The Digital Transition: Are Adults Aged 65 Years or Older Willing to Complete Online Forms and Questionnaires in Patient Portals?

Abstract

Introduction

Patients are being encouraged to complete forms electronically using patient portals rather than on paper, but willingness of older adults to make this transition is uncertain.

Methods

The authors analyzed data for 4105 Kaiser Permanente Northern California 2020 Member Health Survey respondents aged 65-85 years who answered a question about willingness to complete online forms and questionnaires using a patient portal. Data weighted to the Kaiser Permanente Northern California membership were used to estimate percentages of older adults willing to complete patient portal forms and questionnaires. Chi-square tests and log-Poisson regression models that included sociodemographic, internet use, and patient portal variables were used to identify factors predictive of willingness.

Results

Overall, 59.6% of older adults were willing to complete patient portal forms, 17.6% were not willing, and 22.8% were not sure. Adults aged 75–85 (49.5%) vs 65–74 years (64.8%) and Black (51.9%) and Latino (46.5%) vs White (62.8%) adults were less likely to indicate willingness. In addition to racial and ethnic differences and younger age, higher educational attainment, use of the internet alone (vs internet use with help or not at all), having an internet-enabled computer or tablet, and having sent at least 1 message through the patient portal increased likelihood of being willing.

Conclusions

Health care teams should assess older adults’ capabilities and comfort related to completion of patient portal–based forms and support those willing to make the digital transition. Paper forms and oral collection of information should remain available for those unable or unwilling to make this digital transition.

Introduction

As health care practitioners increasingly request that patients complete forms and questionnaires electronically using patient portals rather than on paper, it is important to gauge the willingness of older adults to complete portal-based forms and questionnaires, particularly given known disparities in their use of digital information technologies^1–5 and patient portals.^5–7 The authors used self-reported data from a 2020 health plan member survey to investigate the willingness and the factors affecting willingness of older adults to complete forms and questionnaires online using the health plan’s patient portal.

Methods

Study population

Kaiser Permanente Northern California is a large integrated health care delivery system serving over 4 million members, including more than 700,000 adults aged ≥ 65 years who mostly reside in the Greater San Francisco Bay, Sacramento Valley, Silicon Valley, and Central Valley areas of California. All Kaiser Permanente Northern California members are encouraged to use the health plan’s patient portal for secure messaging with clinicians and member services, viewing of laboratory test results, scheduling of nonurgent appointments, and other tasks, including completion of online forms and clinical questionnaires.

Data source

The Kaiser Permanente Northern California Member Health Survey (MHS) is a periodic research survey that collects sociodemographic and health-related information using a self-administered questionnaire sent to a stratified random sample of health plan adults.^8,9 The 2020 MHS, conducted June through December 2020 during the early months of the COVID-19 pandemic, had a starting sample that was selected from the population of current Kaiser Permanente Northern California adult members who had been members since October 2019 and whose preferred written language in the electronic health record (EHR) was English. Preset numbers of members were randomly sampled within 8 age–sex strata (male, female x ages 25–44, 45–64, 65–74, and 75–90 years), with additional preset numbers of Black, Latino, and Asian/Pacific Islander (Asian/PI) adults identified from EHR race/ethnicity data randomly selected within the same age–sex strata. Members had the option of completing a print questionnaire mailed to their home or an online (DatStat) questionnaire hosted on a Kaiser Permanente Northern California Division of Research secure server that could be linked to from a regular (nonsecure) email sent to everyone with a known email address or using the URL provided in the print survey materials. Members were told that they would be automatically entered into a drawing for 1 of 100 $100 gift cards if they responded. The survey response rate among adults aged 65–85 years, who were the focus of this study, was 52.9% (4338/8200), with 86.7% of these respondents completing a print questionnaire.

Study variables

The 2020 MHS included a question about willingness to complete patient portal questionnaires and forms: “Would you be willing to enter information into an online questionnaire/form on the kp.org website if you were sent a link by email or kp.org secure message? [Yes/No/Not sure].” Survey-derived factors examined for association with willingness to complete patient portal forms and questionnaires included age group, sex, race and ethnicity, educational attainment, internet use, and access to an internet-enabled device. Additionally, the authors used EHR-derived data to identify whether the respondent had an activated patient portal account at the time of the survey and had sent a message to a clinician in the prior 18 months as a proxy for demonstrated ability to communicate health-related information through the patient portal.

Statistical analysis

The authors’ analyses were restricted to the 4105/4338 respondents aged 65–85 years who answered the question about online form completion. All analyses were performed using SAS v 9.4 (SAS Institute, Cary, NC, 2014). Respondents were assigned poststratification weighting factors so that the analytic sample, after weighting, would reflect the age x sex x racial/ethnic (White, Black, Latino, Asian/PI, other) composition of the 2019 Kaiser Permanente Northern California membership aged 65–85 years with English as their written language preference. The individual-level poststratification weights were created by assigning each respondent to a cell based on age (65–74, 75–85 years), sex (male, female), and racial/ethnic group (White, Black, Latino, Asian/PI, Other) and then dividing the total number of Kaiser Permanente Northern California members in that cell by the number of survey respondents in that cell. To prevent variance inflation due to the very large denominators produced from the population weights, the authors normalized the population weights so that the total number in the analytic weighted sample was equivalent to that of the unweighted sample, while maintaining the same age–sex–racial/ethnic composition of the population-weighted sample.

The authors first produced descriptive statistics describing characteristics of the study population overall and by 2 age groups (65–74 and 75–85 years) and 4 racial/ethnic groups (White, Black, Latino, and Asian/PI), as shown in Table 1. The authors next compared percentages of older adults willing (vs not willing or not sure) to complete patient portal forms and questionnaires by age group, racial/ethnic group, educational attainment (high school graduate or less, some college/associate degree, bachelor’s degree or higher), self-reported ability to use the internet (not at all, with help, by self), having an internet-enabled computer (vs tablet or smartphone only or no internet-enabled device), and recent history of sending ≥ 1 patient portal message to a Kaiser Permanente Northern California clinician. The authors used χ² tests to identify factors associated with being willing to complete patient portal forms, with P values < 0.05 considered to be statistically significant. These comparisons are shown in the Figure.

Table 1:

Characteristics of the study population, by age and racial/ethnic groups, based on weighted survey data

Characteristic	Age group			Race and ethnicity
	All	65–74 y	75–85 y	White	Black	Latino	Asian/PI
Age (mean y ± SD)	72.6 ± 3.5	69.3 ± 1.9	79.1 ± 1.7	72.8 ± 3.6	72.3 ± 3.6	72.4 ± 3.8	72.0 ± 3.3
Female	54.7%	54.1%	56.1%	54.7%	58.5%	55.3%	54.4%
Educational attainment
≤ High school graduate	18.6%	16.8%	22.3%a	16.4%	22.8%b	40.4%b	13.7%
Some college/associate degree	34.0%	34.9%	32.2%	34.0%	44.2%	36.0%	28.5%
≥ Bachelor’s degree	47.4%	48.3%	45.5%	49.6%	33.0%b	23.6%b	57.8%b
Ability to use the internet
Uses by self without help	83.5%	88.9%	73.0%a	87.4%	71.8%b	73.5%b	78.8%b
Uses with help or proxy uses	8.4%	5.6%	13.6%	6.4%	12.7%b	13.6%b	12.2%b
Does not use	8.1%	5.5%	13.4%a	6.2%	15.5%b	12.9%b	9.0%b
Access to an internet-enabled device
Desktop or laptop computer	92.2%	94.0%	88.9%a	94.3%	87.6%b	87.5%b	89.4%b
Wi-Fi–enabled tablet or smartphone only	1.4%	1.3%	1.4%	0.7%	1.5%	2.3%b	3.3%b
Nonec	6.4%	4.7%	9.7%a	4.9%	10.9%b	10.2%b	7.3%b
Has an activated patient portal account	94.6%	96.2%	91.6%a	96.3%	87.6%b	91.7%b	94.5%
Sent ≥ 1 message through patient portal	63.8%	66.7%	58.2%a	68.6%	45.4%b	56.9%b	56.8%b

Note: Means and percentages are based on respondent data that were weighted to the age–sex–racial/ethnic group composition of the health plan membership. Table subgroup numbers (n) prior to and [after] weighting are as follows: All: N = 4105 [4105]; 65–74 years: n = 2328 [2725]; 75–85 years: n = 1771 [1380]; White: n = 2770 [2761]; Black: n = 291 [298]; Latino: n = 329 [363]; Asian/PI: n = 682 [625].

^a75–85 year age group significantly differs from 65–74 year age group after controlling for sex and racial/ethnic group at p < 0.01.

^bRacial/ethnic group significantly differs from White group after controlling for sex and age group at p < 0.01.

^cIncludes those who have a cell phone that is not used to access email or the internet.

Asian/PI, Asian or Native Hawaiian/Pacific Islander; SD, standard deviation.

Figure:

Percentages of adults aged 65–85 years willing to complete an online form or questionnaire through the health plan’s patient portal, by sociodemographic characteristics and ability to use the internet, overall and by age group. Asian/PI = Asian or Pacific Islander; assoc. = associate.Legend: P < 0.05 ^aWhite; ^b≥ Bachelor’s degree; ^cSome college/associate degree; ^dUses internet by self; ^eUses internet with help.

The authors used modified log-Poisson regression models to derive adjusted prevalence ratios with 95% confidence intervals (aPR, CI) that reflect the relationship of demographic and internet use factors with willingness to complete patient portal forms, adjusting for the covariates. Because the outcome being modeled was not a rare event, an aPR is a more appropriate measure of association than an adjusted odds ratio derived from a logistic regression model, which can inflate the strength of association.^10,11 Receiver operating characteristic (ROC) curves were created and areas under the ROC curves (AUC) were calculated to compare model fit using the predicted probabilities from the log-Poisson regression models.¹² These results are shown in Table 2.

Table 2:

Factors associated with older adults’ willingness to complete forms and questionnaires using a health plan patient portal

Factor	Model 1a	Model 2b	Model 3c
	aPR [95% CI]	aPR [95% CI]	aPR [95% CI]
Age group, y
65–74	reference	reference	reference
75–85	0.78 [0.73-0.82]	0.87 [0.83-0.92]	0.88 [0.83-0.93]
Sex
Male	reference	reference	reference
Female	0.97 [0.92-1.02]	0.97 [0.92-1.01]	0.97 [0.92-1.02]
Racial/ethnic group
White	reference	reference	reference
Black	0.87 [0.78-0.98]	0.97 [0.87-1.08]	1.00 [0.90-1.11]
Latino	0.84 [0.74-0.95]	0.89 [0.79-0.99]	0.90 [0.80-1.00]
Asian or Pacific Islander	0.89 [0.83-0.96]	0.96 [0.90-1.03]	0.98 [0.92-1.05]
Other	0.77 [0.55-1.09]	0.90 [0.68-1.20]	0.92 [0.69-1.22]
Education
≥ Bachelor’s degree	reference	reference	reference
Some college/associate degree	0.84 [0.79-0.89]	0.91 [0.86-0.96]	0.91 [0.87-0.96]
≤ High school graduate	0.61 [0.55-0.67]	0.78 [0.71-0.85]	0.79 [0.72-0.86]
Ability to use the internet
Uses by self	—	reference	reference
Uses with help or a proxy uses it	—	0.41 [0.34-0.50]	0.43 [0.35-0.52]
Does not use	—	0.11 [0.07-0.18]	0.13 [0.08-0.19]
Sent ≥ 1 message through patient portal in prior 18 mo
No	—	—	reference
Yes	—	—	1.20 [1.13-1.28]
Model fitd	AUC = 0.64	AUC = 0.70	AUC = 0.72

^aModel includes age group, sex, racial/ethnic group, and education.

^bModel includes age group, sex, racial/ethnic group, education, and ability to use the internet.

^cModel includes age group, sex, racial/ethnic group, education, ability to use the internet, and history of sending a message through the patient portal in the prior 18 mo.

^dModel fit is AUC-statistic calculated using logistic regression with the predicted values from the log-Poisson model as the independent variable as described in the statistical analysis subsection of the text.

aPR: Adjusted prevalence ratio;AUC, Area under the receiver operating characteristic curve; CI: Confidence interval; variable = Willing to complete patientDependent variable, Willing to complete patient portal forms/questionnaires vs not willing/not sure.

All reported results are based on weighted data, and all subgroup differences mentioned in the text are significant at P < 0.01. The authors did not adjust for multiple comparisons but report the results of all statistical tests.¹³

Results

After weighting, the analytic sample was 54.7% female, mean age 72.6 ± 3.5 years, and 67.3% were White, 7.3% Black, 8.8% Latino, 15.2% Asian/PI, and 1.4% other race or ethnicity. Over 80% of adults had attended college, with approximately 47% having a bachelor’s or postgraduate degree. Most adults used the internet (83.5% without help and 8.4% with help or proxy user) and had an internet-enabled device (93.2% a computer or tablet, and an additional 2.5% had a smartphone). Approximately 95% had an activated patient portal account at the time of the survey, and about two-thirds had used the patient portal in the past 18 months to send a message to a clinician. As shown in Table 1, adults aged 75–85 (vs 65–74) years were less likely to have a computer, to use the internet by themselves, to have a patient portal account, and to have used the patient portal to send a secure message in the past 18 months. Compared with White adults, Black, Latino, and Asian/PI adults were less likely to use the internet without help or to use the internet at all, to have a computer, to have an activated patient portal account, and to have sent a message through the patient portal.

Among noninternet users, 62.4% did not have access to an internet-enabled device and only 32.6% had access to an internet-capable computer, with no significant difference by age or racial/ethnic group. Noninternet users were less likely than internet users to have a 4-year college degree (17.9% vs 50.1%, P < 0.0001), an activated patient portal account (56.6% vs 98.0%, P < 0.0001), and to have sent ≥ 1 secure message during the prior 18 months (13.4% vs 68.5%, P < 0.0001), differences that retained the same statistical significance level after adjusting for sex, age, and racial/ethnic group.

Overall, 59.6% [CI = 58.1-61.2%] of older adults were willing to complete patient portal forms, 17.6% [CI = 16.4-18.8%] were not willing, and 22.8% [CI = 21.4-24.1%] were not sure (Figure). Adults aged 75–85 (vs 65–74) years were less likely to be willing (49.5% vs 64.8%, P < 0.001) and more likely to be unwilling (25.9% vs 13.4%, P < 0.001). Black and Latino (both age groups) and Asian/PI (older age group) adults were less likely to be willing than White adults. Willingness increased with level of education and the ability to use the internet (both age groups; Figure).

Among adults who used the internet without and with help, 98.0% and 94.0%, respectively, could access the internet using a computer, with the remainder only having access by touchscreen tablet or smartphone. Among adults who used the internet with help, willingness did not differ by mode of access. However, among those who used the internet without help, willingness to complete patient portal forms was significantly greater among those who had internet access using a computer compared with access only by tablet or smartphone (69.1% vs 39.6%, P < 0.01, aPR of 1.71 [CI = 1.03-2.85] after adjusting for sex, age, and racial/ethnic group).

Model 1 (Table 2) shows that differences in willingness to complete patient portal forms remained significant for age group, racial/ethnic group, and educational attainment after controlling for other sociodemographic factors. Not shown is that willingness was significantly lower among women than men only in the 75–85-year age group (aPR = 0.86 [CI = 0.78-0.95]). Model 2 (Table 2) shows that after adjusting for sociodemographic factors, ability to use the internet by oneself increased willingness by approximately 60% compared with those who used the internet with help (aPR = 0.41 [CI = 0.34-0.50]) and by nearly 90% compared with those who do not use the internet at all (aPR = 0.11 [CI = 0.07-0.18]). Model 3 (Table 2) shows that history of having sent ≥ 1 message through the patient portal in the prior 18 months increases likelihood of being willing to complete patient portal forms by 20% after adjusting for sociodemographic and internet use factors. However, the addition of this factor that serves as a proxy for patient portal experience had a negligible effect on the aPRs of the other factors and did not improve accuracy of predicting who would be willing to complete patient portal forms. Not shown is that among adults who used the internet without help, having only smartphone or tablet (vs computer) access decreased likelihood of being willing to complete online forms by approximately 40% (aPR = 0.63 [CI = 0.58-0.67]) after adjusting for sociodemographic characteristics.

To examine whether the association of ability to use the internet by oneself vs only with help predicted willingness to complete online forms was similar across racial/ethnic groups, the authors ran Model 2 (omitting the racial/ethnic group variable) separately for the 4 racial/ethnic groups. The aPRs for White, Latino, and Asian/PI adults (range 0.40–0.47) were consistent with the overall aPR of 0.41, with Black adults having a much lower but unstable aPR.

Discussion

In this study, the authors found that although the majority of older adults indicated willingness to complete online forms and questionnaires using a patient portal, a large percentage were not willing or not sure that they would be willing to do so. The authors found significant differences in willingness by age group, racial/ethnic group, educational attainment, and internet access factors, as well as significant racial/ethnic group differences in internet access factors. After adjusting for internet access factors, demographic differences in willingness to complete patient portal forms decreased (age group, education) or became statistically nonsignificant (racial/ethnic group).

Among older adults who used the internet without help, those who only had internet access via mobile phone were significantly less likely than those who had computer or tablet access to be willing to complete patient portal–based forms. Usability studies have identified barriers commonly experienced by older adults when completing online tasks using a mobile phone, including vision and readability problems (eg, small screen size, small font size, dropdown lists, buttons) and dexterity problems with touch screens and small virtual keyboards.¹⁴ This suggests that improving older adults’ access to the internet and an internet-enabled computer or tablet, along with offering training and other types of technical support to improve general digital skills and skills specific to completing online forms and questionnaires accessed through a patient portal, will likely be needed to promote more equitable use of patient portals to complete forms and questionnaires. This type of training and support began to be offered by health plans during the COVID-19 emergency to encourage patients’ uptake of video visits.¹⁵

Understanding patient preferences, capabilities, and internet access factors is important, as patients unwilling or unable to enter information into electronic forms can disrupt workflow and result in frustration of patients and health care staff, impaired collection of patient-reported outcome data, and biased research and survey results.^16,17 More importantly, these findings have health care access implications for the rapidly growing older adult patient population, as disparities among this population will result in certain groups being more adversely impacted by the growing reliance on digital technology.

To the authors’ knowledge, this is the first study to investigate the willingness of older adults to provide health information through online patient portal forms and questionnaires. Strengths of the study include a large socioeconomically and racially/ethnically diverse study population, providing the option to complete the survey using a print or online questionnaire, and the ability to examine factors associated with willingness derived from contemporaneous self-reported and EHR data. However, the authors acknowledge that this study had some limitations. First, the outcome variable, willingness to complete forms and questionnaires through a patient portal, was based on self-reporting that was not cross-validated with EHR data. Second, data were collected in 2020, shortly after the start of the COVID-19 pandemic, and willingness and capability of older adults to complete forms and questionnaires using the patient portal may have changed since that time. However, these self-reported findings remain useful for understanding barriers and the digital divide. Third, the authors also reported on an aggregated Asian/PI group, which previous research suggests masks Asian ethnic subgroup differences in older adults’ use of digital information technologies and patient portals for health-related purposes.¹⁸ Future studies in larger populations should examine ethnic subgroup differences. Fourth, although the authors used a poststratification weighting factor to make the analytic dataset better reflect the underlying age, sex, and racial/ethnic composition of the member population to which they wanted to generalize the results, they acknowledge that this statistical adjustment may not have overcome all sources of nonresponse bias. Finally, the study sample was restricted to older adults from one health plan who could read and understand English, who were better educated than the general older adult population,² and who reside in a geographic area of the United States that has relatively good high-speed broadband access, which may limit generalizability to other populations.

Conclusion

These findings underscore the importance of having health care teams assess older adults’ capabilities, concerns, and comfort using internet-enabled devices and being ready to assist those who want help or indicate a willingness to learn.⁷ Paper forms and questionnaires and oral collection of information should remain available for those unable or unwilling to make the digital transition.¹⁶ Patients who are willing to try to make the transition but need additional support to do so should be given encouragement by health care staff, individualized skills training and ongoing technical support, and, if needed, the digital technologies to access and enter information into electronic forms and questionnaires using the patient portal.