Trust and Privacy: How Patient Trust in Providers is Related to Privacy Behaviors and Attitudes

Abstract

Healthcare today requires extensive sharing and access to patient health information. The use of health information technology (health IT) exacerbates patients’ privacy concerns because it expands the availability of patient data to numerous members of the healthcare team. Patient concerns about the privacy of their data may be associated with nondisclosure of their information to providers. Patient trust in physicians, a multi-dimensional perception influenced by patient, physician, and situational factors, can facilitate disclosure and use of health IT. Previous work has done little to explore how specific dimensions of trust in physicians are related to patient information-sharing concerns or behavior. Using data from a nationally-representative survey, we show that patients with higher trust in provider confidentiality have significantly lower likelihood of reporting having ever withheld important health information and lower likelihood of thinking it is important to find out who has looked at their medical records. Patient trust in physician competence is related to higher likelihood of thinking it is important for health care providers to share information electronically This work sheds light on the importance of considering multiple dimensions of trust for patient behavior and attitudes related to their information sharing with health care providers.

Introduction

The delivery of healthcare today requires extensive sharing of and access to patient health information. The use of health information technology (IT) expands the availability of patient data to numerous healthcare providers throughout and between healthcare organizations. Since the widespread expansion of electronic health record (EHR) implementation following the distribution of incentives by the HITECH act, use of Internet-enabled applications for health (e.g., patient portals, mHealth apps) has increased steadily over the past decade. Use of health IT has yielded numerous positive benefits for patients, including increased access to data and improved efficiency¹. However, such benefits come with increased concern about the potential for privacy breaches^2,3. Indeed patient concerns about the privacy of their health information can impede their access to health care and hinder disclosure to providers^4-7, creating incomplete medical records. Complete information sharing by patients with providers is necessary to ensure the quality and accuracy of data available in health information systems.

Patient trust in physicians, a multi-dimensional perception,^8-10 can potentially remedy the issue. Patient trust in physicians is influenced by patient, physician, and situational factors, and can facilitate disclosure⁹ and use of health IT¹¹. One dimension, in particular, seems to matter more than others for ameliorating privacy concerns^6,12: trust in physician confidentiality. However, few quantitative studies of trust in physicians explore multiple dimensions and those that do have only single measures of each dimension, leaving critical gaps in knowledge about how trust may matter^8-10.

Systematic disparities in patient trust in physicians also exist, though again the dimensions of trust matter differently for different groups. Some research finds that racial and ethnic minorities have lower levels of certain dimensions of trust in physicians^13,14 and other research shows that patients with devalued health characteristics have lower trust in physician confidentiality specifically¹². Similarly, disparities in use of health IT exist between racial groups and between those with stigmatized health conditions, suggesting the need to further investigate how different dimensions of trust may be related to patient attitudes and behavior related to information sharing.

Here we analyze national survey data to determine how dimensions of patient trust in physicians are related to patient disclosure of health information and to attitudes about information sharing within health care (see Figure 1). Figure 1 situates our analysis in the context of existing work. Our analysis specifically focuses on the association between three dimensions of trust in physicians and patients’ attitudes towards privacy and health information disclosure behavior (represented by the red arrow in Figure 1). Associations between patient demographic characteristics, health status, trust in physician, and health information attitudes and behaviors are indicated by blue directional arrows. Finally, the association between accuracy and quality of data in health IT systems and health information attitudes and behaviors is represented by a dashed line, as this is an indirect consequence of patients’ disclosure behaviors and a needed focus of future work.

Figure 1.

Study Framework: Situating our Analysis

The red arrow represents the focus of our analysis.

Methods

We conducted a cross-sectional, nationally-representative mail survey in 2014 to five thousand households across the United States based on three separate samples of households. For the general population sample, 1800 households were selected. For the Hispanic oversample, 1600 households were selected from the five communities with the highest concentration of Hispanics in the US: East Los Angeles, CA; Laredo, TX; Hialeah, FL; Brownsville, TX; and McAllen, TX. The African-American oversample included 1600 households from the following communities: Detroit, MI; Jackson, MS; Miami Gardens, FL; Birmingham, AL; and Baltimore, MD. In total, 5,000 households were selected. A total of 784 households completed the survey. The response rate of the survey was 15.7%. However, from the original sample of 5,000 households, 481 surveys were returned as undeliverable with no forwarding address available. The survey was conducted by the University of Nebraska Bureau of Sociological Research in partnership with Dartmouth College and was funded by HHS 90TR0003/01. Data weights account for sample design (including the three sampling frames), nonresponse, and population characteristics to be representative of the national population.

Respondents were asked whether they had ever withheld information from their provider because of privacy and security concerns (1=yes/0=no). Respondents were also asked two questions about attitudes towards privacy, namely, how strongly they agreed that it is important 1) to find out who has looked at their medical record and 2) that providers share their health information electronically (for each, strongly agree=1/less than strongly agree=0).

The survey also assessed several dimensions of trust in physicians⁹, based on previous literature: trust in physician confidentiality, trust in physician honesty, trust in physician competence, and overall trust in physician. Each dimension was measured with items using a 5-point Likert scale. Trust in physician confidentiality is an aggregate measure created as the average response to three statements, namely “I trust my doctor to keep personally sensitive information private¹⁵,” “I worry that private information my doctor has about me could be used against me¹⁵,” and “I worry that my doctor may share embarrassing information about me with people who have no business knowing it.^16,17” Trust in physician honesty is an aggregate measure created as the average response to two statements, “I feel my doctor does not do everything s/he should for my medical care¹⁵” and “I trust my doctor to perform only medically necessary tests and procedures.¹⁸” Trust in physician competence is an aggregate measure created as the average response to three statements, “I trust my doctor to offer me high quality care¹⁸,” “my doctor is a real expert in taking care of medical problems like mine¹⁵,” and “my doctor’s medical skills are not as good as they could be” (reverse coded)^16,17. An overall measure of complete trust is included as the response to the statement “all in all, I have complete trust in my doctor^16,17,19.”

Data analysis was conducted using Stata 15.1. Survey data was analyzed using population sample weights. Logistic regression models were fitted to predict the three binary categorical dependent variables: having ever withheld information from providers due to concern about privacy, strongly agreeing that it is important that healthcare providers share my information electronically, and strongly agreeing that it is important that I can find out who has looked at my medical records. Independent variables are the four measures of trust: trust in confidentiality, honesty, competence, and overall trust. Each model controls for respondent demographic characteristics, self-rated health, health insurance status, and care quality (see Table 1 for descriptive statistics).

Table 1.

Descriptive statistics of Consumer Characteristics, Analytic Sample N=542, unweighted

Variables	%
Female	51.6
Race/Ethnicity African American, non-Hispanic	8.8
Hispanic	6.6
White, non-Hispanic	84.6
Age
18 – 29 years of age	8.7
30 – 39 years of age	20.5
40 – 49 years of age	21.9
50 – 59 years of age	22.0
60+ years of age	26.8
Education
High School or less	15.9
Some College	26.1
BA/BS College Degree	35.7
Graduate Degree	22.3
Excellent or Good Self-rated Health	84.9 (0.02)
Has health insurance	96.9
Quality of Health Care in previous 12 months
Poor	1.6
Fair	5.6
Good	32.2
Very Good	42.2
Excellent	18.3

Source: 2014 nationally-representative survey of US Adults. Author’s compilation

Results

Table 1 provides unweighted demographic details about the study sample. The majority of respondents in the sample are white (84.6%), educated with at least a BA/BS degree (58%), have excellent or good self-rated health (84.9%), have health insurance (96.0%), and report receiving very good or excellent health care in the previous year (60.5%).

We observe a significant association between trust in confidentiality and whether patients had ever withheld information from their doctor (OR = 0.20, Table 2). However, no significant associations were found between trust in provider honesty and competence and having ever withheld information. In a model predicting patients’ expectation that it is very important that providers share their health information electronically, we find a significant association with trust in physician competence (OR = 2.87), but no significant relationship with trust in physician confidentiality or honesty. Finally, in a model predicting patients’ expectation that it is very important that they find out who has looked at their medical records, we find a significant association with trust in physician confidentiality (OR = 0.59), but no significant association with trust in physician honesty or competence. No model showed a significant association between overall trust and the three dependent variables. All models control for race, gender, education, and age.

Table 2.

Odd-ratios for Logistic Regression of Trust in Physician dimensions on Patient Behavior and Expectations about Information Access and Sharing, Odds-ratios [95% CI]

	Ever Withheld Information because concerned about privacy		Very important that HC providers share my information electronically		Very important that I find out who has looked at my medical records
		1	2	1	2	1	2
Trust in physician
	0.60	0.45	1.94 *	0.94	0.92	1.63
Overall	[0.32 – 1.1]	[0.15 - 1.4]	[1.2 – 32]	[0.43 - 2.0]	[0.61 – 1.4]	[0.86 - 3.1]
Honesty		2.45		1.31		0.99
		[0.78 - 7.7]		[0.69 - 2.5]		[0.61 - 1.6]
Competence		3.53		2.87 *		0.97
		[0.77 - 16.2]		[1.1 - 7.8]		[0.48 - 1.9]
Confidentiality		0.20 ***		1.01		0.59 *
		[0.10 - 0.42]		[0.48 - 2.2]		[0.35 - 0.99]
With Controls	YES	YES	YES	YES	YES	YES
F-test	3.9 ***	3.3 ***	4.6 ***	4.0 ***	1.6	2.6 ***
Degrees of freedom	14,700	17,697	14,700	17,697	14,700	17,697

Source: 2014 nationally-representative survey of US Adults. Author’s compilation.

Discussion

Trust is an important determinant of patients’ behaviors and experiences of healthcare^9,10,12,13, and here we show how it is associated with information sharing. We provide evidence that specific dimensions of trust in physicians, namely, trust in confidentiality and competency of providers, influence patients’ behaviors and expectations for information sharing in health care. This is in contrast to much previous work which considers trust to be a singular, generalized determinant of patient behavior. We find that specific dimensions of patient trust in physicians have unique associations, namely, trust in confidentiality is associated with lower odds of having ever withheld info and reporting concern about who is accessing their health record information. In addition, patients with higher levels of trust in physician competency have greater odds of agreeing that it is important for providers to share health information with each other. Encouraging patients’ health information sharing, and reducing patients’ withholding, may increase the accuracy and quality of health data available in health IT systems.

This work demonstrates not only that trust in providers is essential for information sharing, but that different dimensions of trust matter differently. This finding further supports previous work showing trust is situational¹². Patients’ trust is not a singular, generalized phenomenon, instead it is a nuanced series of relationships based on specific expectations and behaviors. Moreover, we show that the individual dimensions of trust are particularly important to study, as each contributes in different ways to the patient experience and the physician-patient relationship.

Additionally, patient information disclosure and concerns about access are also multifaceted and nuanced. Patients are concerned about who accesses health data and why the data are used²⁰. Perceptions of how data are used may also influence patients’ trust²⁰. Trust in providers is associated with the quality of the doctor-patient relationship. Subsequently, dimensions of trust may play different roles in ensuring good communication between patients and providers, in patients’ use of healthcare services, and for adhering to treatment^21,22.

Given systematic disparities in patient trust, in which racial and ethnic minorities have lower trust in physicians^13,14, while patients with devalued health characteristics have lower trust in physician confidentiality than others¹², further research is needed to examine how these factors may intersect with the dimensions of trust and their relationships to health information disclosure and concerns as shown here. This paper controls for patient age, gender, race, ethnicity, self-rated health, insurance status, and quality of health care but further study is needed to disentangle the relationships and possible interactions between and among them, dimensions of trust and health information sharing. Further, we have little understanding of how features of health IT applications interact with dimensions of patient trust. Health IT application design may unintentionally diminish trust or exacerbate rather than reduce privacy concerns that hinder use of beneficial health IT

These critical gaps in knowledge mean practitioners are missing when some patients, particularly the most vulnerable, may be more likely to have incomplete or inaccurate health data. The consequences of incomplete data include less effective treatment, more potential medical errors, and possible bias in treatment due to lack of data-representation in decision-making algorithms. Similar to a lack of representation in clinical studies, when patients’ data is absent, practitioners risk not learning which patients may most benefit from precision health treatments.

This study has a number of limitations that must be considered. First, data for this analysis come from a cross- sectional sample collected in 2014. Expansions in health IT use since then, as well as the evolution of available technologies, may change patients’ experiences and perceptions of technology. Additionally, qualitative research would allow more in-depth understanding of how different dimensions of trust influence patients’ behaviors and expectations about physicians, health IT, and privacy.

Conclusion

New information technologies in health care have the potential to enhance care quality and access, yet the widespread information sharing enabled by them also create new risks and concerns. Patients concerned about privacy may be less likely to disclose health information to providers and may be concerned about how their information is being used and accessed within the health care system. While trust is known to be an important factor in health care delivery and doctor patient relationships, previous work has not shown how specific dimensions of trust in physicians contribute to patient information sharing behavior or concerns. Here we show that one particular dimension of trust, trust in provider confidentiality, is associated with patients’ withholding information from clinicians and concerns about who is accessing their records, while a different dimension, trust in physician competency, is associated with agreeing that providers should share information. This is an important contribution given the increasing reliance on databases of patient information to tailor the delivery of care.

Ensuring that new technologies and their benefits are available to all patients, especially the most vulnerable, is necessary to prevent health disparities from widening as new effective health information technologies come in to widespread use. What is needed is an understanding of how dimensions of patient trust matter for health IT use and endorsement, whether particular dimensions of trust alleviate privacy concerns, and how features of health IT influence patient privacy concerns or usage.