 0.13], p = .006, β = .10 and b = 0.46, se = .03, 95% CI [0.41, 0.52], p < .001, β = .60). At the same time, less negative emotions, such as stress and anxiety, were experienced as a result of disclosing information asked for in the scenario (b = −0.12, se = .04, 95% CI [−0.20, −0.05], p = .002, β = −.13) and other personal health-related information (b = −0.31, se = .04, 95% CI [−0.38, −0.23], p < .001, β = −.32).

Regarding RQ3, the fourth and final model revealed some effect of intimacy on emotional states was found. While verbal intimacy did not affect experiencing more or less positive emotions (b = 0.06, se = .05, 95% CI [−0.04, 0.16], p = .244, β = .03) or negative emotions (b = 0.06, se = .08, 95% CI [−0.10, 0.21], p = .455, β = .02), physical intimacy did to some extent. We found that being asked to share physically intimate information (i.e., showing the hip) during video consultations was associated with experiencing more negative emotions as compared to being asked to share physically non-intimate information (i.e., showing the wrist) (b = 0.19, se = .08, 95% CI [0.03, 0.35], p = .019, β = .07). Physical intimacy did not influence positive emotions (b = 0.01, se = .05, 95% CI [−0.09, 0.11], p = .798, β = .01). There were no interaction effects between verbal and physical intimacy on emotional states. Figure 2 presents an overview of the results, combining all variables in one final model³ and Table 2 displays the mean scores grouped by verbal and physical intimacy.

Table 2. Descriptive statistics (Means with standard deviations within parentheses) of all latent variables grouped by verbal and physical intimacy.

	Benefits	PrivCon	CommBar	Trust	DisScen	DisGen	EmoPos	EmoNeg
Overall mean	4.59 (1.34)	4.21 (1.51)	4.53 (1.19)	4.47 (1.35)	4.77 (1.59)	4.63 (1.76)	3.86 (1.40)	3.16 (1.31)
Verbal
Non-intimate	4.55 (1.36)	4.22 (1.51)	4.56 (1.20)	4.42 (1.37)	4.70 (1.63)	4.56 (1.78)	3.78 (1.40)	3.20 (1.34)
Intimate	4.64 (1.31)	4.20 (1.53)	4.49 (1.17)	4.52 (1.32)	4.83 (1.56)	4.71 (1.75)	3.93 (1.38)	3.11 (1.28)
Physical
Non-intimate	4.66 (1.34)	4.12 (1.53)	4.42 (1.23)	4.54 (1.32)	4.90 (1.59)	4.64 (1.77)	3.93 (1.39)	3.06 (1.31)
Intimate	4.53 (1.33)	4.32 (1.49)	4.63 (1.14)	4.39 (1.37)	4.62 (1.59)	4.62 (1.76)	3.78 (1.40)	3.26 (1.30)
Verbal × Physical
Non-intimate (V, P)	4.61 (1.36)	4.16 (1.50)	4.50 (1.21)	4.48 (1.32)	4.84 (1.63)	4.59 (1.77)	3.85 (1.37)	3.09 (1.30)
Non-intimate (V), Intimate (P)	4.49 (1.36)	4.29 (1.51)	4.63 (1.19)	4.36 (1.40)	4.56 (1.61)	4.53 (1.79)	3.72 (1.44)	3.32 (1.38)
Intimate (V), Non-intimate (P)	4.70 (1.32)	4.08 (1.57)	4.35 (1.24)	4.61 (1.31)	4.97 (1.55)	4.70 (1.77)	4.00 (1.40)	3.03 (1.33)
Intimate (V, P)	4.57 (1.29)	4.34 (1.47)	4.63 (1.08)	4.42 (1.33)	4.69 (1.55)	4.71 (1.73)	3.86 (1.37)	3.20 (1.22)

Descriptives are calculated based on the averaged predicted values of the indicators of the latent variables in the model. Abbreviations: V = Verbal; P = Physical.

Discussion

With the digitization of healthcare and the urgency to adapt health care services to current and future pandemic outbreaks, it is crucial to understand the causes and consequences of new ways for patients and doctors to communicate, such as through video-mediated communication. This study set out to investigate how people weigh benefits, (privacy and communication) costs, and trust in their decision to self-disclose intimate information in video consultations and examine the impact on positive and negative emotional states. We used a scenario-based experimental design to systematically test the potential differential effects of verbally and physically intimate self-disclosures, and presented these scenarios to a large representative sample of the Dutch population (N = 2,251).

With regard to our first aim, we were able to replicate and extend earlier findings on the privacy calculus. In line with earlier studies (Bol et al., 2018; Dienlin & Metzger, 2016), we found that people considered both benefits, such as saving time and money, and costs, such as running privacy risks, in their decision to self-disclose. As expected, more benefits were related to a higher willingness to self-disclose and more privacy concerns to a lower willingness to self-disclose. We also showed that people perceive other costs in video-mediated doctor-patient communication, such as having less personal doctor-patient communication, thereby extending ways to conceptualize and operationalize costs as part of the privacy calculus. Specifically, expecting communication barriers during a video consultation resulted in a lower willingness to self-disclose. Trust also played an important role in people’s willingness to self-disclose: people who have more trust in video consultation technology were also more willing to disclose personal information during the video consultation.

The second aim of this study was to understand how verbal and physical intimacy influences the cost-benefit tradeoff people make in their decision to self-disclose. Our results showed that disclosing verbal intimacy only had a small, but significant, effect on communication barriers. People who disclosed verbally intimate information expected less communication barriers. Benefits, privacy costs, trust, and self-disclosure were not influenced by verbal intimacy. This result may be due to the fact that our manipulation of verbal intimacy (expressing concerns about the health situation) was not perceived as sufficiently intimate. For physical intimacy, we found no effects on the cost-benefit tradeoff preceding self-disclosure, but we found that physical intimacy had a small, but significant, direct effect on the willingness to share intimate information. More specifically, being asked to show one’s hip resulted in a lower willingness to do so compared to being asked to show one’s wrist. Following Omarzu’s (2000) DDM, intimate physical self-disclosure might have been perceived as more risky, and thus resulted in a lower willingness to self-disclose.

Regarding our third aim, we found that self-disclosure generally leads to experiencing more positive emotions (e.g., relief and support) and less negative emotions (e.g., stress and anxiety). However, we found that disclosing physical intimate information leads to experiencing more negative emotions. We expected that the lower willingness to disclose physically intimate information would explain these negative emotions experienced after self-disclosure; however, our data did not support a mediating relationship between physical intimacy and experiencing negative emotions via self-disclosure. Alternatively, we could argue that by seeing yourself, while talking to a doctor might enhance self-focus (Joinson, 2001), which makes the patient feel more uncomfortable in the case of showing a more intimate body part. This can particularly be the case for people with lower body satisfaction.

Implications

Our results have several implications for theory and practice. Regarding the privacy calculus, our findings do not only replicate existing notions on how people balance benefits and privacy costs in their decision to self-disclose, but also extend our knowledge on what other factors people consider in this tradeoff. Based on an extensive pretest, we included perceived communication barriers as another factor people would weigh in on their decision to self-disclose. By adding new factors, we contribute to a more comprehensive understanding of how people make self-disclosure decisions. On a critical note, we should acknowledge that our take on the privacy calculus – including its variables we investigated here – aligns very much with the often criticized rational assumption of weighing costs and benefits in self-disclosure decisions. Decisions regarding self-disclosure are often spontaneous and less deliberate and rely more on affective states, such as mood and emotions (Dinev, McConnell, & Smith, 2015). However, there is also empirical evidence that shows that more rational and deliberate weighing of costs and benefits often occurs when it concerns disclosure of sensitive information, such as in the context of healthcare (Smith, Dinev, & Xu, 2011). Nonetheless, as we also showed evidence for the association between self-disclosure and emotional states, further research should unravel which affective processes are considered in the tradeoff to share intimate health-related information in CMC.

Furthermore, this study has important theoretical implications for self-disclosure research. Our study showed intrapersonal effects of self-disclosure such that self-disclosure generally leads to experiencing positive emotions (relief and support) but that the disclosure of physical intimate information can result in experiencing negative emotions (stress and anxiety). These results add to the capitalization process, which means that disclosing personal information can have additional effects on a person’s emotional state, as it increases the salience of the information (Gable & Reis, 2010). This first implies that the valence of the information disclosed (e.g., disclosing something you do versus do not feel ashamed of) has an intrapersonal effect and, second, that the context might interact with this effect by increasing self-awareness and thus making the VMC-user more insecure of disclosing certain types of information. Future studies could further focus on these intrapersonal effects of self-disclosure and further nuance the link between the valence of the disclosed information and the context in which people disclose that information.

Our third contribution to theory relates to nuancing our understanding of the conditions under which people self-disclose. Self-disclosure may depend on several factors, such as the level of disclosure intimacy, which might make a self-disclosing individual feel more or less vulnerable. It is therefore often theoretically proposed that individuals evaluate contextual conditions that make self-disclosure potentially more or less risky (e.g., Nissenbaum, 2010; Omarzu, 2000; Petronio, 2002). For example, a person may be willing to self-disclose during an online video consultation when such self-disclosure is unlikely to be embarrassing or socially unaccepted, whereas this same person may be unwilling when such information contains highly sensitive information. Our results showed that the level of disclosure intimacy indeed played a role in people’s decision to self-disclose, but also in which emotions they experienced afterward. Although self-disclosure intimacy has been extensively studied before, it has mainly been studied as a dependent variable (e.g., Joinson, 2001). By systematically manipulating different types and levels of intimacy in a scenario-based experimental design, we were able to shed new light on how disclosure decisions are being made, the factors that impact them, and the intrapersonal consequences they may have. Such findings help shape the current theoretical state-of-the-art knowledge on the specific rules and conditions under which people make decisions about self-disclosure, which have so far only gained limited attention in empirical research.

Our results also have practical implications for healthcare and the use of communication technologies in doctor-patient communication in particular. Our study showed that not all sorts of consultations are suitable for VMC. Our results suggest that consultations in which more intimate parts of the body should be disclosed are less suitable for VMC. This is because people are less willing to do so, but also because if people disclose physical intimate information, they can experience negative emotions afterward. This is important information to include in policy recommendations for care institutions (e.g., guidelines for healthcare professionals on emotion management when intimate disclosures are requested in VMC).

Limitations and Suggestions for Future Research

Although our scenario-based experimental design allowed us to systematically vary aspects of CMC that otherwise had been impossible or even unethical, we cannot draw strong conclusions due to lacking ecological validity. Based on the privacy paradox (Barnes, 2006), we may expect that privacy attitudes and privacy behaviors are not always aligned. Similarly, how people weighed benefits and (privacy) costs and self-disclose in hypothetical scenarios could be different in real-life. For example, in real-life it might not be clear beforehand what the patient will have to reveal during the consultation, resulting in more spontaneous self-disclosure that is not driven by a rational calculus. Moreover, as most research has used cross-sectional study designs to study the privacy calculus (Abramova, Wagner, Krasnova, & Buxmann, 2017), future research should strive to assess causality in explaining self-disclosure behaviors to capture a true representation of (rational) decision making as proposed by the privacy calculus.

Second, as the covid-19 pandemic forces regular healthcare to fast-track its digitization, leaving patients no choice but to have video consultations with their doctor, it could be that self-disclosure depends on different factors and has different consequences. Perhaps, prior experience with video consultations – which might have become more prevalent in times of covid-19 – might play a facilitating or hampering role in self-disclosure causes (e.g., expecting less or more communication barriers) and consequences (experiencing more or less positive emotions). To better grasp self-disclosure processes, future research should invest in understanding online self-disclosure in more naturalistic settings by, for example, performing direct observations of real video consultations or conducting qualitative interviews with patients and providers.

Third, as our manipulation of verbal intimacy did not lead to differential perceptions of intimacy among our participants, we cannot draw firm conclusions about the (lack of) impact of verbal intimacy on (the causes and consequences of) self-disclosure. Based on Altman and Taylor’s (1973) operationalization of disclosure intimacy, we expected that having to disclose one’s fears and concerns about a potential follow-up surgery would be considered as more intimate than disclosing primary descriptive information about wound care. However, it might be that anxiety toward surgery is common and nothing to be embarrassed about. Future research should therefore assess the construct validity of the experimental manipulations via a pilot study of the created manipulation or by using existing validated manipulations (Chester & Lasko, 2021).

Another important venue for future research is to control for potential individual differences in perceived disclosure intimacy. Although certain topics are generally perceived as more intimate than others (e.g., Pickard, Roster, & Chen, 2016), other factors, such as body (dis)satisfaction (Hassell & Cotton, 2017), interpersonal trust (e.g., Steel, 1991), and urgency (Acquisti, Brandimarte, & Loewenstein, 2015) influence people’s willingness to self-disclose intimate information as well. For example, trust in the doctor and/or the relationship with the doctor affect the amount and depth of self-disclosure (e.g., Wheeless & Grotz, 1977). Moreover, the level of urgency of a health issue may also influence self-disclosure, as privacy research has demonstrated that people tend to engage in more risky disclosures when immediate gratification (e.g., being able to discuss an urgent issue right away) outweighs the delayed, potentially negative future consequences (e.g., a privacy breach) (Acquisti et al., 2015). Exploring such factors may help explain the causes and consequences of intimate self-disclosures. Moreover, investigating a broader selection of factors simultaneously would allow to better speak about overarching patterns and contribute to enhancing generalizability of our findings and understanding of complex interactions in computer-mediated doctor-patient communication (Reeves, Yeykelis, & Cummings, 2016). In a rapidly digitizing society in which healthcare services continuously need to be adapted to maintain quality of care, this study provides important new fundamental knowledge on how video-mediated interactions between doctors and patients are shaped and impact patient well-being.

Open Scholarship

This article has earned the Center for Open Science badges for Open Data and Open Materials through Open Practices Disclosure. The data and materials are openly accessible at https://osf.io/yc4eg/?view_only=866f73554bc64504be85ce9fee9f671b

Funding

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data described in this article are openly available in the Open Science Framework at https://osf.io/yc4eg/?view_only=866f73554bc64504be85ce9fee9f671b

Notes

1. The online supplementary material (OSM) includes the data of the study, the (additional) analyses, and study materials, and can be accessed here: https://osf.io/yc4eg/?view_only=866f73554bc64504be85ce9fee9f671b.

2. Omitting covariates resulted in a significant, negative relationship between communication barriers and self-disclosure of personal health-related information (b = −0.12, se = .04, 95% CI [−0.19, −0.04], p = .002, β = −.09).

3. Omitting covariates resulted in a non-significant relationship between self-disclosure of information asked for in the scenario and experiencing positive emotions (b = 0.03, se = .02, 95% CI [−0.01, 0.08], p = .122, β = .05).