This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
As the US health care system is embracing data-driven care, personal health information (PHI) has become a valuable resource for various health care stakeholders. In particularly, health consumers are expected to autonomously manage and share PHI with their health care partners. To date, there have been mixed views on the factors influencing individuals’ health data–sharing behaviors.
This study aimed to identify a key factor to better understand health information sharing behavior from a health consumer’s perspective. We focused on daily settings, wherein health data–sharing behavior becomes a part of individuals’ daily information management activities. Considering the similarity between health and finance information management, we explicitly examined whether health consumers’ daily habit of similar data sharing from the financial domain affects their PHI-sharing behaviors in various scenarios.
A Web-based survey was administered to US health consumers who have access to and experience in using the internet. We collected individual health consumers’ intention to share PHI under varying contexts, habit of financial information management (operationalized as internet banking [IB] use in this paper), and the demographic information from the cross-sectional Web-based survey. To isolate the effect of daily IB on PHI-sharing behaviors in everyday contexts, propensity score matching was used to estimate the average treatment effect (ATE) and average treatment effect on the treated (ATET) regarding IB use. We balanced the treatment and control groups using caliper matching based on the observed confounding variables (ie, gender, income, health status, and access to primary care provider), all of which resulted in a minimal level of bias between unmatched and matched samples (bias <5%).
A total of 339 responses were obtained from a cross-sectional Web-based survey. The ATET results showed that in terms of sharing contents, those who used IB daily were more likely to share general information (
This study examined whether daily management of similar information (ie, personal financial information) changes health consumers’ PHI-sharing behavior under varying sharing conditions. We demonstrated that daily financial information management can encourage health information sharing to a much broader extent, in several instances, and with many stakeholders. We call for more attention to this unobserved daily habit driven by the use of various nonhealth technologies, all of which can implicitly affect patterns and the extent of individuals’ PHI-sharing behaviors.
As the US health care system is embracing data-driven care, personal health information (PHI) needs to be shared and managed across clinical settings by health consumers [
Health consumers manage not only health information but also nonhealth information on a daily basis. As individuals perform day-to-day functioning in the areas of finance, communication, transportation, socialization, and entertainment [
Against this backdrop, little attention has been paid to explore health consumers’ health data sharing in the context of daily living. We propose that health consumers are likely to be attracted from similar experiences and may have formed a habit with repeated exposure to the similar tasks while executing health and nonhealth tasks [
We recruited participants who were or had the potential to manage and share their own health data electronically. As we particularly focused on individuals who can manage health and financial tasks by the use of relevant technologies, we needed a study sample in which we can capture their daily activities of various data management beyond health care settings. To this end, we contracted with a market research company that had access to the paid panel of health consumers across 50 states in the United States for administering a cross-sectional Web-based survey in 2017. Each participant was incentivized by the completion and quality of their response, which was mainly managed by the market research firm. Our Web-based survey incorporated multiple items that measure health information–sharing behaviors, including health consumers’ intention to share their health information, information sharing contexts, demographic information, and daily technology use such as internet banking (IB) use. An institutional review board approval was obtained before survey distribution. As a result, a total of 339 responses were used for further analysis.
All survey items were sourced from the existing literature, as presented in
Finally, demographic information was captured for gender, marital status, income, education, occupation, race, ethnicity, health condition (chronic disease), and having a primary doctor within the domicile [
Survey items.
Types | Survey itemsa, b | Reference | |
Sharing contents |
General information Current health information Past health information All health information |
[ |
|
Sharing instances |
In all cases and instances For the purposes of care delivery within the clinical setting For the purposes of other than provision of care (eg, research or marketing) In case of medical emergency conditions |
[ |
|
|
|||
|
General constituents |
Other physicians (who are not involved in your care) at hospitals Other community physicians not involved in your care Health administrators (eg, managers), government agencies Health care researchers Health insurance companies |
[ |
Care-related constituents |
Physicians (who are involved in your care) at hospitals Other community physicians involved in your care (treating physicians) Nurses Pharmacists |
[ |
|
Habitual use of internet banking |
Frequency of internet banking usec |
[ |
aWe adopted all items from the study by Whiddett et al [
bInformation-sharing items are measured on a 5-point Likert scale anchoring on 1 (strongly disagree) to 5 (strongly agree).
cFrequency of daily technology use is measured on daily, weekly, and monthly scales adopted from the survey of International Finance Corporation [
The objective of the study was to examine whether frequent use of IB affects when, what, and with whom health consumers are willing to share their own personal health data. However, in observational studies similar to this study, it is often a challenge to isolate the treatment effect because confounding factors can influence both treatment and outcome [
In our data, health consumers’ use of IB and health data sharing can be confounded by known factors, that is, demographic characteristics and health status. Following a step-by-step suggestion from the study by Becker and Ichino [
As shown in
Although our data were obtained from a cross-sectional Web-based survey from US health consumers, we further evaluated the representativeness of our sample compared with established benchmark. The Board of Governors of the Federal Reserve System has conducted a Web-based survey on financial consumers’ use of mobile banking in selective years [
Characteristics of survey participants (total number of responses=339).
Demographic variables | All IBa users (N=339), n (%) | Daily IB users (n=96), n (%) | Weekly IB users (n=170), n (%) | Monthly IB users (n=73), n (%) | |
|
|||||
|
Male | 114 (33.6) | 34 (35) | 54 (31.8) | 26 (36) |
|
Female | 225 (66.4) | 62 (65) | 116 (68.2) | 47 (64) |
|
|||||
|
Married | 188 (55.5) | 51 (53) | 103 (60.6) | 34 (47) |
|
Divorced | 26 (7.7) | 6 (6) | 12 (7.1) | 8 (11) |
|
Separated | 7 (2.1) | 0 (0) | 4 (2.4) | 3 (4) |
|
Never married | 118 (34.8) | 39 (41) | 51 (30.0) | 28 (39) |
|
|||||
|
18-24 | 53 (15.6) | 15 (16) | 21 (12.4) | 17 (23) |
|
25-34 | 128 (37.8) | 39 (41) | 67 (39.4) | 22 (30) |
|
35-44 | 78 (23.0) | 23 (24) | 42 (24.7) | 13 (18) |
|
45-54 | 43 (12.7) | 13 (14) | 24 (14.1) | 6 (8) |
|
55-64 | 27 (8.0) | 5 (5) | 11 (6.5) | 11 (15) |
|
≥65 | 10 (3.0) | 1 (1) | 5 (3.0) | 4 (6) |
|
|||||
|
<20,000 | 51 (15.0) | 13 (14) | 23 (13.5) | 15 (21) |
|
20,000-39,999 | 76 (22.4) | 20 (21) | 30 (17.7) | 26 (36) |
|
40,000-59,999 | 59 (17.4) | 22 (23) | 28 (16.5) | 9 (12) |
|
60,000-79,999 | 53 (15.6) | 14 (15) | 31 (18.2) | 8 (11) |
|
80,000-99,999 | 44 (13.0) | 7 (7) | 31 (18.2) | 6 (8) |
|
>100,000 | 56 (16.5) | 20 (21) | 27 (15.9) | 9 (12) |
|
|||||
|
Less than high school | 9 (2.7) | 4 (4) | 4 (2.4) | 1 (1) |
|
High school graduate | 70 (20.7) | 19 (20) | 29 (17.1) | 22 (30) |
|
Some college | 93 (27.4) | 28 (29) | 46 (27.1) | 19 (26) |
|
2-year degree | 35 (10.3) | 11 (12) | 14 (8.2) | 10 (14) |
|
4-year degree | 85 (25.1) | 21 (22) | 54 (31.8) | 10 (14) |
|
Master’s degree | 40 (11.8) | 10 (10) | 21 (12.4) | 9 (12) |
|
PhD | 7 (2.1) | 3 (3) | 2 (1.2) | 2 (3) |
|
|||||
|
Employed full time | 196 (57.8) | 65 (68) | 102 (60.0) | 29 (40) |
|
Employed part time | 40 (1.8) | 9 (9) | 21 (12.4) | 10 (14) |
|
Unemployed looking for work | 29 (8.6) | 7 (7) | 11 (6.5) | 11 (15) |
|
Unemployed not looking for work | 34 (10.0) | 10 (10) | 15 (8.8) | 9 (12) |
|
Retired | 19 (5.6) | 1 (1) | 9 (5.3) | 9 (12) |
|
Disabled | 21 (6.2) | 4 (4) | 12 (7.1) | 5 (7) |
|
|||||
|
White | 269 (79.4) | 77 (80) | 137 (80.6) | 55 (75) |
|
Black | 35 (10.3) | 6 (6) | 15 (8.8) | 14 (19) |
|
Asian | 23 (6.8) | 9 (9) | 12 (7.1) | 2 (3) |
|
Other | 12 (3.5) | 4 (4) | 6 (3.5) | 2 (3) |
|
|||||
|
Hispanic | 38 (11.2) | 10 (10) | 21 (12.4) | 7 (10) |
|
Non-Hispanic | 301 (88.8) | 86 (90) | 149 (87.7) | 66 (90) |
|
|||||
|
Yes | 113 (33.3) | 24 (25) | 60 (35.3) | 29 (40) |
|
No | 226 (66.7) | 72 (75) | 110 (64.7) | 44 (60) |
|
|||||
|
Within 5 miles | 150 (44.3) | 44 (46) | 77 (45.3) | 29 (40) |
|
Within 10 miles | 129 (38.1) | 32 (33) | 66 (38.8) | 31 (43) |
|
Within 30 miles | 44 (13.0) | 15 (16) | 20 (11.8) | 9 (12) |
|
Not available | 16 (4.7) | 5 (5) | 7 (4.2) | 4 (6) |
aIB: internet banking.
bn=236 for all IB users.
We conducted PSM analysis using Stata version 14.2 software (College Station, Texas). In our analysis, we chose income, race, health status, and having a primary care doctor within close proximity as our confounding variables, among others, which are likely to influence both IB use and health information sharing behavior. Subsequently, the propensity score was calculated for each block using a logit model.
Distribution of propensity score between treated and untreated groups.
Covariate balance before and after propensity score matching.
Variables | Unmatched sample | Matched sample | ||||||||
|
Mean, Treated | Mean, Untreated | Bias (%) | Mean, Treated | Mean, Untreated | Bias (%) | ||||
|
||||||||||
|
Female | 0.65 | 0.67 | −5 | .69 | 0.65 | 0.65 | 0 | >.99 | |
|
||||||||||
|
20,000-39,999 | 0.22 | 0.23 | −1.6 | .90 | 0.22 | 0.22 | 0 | >.99 | |
|
40,000-59,999 | 0.21 | 0.15 | 15.4 | .21 | 0.21 | 0.21 | 0 | >.99 | |
|
60,000-79,999 | 0.16 | 0.16 | 1.1 | .93 | 0.16 | 0.16 | 0 | >.99 | |
|
80,000-99,999 | 0.08 | 0.15 | −21.8 | .10 | 0.08 | 0.08 | 0 | >.99 | |
|
>100,000 | 0.22 | 0.15 | 19.4 | .11 | 0.22 | 0.22 | 0 | >.99 | |
|
||||||||||
|
No | 0.72 | 0.63 | 17.9 | .16 | 0.72 | 0.72 | 0 | >.99 | |
|
||||||||||
|
Within 5 miles | 0.42 | 0.44 | −2.6 | .84 | 0.42 | 0.42 | 0 | >.99 |
Density plots in health information sharing behavior. Treated sample comprised daily users of internet banking, and the rest of the users were included in the untreated group.
Finally, we estimated ATE and ATET as displayed in
In
Average treatment effect of daily internet banking use for matched pair sample.
Outcomes | Coefficient | SE | Z score | 95% CI | |||||||||
|
|||||||||||||
|
General information | 0.324 | 0.124 | 2.61 | .009 | 0.081 to 0.567 | |||||||
|
Current information | 0.364 | 0.125 | 2.91 | .004 | 0.119 to 0.61 | |||||||
|
Past information | 0.17 | 0.127 | 1.34 | .18 | −0.08 to 0.42 | |||||||
|
Full information | 0.215 | 0.153 | 1.4 | .16 | −0.085 to 0.514 | |||||||
|
|||||||||||||
|
All cases and situations | 0.281 | 0.151 | 1.87 | .06 | −0.014 to 0.577 | |||||||
|
Care purposes | 0.131 | 0.111 | 1.18 | .24 | −0.086 to 0.349 | |||||||
|
Noncare purposes | 0.086 | 0.178 | 0.48 | .63 | −0.263 to 0.435 | |||||||
|
Medical emergency | 0.133 | 0.095 | 1.39 | .16 | −0.054 to 0.32 | |||||||
|
|||||||||||||
|
|
||||||||||||
|
|
Your physician | 0.084 | 0.114 | 0.74 | .46 | −0.139 to 0.307 | ||||||
|
|
Involving community physician | −0.064 | 0.122 | −0.53 | .60 | −0.304 to 0.175 | ||||||
|
|
Nurses | 0.119 | 0.139 | 0.85 | .39 | −0.154 to 0.393 | ||||||
|
|
Pharmacists | −0.023 | 0.174 | −0.13 | .89 | −0.364 to 0.317 | ||||||
|
|
||||||||||||
|
|
Noninvolving physician at hospital | 0.278 | 0.191 | 1.45 | .15 | −0.097 to 0.653 | ||||||
|
|
Noninvolving community physician | 0.234 | 0.187 | 1.25 | .21 | −0.132 to 0.601 | ||||||
|
|
Health administrators (eg, managers) | 0.347 | 0.174 | 1.99 | .05 | 0.005 to 0.688 | ||||||
|
|
Government | 0.23 | 0.189 | 1.22 | .22 | −0.14 to 0.601 | ||||||
|
|
Health care researchers | 0.179 | 0.185 | 0.96 | .34 | −0.185 to 0.542 | ||||||
|
|
Insurance | 0.169 | 0.179 | 0.94 | .35 | −0.183 to 0.521 |
Average treatment effects on the treated of daily internet banking use.
Outcomes | Coefficient | SE | Z score | 95% CI | |||||||||
|
|||||||||||||
|
General information | 0.346 | 0.140 | 2.470 | .01 | 0.071 to 0.621 | |||||||
|
Current information | 0.399 | 0.134 | 2.960 | .003 | 0.135 to 0.662 | |||||||
|
Past information | 0.208 | 0.145 | 1.430 | .15 | −0.076 to 0.492 | |||||||
|
Full information | 0.334 | 0.160 | 2.090 | .04 | 0.021 to 0.647 | |||||||
|
|||||||||||||
|
All cases and situations | 0.319 | 0.139 | 2.300 | .02 | 0.047 to 0.591 | |||||||
|
Care purposes | 0.192 | 0.117 | 1.640 | .10 | −0.037 to 0.421 | |||||||
|
Noncare purposes | 0.156 | 0.200 | 0.780 | .44 | −0.236 to 0.547 | |||||||
|
Medical emergency | 0.179 | 0.104 | 1.710 | .09 | −0.026 to 0.383 | |||||||
|
|||||||||||||
|
|
||||||||||||
|
|
Your physician | 0.058 | 0.136 | 0.430 | .67 | −0.208 to 0.324 | ||||||
|
|
Involving community physician | −0.119 | 0.139 | −0.860 | .39 | −0.392 to 0.153 | ||||||
|
|
Nurses | 0.184 | 0.139 | 1.320 | .19 | −0.089 to 0.457 | ||||||
|
|
Pharmacists | 0.021 | 0.155 | 0.140 | .89 | −0.283 to 0.326 | ||||||
|
|
||||||||||||
|
|
Noninvolving physician at hospital | 0.331 | 0.200 | 1.660 | .10 | −0.060 to 0.722 | ||||||
|
|
Noninvolving community physician | 0.201 | 0.199 | 1.010 | .31 | −0.188 to 0.590 | ||||||
|
|
Health administrators (eg, managers) | 0.350 | 0.177 | 1.980 | .05 | 0.003 to 0.698 | ||||||
|
|
Government | 0.232 | 0.168 | 1.380 | .17 | −0.097 to 0.561 | ||||||
|
|
Health care researchers | 0.146 | 0.177 | 0.820 | .41 | −0.202 to 0.493 | ||||||
|
|
Insurance | 0.249 | 0.152 | 1.640 | .10 | −0.049 to 0.548 |
Despite growing expectation of individuals’ responsibility for sharing PHI for data-driven care, there have been mixed results regarding factors influencing health consumers’ sharing intention under various circumstances. Given that management of financial information resembles that of health information, this study called for more attention on individuals’ daily use of financial information management as a proxy for health information sharing and hypothesized that the more frequently individuals managed their financial information, the more likely they were willing to share PHI under various sharing conditions. Our PSM results revealed that daily IB users were more willing to share the large extent of health information for all instances, even with personnel who were not directly involved in their care process. This is one of the first studies that explores the role of a daily habit of financial information management to predict individuals’ intention in sharing their health information.
Although we presented important findings on the role of daily habit of IB use, our results should be interpreted with caution because of their limitations. First, our study is conducted via a cross-sectional Web-based survey. Although our research question was aligned with our study design, tracking health consumers’ information management behavior over time can provide an in-depth view and further identify contextual factors in the daily living context. The cross-section time series information on health consumers’ information management in daily living would be beneficial in the future to capture granular level of measures and to control unobserved heterogeneity among individuals. Second, we conceptualized our treatment effect as habitual use of financial data management and operationalized it by the daily use of IB. Although our unidimensional, binary measure of IB use was appropriate for PSM methodology, future research can incorporate multi-item measures to capture multidimensional aspects of financial data management for health consumers in the richer research models. Finally, we acknowledge that majority of survey respondents in our study have no immediate health issues (237/339, 69.9%); therefore, the results of this study may not be generalized to patients who have various health conditions and statuses. As we assumed that the same individuals can be both health and financial consumers, the findings of this study can be a baseline information to compare individuals’ behaviors in medical situations in subsequent research.
In this paper, we first examined the daily habit that has not been of focus in health care research. More specifically, this paper juxtaposed the similarity between health and financial data and identified financial data habit as a key factor in understanding health data sharing from the same individuals. Our approach assumed that the same individuals are customers for both health and financial services; therefore, such individual-level behaviors can be closely related. Theoretically, it is known that when people cope with a new event, their reaction might be predictable simply because there are likely to base their reaction on past experience or their knowledge of similar situations [
Although prior health care literature highlighted the importance of individuals’ habit to understand health behaviors, habit has been mainly defined within the context of health care. For example, health consumers’ exposure to personal health records technology influenced individuals’ share of PHI with care providers and non–care-related providers [
This paper also showed that frequent exposure to IB is positively related to health data–sharing behavior. This finding is in line with and extends prior health research in two ways. First, although the effect of internet use has been widely discussed to understand health information sharing behavior, the influence of habitual internet use is lesser known [
For future research, it will be worthwhile to revisit this research model in clinical setting and explore the effect of habit on health data sharing in the clinical setting for those who have chronic conditions or medical urgency. Health data sharing is a complex and variable phenomenon, and more interdisciplinary research is indispensable. As health consumers are attaining more ownership to manage and share their own health information via websites, wearables, and mobile apps, it is important to determine whether they are capable of dealing with this volume of data [
This research has practical implications. First, health technology vendors may design health information management tools modeled after financial information management tools [
The objective of this study was to examine the effect of daily use of financial technology on health information sharing. Considering the similarity between health and financial technology and the characteristics of such information, this study proposes that the unobserved habit of managing sensitive information daily can further affect managing and sharing another type of sensitive information—PHI. Results from PSM reveal that frequent users of financial technology are more prone to share their entire health information in all instances, even with non–care-related stakeholders. Subsequent research can explore more granular types of habits in various life domains to better understand health consumers’ readiness to manage self-health information for realizing consumer-centered care in the future.
National profiles of mobile banking users.
average treatment effect
average treatment effect on the treated
exploratory factor analysis
internet banking
personal health information
propensity score matching
We would like to thank Pouyan Zadeh for the support in data collection.
None declared.