Trends in e-visit adoption among U.S. office-based physicians: Evidence from the 2011–2015 NAMCS

International Journal of Medical Informatics 129 (2019) 260–266

Contents lists available at ScienceDirect

International Journal of Medical Informatics journal homepage: www.elsevier.com/locate/ijmedinf

Trends in e-visit adoption among U.S. office-based physicians: Evidence from the 2011–2015 NAMCS

T

⁎

Young-Rock Honga, , Kea Turnera, Sandhya Yadava, Jinhai Huoa, Arch G. Mainous IIIa,b a b

Department of Health Services Research, Management and Policy, College of Public Health and Health Professions, University of Florida, Gainesville, FL, United States Department of Community Health and Family Medicine, University of Florida, Gainesville, FL, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Electronic office visit E-Visit Health information exchange Secure messaging Patient-physician communication

Background: Electronic visits (e-visits) have the potential to expand patients’ access to care and reduce healthcare costs. We aimed to describe trends in e-visit adoption among the U.S. office-based physicians and examine physician-and practice-level factors associated with e-visit adoption. Methods: This was a retrospective observational study of 2011–2015 National Ambulatory Medical Care Survey. We used the Cochran-Armitage tests to evaluate trend changes in e-visit adoption among the U.S. office-based physicians. Multivariable logistic regression was used to calculate the odds of adopting e-visits adjusting for physician and practice characteristics. Results: Our sample included 10,767 respondents, representing 327,836 office-based physicians in the U.S. Our analysis indicated that, in 2015, 15.9% of physicians adopted e-visits, which is a minor increase of 2.2% in total utilization of 13.7% in 2011. The likelihood of adopting e-visits was 2.7 times higher for physicians who have fully implemented electronic health records systems compared (odds ratio, 2.66, [95% CI, 2.16–3.28]) to physicians who have not implemented EHRs. Other predictors of e-visit adoption included primary care rather than specialty care, capitated payment model, and having a secure messaging capability. Conclusions: Our study demonstrates that overall e-visit adoption is low and has not been implemented as rapidly as other health information technologies. While use of secure information technology could be a facilitator for e-visit implementation, there are other barriers affecting widespread adoption. E-visits are a promising strategy for increasing patients’ access to care. Future research is needed to explore implementation barriers that might be impeding e-visit adoption.

1. Introduction Electronic visits (e-visits) have the potential to expand patients’ access to care and reduce cost while maintaining the same quality of care provided through in-person visits. E-visits, for example, can eliminate the need for transportation to an in-person visit and allow care to be accessed outside of a healthcare systems’ normal hours of operation or on weekends [1]. Several studies have reported that evisits are significantly cheaper than in-person visits (e.g., patient out-ofpocket costs and total visit costs) across a number of healthcare conditions [1,2]. E-visits have also demonstrated a similar level of quality compared to in-person visits, such as number of follow-up visits (e.g., an indicator for misdiagnosis or ineffective treatment) and patient satisfaction [3,4]. Despite notable benefits, studies suggest that e-visit adoption has been low [4,5].

E-visits can be defined as asynchronous electronic patient-provider communication including online consultations, exchanging health information, completing administrative forms, and managing medication through electronic health record (EHR) system or secured web portals [1–12]. E-visits typically include structured collection of patient information, diagnosis of a condition, and the creation of a treatment plan by a physician or other healthcare provider [6,7]. Studies that have explored e-visits have reported low usage. For example, healthcare systems that have implemented e-visits have found that less than ten percent of visits were categorized as e-visits [3,8]. Researchers have reported numerous barriers to implementation that may account for low usage, such as reimbursement, concerns about privacy and security, and lack of adequate technology [9]. In recent years, however, EHR and patient portal adoption has increased, which may expand physicians’ access to secure web-messaging tools and thus reduce some barriers to

⁎ Corresponding author at: Department of Health Services Research, Management and Policy, College of Public Health and Health Professions, 1225 Center Drive, HPNP 3118, University of Florida, Gainesville, FL, 32611, United States. E-mail address: [email protected]fl.edu (Y.-R. Hong).

https://doi.org/10.1016/j.ijmedinf.2019.06.025 Received 18 March 2019; Received in revised form 5 June 2019; Accepted 24 June 2019 1386-5056/ © 2019 Elsevier B.V. All rights reserved.

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

physicians response options included "Yes, all electronic" (defined as fully use), "Yes, part paper and part electronic" (as partly use), and "No". For the health information exchange (HIE) capability, response options were "Yes, used routinely", "Yes, but turned off or not used", and "No". The respondents who refused to answer or chose ‘Don't know’ were classified as “unknown.”

e-visit usage [10–12]. Given the potential for e-visits to expand access to care and reduce costs [1–5], it is important to examine temporal trends in physicians’ evisit adoption. To date, studies examining e-visit adoption have been conducted within individual health systems and not on a national scale [1–6,8–10]. Moreover, relatively little is known about what physicianor practice-factors are associated with e-visit adoption. By understanding the provider- and practice-level factors associated with evisits, future interventions and policies can be developed to increase adoption among non-users [13]. Our study has three objectives: 1) to describe trends in e-visit adoption from 2011 to 2015 among a nationally representative sample of U.S. office-based physicians; 2) to examine whether physicians’ adoption of other health information technologies, such as EHRs, is associated with e-visit adoption; and 3) to examine other practice- and physician-level factors associated with e-visit adoption (e.g., specialty, payer mix, reimbursement model).

2.2.3. Physician and practice characteristics The survey collected information on physician and practice characteristics that include practice specialty (primary care and other medical specialty); type of office setting (solo practice, physician group practice, other setting [academic centers, other hospitals, community health centers, and other]); ownership of practice (physician/physician group, academic centers/other hospitals, and insurance company/ health plans/other); physician employment status (owner of practice and employee/contractor); percentage of the patient care revenue received from Medicare, Medicaid, and capitation payments (less than 25%, 25%–50%, 51%–75%, more than 75%); census region (Northeast, Midwest, South, West); and Metropolitan Statistical Area (MSA and non-MSA).

2. Methods 2.1. Data and analytic sample

2.3. Statistical analyses We performed a retrospective observational study of the 2011–2015 National Ambulatory Medical Care Survey (NAMCS). The NAMCS is an annual survey of nationally representative office-based physicians providing direct patient care, which is conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention with an average response rate of 43.1% [14,15]. A sample of physicians taken from the master files of the American Medical Association (AMA) are randomly assigned to one of the 52 reporting weeks in the survey year. In this period, data on patient characteristics and medical care services a physician provided are recorded by trained interviewers. Physician and practice data are also collected through inperson interviews and mailed or online survey on electronic health record (EHR) systems [14,15]. Our initial sample consisted of 11,170 office-based physicians. We excluded those who refused to respond or those with missing values on key study variables (provision of an online consultation with patients, n = 316; adoption and use of EHR system, n = 87), resulting in a total analytic sample of 10,767 office-based physicians (weighted sample size: 327,836 physicians in the U.S). Use of this publicly available dataset does not constitute human subjects research, and therefore this study was deemed exempt from review by our Institutional Review Board (IRB201802666).

We first assessed trends in the adoption of e-visit and the prevalence of having HIT capabilities (EHR system and the computerized capabilities for HIE) from 2011 to 2015. We then examined temporal trends in the type of patient office visits between physicians who adopted e-visit services and who did not. For these trend analyses, we used linear regression models to estimate rates of each indicator and the Cochran-Armitage tests to evaluate the significance of the trend changes over the study period. To examine the association of the e-visit adoption with the use of HIT and HIE capabilities, we used chi-square tests for univariate association and multivariable logistic regressions to calculate odds of e-visit adoption adjusting for physician and practice characteristics including physician specialty, type of office settings, practice ownership, physician employment status, source of patient care revenue, census region, MSA, and survey year. For all analyses, we used PROC SURVEY procedures in SAS 9.4 (SAS Institute) to account for the stratified, clustered sample design used by NAMCS and used recommended physician-level weights to account for nonresponse bias and to produce national estimates of office-based physicians in the U.S. 3. Results 3.1. Trends in E-visit adoption and computerized capabilities for HIE

2.2. Measures Among 10,767 office-based physicians, 15.9% reported having adopted e-visit services for their patients in 2015, increased slightly from 13.7% in 2011 (Fig. 1). However, this change in trends did not reach statistical significance (Ptrend = .055). By 2015, 84% of officebased physicians adopted and used any type of EHR system, increased by 39.1% since 2011 (Ptrend < .001). Having the capabilities for HIE in office-based practice also increased and reached 54.4%–66.8% in 2015, more than doubling increase from 2011 (Ptrend < .001).

2.2.1. Adoption of E-visit services Given the pre-defined the concept of e-visit [1–12], we determined physicians who adopted e-visit services if they ever engaged in onlinebased consultations with patients in the reporting period. We used a question asking “During last normal week of practice, did you have any Internet/email consults with patients?” to measure the e-visit adoption. 2.2.2. Health information technology (HIT) capabilities The use of any EHR system was determined by a "Yes" response to a question asking "Does your practice use electronic medical or health records?—does not include billing record systems." To assess the capability of health information exchange, we used 2 questions asking if a practice has computerized capability for: (1) exchanging secure messages with patients; and (2) providing patients with clinical summaries for their visit. These questions on computerized capabilities were asked regardless of whether physicians actually used EHR system in practice [15]. Considering the core criteria for the Meaningful Use of Health Technology [9,16,17], we focused on whether these computerized capabilities were used functionally; for the use of EHR system, recorded

3.2. Association between E-visit adoption and computerized capabilities for HIE We compared the adoption of e-visit services by practice and physicians’ capabilities for health information exchange (Table 1). The prevalence of the e-visit adoption was 18.8% (95% confidence interval [CI], 17.2%–20.3%) and 13.3% (95% CI, 10.0%–16.6%) among officebased physicians with EHR in fully use and in partly use, respectively. When compared to those without EHR use, likelihood of e-visit adoption was higher for those with EHR in fully use (Odds ratio [OR], 2.66, [95% CI, 2.16–3.28]) and EHR in partly use (OR, 1.77, [95% CI, 261

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

Fig. 1. Trends in Adoption of E-visit Services and Capabilities of Health Information Technology among Office-Based Physicians in the U.S.2011–2015. Note. Percentages are weighted to represent national estimates of office-based physician characteristics and their practices. EHR adoption and use includes both partial and full electronic health records systems.

Table 1 Association of Health Technology Information Use with the Adoption of E-Visit among U.S. Office-Based Physicians. E-Visit Adoption

Sample N (National Estimate)

Total No. 10767 (327836)

Yes No. 1480 (49946)

Health Information Technology EHR System Use No EHR system

2767

225

Partly use; part paper and part electronic

1248

137

Fully use; all electronic

6752

1118

3454

308

Yes

6749

1101

Yes; but do not use (turned off)

430

45

134

26

5442

413

Yes

4145

962

Yes; but do not use (turned off)

889

70

291

35

Having Capabilities for: Providing Electronic Summary of Visit No

Unknown

d

Exchanging Secure Messages No

Unknown

d

Odds of Adopting E-Visit

Weighted %, (95% CI) 15.9 (12.5–19.3)

a

No No. 9287 (277890)

Weighted %, (95% CI) 84.1 (80.7–87.5)

a

P

b

OR (95% CI)

AOR c (95% CI)

1.00

1.00

1.77** (1.27–2.47) 2.66*** (2.16–3.28)

1.64* (1.06–2.56) 1.61** (1.12–2.32)

1.00

1.00

2.27*** (1.88–2.75) 1.13 (0.68–1.88) 3.41*** (1.76–6.60)

0.86 (0.60–1.23) 0.76 (0.38–1.50) 2.24 (0.87–5.75)

1.00

1.00

4.17*** (3.49–4.99) 1.37 (0.90–2.08) 1.61 (0.95–2.74)

3.47*** (2.78–4.32) 1.24 (0.81–1.90) 1.42 (0.81–2.47)

< .001 8.0 (6.7–9.3) 13.3 (10.0–16.6) 18.8 (17.2–20.3)

2542 1111 5634

92.0 (90.7–93.3) 86.7 (83.4–90.0) 81.2 (79.7–82.8) < .001

9.2 (7.8–10.5) 18.7 (17.1–20.2) 10.3 (5.9–14.6) 25.6 (13.5–37.7)

3146 5648 385 108

90.8 (89.5–92.2) 81.3 (79.8–82.9) 89.7 (85.4–94.1) 74.4 (62.3–86.5) < .001

7.8 (6.8–8.8) 26.1 (23.9–28.3) 10.4 (6.7–14.1) 12.0 (6.7–17.3)

5029 3183 819 256

92.2 (91.2–93.2) 73.9 (71.7–76.1) 89.6 (85.9–93.3) 88.0 (82.7–93.3)

Abbreviations: EHR, electronic health record. *P < .05 **P < .01 ***P < .001. a Percentages are weighted to represent national estimates of office-based physician characteristics and their practices. b Statistically significant differences between groups were detected by Chi-square tests. c Adjusted for all other physician and practice characteristics. d Physicians refused to answer, answered “Don’t know”, or had missing values were treated as “Unknown.”. 262

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

Table 2 Association of Physician/Practice Characteristics with the Adoption of E-Visit Services among U.S. Office-Based Physicians. E-Visit Adoption

Odds of Adopting E-Visit

Yes Physician and Practice Characteristics

Total No.

No.

Sample N (National Estimate)

10767 (327836)

1480 (49946)

Physician Specialty Primary Care

4186

701

6581

779

3367

405

6212

754

1188

321

8061

967

Medical/Academic Center or Other Hospital

950

162

Insurance Company, HMO, Other Health Care Corporation, or Other Unknown d

1378

304

378

47

6771

775

3996

705

4448

700

26%-50%

3319

377

51%-75%

1354

125

232

17

1414

261

7637

1052

26%-50%

1162

127

51%-75%

400

33

153

8

1415

260

7311

917

26%-50%

360

63

51%-75%

165

31

> 75%

253

67

2678

402

1783

187

Specialty Care Type of Office Setting Solo Practice Physician Group Practice Other Setting

e

Ownership of Practice Physician or Physician Group

Physician Employment Status Owner of Practice Employee or Contractor Source of Patient Care Revenue Percent of Revenue from Medicare ≤25%

> 75% Unknown

d

Percent of Revenue from Medicaid ≤25%

> 75% Unknown

d

Percent of Revenue from Capitation ≤25%

Unknown

d

Census Region Northeast

No Weighted %, (95% CI) a 15.9 (12.5–19.3)

No. 9287 (277890)

Weighted %, (95% CI)a 84.1 (80.7–87.5)

P

b

OR (95% CI)

AOR c (95% CI)

1.43*** (1.21–1.69) 1.00

1.38*** (1.16–1.64) 1.00

1.00

1.00

0.94 (0.77–1.15) 3.32*** (2.57–4.29)

0.62*** (0.50–0.77) 1.28 (0.88–1.85)

1.00

1.00

1.39* (1.03–1.87) 2.24*** (1.78–2.80) 0.83 (0.54–1.28)

1.39 (0.87–2.23) 1.15 (0.77–1.68) 0.71 (0.43–1.16)

1.00

1.00

1.65*** (1.39–1.95)

1.15 (0.89–1.49)

1.00

1.00

0.70** (0.57–0.86) 0.67* (0.48–0.93) 0.46** (0.26–0.81) 1.31* (1.02–1.69)

0.69* (0.55–0.86) 0.62* (0.43–0.89) 0.47* (0.24–0.91) 8.97** (1.90–42.5)

1.00

1.00

0.72* (0.54–0.98) 0.39*** (0.23–0.64) 0.18*** (0.07–0.42) 1.42** (1.11–1.80)

0.62** (0.43–0.89) 0.32*** (0.17–0.58) 0.13*** (0.05–0.31) 0.12** (0.02–0.55)

1.00

1.00

1.34 (0.90–2.02) 1.82 (0.98–3.38) 2.83*** (1.76–4.53) 1.16 (0.96–1.41)

1.36 (0.86–2.15) 1.41 (0.76–2.62) 1.77* (1.15–2.73) 0.79** (0.64–0.99)

1.00

1.00

< .001 17.6 (16.0–19.3) 13 (11.6–14.4)

3485 5802

82.4 (80.7–84) 87.0 (85.6–88.4) < .001

13.3 (11.5–15.2) 12.6 (11.4–13.9) 33.8 (29.3–38.2)

2962 5458 867

86.7 (84.8–88.5) 87.4 (86.1–88.6) 66.2 (61.8–70.7) < .001

13.4 (12.2–14.6) 17.6 (13.7–21.6) 25.6 (21.7–29.5) 11.3 (7.1–15.5)

7094 788 1074 331

86.6 (85.4–87.8) 82.4 (78.4–86.3) 74.4 (70.5–78.3) 88.7 (84.5–92.9) < .001

12.8 (11.5–14.1) 19.4 (17.4–21.4)

5996 3291

87.2 (85.9–88.5) 80.6 (78.6–82.6) < .001

16.8 (15.1–18.5) 12.4 (10.6–14.2) 11.9 (8.7–15.0) 8.4 (4.0–12.8) 20.9 (17.3–24.6)

3748 2942 1229 215 1153

83.2 (81.5–84.9) 87.6 (85.8–89.4) 88.1 (85.0–91.3) 91.6 (87.2–96.0) 79.1 (75.4–82.7) < .001

15.6 (14.3–16.9) 11.8 (8.8–14.7) 6.6 (3.5–9.8) 3.2 (0.5–5.8) 20.7 (17.1–24.3)

6585 1035 367 145 1155

84.4 (83.1–85.7) 88.2 (85.3–91.2) 93.4 (90.2–96.5) 96.8 (94.2–99.5) 79.3 (75.7–82.9) .006

13.9 (12.7–15.2) 17.8 (12.1–23.6) 22.7 (12.0–33.4) 31.4 (21.5–41.2) 15.8 (13.6–18.0)

6394 297 134 186 2276

86.1 (84.8–87.3) 82.2 (76.4–87.9) 77.3 (66.6–88.0) 68.6 (58.8–78.5) 84.2 (82.0–86.4) < .001

10.5 (8.8–12.3)

1596

89.5 (87.7–91.2)

(continued on next page)

263

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

Table 2 (continued) E-Visit Adoption

Odds of Adopting E-Visit

Yes Physician and Practice Characteristics

Total No.

No.

Sample N (National Estimate)

10767 (327836)

1480 (49946)

Midwest

2682

297

South

3832

451

West

2470

545

9631

1368

Non-MSA

1136

112

Survey Year 2011

1251

148

2012

3479

420

2013

2594

380

2014

2103

353

2015

1340

179

Metropolitan Statistical Area (MSA) MSA

No Weighted %, (95% CI) a 15.9 (12.5–19.3) 12.1 (10.1–14.2) 11.9 (10.3–13.4) 26.3 (23.3–29.4)

No. 9287 (277890) 2385 3381 1925

Weighted %, (95% CI)a 84.1 (80.7–87.5)

P

b

87.9 (85.8–89.9) 88.1 (86.6–89.7) 73.7 (70.6–76.7)

OR (95% CI)

AOR c (95% CI)

1.17 (0.89–1.53) 1.14 (0.90–1.45) 3.03*** (2.37–3.88)

0.97 (0.74–1.28) 1.04 (0.80–1.33) 2.01*** (1.49–2.72)

1.00

1.00

0.73 (0.51–1.06)

0.78 (0.54–1.13)

1.00

1.00

0.96 (0.73–1.26) 1.17 (0.90–1.52) 1.35* (1.04–1.76) 1.19 (0.85–1.68)

0.98 (0.74–1.30) 1.01 (0.75-1.34) 1.05 (0.78–1.40) 0.84 (0.59–1.19)

.066 15.5 (14.4–16.7) 11.8 (8.1–15.6)

8263 1024

84.5 (83.3–85.6) 88.2 (84.4–91.9) .017

13.7 (11.1–16.4) 13.2 (11.5–15.0) 15.6 (13.8–17.5) 17.7 (15.6–19.7) 15.9 (12.5–19.4)

1103 3059 2214 1750 1161

86.3 (83.6–88.9) 86.8 (85.0–88.5) 84.4 (82.5–86.2) 82.3 (80.3–84.4) 84.1 (80.6–87.5)

*P < .05 **P < .05 ***P < .001. a Percentages are weighted to represent national estimates of office-based physician characteristics and their practices. b Statistically significant differences between groups were detected by Chi-square tests. c Adjusted for all other physician and practice characteristics. d Physicians refused to answer, answered “Don’t know”, or had missing values were treated as “Unknown.”. e Includes academic centers, other hospitals, community health centers, and other types of health providers.

Medicare or Medicaid appeared to have a lower likelihood of e-visit adoption. Compared to those with less than or equal to 25% of revenue from Medicare, those with more than 25% or more had 31%–53% decrease in the odds of adopting e-visit services. Similarly, having a greater portion of revenue from Medicaid was associated with 38%–87% lower odds of e-visit adoption. Conversely, those with the percent of revenue from capitation > 75% had higher odds of adopting e-visit services (AOR 1.77, 95% CI 1.15–2.73) than those less than or equal to 25% of revenue from capitation payments. Lastly, by geographic region, physicians and practices in the West region had a higher likelihood of adopting e-visit services (AOR 2.01, 95% CI 1.49–2.72) than those in other census regions. We found no significant difference between MSA and non-MSA.

1.27–2.47]). In fully adjusted models, albeit attenuated, their associations remained consistent and having EHR system was a strong predictor of the adoption of e-visit services (both P-values < 0.05). For the association between capabilities for HIE and e-visit adoption, we found that 18.7% (95% CI, 17.1%–20.2%) of those having computerized capabilities for providing electronic summary of visit for patients adopted e-visit services and their likelihood of e-visit adoption was 2.3 times higher (OR 2.27, [95% CI, 1.88–2.75]) than those without the capabilities. However, when adjusting for other practice and physician characteristics, this association became no longer significant (P = .388). Notably, the prevalence of the e-visit adoption was threefold higher among those having capabilities for secure messaging with patients (26.1%, [95% CI, 23.9%–28.3%]) than those without the capabilities (7.8%, [95% CI, 6.8%–8.8%]). Even after adjusting for other characteristics, the likelihood of e-visit adoption was more than three times higher (adjusted odds ratio [AOR], 3.47, [95% CI, 2.78–4.32]) among those having secure messaging capabilities than those without.

4. Discussion We conducted this study to understand the trends in e-visit adoption and examine the role of HIT capabilities and other practice-and physician-level factors on e-visit adoption among US office-based physicians. We found a minor increase in e-visit adoption from 2011 to 2015. By 2015, only 16% of physicians reported that they provided e-visits, which is an increase of 2% since 2011. To our knowledge, this is the first study to produce a nationally representative estimate of e-visit adoption among physicians. While physician adoption of other HIT grew substantially during the study period—40% increase in using EHR system and 135% increase in having capabilities for secure messaging with patients—, e-visit adoption barely changed from 2011 to 2015. Our estimate of e-visit adoption among physicians was similar to the utilization of telemedicine for interactions between patients and physicians reported by the American Medical Association’s Physician

3.3. Practice-physician characteristics associated with the E-visit adoption Table 2 shows the associations of practice and physician characteristics with the adoption of e-visit services. Compared to specialty care, primary care physicians had a greater likelihood of e-visit adoption (AOR 1.38, [95% CI, 1.16–1.64]). By practice ownership, practices owned by insurance company/health management organization/other health care corporations had a higher likelihood of e-visit adoption (OR 2.24, [95% CI, 1.78–2.80]); however, this association was no longer significant after adjusting for other characteristics. Concerning source of patient care revenue, having a greater portion of revenue from 264

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

implementation of e-visits [29]. Such a program could help physicians to overcome implementation barriers that might be preventing broader usage of e-visits. Our findings highlight that insurers and policymakers who seek to promote such interactions between patients and providers further should develop standardized guidelines for secure e-visit services and incentivize for better use of the established health information infrastructure [26,30]. This study has some limitations. First, information on EHR and HIT use was self-reported in NAMCS. While we tried to address nonresponse bias [15], it is still possible that practices and physicians who already adopted were more likely to respond the survey questions, leading to underestimates of non-adopters. Second, unmeasured practice-level or patient-level factors (e.g., patient mix) may be associated with HIT and e-visits utilization. For example, studies have found differences in adoption of e-visits based on patient condition (e.g., conjunctivitis, urinary tract infection) and patient-level characteristics (e.g., gender, race/ethnicity, age) [2,6,8,19,22]. Future studies could use EHR data to explore how provider e-visit adoption varies based on an aggregate score of their patients’ characteristics and conditions. Third, the adoption of e-visit services was measured based on physicians’ one normal week of practice, not a full year. However, we used the recommended NAMCS weighting adjustments to account for variations in the typical number of weeks physicians practiced during a year to generate representative annual estimates [15]. Fourth, NAMCS did not provide estimates of the frequency of e-visit interactions; thus, we were not able to measure the intensity of e-visits adoption. Lastly, although we used a nationally representative sample of U.S. office-based physicians, our results may not be generalizable to physicians who were not officebased or having other specialties that were not included in the AMA database. Despite these limitations, to our knowledge, this is the first study to broadly examine practice and physician level factors associated with the adoption of e-visit services and provides the first, national level e-visit adoption rate in the U.S.

Practice Survey [18]. In 2016, 15.4% of physicians in practice reported they used online-based modalities (including remote patient monitoring and data exchange) to interact with patients, and 11.2% used them to interact with other health care professionals [18]. The results of our study also suggest that physicians’ HIT infrastructure, such as full implementation of EHRs and secure messaging capability, care associated with e-visit adoption. In particular, secure messaging (adjusted odds ratio [AOR], 3.47, [95% CI, 2.78–4.32]) had the strongest impact on e-visit adoption. This finding suggests that access to secure messaging technology may be a facilitator for e-visit implementation but other implementation barriers are affecting widespread adoption. For example, studies have reported that having a structured questionnaire can improve diagnostic accuracy of e-visits [19]. Some physicians may not work in settings that have developed structured questionnaires for e-visits, making it difficult for the physician to implement e-visits. Additionally, studies suggest that e-visits may be easier to implement for conditions that have clear guidelines about when the condition can be treated without examination and under what circumstance an in-person examination is needed (e.g., urinary tract infections) [2]. Future studies should explore how implementation guidance and structured questionnaires affect physician adoption and satisfaction with e-visits. The findings of our study also suggest that payer mix and reimbursement model are important predictors of e-visit adoption. Although reimbursement of e-visits has improved over time, since the American Medical Association approved of online consultation billing and the development of a Current Procedural Terminology code for evisits, studies still report e-visit reimbursement as a barrier to adoption [5,20–22]. For example, one study found that a patients’ insurance coverage of e-visits was a significant predictor of patients’ use of e-visits [22]. Our study also found that capitated payment models were positively associated with e-visit adoption. As other flexible payment models are expanded, such as value-based payment, it is possible that evisit adoption will be easier to implement. For example, Medicare has traditionally defined telemedicine as two-way, real-time communication—not asynchronous communication—limiting reimbursement of e-visits [23]. However, in 2013, the Centers for Medicare and Medicaid Services started introducing payment for asynchronous forms of telemedicine through the Transitional Care Management and Chronic Care Management programs [24,25]. Future studies should explore how changes to payment models affect e-visit adoption. It is also important to better understand how state-level regulations may affect implementation. Currently, some state regulations limit telemedicine to face-to-face encounters or have not passed Telehealth parity laws [23,26]. Future studies could examine how e-visit adoption rates vary across states. E-visit adoption may also change in the next few years in response to newer policies that promote patient engagement with electronic health information. For example, in 2017 CMS proposed quality measures for the Medicare Access and CHIP Reauthorization Act (MACRA) that included patient engagement with digital health tools. The guidance specifically references digital tools that can help patients to manage their healthcare outside of in-person visits [25,27]. Stage 3 Meaningful Use requirements also focus on patient engagement and specifically engagement through patient portals (e.g., provide patientspecific health education resources for 35% of patients through the patient portal) [28]. It is possible that increased quality measurement and policy incentives that focus on patient engagement may increase physician adoption of e-visits over time. Finally, future research is needed to determine what constitutes meaningful use of e-visits and what implementation support is needed to effectively implement e-visits. For EHR implementation, regional extension centers provided significant implementation support to physicians adopting EHR systems. The Office of the National Coordinator for Health IT could develop a similar program to test whether providing implementation support to physicians improves uptake of and effective

5. Conclusions E-visits are a promising strategy for increasing patients’ access to care. Our study suggests that overall e-visit service provision is low and has not been adopted as rapidly as other HITs, such as EHRs. There are likely remaining implementation barriers that need to be addressed, such as lack of guidance for implementation and lack of quality measures. To support increasing policy initiatives and payment models that emphasize patient engagement with health information, future research is needed to identify and test effective implementation strategies that promote e-visit service adoption among office-based physicians. Author’s contribution

• Study conception and design: Hong, Turner, Huo, and Mainous. • Acquisition, analysis, or interpretation of data: All authors. • Drafting of manuscript: Hong, Turner, and Yadav. • Critical revision of the manuscript for important intellectual content: Turner, Huo, and Mainous. • Statistical analysis: Hong and Huo. • Administrative, technical, or material support: Hong and Yadav. • Study supervision: Mainous and Turner. Source of funding There was no external funder for this study.

Summary Points have the potential to expand patients’ access to care • E-visits and reduce healthcare costs. 265

International Journal of Medical Informatics 129 (2019) 260–266

Y.-R. Hong, et al.

found that overall e-visit adoption is low and has not been • We implemented as rapidly as other health information tech-

• •

[14]

nologies among U.S. physicians. There are likely remaining implementation barriers that need to be addressed. Insurers and policymakers who seek to promote e-visits should develop standardized guidelines for secure services and incentivize for better use of the established health information infrastructure.

[15]

[16]

[17]

References [18]

[1] J.E. Rohrer, K.B. Angstman, S.C. Adamson, M.E. Bernard, J.W. Bachman, M.E. Morgan, Impact of online primary care visits on standard costs: a pilot study, Popul. Health Manag. 13 (2) (2010) 59–63, https://doi.org/10.1089/pop.2009. 0018. [2] P.T. Courneya, K.J. Palattao, J.M. Gallagher, Health partners' online clinic for simple conditions delivers savings of $88 per episode and high patient approval, Health Aff. 32 (2) (2013) 385–392, https://doi.org/10.1377/hlthaff.2012.1157. [3] A. Mehrotra, S. Paone, G.D. Martich, S.M. Albert, G.J. Shevchik, A comparison of care at E-visits and physician office visits for sinusitis and urinary tract infection, JAMA Intern. Med. 173 (1) (2013) 72, https://doi.org/10.1001/2013. jamainternmed.305. [4] R. Hickson, J. Talbert, W.C. Thornbury, N.R. Perin, A.J. Goodin, Online medical care: the current state of “eVisits” in acute primary care delivery, Telemed e-Health 21 (2) (2015) 90–96, https://doi.org/10.1089/tmj.2014.0022. [5] T.F. Bishop, M.J. Press, J.L. Mendelsohn, L.P. Casalino, Electronic communication improves access, but barriers to its widespread adoption remain, Health Aff. 32 (8) (2013) 1361–1367, https://doi.org/10.1377/hlthaff.2012.1151. [6] K.S. Penza, M.A. Murray, J.L. Pecina, J.F. Myers, J.W. Furst, Electronic visits for minor acute illnesses: analysis of patient demographics, prescription rates, and follow-up care within an asynchronous text-based online visit, Telemed e-Health 24 (3) (2018) 210–215, https://doi.org/10.1089/tmj.2017.0091. [7] The American Academy of Family Physicians, Virtual e-Visits, (2017) Published, Accessed August 8, 2018 https://www.aafp.org/about/policies/all/virtual-visits. html. [8] A. Mehrotra, S. Paone, G.D. Martich, S.M. Albert, G.J. Shevchik, Characteristics of patients who seek care via eVisits instead of office visits, Telemed e-Health 19 (7) (2013) 515–519, https://doi.org/10.1089/tmj.2012.0221. [9] P. Whitten, L. Buis, B. Love, Physician-patient e-visit programs: implementation and appropriateness, Dis. Manag. Heal Outcomes 15 (4) (2007) 207–214, https://doi. org/10.2165/00115677-200715040-00002. [10] V. Patel, W. Barker, E. Siminerio, Disparities in individuals’ access and use of health information technology in 2014, ONC Data Br. (34) (2016) 1–12 https://www. healthit.gov/sites/default/files/briefs/disparities_in_consumer_access_and_use_ 2014_databrief_2_24_16.pdf. [11] K. Turner, Y.R. Hong, S. Yadav, J. Huo, A.G. Mainous, Patient portal utilization: before and after stage 2 electronic health record meaningful use, J. Am. Med. Inform. Assoc. (April) (2019), https://doi.org/10.1093/jamia/ocz030. [12] C.-J. Hsiao, A.K. Jha, J. King, V. Patel, M.F. Furukawa, F. Mostashari, Office-based physicians are responding to incentives and assistance by adopting and using electronic health records, Health Aff. 32 (8) (2013) 1470–1477, https://doi.org/10. 1377/hlthaff.2013.0323. [13] H. Daniel, L.S. Sulmasy, D.M. Delong, et al., Policy recommendations to guide the use of telemedicine in primary care settings: an American College of Physicians

[19]

[20]

[21]

[22]

[23] [24]

[25]

[26]

[27]

[28]

[29]

[30]

266

position paper, Ann. Intern. Med. 163 (10) (2015) 787–789, https://doi.org/10. 7326/M15-0498. Centers for Disease Control and Prevention, National Ambulatory Medical Care Survey, (2017) Published, Accessed October 13, 2018 https://www.cdc.gov/nchs/ ahcd/about_ahcd.htm. National Center for Health Statistics, NAMCS Micro-Data Documentation, (2015) Published 2018. Accessed November 5, 2018 ftp://ftp.cdc.gov/pub/Health_ Statistics/NCHS/Dataset_Documentation/NAMCS/doc2015.pdf. M.F. Furukawa, J. King, V. Patel, C.J. Hsiao, J. Adler-Milstein, A.K. Jha, Despite substantial progress in EHR adoption, health information exchange and patient engagement remain low in office settings, Health Aff. 33 (9) (2014) 1672–1679, https://doi.org/10.1377/hlthaff.2014.0445. B.A. Kash, J. Baek, E. Davis, T. Champagne-Langabeer, J.R. Langabeer, Review of successful hospital readmission reduction strategies and the role of health information exchange, Int. J. Med. Inform. 104 (February) (2017) 97–104, https:// doi.org/10.1016/j.ijmedinf.2017.05.012. C.K. Kane, K. Gillis, The use of telemedicine by physicians: still the exception rather than the rule, Health Aff. 37 (12) (2018) 1923–1930, https://doi.org/10.1377/ hlthaff.2018.05077. M. Player, E. O’Bryan, E. Sederstrom, J. Pinckney, V. Diaz, Electronic visits for common acute conditions: evaluation of a recently established program, Health Aff. 37 (12) (2018) 2024–2030, https://doi.org/10.1377/hlthaff.2018.05122. P.C. Tang, W. Black, C.Y. Young, Proposed criteria for reimbursing eVisits: content analysis of secure patient messages in a personal health record system, AMIA. Annu. Symp. Proc. AMIA Symp. 25 (4) (2006) 764–768 http://www.ncbi.nlm.nih.gov/ pubmed/1839266. Y.Y. Zhou, T. Garrido, H.L. Chin, A.M. Wiesenthal, L.L. Liang, Patient access to an electronic health record with secure messaging: impact on primary care utilization, Am. J. Manag. Care 13 (7) (2007) 418–424 http://www.ncbi.nlm.nih.gov/ pubmed/17620037. C. Jung, R. Padman, G. Shevchik, S. Paone, Who are portal users vs. early e-Visit adopters? A preliminary analysis, AMIA. Annu. Symp. Proc. AMIA Symp. 2011 (2011) 1070–1079 http://www.ncbi.nlm.nih.gov/pubmed/22195168. Y.T. Yang, Telehealth Parity Laws. Health Affairs Blog, (2019), https://doi.org/10. 1377/hpb20160815.244795. J. Schurrer, A. O’ malley, C. Wilson, N. Mccall, N. Jain, Evaluation of the Diffusion and Impact of the Chronic Care Management (CCM) Services: Final Report, Baltimore, MD (2017) https://innovation.cms.gov/Files/reports/chronic-caremngmt-finalevalrpt.pdf. D. Flannery, R. Jarrin, Building a regulatory and payment framework flexible enough to withstand technological progress, Health Aff. 37 (12) (2018) 2052–2059, https://doi.org/10.1377/hlthaff.2018.05151. Lin C-CC, A. Dievler, C. Robbins, A. Sripipatana, M. Quinn, S. Nair, Telehealth in health centers: key adoption factors, barriers, and opportunities, Health Aff. 37 (12) (2018) 1967–1974, https://doi.org/10.1377/hlthaff.2018.05125. Quality Payment Program. Explore Measures: How to Use This Tool, (2019) Published. Accessed March 9, 2019 https://qpp.cms.gov/mips/explore-measures/ improvement-activities?py=2018&measureSubcategoryName= BeneficiaryEngagement&measureWeighting=High. Centers for Medicare and Medicaid Services, EHR Incentive Program Eligibility, (2016) Published, Accessed January 11, 2019 https://www.cms.gov/regulationsand-guidance/legislation/ehrincentiveprograms/eligibility.html. M.M. Casey, I. Moscovice, J. McCullough, Rural primary care practices and meaningful use of electronic health records: the role of regional extension centers, J. Rural Health 30 (3) (2014) 244–251, https://doi.org/10.1111/jrh.12050. J. Ye, G. Rust, Y. Fry-Johnson, H. Strothers, E-mail in patient–provider communication: a systematic review, Patient Educ. Couns. 80 (2) (2010) 266–273, https:// doi.org/10.1016/j.pec.2009.09.038.