Does Provider’s Supportive Communication Enhance Trust and Treatment Adherence Attitudes in Mild and Severe Dysfunction Scenarios? A Web-Based Experimental Study

Abstract

Purpose

The way healthcare providers communicate with patients can influence patients’ trust in providers and their attitudes toward treatment adherence. The present study investigated whether such influences of providers’ communication can be modulated by dysfunction severity and participant gender.

Participants and Methods

In an online experiment, participants (N = 508) were instructed to imagine having either mild or severe olfactory dysfunction (OD). They were then randomized to watch a video of a healthcare provider delivering an olfactory training (OT) as a treatment for OD, with either neutral or emotionally supportive communication. Participants then rated their perceived trust in the provider, attitudes toward treatment effectiveness, treatment engagement, treatment unpleasantness, and treatment risk. State anxiety was measured before and after watching the videos.

Results

Supportive communication elicited higher trust in the provider than neutral communication, with a small to medium effect size (Cohen’s d = 0.36). While supportive communication did not directly affect anxiety, exploratory analyses showed that higher levels of trust in the supportive communication group were linked to lower anxiety. Moreover, supportive communication elicited higher treatment effectiveness attitude. No other treatment adherence attitudes were affected by the communication. Dysfunction severity did not affect trust in provider and treatment adherence attitudes. Men reported lower perceived treatment unpleasantness than did women, and no other gender differences emerged.

Conclusion

Supportive communication may elicit higher trust in provider independent of dysfunction severity and gender, however, likely with a smaller effect than previously assumed, and with no influence on attitudes toward adherence to treatment. Gender differences in perceived attitudes toward treatment adherence appear to be limited.

Introduction

This study examined the effects of emotionally supportive communication on perceived trust in healthcare providers and attitudes toward treatment adherence in relation to dysfunction severity and gender.

Healthcare delivery is often a communicative process between providers and patients,¹ and communication shapes how patients perceive both their care and the quality of that care.² Among different communication styles,³ a patient-centered communication is widely regarded as the gold standard,^4–6 and is often implemented through emotionally supportive communication.⁷ An emotionally supportive communication encompasses the expression of care, concern, sympathy and empathy,⁸ active listening, and showing compassion through verbal and non-verbal behaviors.^9,10

Patients must trust their provider to disclose personal information, undergo testing, and take potentially risky medication.¹¹ Emotionally supportive communication has been shown to enhance trust in treatment and the provider of that treatment.^9,10,12–14 Effective treatment requires patients to follow treatment recommendations.^15,16

Treatment adherence refers to the extent to which patients follow such recommendations,¹⁷ for instance, behaviors such as consuming medication, following a nutrition plan or physical therapy.¹⁸ Adherence attitudes present a key cognitive component of adherence, and reflect patients’ perception of the associated benefits and burden of adhering to a treatment regimen.¹⁸ Patients’ attitudes toward the treatment can, in turn, shape their intention to adhere to it,^19–21 which may itself be influenced by provider communication. According to Street et al’s (2009) model of patient-provider communication and health outcomes,²² communication affects treatment outcomes through proximal and intermediate factors around the treatment. For instance, communication between provider and patient could help the patient build trust in their provider and develop more positive attitudes toward treatment, which can in turn affect behavioral outcomes such as the patient actively adhering to treatment and having better emotional management. Consistent with this, communication with a warm provider with positive expectations has been shown to increase positive treatment expectations.^23–25 Barnes et al (2023)²⁶ found that healthy participants undergoing a three-day sham nasal drop treatment reported greater psychological well-being when interacting with a highly warm practitioner, whereas those who interacted with a less warm practitioner reported more side-effects. The present study focuses on adherence attitudes as a proximal outcome of communication, that is the cognitive component of adherence behavior. Attitudes are a potential determinant of future adherence behavior.^19–21

The impact of supportive communication on trust and attitudes toward treatment adherence may depend on situational factors such as the severity of the patient’s dysfunction.²⁷ However, severity has mostly not been considered in previous research on communication outcomes. Greater severity of dysfunction is often accompanied by emotional distress and increased need for emotional support, especially when treatment does not promise immediate improvement.²⁸ Patients presenting with more severe health problems are likely to perceive greater illness threat, uncertainty, and consequently higher emotional distress and anxiety.^29–31These emotional responses may influence how patients trust their provider.^28,32 However, surprisingly only few studies have examined this and the available evidence is mixed, as some studies suggest that more severe dysfunctions may lead to higher trust in providers,³³ whereas others³⁴ report greater trust in less severe dysfunctions. Peerdeman et al (2021)³⁵ found that portraying a physician as competent and empathic increased analog patients’ trust in the provider, independent of the severity of an imagined skin condition (a mole or melanoma). Moreover, little is known about the impact of dysfunction severity on the link between supportive communication and attitudes toward treatment adherence. In general, patients with more severe dysfunction tend to adhere less to treatment than those with lower dysfunction severity.^36,37

Previous studies suggest potential gender differences in patient-provider communication.³⁸ For instance, women are more likely than men to prefer a provider using emotional communication styles,³⁹ but evaluate supportive messages as more critical than men, and are less satisfied with the support they receive.⁴⁰ Moreover, women tend to show higher trust in their physicians than men,^41–43 whereas men with chronic conditions are more likely to adhere to treatment than women.^44,45 Such effects may be modulated by anxiety levels and patient-provider communication.^{24,25,46–48}

Despite growing research on patient-provider communication, trust, and attitudes toward treatment adherence, several important gaps remain. Characteristics such as dysfunction severity and gender remain underexplored, despite their potential to modulate the relationship between communication, trust, and attitudes toward treatment adherence.^27,35 Moreover, existing studies are mostly observational and lack mechanistic insight.^10,12,14 The few available experimental studies are limited¹³ and often rely on written depictions of communication.³⁵ This web-based experimental vignette video study was conducted to address these gaps.

Healthy participants, acting as analog patients,⁴⁹ were instructed to imagine having a reduced sense of smell (olfactory dysfunction, OD) following a COVID-19 infection based on a vignette describing two severity ODs of either mild or severe dysfunction. Participants were then randomized to watch a videotaped provider describing an olfactory training (OT) program with either neutral or supportive communication. Participants rated their trust in provider and their attitudes toward treatment adherence, including perceived treatment effectiveness, engagement, unpleasantness, and risk. Prior to data collection, the dysfunction severity vignettes were validated by a physician, and the communication videos were validated by a group of trained coders, based on a study-specific coding log (See Methods for details).

The present study used analog patients. Although this approach relies on hypothetical rather than real clinical experiences, it has been shown to provide valuable insights for research and clinical practice.^49–51 Moreover, analog patient studies offer a unique opportunity to manipulate and study factors that are often difficult to investigate in patient samples.⁴⁹ Nevertheless, because participants responded to a hypothetical scenario rather than a lived experience, the findings may not fully generalize to clinical populations or to long-term outcomes of patient-provider communication.⁵¹

The following hypotheses were tested: supportive communication increases perceived 1) trust in provider, 2) attitudes toward treatment effectiveness, 3) treatment engagement, and reduces perceived 4) treatment unpleasantness and 5) treatment risk. Furthermore, in severe dysfunction, supportive communication leads to 6) higher perceived trust in provider, 7) higher attitudes toward treatment effectiveness, and 8) higher treatment engagement . 9) Women in the supportive communication group report higher perceived trust in provider than men, and 10) men report better perceived adherence attitudes (across all items) than women. Lastly, 11) supportive communication reduces anxiety, and 12) communication style modulates the relationship between anxiety, perceived trust in provider and treatment adherence attitudes.

Material and Methods

Participants

Healthy adults (≥ 18 years old) who were proficient in Norwegian participated in this study. No further exclusion criteria were applied. Participants were recruited through YouGov Omnibus, an online polling agency in Norway. As a compensation for participation, participants received points that could be redeemed as gift cards. Data were collected on YouGov between June 19 and June 26, 2025.

The sample consisted of 508 healthy participants, including 267 women (52.6%) and 241 men (47.4%). Age range was 18–86 years old, with a mean age of 49.78 years (SD = 16.74). A total of 36.4% (n = 185) of the sample reported education at high school level, 30.3% (n = 154) reported short university education (1–3 years), 25.4% (n = 129) reported long university education (4+ years), 6.9% (n = 35) reported primary school education, and 1.0% (n = 5) did not want to disclose their highest education level. A total of 30.7% (n = 156) of the sample were from Innlandet, Østfold, Akershus, and Buskerud, 25.4% (n = 129) were from Vestlandet, 16.5% (n = 84) were from Trøndelag or Nord-Norge, 14.4% (n = 73) were from Agder or Sør-Østlandet, and 13.0% (n = 66) were from Oslo.

Online Survey

An online survey was administered on the YouGov Omnibus website. Participants first entered their demographic information (gender, age, education, and place of residence) and gave their consent. They were then navigated to an audio control page to verify whether their devices could play the sounds. Afterwards, participants reported their anxiety (state and trait) and were randomly assigned to mild or severe OD vignettes and then to a videotaped provider with either supportive or neutral communication. Finally, participants rated their perceived trust in provider, their attitudes toward treatment adherence, and their state anxiety. The average survey completion time was 12.86 minutes (M = 12.86, SD = 5.07). For more details, see the Procedure.

Randomization

This study included two randomizations performed by YouGov. First, the participants were randomized to either a mild or a severe OD severity vignette. Next, they were randomly assigned to watch a provider with either a neutral or a supportive communication style. Randomization blocks (Dysfunction severity × Communication) were compared for age, gender, education, region, and baseline state and trait anxiety using ANOVA for continuous variables and Pearson’s Chi-square tests for ordinal variables. No significant differences were found in age, gender, region, education, or baseline state or trait anxiety (for more details, see Supplementary Materials S1).

Olfactory Dysfunction (OD) Severity

Two written vignettes were developed to describe the OD following COVID-19 infection. The OD vignettes differed in the severity, one depicting a mild case of OD with partial loss of smell (hyposmia), and the other portraying a severe case of OD with complete loss of smell (anosmia).⁵² Both vignettes followed the same structure and addressed the same theme, with wording modified only where it was necessary to imply differences in severity.

Participants were first asked to imagine themselves in a situation where they experienced either mild or severe OD following the COVID-19 infection. Each vignette illustrated how these changes in olfaction diminished food enjoyment, ability to detect hazards and increased worries about personal odors. These themes were chosen based on typical reported impairments of individuals experiencing olfactory dysfunction.^53,54 The vignettes described these situations as either mild or severe dysfunctions, and concluded with a description that the participant decided to join an OT program study designed to improve the sense of smell post-COVID-19 infection. This served as an introduction to the next phase of the study, where participants were asked to watch a video featuring a provider showing either neutral or emotionally supportive communication while carrying out the OT program (For a detailed description of the vignette development and content, see Supplementary Materials S2, Table S2).

Mild Dysfunction

Participants were instructed to imagine a reduced sense of smell. They can still detect strong odors; however, subtle smells are more difficult to detect. They experience a reduced sense of taste and their ability to detect hazards such as smoke is reduced. They sometimes worry about their personal odor and sometimes need to ask their friends and family about whether they smell unpleasant.

Severe Dysfunction

Participants were instructed to imagine that their sense of smell had completely disappeared. They cannot smell anything, and even strong odors are impossible to detect. Their sense of taste is severely reduced, and they are unable to detect hazards such as smoke. They are constantly worried about their personal odor, and often ask friends and family whether they smell unpleasant as they cannot tell themselves.

Validation Checks

To validate the OD vignettes, a physician reviewed them and confirmed that the vignettes accurately depicted OD conditions that varied appropriately in severity from mild to severe. The physician provided minor feedback on the vignettes, and the feedback was applied to the vignettes before they were used in the survey.

Provider Communication

Participants were randomized to watch videotaped recordings of a female provider who demonstrated the OT program with either neutral or supportive communication. The videos depicted a female provider (played by a professional actor) interacting with a patient (portrayed by a confederate). The confederate was neither visible nor audible in the recording; however, the scene was structured to imply that the provider actively listened and responded to the patient. First, the provider welcomed the patient and asked them to sign a consent form. The provider then described a study that tested the effects of an OT program on OD. Next, the provider informed them that they needed to examine the olfactory function of the patient using a test.⁵⁵ Finally, the provider gave the patient instructions on how to perform the OT program. The video recordings followed the same storyline, conveyed the same information, and differed only in communication style, which was either neutral or supportive (Figure 1). The female actor (the ‘provider’) received approximately 25 hours of training on how to convey the intended communication styles. Non-verbal behaviors were guided during training, but not scripted, allowing the videotapes to remain as natural as possible. Transcripts of the provider communication videos content are provided in Supplementary Materials S3, Table S3.

Figure 1.

Neutral and Supportive Communication Styles.

Notes: The upper row: Neutral communication; The lower row: Supportive communication.

Supportive Communication

Following an approach similar to that of Kaptchuk et al (2008),⁵⁶ the provider expressed empathic statements (eg, “I understand that this must be a difficult concern to have”), communicated positive expectations about the effectiveness of OT (eg, “I believe that this training can help you get better if you follow the training routine”; “Previous research has shown that this training is effective”), actively listened to the patient (eg, by repeating the patient’s words), showed an interest in the patient’s olfactory trajectory and asked personal questions how the OD impacted their quality of life (eg, “When did you get infected?”; “How has your sense of smell impacted your everyday life?”). Moreover, the provider gave more practical tips and examples on how to undergo the training (eg, “It can be helpful to do this in front of a mirror”; “Maybe this odor reminds you of Christmas”). Furthermore, the provider showed positive non-verbal behaviors, such as enhanced eye contact, smiling, open and positive body movements (eg, leaning forward), friendly tone of voice, and positive facial expressions.⁵⁷

Neutral Communication

In neutral communication, the provider delivered essential information concisely and in a matter-of-fact manner. The provider did not include empathic statements, expressions of expectations regarding treatment effectiveness, or personal inquiries about the patient’s experiences. Interaction was limited to the provision of basic instructions related to the OT without elaboration, examples, or additional guidance. The provider did not engage in reflective listening (eg, repeating the patient’s words) or explore the patient’s condition, history, or its impact on daily life. As for non-verbal behaviors, the provider showed limited eye contact, smiling and body movements and used a monotonous, flat tone of voice. Importantly, positive non-verbal behaviors were reduced in this communication, but not completely absent, to avoid creating negative communication.^58–60

Proof-of-Concept

To ensure the validity and reliability of the communication styles, the videos were evaluated by six (three women) coders with backgrounds in psychology, social work, and healthcare. The coders received training from the first author (RM) on how to code for verbal and non-verbal communication. Coders received 30-minute individual training based on a study-specific guideline covering the study aim, the purpose of coding, and definitions of all coding categories (supportiveness, warmth and friendliness, empathy, interest in patient, expressiveness, and positivity), without being informed about the experimental manipulations. Coding was performed following the “general impression approach”⁶¹ on a coding log designed for the present study, following a similar approach to Daniali et al (2023, 2024).^60,62 Results of the proof-of-concept showed that supportive communication received higher ratings on all coding categories compared to neutral communication, confirming the construct validity that communication videos differed in their level of supportiveness. The coders had excellent inter-rater reliability, and the coding cateogries had good to excellent reliability. Descriptions of the coding log, procedure, and results of the proof-of-concept are provided in the Supplementary Materials S4 and S5, including Table S4, Figure S4, Tables S5A and B.

Measures

Demographics

Age, gender, place of residence and highest education level were collected as demographic data.

Perceived Trust in Provider

Perceived Trust in Provider was measured by Trust in Physician Scale⁶³ which is an 11-item anchored on a five-point Likert Scale ranging from “strongly agree” to “strongly disagree” measuring trust in provider based on statements such as “I doubt that my doctor would really care about me as a person”. The scale yields three dimensions: dependability (looking out for patients’ best interests), confidence in physicians’ knowledge and skills, and confidentiality and reliability of information between physicians and patients.⁶³ Items 1, 5, 7, and 11 are reverse-coded, and the total score ranges are 11–55. The original items were adapted for use in the present study. For instance, item 1, “I doubt that my doctor would really care about me as a person” was changed to “I doubt that the provider would really care about me as a person.” Item 9 of the original scale “My doctor is a real expert in taking care of medical problems like mine” was changed to “This provider is a real expert in taking care of olfactory dysfunction”. The term “doctor” was replaced in all items with the term “provider.” To have consistent wording throughout the questionnaire with other scales in the battery, the original anchors “agree/disagree” were changed to “somewhat agree/somewhat disagree”, and “neutral” was changed to “neither agree nor disagree.” The scale was translated into Norwegian, followed by the first and third authors (RM and MAF, respectively). The internal reliability of the scale in the present study was Cronbach’s α = 0.90.

Treatment Adherence Perception Questionnaire (TAP-Q)

TAP-Q measures attitudes toward treatment adherence through perceived treatment behavior, perceived treatment benefit and perceived burden of treatment.¹⁸ Three items of TAP-Q were used in this study. Item “3” (“Over the next week, how well do you think you are likely to follow these recommended actions?”) was used to measure attitudes of perceived treatment engagement, item “8” (“Assuming you do the recommended actions, how effective do you think these actions are in accomplishing the things they are supposed to accomplish?”) was used to measure perceived treatment effectiveness, and item “14” (“How much would you agree that doing the recommended actions can be annoying or bothersome for you?”) was used to measure perceived treatment unpleasantness. The three items were slightly changed to match the context of the study. Item “3” was changed to “How well do you think you would follow the recommended actions (training three times a day for twenty days)?”. The anchoring of the item was also changed to a percentage slider from “0%” to “100%”; item “8” was changed to “How effective do you think this training is in improving olfaction” and participants provided their response on a scale from “not effective at all” corresponding to “1” to “extremely effective” anchored to “5”; and item “14” was changed to “How much would you agree that doing the olfactory training could be annoying or bothersome for you?”. The response scale was changed to “Strongly disagree” anchored to “1”, to “Strongly agree” anchored to “5” to have consistent wording throughout the questionnaire.

Treatment Expectations Questionnaire (TEX-Q)

TEX-Q measures treatment expectations, including harm expectations such as perceived treatment risk.⁶⁴ Item “7” of TEX-Q (“To what extent do you expect risks from the treatment”) was translated to Norwegian and used in this study to measure perceived treatment risk after slightly adjusting the wording and response rating. The question was changed to “How much risk do you think olfactory training might involve”. The response rating was also changed to a slider ranging from “0 – no risk” to “10 – extreme risk.”

State and Trait Anxiety (STAI)

State and Trait Anxiety were measured using STAI.⁶⁵ The state scale measures the intensity of how anxious individuals feel at a specific moment in terms of tension, nervousness, worry, and apprehension, whereas the trait scale measures a person’s general tendency toward anxiety-proneness.⁶⁶

The original STAI has 40 items: 20 for state and 20 for trait anxiety. The items are measured on a 4-point Likert scale, for state anxiety ranging from “Not at all” to “Very much so,” and for trait anxiety from “Almost never” to “Almost always”.⁶⁵ Scores for the state and trait scale each can range from 5 (lowest anxiety) to 20 (highest anxiety). Following Zsido et al (2020),⁶⁷ a short version of STAI was used in this study, including items 6, 9, 12, 13 and 18 from the state anxiety scale and item 28, 29, 37, 38 and 40 from the trait anxiety scale.⁶⁷ A Norwegian translation was used.⁶⁸ Trait anxiety was measured before watching the video of the provider and state anxiety before and after watching the assigned video in this study.

Perceived Social Support

Participants rated the perceived availability of social support in their personal lives using the Multidimensional Scale of Perceived Social Support (MSPSS).⁶⁹ The MSPSS consists of 12 items assessing perceived social support from family, friends, and significant others. Each item is rated on a 7-point Likert scale ranging from “1” (very strongly disagree) to “7” (very strongly agree). Scores can be averaged to yield an overall perceived social support score ranging from “1” to “7”, with higher scores indicating greater perceived social support. A Norwegian translation was used in this study.⁷⁰ The results were not investigated in the present study.

Procedure

Participants were recruited through YouGov Omnibus. They were first asked to state their gender, age, education, and place of residence before they were directed to a consent form. As part of the consent form, participants were informed that the study aimed to investigate how illness conditions, provider characteristics, and personal traits influence perceptions of trust and treatment expectations. They were further told that the study involved completing a questionnaire based on a hypothetical scenario which they were asked to imagine themselves in, and that participation included watching a video, therefore requiring a device capable of playing sound.

To prevent bias, no information about the different types of dysfunction severity and communication was passed to the participants.

After providing consent, participants completed an audio control page to ensure their device was capable of playing sound. They watched a video showing a pig making sheep sounds (bleats). Following this, participants were asked to identify the animal they saw and what sound the animal was making. This audio control page was included to ensure that participants could complete the study as intended. Participants who answered incorrectly were excluded from this study.

Following the audio control page, participants reported state and trait anxiety and perceived availability of social support. They were then directed to a page stating that they will now be presented with a scenario to imagine themselves in. It was made clear that it was important to carefully read the following information to answer the remaining questions. Subsequently, participants were randomly assigned to either the mild or severe dysfunction severity vignette. To maintain participants’ attention, the vignette was divided into two pages. The first page described the severity of dysfunction and the second page introduced the OT program. Each page was displayed for at least 10 seconds before the participants could proceed to the next page.

After reading their assigned vignettes, participants were randomly assigned to either the neutral or supportive communication videos. Participants had to watch the video in its entirety before moving on to the next page. Following the video, participants rated their perceived trust in provider, perceived attitudes toward adherence to the OT program, and their state anxiety level.

Power and Sample Size Calculation

Through a similar study design, Peerdeman et al (2021)³⁵ compared low- and high-empathy provider communication scenarios in analog patients and reported a partial eta squared (η_p²) of 0.14 corresponding to a large effect size of Cohen’s d = 0.81.⁷¹ However, because our study differed in several methodological respects, eg, hypothetical condition (olfactory dysfunction), a broader communication manipulation (verbally and non-verbally), and the use of videotaped providers, we conservatively assumed a medium effect size of 0.25. Accordingly, a priori power analysis was conducted in G*Power⁷² for a between-subjects ANOVA (8 groups; fixed effects, special, main effects and interactions), using an effect size of f = 0.25, α = 0.05, and power = 0.95. The analysis indicated that a minimum of 210 participants would be required. However, to provide greater power for detecting interaction effects, account for potential error margins, attrition probability, mitigating violation of assumptions, and the online nature of the study, we aimed to recruit a larger sample of approximately 500. The final sample consisted of 508 respondents.

Design and Statistics

Data analysis was performed using JASP.96. The significance level was set at p < 0.05.

The study employed a factorial between-subjects design with communication (supportive vs neutral) × dysfunction severity (mild vs severe) × gender (men vs women) 2×2×2 factors.

To examine the main and interaction effects of communication, dysfunction severity, and gender, five separate three-way ANOVAs were conducted with ‘perceived trust in provider’, attitudes toward ‘perceived treatment effectiveness’, ‘perceived treatment engagement’, ‘perceived treatment unpleasantness’, and ‘perceived treatment risk’ as dependent variables. The normality of residuals was tested using the Kolmogorov–Smirnov test with Lilliefors correction, and homogeneity of variances was tested using Levene’s test. Residuals were normal for perceived trust in provider, but not for perceived treatment effectiveness, engagement, unpleasantness, and risk. However, because homogeneity of variances was met for all outcomes and given the large sample size and balanced design, and ANOVA’s robustness to non-normality,^73,74 ANOVA analyses were conducted as planned.

To investigate whether communication affected state anxiety, a repeated-measures ANOVA was conducted with (state) anxiety measured before and after watching the video as the within-subject factor, and communication (two levels), dysfunction severity (two levels), and gender (two levels) as the between-subject factors.

All significant interaction effects were followed by pairwise comparisons using Bonferroni-corrected estimated marginal mean values.

To verify the effect sizes, we conducted bootstrapped analyses (1,000) samples and inspected bootstrapped 95% confidence intervals for mean scores and mean differences for each significant main and interaction effect to rule out Type I errors. The effect sizes were interpreted using Cohen’s thresholds for small (d = 0.2), medium (d = 0.5), and large (d = 0.8).⁷¹

For exploratory investigations, correlations between all dependent variables in the overall data and within each communication group were assessed using Pearson’s two-tailed correlations. Next, to investigate the relationship between the dependent variables (trust in provider, treatment adherence attitudes and state anxiety) across communication, moderation analyses were conducted using moderation analysis with PROCESS macro⁷⁵ applying model 1 with 5000 bootstrap samples.

Ethics

This study was approved by the Ethics Committee of Department of Psychology at NTNU (project number: C-REP2026/008). All participants provided written informed consent on participation and publication of anonymized responses before participation. The actor provided written informed consent for participation and publication of the image.

Results

Descriptive Statistics

Means and standard deviations for perceived trust in provider, attitudes toward treatment effectiveness, treatment engagement, treatment unpleasantness, and treatment risk across communication, dysfunction severity, and gender are reported in Table 1.

Table 1.

Perceived Trust in Provider and Attitudes Toward Treatment Adherence Across Communication, Dysfunction Severity and Participant Gender (Mean, SD)

Group Outcome	NC (n = 259)	SC (n = 249)	MD (n = 251)	SD (n = 257)	Female (n = 267)	Male (n = 241)
Trust in provider (Scale 11–55)	38.24 (7.62)	41.02 (7.79)	39.48 (7.85)	39.72 (7.81)	39.24 (7.86)	40.01 (7.77)
Treatment effectiveness (Scale 1–5)	2.57 (0.85)	2.87 (0.90)	2.78 (0.83)	2.65 (0.94)	2.72 (0.89)	2.71 (0.89)
Treatment engagement (Scale 0–100)	72.29 (27.97)	74.22 (28.38)	71.41 (28.74)	75.01 (27.53)	72.93 (27.96)	73.57 (28.45)
Treatment unpleasantness (Scale 1–5)	3.42 (1.12)	3.30 (1.14)	3.40 (1.07)	3.33 (1.18)	3.49 (1.08)	3.22 (1.17)
Treatment risk (Scale 0–10)	1.67 (1.96)	1.70 (2.02)	1.87 (2.07)	1.51 (1.89)	1.69 (1.98)	1.68 (2.01)

Note: N = 508.

Abbreviations: NC: Neutral Communication; SC: Supportive Communication; MD: Mild Dysfunction, SD: Severe Dysfunction.

Trust in Provider

The main effect of communication (F(1, 500) = 16.26, p < 0.001, η_p² = 0.03) showed that participants reported higher perceived trust in supportive communication (M = 41.00, SE = 0.49) than in neutral communication (M = 38.22, SE = 0.48) (Mean diffs ≥ 2.78, SE ≥ 0.69, p < 0.001; Cohen’s d = 0.36; 95% bootstrapped CI [−4.06, −1.47]). (Figure 2, left panel). No other main or interaction effects were observed (ps > 0.05).

Figure 2.

Perceived Trust in Provider and Treatment Effectiveness by Communication Style.

Notes: Supportive communication (green bar) elicited higher perceived trust in provider (left-hand panel), and larger perceived treatment effectiveness (right-hand panel) than the neutral communication (orange bar). *** p <0.001; Error bars ± 2 SE.

Treatment Adherence Attitudes

Treatment Effectiveness Attitude

The main effect of communication (F(1, 500) = 13.88, p < 0.001, η_p² = 0.03) showed that participants reported higher perceived treatment effectiveness in the supportive communication group (M = 2.86, SE = 0.06) than in the neutral communication group (M = 2.57, SE = 0.06) (Mean diffs ≥ 0.29, SE ≥ 0.08, p < 0.001; Cohen’s d = 0.33; 95% bootstrapped CI [−0.45, −0.13]) (Figure 2, right panel). No other main or interaction effects were observed (ps > 0.05).

Treatment Engagement Attitude

No main or interaction effects of communication (F(1, 500) = 0.84, p = 0.36), dysfunction severity (F(1, 500) = 2.27 p = 0.13), or gender (F(1, 500) = 0.02, p = 0.89) were significant.

Treatment Unpleasantness Attitude

There were no main effects of communication (F(1, 500) = 1.87, p = 0.17) or dysfunction severity (F(1, 500) = 0.45, p = 0.50). However, there was a main effect of gender (F(1, 500) = 6.65, p = 0.01, η_p² = 0.01), which showed that compared to women (M = 3.48, SE = 0.07), men (M = 3.22, SE = 0.07) perceived less treatment unpleasantness (Figure 3) (Mean diffs ≥ −0.26, SE ≥ 0.10, p = 0.01; Cohen’s d = 0.23; 95% bootstrapped CI [−0.45, −0.06]. No other main or interaction effects were observed (ps > 0.05).

Figure 3.

Treatment Unpleasantness by Gender.

Notes: Men (pink bar) reported lower perceived treatment unpleasantness than women (blue bar). ** p = .01; Error bars ± 2 SE.

Treatment Risk Attitude

There was no main effect of communication (F(1,500) = 0.00, p = 0.99) or gender (F(1,500) = 0.00, p = 0.99); however, a main effect of dysfunction severity (F(1, 500) = 3.99, p = 0.046, η_p² = 0.008) emerged. Participants who were instructed to imagine having a mild dysfunction perceived the OT as riskier (M = 1.87, SE = 0.13) than participants instructed to imagine having a severe dysfunction (M = 1.51, SE = 0.13) (Mean diffs ≥ 0.36, SE ≥ 0.18, p = 0.046). However, the effect size was unreliable (Cohen’s d = 0.18); 95% bootstrapped CI [−0.69, 0.02]. No other main or interaction effects were observed (ps > 0.05).

State Anxiety

The main effect of time (F(1, 500) = 101.99, p < 0.001, η_p² = 0.17; Cohen’s d = 0.90), showed that state anxiety decreased from before to after watching the videos with no difference between neutral and supportive communication (p = 0.86). No other theoretically relevant main or interaction effects were observed (ps > 0.05).

Exploratory Analysis

The associations among perceived trust in provider, treatment adherence attitudes, and state anxiety (measured after watching the videos) in the overall data and across communication are presented in Table 2.

Table 2.

Correlation Between Perceived Trust in Provider, Attitudes Toward Treatment Adherence and State Anxiety After Watching the Video

Overall (N = 508)	1	2	3	4	5
1)Trust in provider
2)Treatment effectiveness	0.48***
3)Treatment engagement	0.49***	0.41***
4)Treatment unpleasantness	−0.25***	−0.22***	−0.30***
5)Treatment risk	−0.22***	−0.09*	−0.15***	0.07
6)Anxiety	−0.14**	0.06	−0.11*	0.10*	0.38***
In Supportive Communication (N = 249)
1)Trust in provider
2)Treatment effectiveness	0.51***
3)Treatment engagement	0.52***	0.42***
4)Treatment unpleasantness	−0.23***	−0.21***	−0.27***
5)Treatment risk	−0.32***	0.05	−0.22***	0.13*
6)Anxiety	−0.24***	0.08	−0.10	0.16*	0.49***
In Neutral Communication (N = 259)
1)Trust in provider
2)Treatment effectiveness	0.43***
3)Treatment engagement	0.47***	0.39***
4)Treatment unpleasantness	−0.26***	−0.22***	−0.32***
5)Treatment risk	−0.14*	−0.13*	−0.09	0.02
6)Anxiety	−0.05	0.05	−0.11	0.05	0.27***

Notes: 2-tailed, * p <0.05, ** p <0.01,***p <0.001.

Overall, perceived trust in provider was positively associated with treatment effectiveness attitude and engagement, and negatively correlated with treatment unpleasantness and risk attitudes. In addition, treatment effectiveness attitude was positively correlated with treatment engagement and negatively correlated with unpleasantness attitudes in both groups. Treatment engagement was also negatively correlated with treatment unpleasantness in both groups.

In supportive communication, perceived trust in provider was negatively associated with state anxiety. Moreover, in supportive communication, treatment risk attitude was negatively correlated with treatment engagement attitude and positively correlated with treatment unpleasantness attitude. However, these associations were not observed in neutral communication. In neutral communication, treatment risk attitude was negatively correlated with treatment effectiveness attitude. To understand why these associations differed between communication groups, we conducted follow-up moderation analyses using communication as a moderating factor.

Trust in Provider and Anxiety

The negative association between anxiety and trust in provider was observed only in supportive communication (Table 2). We conducted a moderation analysis with perceived trust in provider as the predictor, communication as the moderator, and (post-video) anxiety as the outcome. This model was selected based on previous studies.^24,25,46

The overall model, R² = 0.02, F(3, 504) = 4.95, p = 0.002, indicated an interaction of perceived trust by communication (b = −0.01, SE = 0.01, t = −2.06, p = 0.039, 95% CI [−0.02, −0.00]. Examination of conditional effects showed that in neutral communication, perceived trust in the provider was not related to anxiety (b = −0.00, SE = 0.00, t = −0.83, p = 0.407). However, in supportive communication, higher perceived trust was significantly associated with lower anxiety (b = −0.02, SE = 0.00, t = −3.75, p < 0.00) (Figure 4).

Figure 4.

Trust in Provider and Anxiety.

Notes: Trust in P: Trust in Provider; Moderation analysis with ‘trust in provider’ as predictor, ‘communication style’ as the moderator, and ‘anxiety’ as the outcome. Higher percieved trust in provider was associated with lower anxiety, but only in the supportive communication group.

Trust in Provider and Perceived Treatment Effectiveness

Since participants in the supportive communication group reported higher perceived attitude toward treatment effectiveness, and as treatment effectiveness was correlated with perceived trust in provider in both groups, a moderation analysis was performed with perceived trust in provider as a predictor, treatment effectiveness attitude as the outcome, and communication as the moderator. The results indicated that the interaction between perceived trust in provider and communication groups was not significant (b = 0.01, SE = 0.00, t = 1.15, p = 0.24).

Communication did not moderate any other significant association with treatment adherence attitudes.

Discussion

The present study examined the impact of emotionally supportive communication on perceived trust in providers and treatment adherence attitudes in relation to dysfunction severity and gender.

Several findings emerged. Participants who watched the provider with supportive communication reported higher perceived trust in provider and higher perceived treatment effectiveness than participants who watched the provider with neutral communication. However, supportive communication did not influence attitudes toward perceived treatment engagement, perceived treatment unpleasantness, and perceived treatment risk. Moreover, although supportive communication did not directly reduce anxiety, exploratory moderation analysis showed that higher perceived trust in provider in this group was associated with lower anxiety. In supportive communication, attitude toward treatment risk was negatively associated with treatment engagement attitude and positively associated with treatment unpleasantness attitude. In neutral communication, attitude toward treatment risk was negatively correlated with treatment effectiveness attitude. However, none of these associations were moderated by communication. Dysfunction severity did not affect perceived trust in provider or any attitudes toward treatment adherence. Men perceived treatment as less unpleasant than women, and no other gender differences emerged.

Compared to the neutral communication, supportive communication elicited higher perceived trust in provider, confirming hypothesis 1, and previous research linking supportive communication to higher trust in provider.^{9,10,12–14,35} Trust in provider has been described as a direct outcome of patient-centered communication,^22,27 and being emotionally supportive is part of patient-centered care.⁷ The supportive provider in the current study expressed empathic statements, displayed active listening and signaled interest in the patient, which are characteristics recognized as qualities that help patients to build trust.⁷⁶ The experimental design of the present study provides support for what has been previously reported through observational investigations.

While Peerdeman et al (2021)³⁵ reported a large effect of communicating empathy on trust in provider (N = 237; Cohen’s d = 0.81 for empathy manipulation), the present study found a small-to-medium effect (Cohen’s d = 0.36) of supportive communication on perceived trust in provider. Although both studies used healthy participants as analog patients, their focus and methodology differed. Peerdeman et al (2021)³⁵ used written descriptions to portray competence and empathy, while the present study illustrated supportive communication through a video recording of a real person displaying emotional support through both verbal and non-verbal behaviors. In addition, the present study included a sample roughly twice as large (N = 508, 52.6% female, mean age of 49.78 years), compared to the younger and predominantly female sample in Peerdeman et al (2021)³⁵ (89% female, 94% ≤ 26 years). Taken together, the present study extends previous research by examining supportive communication using a more ecologically valid manipulation that incorporates both verbal and non-verbal communication. The observed effects here were smaller than previously reported.³⁵ Therefore, the present findings suggest that the influence of supportive communication may be more modest, especially when the communication is tested in a more holistic approach. However, other factors such as use of analog patients, cultural differences or disparities in the communication styles between the studies (empathic vs supportive) may be taken into account.

The present study proposes a methodological framework for investigating provider communication that integrates both verbal and non-verbal behaviors while maintaining experimental control. The study successfully developed and tested the desired communication styles through the use of verbal and non-verbal communication scenarios, provider training, and validation of communication style (see Supplementary Materials S3–S5 including Tables S3, S4, Figure S4, Tables S5A and B for details). This suggests that the experimental investigation of provider communication, which is often considered challenging due to it’s multidimensional nature,⁷⁷ is possible through a structured, holistic approach. The present study provides a pragmatic framework for experimentally examining communication in a more realistic manner.

Anxiety levels were the same after watching the videos in both communication groups, disconfirming hypothesis 11. This is in contrast with previous studies reporting that supportive communication may reduce anxiety,^24,25 which could be partly due to the healthy nature of the participants, who inherently did not have high anxiety levels. An exploratory moderation analysis, however, showed that, in supportive communication group, higher trust was associated with lower state anxiety, partly confirming hypothesis 12. Aligned with previous studies,^24,25,46,47 this finding suggests that higher trust in provider may facilitate the mitigating effects of trust in providers on anxiety levels. However, since trust was not manipulated in this experiment, firm causality could not be established. Moreover, the model accounted for only 2% of the variance in anxiety, which may be attributable to the low baseline anxiety levels due to the online and analog nature of the study. Although modest in size, consistent with previous studies^24,25,46,47 this model indicates that supportive communication may contribute to the effects of trust in provider on anxiety levels. However, due to their exploratory nature, the findings should be interpreted with caution and further investigation is warranted.

Compared to the neutral communication, participants in the supportive communication group reported higher attitude toward treatment effectiveness, confirming hypothesis 2, and aligned with previous research,^23–26 that suggested supportive communication may enhance expectations of positive treatment outcomes. Importantly, perceived treatment effectiveness was the only attitude that was explicitly addressed verbally by the videotaped provider. At the same time, supportive communication did not moderate the relationship between trust in provider and treatment effectiveness, disconfirming hypothesis 12, indicating that the higher perceived trust observed in the supportive communication group did not translate into higher perceptions of treatment effectiveness. This implies that supportive communication was likely not responsible for the higher perception of treatment effectiveness, but the verbal expression of positive outcome expectations in this group. Consistent with this, van Osch et al (2017)²⁵ found that warm and empathic communication alone did not raise positive expectations about treatment, unless they were verbally raised by the videotaped physician. Moreover, in the neutral communication group, attitude toward treatment effectiveness was negatively correlated with treatment risk attitude, whereas this association was absent in the supportive communication group, meaning that higher attitude toward treatment effectiveness was not linked to the perception of lower treatment risk attitude, suggesting that the absence of articulation of positive expectations of outcomes might be linked to perceptions of higher treatment risk attitude. However, as correlations were inspected for exploratory reasons, the findings should be interpreted with caution.

Communication did not affect attitudes toward treatment engagement, treatment unpleasantness, or treatment risk, disconfirming hypotheses 3, 4, and 5. This supports previous studies²⁵ that communication style may not influence treatment adherence attitudes.

Dysfunction severity did not show an interaction with communication, either on trust or treatment adherence attitudes, disconfirming hypotheses 6, 7 and 8, respectively. The link between dysfunction severity and patient-provider communication is mixed. Hall et al (2001)³³ argued that experiencing a medical condition makes a patient vulnerable and dependent on the provider, and that a higher vulnerability, consequently, can increase the potential for higher trust. In contrast, Liu et al (2024)³⁴ showed that when patients perceived less severe threats from their disease, the impact of doctor-patient communication (measured as communication focusing on content, empathy and outcome when talking to a patient) on raising trust was stronger than when they perceived the threat of their disease as more severe. Others, such as Peerdeman et al (2021)³⁵ showed that written descriptions of empathy increased participants’ trust in physicians, regardless of whether participants were presented with a low- or high-severity skin condition. The present findings align with Peerdeman et al (2021)³⁵ and suggest that supportive communication may enhance perceived trust in provider regardless of dysfunction severity, indicating that the beneficial effects of a supportive communication may not be limited to severity levels of health-related concerns. However, the absence of interaction effects between communication and dysfunction severity should be interpreted with caution, as the present study may have been underpowered to detect small interaction effects, at least within the context of analog patients.

To the best of our knowledge, this is the first study to investigate the role of dysfunction severity in the relationship between supportive communication and treatment adherence attitudes, with the results suggesting no effect. Gao et al (2000)⁷⁸ reported no relationship between disease severity and perceptions of treatment benefits (eg, expectations of medication prolonging life or decreasing chance of getting worse) or treatment barriers (eg, stress of treatment and side effects) in HIV/AIDS patients. Although behavioral treatment adherence rates have been reported to be lower in severe health conditions,³⁷ the present study suggests that treatment adherence attitudes¹⁸ may not be affected by the severity of dysfunction. However, it is important to distinguish that the present study measured immediate perceptual outcomes and attitudes related to an imagined condition, and not subsequent behavioral adherence over time. As noted above, the null findings should be interpreted with appropriate caution.

Gender did not modulate the impact of communication on participants’ perceived trust in provider. There was no difference in perceived trust in provider between women and men, disconfirming Hypothesis 9. These results contrast studies showing gender differences in trust in providers.^41–43 Previous research on gender differences in patient–provider communication has yielded mixed findings.^{38–40,79,80} However, previous studies did not account for the role of communication in gender differences in trust in providers. The current findings suggest that supportive communication may have similar effects in both men and women in their trust to their provider. In line with this, Nwosu et al (2025)⁸¹ recently reported that gender did not moderate the observed positive relation between perceived physician communication (how patients evaluated their physicians skill to communicate information, show empathy and respond to concerns) on emotional and cognitive trust in provider. They argued that both genders value patient-centered communication from providers.

Men perceived the treatment as less unpleasant than women, while there were no gender differences across the remaining treatment adherence attitudes. Therefore, Hypothesis 10 was partly confirmed. The results suggest that there may be gender differences in some aspects of adherence attitudes, in this case unpleasantness, but not across all aspects. While previous studies have shown that men are more likely to adhere to treatment than women,^44,45 the present study showed that men may also perceive less treatment unpleasantness than women, which is partly in line with these documents. However, a potential factor contributing to the observed gender differences in treatment unpleasantness could be that men were exposed to a female (videotaped) provider. Prior research has shown that male patients report higher satisfaction to female providers⁸² and have a tendency to report lower symptoms such as pain to the provider of the opposite sex.⁸³ The use of only one female provider in the present study might have contributed to men perceiving the treatment as less unpleasant than women. Therefore, the results may be context-specific and inconclusive, as the present study was unable to investigate a similar gender interaction effect for female participants due to the lack of a male provider. This aspect remains to be investigated in future studies.

Limitations, Strengths, and Directions for Future Research

The present study has several points of limitation and strength.

One limitation of this study was the use of analog patients who imagined having a health condition based on a written vignette. Although these severity vignettes were validated by a physician, participants did not have lived experience with OD. Findings from analog patients might not be best generalized to clinical patients and long-term outcomes of patient-provider relationships.⁵¹ The external validity of the present findings should therefore be considered carefully. Nevertheless, the use of analog patients has been shown to produce valid and reliable results that highly align with perceptions of clinical patients in patient-provider communication research.^49–51 This approach also offers methodological advantages that are difficult to achieve in clinical settings, including the ability to recruit a large sample⁴⁹ and to minimize ceiling effects, as patients may be more prone to social desirability and evaluate their providers more positive or trustable.^49,51,84 Additionally, dysfunction severity, which was manipulated in the present study, would be difficult to isolate in clinical settings, where severity varies across patients and cannot easily be separated from other patient characteristics.

The use of a videotaped provider does not reflect two-way communication between healthcare providers and patients.⁸⁵ However, experimentally testing genuine two-way communication in laboratory settings is very difficult, if not impossible. That said, using a videotaped provider provides a high level of control over verbal and non-verbal behaviors.^62,85 Unlike many previous studies that have manipulated single aspects of communication, the present study integrated both verbal and non-verbal behaviors into a supportive communication style. This represents a key strength, as it more closely resembles how patients encounter provider communication in real clinical interactions.

Furthermore, as this study used a videotaped female provider, the effects of male providers using supportive communication remain unexplored. Previous research has shown that female and male healthcare providers communicate differently⁶ and that patients express gender preferences when choosing a provider for their medical care.⁸⁶ In addition, the use of only one actor is another limitation of this study. Although the proof-of-concept (See Supplementary Materials S4, including Table S4 and Figure S4) confirmed the actual difference between the communication styles, it cannot be ruled out that the actor's personal characteristics may have influenced the participants’ responses. Nevertheless, previous studies have used single-provider paradigms succesfully.⁸⁷

The supportive communication video was compared to the neutral. Even though the supportive style was conveying the desired characteristics, as established through the proof-of-concept procedure, the neutral style video could potentially still elicit emotions and might not be truly “neutral”.⁶⁰

The observed effect sizes were small to medium. However, the reliability of the effect sizes was double-checked through bootstrapping, and only findings that became significant in bootstrapping analyses were accepted as valid effects. Moreover, the absence of interactions effects by communication, gender, and dysfunction severity, should be interpreted with caution. Despite the large sample, the study may have still been underpowered to detect small, higher-order interaction effects.

The use of single-item measures adapted from broader scales to measure different aspects of treatment adherence attitudes is another limitation. However, single-item measures are commonly used in research to assess immediate outcomes, and to reduce participant burden and risk of non-completion.^88,89 Given the online nature of the present study, minimizing participant burden and fatigue were important considerations. Importantly, the observed correlations between the adherence attitude aspects and also with perceived trust in provider supported the validity of single-item measurements. Nevertheless, future studies should use more valid measures to investigate the outcomes of patient-provider communication on treatment adherence attitudes.

Future research should extend the present findings by examining supportive communication across different types of illnesses and in actual clinical settings. The impact of dysfunction severity on communication may further change over time, as patients gain more experience living with their condition.⁹⁰ Given the cross-sectional design, the present study only allows for conclusion about immediate perceptual outcomes and attitudes, and not subsequent behavioral adherence over time. Accordingly, a single measurement might not be suitable for further investigation. Therefore, longitudinal studies on how supportive communication impacts trust in provider and attitudes toward treatment over time, and how these attitudes might translate into actual adherence behaviors in different patient populations are warranted. Lastly, the present study provides a methodological framework for future research to experimentally investigate communication in a more realistic manner by systematically manipulating both verbal and non-verbal behaviors.

Conclusion

The present study found that supportive communication increased participants’ perceived trust in provider across different imagined dysfunction severities in both men and women. Consistent with previous research, these findings suggest that supportive communication facilitates trust in provider, regardless of the severity of the imagined health concern. However, the magnitude might be smaller than previously assumed, especially when communication is investigated using a more holistic approach.

Moreover, a supportive communication may facilitate the mitigating effects of trust in providers on anxiety levels. In contrast, a supportive communication may not influence attitudes toward adherence.

The findings also suggest potential gender differences in some aspects of treatment adherence attitudes. These findings should be interpreted cautiously given the exploratory and analog design of the study.

Funding Statement

This study was funded by the Norwegian University of Science and Technology (NTNU) and the Bial Foundation (Grant number: 36/20).

Disclosure

The authors report no conflicts of interest in this work.