The Effect of Artificial Intelligence on Patient-Physician Trust: Cross-Sectional Vignette Study.

BACKGROUND: Clinical decision support systems (CDSSs) based on routine care data, using artificial intelligence (AI), are increasingly being developed. Previous studies focused largely on the technical aspects of using AI, but the acceptability of these technologies by patients remains unclear. OBJECTIVE: We aimed to investigate whether patient-physician trust is affected when medical decision-making is supported by a CDSS. METHODS: We conducted a vignette study among the patient panel (N=860) of the University Medical Center Utrecht, the Netherlands. Patients were randomly assigned into 4 groups-either the intervention or control groups of the high-risk or low-risk cases. In both the high-risk and low-risk case groups, a physician made a treatment decision with (intervention groups) or without (control groups) the support of a CDSS. Using a questionnaire with a 7-point Likert scale, with 1 indicating "strongly disagree" and 7 indicating "strongly agree," we collected data on patient-physician trust in 3 dimensions: competence, integrity, and benevolence. We assessed differences in patient-physician trust between the control and intervention groups per case using Mann-Whitney U tests and potential effect modification by the participant's sex, age, education level, general trust in health care, and general trust in technology using multivariate analyses of (co)variance. RESULTS: In total, 398 patients participated. In the high-risk case, median perceived competence and integrity were lower in the intervention group compared to the control group but not statistically significant (5.8 vs 5.6; P=.16 and 6.3 vs 6.0; P=.06, respectively). However, the effect of a CDSS application on the perceived competence of the physician depended on the participant's sex (P=.03). Although no between-group differences were found in men, in women, the perception of the physician's competence and integrity was significantly lower in the intervention compared to the control group (P=.009 and P=.01, respectively). In the low-risk case, no differences in trust between the groups were found. However, increased trust in technology positively influenced the perceived benevolence and integrity in the low-risk case (P=.009 and P=.04, respectively). CONCLUSIONS: We found that, in general, patient-physician trust was high. However, our findings indicate a potentially negative effect of AI applications on the patient-physician relationship, especially among women and in high-risk situations. Trust in technology, in general, might increase the likelihood of embracing the use of CDSSs by treating professionals.

Comparison of the Response to an Electronic Versus a Traditional Informed Consent Procedure in Terms of Clinical Patient Characteristics: Observational Study.

BACKGROUND: Electronic informed consent (eIC) is increasingly used in clinical research due to several benefits including increased enrollment and improved efficiency. Within a learning health care system, a pilot was conducted with an eIC for linking data from electronic health records with national registries, general practitioners, and other hospitals. OBJECTIVE: We evaluated the eIC pilot by comparing the response to the eIC with the former traditional paper-based informed consent (IC). We assessed whether the use of eIC resulted in a different study population by comparing the clinical patient characteristics between the response categories of the eIC and former face-to-face IC procedure. METHODS: All patients with increased cardiovascular risk visiting the University Medical Center Utrecht, the Netherlands, were eligible for the learning health care system. From November 2021 to August 2022, an eIC was piloted at the cardiology outpatient clinic. Prior to the pilot, a traditional face-to-face paper-based IC approach was used. Responses (ie, consent, no consent, or nonresponse) were assessed and compared between the eIC and face-to-face IC cohorts. Clinical characteristics of consenting and nonresponding patients were compared between and within the eIC and the face-to-face cohorts using multivariable regression analyses. RESULTS: A total of 2254 patients were included in the face-to-face IC cohort and 885 patients in the eIC cohort. Full consent was more often obtained in the eIC than in the face-to-face cohort (415/885, 46.9% vs 876/2254, 38.9%, respectively). Apart from lower mean hemoglobin in the full consent group of the eIC cohort (8.5 vs 8.8; P=.0021), the characteristics of the full consenting patients did not differ between the eIC and face-to-face IC cohorts. In the eIC cohort, only age differed between the full consent and the nonresponse group (median 60 vs 56; P=.0002, respectively), whereas in the face-to-face IC cohort, the full consent group seemed healthier (ie, higher hemoglobin, lower glycated hemoglobin [HbA1c], lower C-reactive protein levels) than the nonresponse group. CONCLUSIONS: More patients provided full consent using an eIC. In addition, the study population remained broadly similar. The face-to-face IC approach seemed to result in a healthier study population (ie, full consenting patients) than the patients without IC, while in the eIC cohort, the characteristics between consent groups were comparable. Thus, an eIC may lead to a better representation of the target population, increasing the generalizability of results.

Asking informed consent may lead to significant participation bias and suboptimal cardiovascular risk management in learning healthcare systems.

BACKGROUND: The Utrecht Cardiovascular Cohort - CardioVascular Risk Management (UCC-CVRM) was set up as a learning healthcare system (LHS), aiming at guideline based cardiovascular risk factor measurement in all patients in routine clinical care. However, not all patients provided informed consent, which may lead to participation bias. We aimed to study participation bias in a LHS by assessing differences in and completeness of cardiovascular risk management (CVRM) indicators in electronic health records (EHRs) of consenting, non-consenting, and non-responding patients, using the UCC-CVRM as an example. METHODS: All patients visiting the University Medical Center Utrecht for first time evaluation of a(n) (a)symptomatic vascular disease or condition were invited to participate. Routine care data was collected in the EHR and an informed consent was asked. Differences in patient characteristics were compared between consent groups. We performed multivariable logistic regression to identify determinants of non-consent. We used multinomial regression for an exploratory analysis for the determinants of non-response. Presence of CVRM indicators were compared between consent groups. A waiver (19/641) was obtained from our ethics committee. RESULTS: Out of 5730 patients invited, 2378 were consenting, 1907 non-consenting, and 1445 non-responding. Non-consent was related to young and old age, lower education level, lower BMI, physical activity and haemoglobin levels, higher heartrate, cardiovascular disease history and absence of proteinuria. Non-response increased with young and old age, higher education level, physical activity, HbA1c and decreased with lower levels of haemoglobin, BMI, and systolic blood pressure. Presence of CVRM indicators was 5-30% lower in non-consenting patients and even lower in non-responding patients, compared to consenting patients. Non-consent and non-response varied across specialisms. CONCLUSIONS: A traditional informed consent procedure in a LHS may lead to participation bias and potentially to suboptimal CVRM, which is detrimental for feedback on findings in a LHS. This underlines the importance of reassessing the informed consent procedure in a LHS.