key: cord-0320825-dw66oe4a
authors: Ford, S.; Oberst, K.; Ilardo, J.; An, H. S.; Jones, N.; Hong, H.; Clark, K.; Luo, Z.
title: Offering an Internet Survey Response Modein a Flint Water Crisis Medicaid Population
date: 2020-11-23
journal: nan
DOI: 10.1101/2020.11.22.20236497
sha: 9f323821467284fbd566be37a29eecf95aa87456
doc_id: 320825
cord_uid: dw66oe4a

Abstract Objective: To examine the preferred response mode (internet, phone, paper)to a Medicaid enrollee survey. Data Source: Data was obtained from survey responses for a sub-sample of Flint Water Crisis Medicaid Expansion Waiver enrollees (N=2584). Study Design: Enrollees were offered the choice of utilizing the internet, telephone, or mail to respond to a survey evaluating health services. Analyses were stratified by age, residency, race, and income. Chi-square was utilized to detect categorical differences. Principal Findings: The majority (p<0.01) of participants responded by internet (55.46%), followed by mail (39.36%), and telephone (5.19%). Of those responding by internet, 75% used smart phones for connectivity. Weighted participation estimates for available survey modes showed variation by age, residence, race/ethnicity and poverty status. A smaller proportion (p<0.01) of ethnicities classified as Hispanic and Other used telephone participation compared to White or Black respondents. Respondents at 200% Federal Poverty Level (FPL) or higher used internet at greater rates than those below 200% FPL (p<0.01). Conclusions: Overall, this low-income population preferred the web-based response mode compared to paper or telephone, the variation by race and FPL suggests the continued presence of a digital divide in the population. Nevertheless, our findings revealed greater utilization of web-based tools for survey administration in vulnerable populations than was expected. This is a valuable finding that may inform future health programming and telehealth efforts particularly in the advent of COVID-19.

• Active health care participation using internet connectivity is increasing and may be associated with improved quality and costs of health care services yet access and usage vary among populations.

• Research suggests that persons of low income and resources are unable to participate with internet based activities due to access and literacy issues that may further exacerbate health disparities.

• This study revealed enrollees of the Flint Michigan Section 1115 Demonstration utilized the internet for survey participation at a greater rate than responding by mail or phone.

• Finding that vulnerable populations have greater access to the internet via digital technology than assumed may expand opportunities to participate in their health care through health information portals and telehealth.

Recent studies suggest low-income populations do not use or have limited access to the internet compared to higher income populations. [1] [2] Approximately 30% of non-elderly, non-dually insured Medicaid adults have never used a computer or the internet and 40% do not use email. 3 Low computer literacy, unaffordable internet services, and lack of access to internet ready devices (e.g. mobile phones, computers, tablets, etc.) have been suggested as reasons for this observation. [4] [5] [6] Reports indicate those who make less than $30,000 equating to over 230% Federal Poverty Level(FPL) per year are less likely to adopt or have access to internet technology. 1, 7 Thus, individuals having fewer economic resources are believed to have low access to internet services and equipment.

Federal programs have been implemented to address the disparity in internet access. Among these are ConnectHomeUSA, 8 which is intended to provide internet accessibility in public spaces and low-income housing. In addition, the Federal Communications Commission (FCC) Lifeline Support for Affordable Communications (Lifeline) program offers reduced fees for mobile phones and internet services for individuals at or below 135% of the poverty level. 9 ,10 This includes individuals and families who qualify for Medicaid, Supplemental Nutrition Assistance Program (SNAP), Supplemental Security Income (SSI), and/or Public All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11. 22.20236497 doi: medRxiv preprint Housing Agency (PHA) programs. 10-12 While these programs address some access issues, barriers still exist in the form of internet related literacy and affordability of internet ready devices. 13 A study conducted by Deloitte 14 found that Medicaid enrollees have increased accessibility to smart phones and tablets, but at a significantly lower rate than employed medically insured persons. Likewise, fewer low-income families have access to personal computers in the home, and instead, primarily use smart phones to access the internet compared to those with middle and high incomes. 15 The coronavirus pandemic has made this digital divide even more stark. 16, 17 This continued disparity can be problematic for low-income patients accessing healthcare portals that are optimized for computer monitor screens and in turn, can limit usability for people who do not own a computer. 18 Few empirical and healthcare quality studies examine Medicaid enrollees' access to the internet and related devices and how they utilize the service. Furthermore, academic reports on technology use for personal health quickly become outdated because of frequent changes in computer technology. 1, 19 Internet service now provides access to a variety of tasks formerly conducted via telephone or paper. In particular, the ability to administer surveys online provides individuals with means to participate at their convenience in a secure, anonymous All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. to complete a survey in more than one sitting resulting in improved response rates. These platforms further provide a seamless approach to ensure participants accurately follow complex skip patterns and jumps based on key responses. Thus, internet-based surveys allow for increased completion rates and minimize the occurrence of invalid responses. 20 Another benefit to online survey methodology is the immediate availability of analyzable datasets. Online survey participants contribute not only to direct data entry but reducing delay from data collection to analysis. Due to the minimization of subsequent data entry and the need for data entry validation, online survey processes are more resource efficient. This rapid turnaround of findings may support more timely program evaluation efforts and facilitate more real-time assessments of program modifications. [20] [21] We planned an enrollee survey as a component of a Medicaid Waiver authorization evaluation. Adding an online response method provided the evaluation team with an opportunity to examine response rates of different participation modes. Based on current reports of lower internet use in low-income Medicaid populations, we expected internet survey responses would be All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint lower than mail or telephone participation rates. Further, we were interested to identify how differences in response mode varied by race/ethnicity, income, and geographic residency. The waiver offers additional family support coordination services for pregnant women and children exposed to lead during the Flint Water Crisis. In addition to expanded healthcare services, the qualifying income increased from at or below 133% of the FPL to 400% FPL. Individuals above 400% FPL may also enroll in the waiver through cost-sharing.

The 1115 waiver authorization required a formal evaluation of the success to which key objectives were achieved. The Institute for Health Policy (IHP) at Michigan State University managed the evaluation process. One component of the evaluation plan was to conduct enrollee surveys to gather information about All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint the utilization of and satisfaction with waiver services and participation. The survey also served to provide an opportunity for participants to report on changes in health outcomes attributed to waiver enrollment.

Hypotheses included in the approved evaluation proposal guided the enrollee survey design and question development. Due to concerns about internet accessibility, the evaluation team originally developed a plan including only telephone and paper response options. Prior to survey implementation, the evaluation team sought community feedback All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. The study design included three waves of enrollee surveys occurring at nine-month intervals. Wave 1 represented baseline measurements with Wave 2 (9-month) and Wave 3 (18-month) followup points planned. Based on the Flint residents' feedback, revised survey mode options included 1) web-based participation using two factor authentication, 2) telephone call-in to the survey center, and 3) postal mail option for paper surveys. This report describes response rates by mode for the Wave 1 survey.

To increase the odds of internet response a short bitly web address and assigned study IDs and passcodes were utilized. This information was disseminated to individuals selected for the sample through the mailed introductory letters. The baseline survey provided an opportunity for individuals to enter an email All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint address and telephone number with permission to contact them by SMS so that invitations to upcoming survey waves could be more efficiently directed by the online survey system.

The sampling pool (N = 24,082) included all waiver enrollees having at least six months continuous enrollment at the time of the survey (i.e., the target population). The data for the sampling pool was acquired from the MDHHS Data Warehouse compliant with MSU HRPP. Stratified random sampling was used to select the survey sample. Table 1 shows the sampling proportions by strata.

Enrollee age class, Genesee County residency, race/ethnicity, and FPL as documented in the Data Warehouse administrative files were used as strata. Three age groups were established: 0-6 years, 7-17 years, and 18+ years of age at the time of waiver enrollment. The residency category included three groups: Always lived in Genesee County, Lived In & Out of Genesee County and Never lived in Genesee County. Individuals exposed to the Flint Water did not necessarily live within the City of Flint or in Genesee County. Exposures through employers, daycare or schools, and healthcare facility location(s) were documented. In recognition of the potential for geographic dispersion, the waiver further permitted ongoing coverage for individuals relocating outside of the greater Flint area after All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Table 1 displays the characteristics of total eligible enrollees by strata and the count of selected sample individuals. Generally, the characteristics of the selected sample reflect the population except for residence class where over-sampling was intentional due to small cell size.

Another concern leading to the large sampling frame was the national interest in conducting research on persons in the region after the water crisis. Many research projects were simultaneously recruiting community participation and the success of encouraging participation was worrisome. This also led the evaluation team to secure approval to provide small monetary incentives of $10 per completed survey to promote participation. The incentive payments were structured so that if All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint a participant submitted all three survey waves, they would receive a $20 bonus payment.

After sample selection, the MDHHS Data Warehouse was accessed to extract most currently available enrollee mailing

information. An honest broker approach was used to obtain the identifiable data necessary to execute the mailing. This identifiable data file was hand-delivered on an encrypted and password protected external drive to the OSR for survey implementation. Survey response data was returned to the evaluation team with a random survey ID assigned by OSR.

The survey was programmed into Qualtrics (Qualtrics Inc.)

for the web-based option. Qualtrics is an online survey platform that organizes data intended for statistical analyses. A short web address was created to reduce the keystrokes required of participants to reach the survey page. Access to the survey was controlled using the unique, single-use study ID to prevent participants from completing the survey more than once. The online survey was further protected from non-waiver enrollee participation by the use of two-factor authentication along with restrictions imposed on the ability of internet search engines to locate the survey. All survey materials regardless of mode were provided in English.

OSR used a batch number procedure to manage the mailing process due to the large sample. The sampled enrollees were All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint

All paper surveys were entered by OSR staff using a blinded, double data entry process. Surveys completed by telephone were monitored by OSR supervisory staff with data entry managed through a Computer Assisted Telephone Interview (CATI) system. Web-based survey responses were directly entered by participants and captured by Qualtrics software in a format readily exportable to spreadsheet and statistical software. All forms of data were compiled and cleaned using SAS 9.4 software (SAS Institute, Inc.)

The randomness of the survey sample was assessed by comparing the population to the sampled survey participants by age class, residence class, race/ethnicity, and FPL. Response rate was also compared by participant's characteristics using chi-square test. The association between the respondent's characteristics and the response mode was tested using a chisquare test. In the case of small cell size less than 5, we used the exacted p-value estimated by Monte Carlo simulation. All chi-square and exact tests were performed by SAS 9.4 software.

The selected survey sample was representative of the population as presented in Table 1 . There were no significant differences of proportion between the eligible population and All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint

Since there were significant differences in response rates by demographic characteristics for age, race, and FPL, weights were used to adjust for sampling bias. Table 3 presents associations between population characteristics and response mode based on the adjusted weight. The overall proportion of online participation exceeded other survey modes among all responders. More than half (55.6%) completed the survey using the internet method. Nearly 40% (39.3%) participated using the postal mail option while just 5.2% used the telephone option.

Significant differences in response mode were observed for all demographic characteristics. Regardless of demographic characteristics, internet response showed the highest rate compared to paper or telephone.

In the case of age class, adults (61.8%) showed the highest rate of internet response with nearly a 10% difference between adult and 7-17 years group. With respect to residence class, the In and out of Genesee group had significantly different response modes compared with Always in Genesee and Never in Genesee. The use of internet and paper response among the In and Out of Genesee was similar with less than 5% difference. Chi-square revealed these categorical differences a significant (p<0.01).

Black enrollees had higher response rates using internet and telephone and were less likely overall to respond by paper than All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint individuals in other categories. Conversely, individuals classified as Hispanic and Other/Unknown race were more likely to respond by paper than White or Black individuals (p<0.01). An operational benefit from internet participation was that over 70% of respondents provided email addresses and/or mobile numbers to receive email or text messaging to notify them of future survey waves.

In this report, we documented higher participation with a web-based survey compared to paper or telephone options.

Medicaid Expansion Waiver. The ordered preference for survey participation method was consistent regardless of race/ethnicity, residence category, or FPL of selected enrollees. Of the 2,584 respondents, the majority (55.5%) responded using the internet option, followed by mailed paper All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint surveys (39.4%) with fewer respondents (5.2%) choosing the telephone option.

While the greatest proportion of respondents within each race/ethnicity category used the online option, the rates of postal mail versus telephone option did vary. Of note, those who were identified as Hispanic or categorized as "Other/Unknown" based on the MDHHS administrative enrollment data were significantly less likely to use the telephone option compared to Black or White. This may reflect the availability of and or comfort with conducting the survey using English as a spoken language. Those having English as a second language may prefer options that permit them more time to look up unfamiliar words and/or take advantage of friends, family or other tools to facilitate translation. Unfortunately, since this data was not collected, we cannot ascertain this premise. This waiver allowed those at higher income levels (>212% FPL) to enroll without cost. Even individuals exposed to the Flint water exceeding 400% FPL were eligible to enroll by paying a monthly fee. This enabled the evaluation team to assess for participation differences among varying income levels. Across all income levels, the internet participation mode was most prevalent.

Prevailing opinions are that accessible survey methods for a vulnerable population include only paper and telephone. 1-3 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint Thus, inclusion of web-based survey methods are generally overlooked due to perceptions regarding lack of access to internet services and technology. This evaluation team's original survey plans followed prevailing opinion. However, input from community advisory groups recommended the evaluation team consider internet and smart phone technology as a means to access the survey. The recommendation proved fruitful.

In addition to concerns regarding internet access, another potential barrier identified with an online survey process was the lack of email addresses that are used by these systems to deliver invitations and participation links. Despite not having email addresses for enrollees for Wave 1, the survey coordinators used the available postal contact information to distribute web address information and found that many respondents used the web-based survey as well. The finding that 70% offered their email address for follow-up waves of the survey revealed access and willingness to use digital communications.

Among the majority who completed the online version of the survey, we documented 75% using smart phones as the means of access. Since most of the respondents who accessed the internet using smart phone technology there is great the potential to address health status and disparities using digital platforms. 15 In addition, smart phone access can mitigate the need for more All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. As an example, this evaluation process was able to maximize technology to reduce administrative costs associated with printing, stuffing, mailing, and manually entering survey data.

These findings are encouraging but further evidence is needed to discern associations between uses of technology and specific characteristics of medically underserved populations.

These results show the degree to which this particular cohort All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint was able to overcome the first presumed barrier of access to the internet. If a reasonable likelihood exists that access is available, these findings may support testing technology with the goal of increasing health literacy, communicating with health providers and insurers, and advocating for health concerns.

Substantial gaps will continue to exist for those Medicaid enrollees who lack direct access to or use the internet 18, 19 and we see continued evidence of the digital divide in health disparities during the COVID-19 pandemic. 16 However, it is imperative that internet technology be made available universally for vulnerable populations to reduce health disparities and reduce administrative costs that could be reinvested into direct care. In order to fully test this, we suggest further investigation among those who still remain unable to connect and interact with technology. Solutions may include individual education about gaining access to the internet as well as social supports to address socio-economic hurdles such as subsidized smart phones or public computers and wireless connectivity.

The use of online resources also has the potential to build trust for individuals who may have difficulty communicating with their physician or disclosing adverse or risky health behaviors. 18 Eliminating the shame and embarrassment of All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint disclosing health concerns preliminarily via secure healthcare internet portals, may mitigate fear and discomfort of talking about health concerns when meeting health providers in person. 24, 25 Limitations We describe participation rates by survey mode using a natural experiment associated with implementing a survey with Flint Water Medicaid Expansion enrollees. This evaluation was not designed, nor participants randomly assigned to empirically compare the utilization of various survey response options. All selected individuals in the sample were offered an opportunity to select among all three response options (internet versus telephone versus postal mail). Individuals self-selected their preferred method.

Another limitation was the inability to determine the influence of responder age. The available age categories reflected the enrollee's age of the enrollee rather than that of the individual completing the survey on behalf of the enrollee.

This prevented the evaluation team from evaluating the impact of differences in internet use by age cohorts which documents younger individuals are more apt to utilize the internet to attend to daily activities. 26 While it may be reasonable to expect those eighteen or older answering on their own behalf, we cannot say anything about the surveys that targeted minors as All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint these were directed to the parent or guardian as reflected in the state's administrative data.

The evaluation process was able to take advantage of a community-based suggestion to measure the degree to which a vulnerable population would participate in an online survey. The outcomes of this experience have practical implications for further exploration of digital health equity initiatives targeting lower-income families. 22 The methodological design of this study may be helpful to other entities involved in quality improvement and evaluation of health care delivery. The online participation option provided a number of benefits to the evaluation process. Specifically, the timeliness of the information was improved compared to the paper survey option. Further, complex skip patterns were embedded in the online survey and seamless to the end-user resulting in only All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint applicable questions being asked. Additionally, fewer resources were needed to collect these data as individuals carried out their own data entry. Demonstration programs needing information in a relatively rapid timeframe or wanting to adjust programs in real-time may wish to consider this option.

Providing low-income populations with internet access options can improve their access to appropriate health services as well as provide pertinent health information and education. [27] [28] [29] Additionally, the need to connect low-income individuals to internet technology is critical as it becomes an increasingly integral part of the fabric of our environment as demonstrated during the COVID-19 pandemic.

All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint Facebook, and physician websites to communicate with physicians: a national online survey of retail pharmacy All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint 

All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint 

0-99% 100-199% 200% over All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 23, 2020. ; https://doi.org/10.1101/2020.11.22.20236497 doi: medRxiv preprint

Mobile Technology and Home Broadband

From internet access to internet skills: digital inequality among older adults

Understanding the intersection of Medicaid and work: what does the data say? Issue Brief

The contemporary US digital divide: from access to technology maintenance

Gender, mobile, and mobile internet maintenance affordances, capabilities and structural inequalities: mobile phone use by low-income women

13 percent of Americans have no use for the internet. Praxis, Pew Research Center

Lifeline Program for Low-Income Consumers

Crowd-out or affordability? The lifeline expansion's effect on wireless service spending

Accumulating phones: aid and adaptation in phone access for the urban poor

The digital divide

Consumers are bored with virtual health options: can the health care system deliver? Deloitte Insights

Web surveys by smartphones and tablets: effects on data quality

53% of Americans say internet has been essential during COVID-19 outbreak

s digital divide "is going to kill people" as COVID-19 exposes inequalities. The Guardian

Use of an electronic patient portal among disadvantaged populations

Patient use of email, users

The value of online surveys

How much time do you spend online? Understanding and improving the accuracy of self-reported measures of internet use

As cities face COVID-19, the more digital divide becomes more acute

What the coronavirus reveals about the digital divide between schools and communities. The Brookings Institution

Health literacy and use and trust in health information

All rights reserved. No reuse allowed without permission

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

Has the digital health divide widened? trends of health-related internet use among older adults from 2003 to 2011

der Vaart R, Drossaert C. Development of the digital health literacy instrument: measuring a broad spectrum of health 1.0 and health 2.0 skills

Are patients electronically accessing their medical records? Evidence from National hospital data

New options for national population surveys: The implication of internet and smartphone coverage

All rights reserved. No reuse allowed without permission

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the

All rights reserved. No reuse allowed without permission

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the

Residence Class