Frequency of and Factors Associated with Care Partner Proxy Interaction with Health Care Teams Using Patient Portal Accounts

Abstract

Introduction:

Proxies can communicate with health care teams through patient portals either by using proxy login credentials or a patient's login credentials. The frequency of proxies using patient login credentials is unknown.

Methods:

A random selection of 3,000 portal messages sent in through adult patients' own portal account was reviewed for indicators (referring to the patient in the third person) that someone other than the patient was using the patient portal account.

Results:

Of the reviewed 3,000 portal messages sent through patient portal accounts, 221 (7.4%) appeared to be sent in by a proxy, 2,512 (83.8%) appeared to have been sent in by the patient and for 266 (8.9%) portal messages reviewed it was unclear who sent in the message. There was no difference in mean age between patients who had proxy messages sent through patient portal accounts versus proxy portal accounts. Patients who had proxies send messages through patient accounts were more likely to be married and male. Out of 221 manually reviewed messages apparently sent by proxies through patient portal accounts there were 113 (51%) where the proxy included their name and 56 (25.3%) where they reported their relationship to the patient. During the study period, 0.7% of total messages on adult patients were sent through proxy accounts.

Discussion:

Proxies appear to use patient portal accounts much more frequently than proxy accounts to communicate with the health care team on adult patients; however, when using patient accounts they only identify themselves approximately half of the time.

Introduction

Patient portals can be used by patients and their care partners to review their personal health record (PHR), request appointments, medication refills, manage chronic and preventative health care needs, and communicate securely with their health care team.^1

–4 Use of patient portals is increasing with patients generally reporting they are satisfied with patient portals and the greatest satisfaction is with secure messaging features.^5

–9

Some patients choose to share access to their PHR with family members or other care partners (called a “proxy”). This sharing can be done by setting up a proxy portal account, through which the proxy can view a patient's PHR and communicate with the patient's health care team using portal secure messaging. Proxy accounts are set up using login and password credentials that are distinct from the login credentials used for the patient's own portal account. We define a “patient portal account” as a portal account set up by the patient using the patient's login and password information, whereas a “proxy portal account” is a portal account set up by another person (a proxy) to access a patient's PHR using proxy login and password information that is distinct from the patient's login credentials. Proxy accounts can be provided for parents of minor children, legal guardians, and to other nonguardian adults with consent from the adult patient.^10,11 Potential benefits of having a proxy account can include care partners being able to help manage patients' care by communicating with the health care team, increasing care partners' awareness of specific health issues, and allowing proxies to help in managing appointments, health care information, and prescription refills.¹² A survey of mostly older adult patients found that 79% of adults surveyed were interested in sharing access to their PHR with care partners, most commonly with family members.¹³

Although many health care systems allow setting up proxy accounts to access patients PHR through proxy portal accounts,¹⁴ patients report allowing proxies to access their PHR using the patient's own portal account and login credentials.^15,16 This is potentially problematic for several reasons. First, although setting up proxy access to limited portions of the PHR is possible with proxy accounts, proxies viewing patient's PHR using the patient's own login credentials (through a patient account) gives the proxy access to all information available to the patient, including billing information and potentially sensitive or stigmatizing health information. Second, up to 59% of people report using the same password for many or all online accounts.¹⁷ Providing login credentials and password information for a patient portal account could give proxies unintended access to other accounts. Third, messages sent by a proxy through a patient portal to a health care provider using the patient's own login credentials could lead to confusion as it may not be clear who is sending the message.¹⁸

The extent to which proxies access PHR using the patient's own login credentials (i.e., using a “patient portal account”) is unknown as is whether there are any demographic or other factors associated with proxies accessing PHR using the patient's own login credentials versus logging into a proxy account.

The purpose of this study was to evaluate the frequency of portal messages sent by proxies through patient portal accounts using a patient's own login credentials and to compare patient demographic data for messages sent through a proxy account with those that appear to be sent by proxies through a patient's own account.

Methods

This study took place at the Mayo Clinic in Rochester, Minnesota, a multispecialty academic practice having >2,400 physicians and scientists. All secure messages sent from January 1, 2013 to June 30, 2017 were captured except those associated with patients under age 18. A random set of 3,000 portal messages sent through patient accounts was reviewed for indications that the message was being sent by a proxy using a patient portal account rather than a proxy account. Messages sent through patient portal accounts were coded in the following way:

Message sent by patient through patient portal account.

f the message referred to the patient in the first person (e.g., “I need a refill on my medication”) the message was coded as being sent by the patient through a patient portal account.

Message sent by proxy through the patient portal account.

Messages were coded as presumably being sent by a proxy through the patient portal account if the message referred to the patient in the third person either by name or by pronouns (e.g., “John needs a refill on his medication”).

Message sent by unclear person(s) through the patient portal account.

Messages were coded in this category when there were either no pronouns used or plural pronouns (e.g., “Need refill” or “We need to request a refill on a medication”).

Messages that appeared to be sent by a proxy through the patient's own portal account were analyzed to determine if there was any indication who had sent the message such as a name or a relationship to the patient or both. One author (J.P.) reviewed all 3,000 messages manually and a research assistant reviewed a random subset of 200 messages to evaluate an agreement statistic in designating messages as being sent by a patient, proxy, or not clearly identifiable as patient or proxy. For messages identified as being sent by a proxy through a patient account, two authors (J.P. and M.J.D.) reviewed the messages for indications that the proxy identified themselves by name or relationship to the patient and an agreement statistic was determined.

We compared demographics, receiving health care team (specialty vs. primary care) and whether the message was sent during office hours or not of patients who had messages that appeared to have been sent by proxies through the patient's own portal account to patients who had proxy messages sent by proxies through proxy login accounts to evaluate for any differences between groups. Demographic information, including age, gender, marital status, and race of patients, was collected using the Mayo Clinic Advanced Cohort Explorer search tool that searches electronic medical records electronically and captures demographic data on patients. Whether the message was sent to a primary care team versus specialty team and whether message was sent during typical office hours or not was also electronically collected.

JMP Pro 14.1.0 was used for randomization and statistical tests. Chi-square test was used for categorical analysis and t test was used for differences in means. The study was approved by the Mayo Clinic Institutional Review Board (17-004807).

Results

There were 752,551 portal messages sent on adult patients between January 1, 2013 through June 30, 2017 with 747,581 (99.3%) messages sent using patient login credentials (patient portal accounts) and 4,970 (0.7%) sent using proxy login credentials (proxy accounts). Of the 3,000 randomly selected portal messages sent through patient's portal accounts 221 (7.4%) appeared to be sent in by someone other than the patient and 2,512 (83.8%) appeared to have been sent in by the patient. For 266 (8.9%) messages it was unclear who authored the message. Patient demographics comparing proxy messages sent through proxy accounts with those for proxy messages sent through patient accounts are shown in Table 1. There was no difference in mean age between patients who had messages apparently sent by proxies through patient portal accounts versus proxy portal accounts. Patients who had proxies send messages through patient accounts were more likely to be married and male compared with patients whose proxies used proxy accounts to send messages. When comparing proxy messages sent through proxy accounts with those sent through patient accounts for type of health care team the message was directed toward (primary care vs. specialty care), significantly more messages were sent to specialty care by proxies using the patient's account (183/221 [82.8%]) versus proxies using a proxy account (2,099/4,970 [42.2%]; p < 0.0001). There was no significant difference in whether messages were sent after clinic hours (on the weekend or between 5 pm and 8 am on weekdays) between messages sent by proxies using a patient account (76/221 [34.4%]) compared with messages sent by proxies using a proxy account (1,449/4,970 [29.1%]; p = 0.09).

Table 1.

Demographics of Patients for Proxy Messages Sent Through Patient Accounts Versus Proxy Messages Sent Through Proxy Accounts

DEMOGRAPHIC OF PATIENT	SENT THROUGH PATIENT ACCOUNT	SENT THROUGH PROXY ACCOUNT	p
DEMOGRAPHIC OF PATIENT	TOTAL N = 221	TOTAL N = 4,970	p
Age (years) average	57.8	59.9	0.23
Marital status (married) % (n)	70 (154)	44 (2,174)	<0.0001
Gender (female) % (n)	32 (70)	49 (2,419)	<0.0001
Race (white) % (n)	92 (203)	94 (4,686)	0.15

Of the 221 portal messages that appeared to be from proxies sent through patient login accounts, the frequency and type of identifying information included in the messages are listed in Table 2. Out of 221 manually reviewed messages apparently sent by proxies through patient login accounts, there were 113 (51.1%) where the proxy included their name (full name 66 [29.9%] and first name only 47 [21.3%]). The relationship of the proxy to the patient was included in 56/221 (25.3%). Of these 56, the following relationships were cited: child (52%), spouse (27%), parent (14%), sibling (4%), niece (2%), and grandchild (2%).

Table 2.

Type of Identifying Information Included in 221 Proxy Messages Sent Through Patient Accounts

IDENTIFYING INFORMATION OF PROXY IN MESSAGE	% (n)
No identifying information	42 (94)
Name only	32 (71)
Relationship only	6 (14)
Name and relationship	19 (42)

A Cohen kappa of 0.97 was found for agreement between manual reviewers in assigning whether patient account messages were sent by the patient, a proxy, or if the source of the message was not clearly identifiable as proxy or patient. Cohen kappas of 0.96 (name) and 1 (relationship) were found for agreement between manual reviewers on what identifying information was included in messages sent by proxies through patient accounts.

Discussion

The current literature describes the phenomenon of proxies signing into patient portal accounts using patient login credentials; however, the frequency of this occurring is unknown. This study suggests the frequency could be anywhere from 7% (the percentage of randomly reviewed messages sent through patient accounts appearing to be from proxies) to as frequently as 16% (the percentage of messages that appear to be sent in by proxies plus the messages for which no determination of origin could be made). In contrast, only 0.7% of total messages during the study time period were sent through proxy accounts. This suggests it is much more common for proxies of adult patients to use patient login credentials rather than proxy accounts to access patient's PHR and communicate with a patient's health care team.

Extrapolating findings to our entire message volume, it is estimated that there were ∼55,000 proxy messages that came through patient accounts during the study period. Based on study results, ∼23,000 of these messages (42%) appeared to be from a proxy who was not immediately identifiable. From the health care provider perspective, answering these messages can be daunting. The responding health care team has to decide how to word their responses. Do they ignore the clues that the message appears to be from a proxy and address the message sender as “you” or should the health care team reply by referring to the patient in the third person? The overall etiquette for answering messages may deserve further study.

When proxies use a proxy account to send a portal message about a patient, the portal message arrives for viewing by the health care team in the patient's chart in the electronic medical record, but there is an indication it was sent by a proxy account using proxy login credentials. Text is displayed stating that the message was sent “by proxy name on behalf of patient name” making it transparent to the health care team that they are communicating with someone other than the patient, and identifying exactly who the proxy is. In contrast, when proxies use a patient portal account it is potentially less clear who is sending the message unless the proxy specifically lists their first and last name in the message. In the 221 messages that appear to be sent by proxies through patient portal accounts, the full (first and last) name of the person sending the message was only included 30% of the time. Approximately half of the proxy messages sent through patient accounts included mention of a relationship to the patient (e.g., “My mother has been feeling ill”) without including the name of the person sending the message. For someone citing a spousal relationship it would be clear who was sending the message, but without an accompanying name for other relationships it could still be uncertain who was sending the message if the patient has multiple relations of that type (such as multiple children or siblings).

It is unclear what prompts proxies to choose to communicate with the health care team through the patient's own portal account rather than a proxy account. Convenience could be one driver in that it may be easier for proxies to use a pre-existing patient account rather than setting up a separate proxy account. However, there may be other factors that contribute to this practice based on findings from Wolff et al. who found that in a survey of 389 care partners who had proxy accounts for another person that although 89% of the care partners reported accessing the PHR through their proxy account, 23% of care partners with proxy accounts also reported that they had accessed the PHR using the patient's own portal account.¹⁶ As almost one quarter of care partners chose to use the patient's own portal account rather than an already existing proxy account, there may be additional factors that drive proxies to use patient accounts and login credentials. The study design by Wolff et al. did not include investigation of the reasons for this practice, although they theorize there may be barriers to proxy portal use, such as additional steps to access the account, which could lead to proxy portal accounts being perceived as less convenient. In a survey of patient satisfaction with patient portals, Neuner et al. found that although >66% of respondents reported being very satisfied with the patient portal, of those who responded to questions specifically about proxy portal accounts, less than half of respondents reported being very satisfied with proxy portal accounts.⁵ Thus, there may be some factors specific to proxy portal accounts that make them less attractive.

This study found that proxies used patient portal accounts to communicate with specialty health care teams more frequently than with primary care health care teams. This may be a reflection of the institution serving as a major referral center. Many patients who access specialty care at the Mayo Clinic do not live close by and likely receive primary care elsewhere. It may be that these patients and their proxies anticipate that their communications with the specialty health care team will be short lived (as they often serve in a consultative manner) and thus may feel less inclined to set up a proxy account, whereas proxies of patients receiving primary care at our institution may anticipate a more ongoing long-term relationship prompting them to have a proxy account.

Patients whose proxies messaged the health care team through the patient's own portal account were more likely to be married and male compared with patients whose proxies used a proxy portal account to communicate with the health care team. The majority of proxies (75%) who sent a message through patient accounts did not cite their relationship with the patient in their message; however, the finding that many of these patients were married men may suggest that it is a spouse who is using the patient's account and that spouses may be comfortable sharing login information. We did not see any differences in race, average age, or messages being sent during or after clinic hours between patients whose proxies used proxy accounts and those whose proxies used patient accounts.

Crotty et al. investigated shared PHR access and information sharing concerns with patients and caregivers of patients >75 years. Through this investigation they learned that patients prefer that information sharing preferences ideally should be dynamic (changeable over time). In addition, a key finding was that multiple options and models should be made available if possible to adhere to individual preferences in regard to what type of information can be shared.¹⁹ A study by Latulipe et al. that used a focused interview with older adult patients to explore their preferences regarding PHR access sharing with proxies revealed that many patients were not aware of the extent of information that might be available to proxies logging in using the patient's own portal. In addition, many stated that they were not comfortable with proxies having access to billing information and potentially stigmatized health information.¹⁵ This desire on the part of many patients to choose which aspects of their PHR their proxies have access to can only be achieved by promoting proxy accounts.

In an environmental scan of proxy access, Wolff et al. found that all 20 health care systems in a geographically diverse region offered proxy portal accounts although only half had the information described on their portal registration page.¹⁴ In offering proxy accounts institutions may increase their uptake by actively promoting and advertising them and attempting to balance adequate security features while still maintaining convenience as much as possible. Future research should focus on what factors prompt proxies to use patient portal accounts preferentially over proxy accounts and using that information to promote increased utilization of proxy accounts.

Limitations

Since only portal messages were reviewed, this study does not provide any information on the frequency of proxies logging in with patient credentials to access other aspects of the PHR, including billing, reviewing office notes and test results, or to review communications between the health care team and the patient. As noted earlier, the study setting was a tertiary medical center that also provides primary care. Practices with a higher percentage of primary care may have different results. In addition, this study would not capture messages sent through patient portal accounts by proxies referring to the patient in the first person. Finally, there were many instances in which it was unclear whether the patient was sending the message or not.

Footnotes

Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

References

North

, Crane

, Stroebel

, et al. Patient-generated secure messages and eVisits on a patient portal: Are patients at risk?. J Am Med Inform Assoc, 2013; 20:1143–1149.

Henry

, Shen

, Ahuja

, et al. The online personal action plan: A tool to transform patient-enabled preventive and chronic care. Am J Prev Med, 2016; 51:71–77.

Turner

, Osterhage

, Hartzler

, et al. Use of patient portals for personal health information management: The older adult perspective. AMIA Annu Symp Proc, 2015; 2015:1234–1241.

Osborn

, Rosenbloom

, Stenner

, et al. MyHealthAtVanderbilt: Policies and procedures governing patient portal functionality. J Am Med Inform Assoc, 2011; 18 Suppl 1:i18–i23.

Neuner

, Fedders

, Caravella

, et al. Meaningful use and the patient portal: Patient enrollment, use, and satisfaction with patient portals at a later-adopting center. Am J Med Qual, 2015; 30:105–113.

Kruse

, Bolton

, Freriks

. The effect of patient portals on quality outcomes and its implications to meaningful use: A systematic review. J Med Internet Res, 2015; 17:e44.

Woods

, Schwartz

, Tuepker

, et al. Patient experiences with full electronic access to health records and clinical notes through the My HealtheVet Personal Health Record Pilot: Qualitative study. J Med Internet Res, 2013; 15:e65.

Murphy-Abdouch

, Dolezel

, McLeod

. Patient access to personal health information: An analysis of the consumer's perspective. Perspect Health Inf Manag, 2017;Summer 2017:1–19.

Redelmeier

, Kraus

. Patterns in patient access and utilization of online medical records: Analysis of MyChart. J Med Internet Res, 2018; 20:e43.

10.

Bourgeois

, Taylor

, Emans

, et al. Whose personal control? Creating private, personally controlled health records for pediatric and adolescent patients. J Am Med Inform Assoc, 2008; 15:737–743.

11.

Calman

, Pfister

, Lesnewski

, et al. Electronic access to adolescents' health records: Legal, policy, and practice implications. Fam Pract Manag, 2015; 22:11–14.

12.

Reed

, Huang

, Brand

, et al. Communicating through a patient portal to engage family care partners. JAMA Intern Med, 2018; 178:142–144.

13.

Zulman

, Nazi

, Turvey

, et al. Patient interest in sharing personal health record information: A web-based survey. Ann Intern Med, 2011; 155:805–810.

14.

Wolff

, Kim

, Mintz

, et al. An environmental scan of shared access to patient portals. J Am Med Inform Assoc, 2018; 25:408–412.

15.

Latulipe

, Quandt

, Melius

, et al. Insights into older adult patient concerns around the caregiver proxy portal use: Qualitative interview study. J Med Internet Res, 2018; 20:e10524.

16.

Wolff

, Berger

, Clarke

, et al. Patients, care partners, and shared access to the patient portal: Online practices at an integrated health system. J Am Med Inform Assoc, 2016; 23:1150–1158.

17.

LastPass. Neglect is Helping Hackers Win. Available at https://blog.lastpass.com/2018/05/psychology-of-passwords-neglect-is-helping-hackers-win.html (last accessed December 1, 2018 ).

18.

Sarkar

, Bates

. Care partners and online patient portals. JAMA, 2014; 311:357–358.

19.

Crotty

, Walker

, Dierks

, et al. Information sharing preferences of older patients and their families. JAMA Intern Med, 2015; 175:1492–1497.