Professional Documents
Culture Documents
T
he rapid development of new com- A hospital’s decision about whether to invest
munications technologies, coupled in telehealth is likely shaped by a range of fac-
with growing pressure to develop tors. These include specific features of the hos-
more-efficient health care delivery pital, such as its teaching mission, its affiliations
models, has focused renewed atten- with other institutions, and its technological ca-
tion on telehealth.1 Telehealth encompasses a pabilities. Market factors, including rurality and
wide array of applications—for example, remote degree of competitiveness, may also shape hos-
patient monitoring, electronic intensive care pitals’ decisions about investing in telehealth.
units (eICUs), and video visits—all of which For example, hospitals in more rural areas have
are designed to take advantage of communica- access to federal funds to improve care accessi-
tion technologies to support care from a bility, and as a result they may be more likely to
distance. provide telehealth than hospitals in more urban
The evidence is still emerging on whether or areas.9 Hospitals in more competitive markets
not telehealth consistently leads to better care at may see telehealth as a way to distinguish them-
lower cost.2–4 In addition, the degree to which selves. Finally, and perhaps most importantly,
telehealth technologies have been adopted and state policies affecting reimbursement and reg-
the factors that influence whether or not insti- ulation likely affect how interested hospitals are
tutions adopt them are not well understood.5–8 in offering telehealth services.
Gaining a better understanding of why some ulations. Hospitals in Guam, Puerto Rico, and
hospitals choose to adopt telehealth while others other US territories were excluded because they
do not is critically important. If certain types of are subject to different rules and regulations.
hospitals, such as those that are not part of a We merged the IT Supplement data with re-
larger system, find it particularly difficult to en- sults from the AHA annual survey to develop
gage in telehealth, concerted, targeted efforts measures at the market level and to capture ad-
may be useful once evidence for the benefits of ditional hospital characteristics. We defined a
telehealth is more robust. And if certain state market as a Hospital Referral Region,12 a desig-
policies facilitate or hinder telehealth uptake, nation developed for the Dartmouth Atlas of
policy makers could catalyze adoption through Health Care to identify health care delivery mar-
new legislation. Empirical evidence here would kets based on Medicare beneficiaries’ travel pat-
be very helpful. terns for tertiary hospital care. The Area Re-
In this study we sought to identify the hospi- source File, Medicare inpatient claims, the
tal-, market-, and state-level factors associated Health Resources and Services Administration’s
with telehealth adoption among US hospitals. Health Professional Shortage Area files, and the
We specifically sought to answer the following Dartmouth Atlas were all used for additional mar-
questions: What percentage of US hospitals cur- ket-level measures.
rently have telehealth capability? Which charac- Telehealth Adoption The IT Supplement
teristics of hospitals and markets are associated asks hospitals to report the extent to which they
with hospital telehealth adoption? Is there sub- have implemented a set of health information
stantial variation in adoption of telehealth and communication technology functions, one
across states? And are certain state policies as- of which is telehealth (the survey question can be
sociated with the greater likelihood of telehealth found in the online Appendix).13 We considered a
adoption among hospitals? hospital to have telehealth capability if tele-
To address the rapidly changing environment health was fully implemented in at least one unit.
of health care delivery, we complement our as- Hospital Characteristics We assessed
sessment of current telehealth adoption among whether hospital telehealth adoption was asso-
US hospitals with a discussion of the most prom- ciated with the presence of a cardiac ICU (indi-
ising and innovative applications of telehealth, cating a certain level of technological capability)
focusing on those that target the priorities set and IT resources (at least a basic electronic
out in the Affordable Care Act, such as reductions health record),14 because telehealth can comple-
in thirty-day readmissions to hospitals. These ment these technologies and because it may re-
results provide a timely picture of the factors that flect a broader strategy of investment in ad-
appear to be shaping hospitals’ decisions to in- vanced technology. We also assessed whether
vest in telehealth, and they suggest potential hospitals that were affiliated with a larger system
strategies to help policy makers increase tele- would be more likely than others to have imple-
health adoption. mented telehealth, because the technology of-
fers a way to deliver a full range of services to
patients from any location, increasing access as
Study Data And Methods well as revenue.
We used national data from the Information In addition to these three focal hospital char-
Technology (IT) Supplement to the American acteristics, we examined size, teaching status,
Hospital Association (AHA) 2012 Annual Survey ownership (for-profit, nonprofit private, or pub-
of Hospitals,10 which was sent at the end of 2012 lic), and proportion of Medicaid admissions, be-
to all acute care hospitals in the United States cause these might also be directly related to tele-
and its territories. The IT Supplement was de- health adoption or confound the relationship
signed to capture key functions for which hospi- between the focal characteristics and telehealth
tals used health information and communica- adoption.
tions technologies. The survey was sent to Market Characteristics At the market level
each hospital’s CEO, who was asked to assign we focused on the following four dimensions of
the most knowledgeable person in the institu- the environment that might promote hospital
tion (typically the chief information officer or an investment in telehealth: rurality, population
equivalent person) to complete it. The survey density, designation as a Health Professional
achieved a 63 percent response rate.11 Shortage Area, and the competitiveness of the
Our analytic sample was limited to the 2,891 hospital market. Rurality was measured using
acute care, nonfederal hospitals in the fifty states rural-urban commuting area categories (urban,
and the District of Columbia that responded to suburban, large rural, or small town). Popula-
the IT Supplement. Federal hospitals were ex- tion density was measured based on population
cluded because they are not subject to state reg- per square mile. We considered a market to be a
42
had policies in place specifying reimbursement to have adopted telemedicine if it reported one or
to the telehealth provider for transmission, a more telemedicine applications to be “live and
facility fee, or both. The fifth measure captured operational” in at least one department.13 Ninety- %
policies that restricted reimbursable services by seven percent of the hospitals in our analytic
Of hospitals
location, either to rural or underserved areas or sample were represented in the HIMSS Analytics
We found that 42 percent
by specific types of facilities, primarily hospitals data. of hospitals across the
and physician practices. The sixth measure cap- Limitations Our work has important limita- country had adopted
tured state policies that require private payers to tions. First, the IT Supplement survey asked telehealth as of late 2012.
reimburse telehealth services to the same extent about telehealth only as a single category. Be-
as face-to-face services. Some policies specify cause the survey did not ask about the extent
that private payers must reimburse for telehealth of use or about different types of telehealth (such
in certain situations, such as for rural popula- as remote patient monitoring), our ability to
tions or mental health services. assess specific telehealth applications was limit-
In addition to reimbursement-related policies, ed. In addition, some respondents may have in-
we examined two other types of policies—patient terpreted the question to include a wide array of
applications, while others may have interpreted hospitals with greater technological capabili-
it more narrowly. This prompted us to conduct ties—as measured by the presence of a cardiac
our robustness assessment using an indepen- ICU—were more likely than others to have
dent source of data on telehealth adoption (dis- adopted telehealth. Hospitals that were part of
cussed above). a larger system were also more likely to have
Second, we attempted to include a comprehen- adopted telehealth, as were teaching hospitals.
sive set of variables, but we were unable to mea- Finally, for-profit hospitals were much less likely
sure some that could have affected our key re- than their nonprofit counterparts to have
sults. For example, we were unable to capture adopted telehealth.
data on the financial position of hospitals. More Market-level factors also appeared to shape
broadly, there are many factors that likely affect a hospitals’ decisions about whether or not to
hospital’s decision to invest in telehealth, and we adopt telehealth. Rurality was significantly asso-
were not able to examine all of them. ciated with telehealth adoption overall—an asso-
Finally, we were not able to assess causality ciation that was driven by the greater likelihood
and could assess only associations. However, of adoption among hospitals in large rural areas
many other studies have qualitatively described compared to those in urban areas (Exhibit 2).
how policy barriers limit the adoption of tele- Population density also appeared to be indepen-
health and similar technologies.18–20 dently associated with the odds of telehealth
adoption, with the odds decreasing as density
increased. However, that association was only
Study Results marginally significant.
We found that 42 percent (1,208 of the 2,891 We did not find a significant difference in tele-
hospitals in our sample) of hospitals across health adoption based on whether the hospital
the country had adopted telehealth as of late was in a market that included a designated pri-
2012. When we examined telehealth adoption mary care Health Professional Shortage Area
based on key hospital and market characteris- (Exhibit 2). Finally, the competitiveness of the
tics, we found several differences. market appeared to shape telehealth adoption.
We failed to find a significant association be- Hospitals in the least competitive markets (those
tween telehealth and the adoption of an electron- in the top quartile of the Herfindahl-Hirschman
ic health record system (Exhibit 1). However, Index) were slightly more than half as likely to
Exhibit 1
SOURCE Authors’ analysis. NOTES CI is confidence interval. ICU is intensive care unit. EHR is electronic health record. Ref is reference
category. aNot applicable.
SOURCE Authors’ analysis. NOTES CI is confidence interval. HPSA is Health Professional Shortage Area. Ref is reference category. aNot
applicable. bBased on the Herfindahl-Hirschman Index. cThe proportion of discharged patients readmitted to a nonaffiliated hospital in
the same market.
have adopted telehealth as hospitals in the most ment survey. These facts suggest that factors be-
competitive markets. yond the number of hospitals in the state, such as
When we examined the rates of hospital tele- state policies, affect the proportion of hospitals
health adoption across states, we found substan- that offer telehealth.
tial variation (Exhibit 3). The states with the In fact, we found significant associations be-
broadest adoption were Alaska, where 75 percent tween state policies and hospitals’ adoption of
of the hospitals had telehealth; Arkansas (71 per- telehealth. Specifically, state policies that re-
cent); South Dakota (70 percent); and Maine quired private payers to reimburse for telehealth
(69 percent). In contrast, Rhode Island had 0 per- services to the same extent as face-to-face ser-
cent of its hospitals participating in telehealth, vices made hospitals more likely to adopt tele-
and Utah had only 13 percent. health (Exhibit 4). In contrast, state policies that
The highest- and lowest-adopting states all had required out-of-state providers to have special
fairly small populations, and each had fewer licenses when delivering telehealth services to
than fifty hospitals responding to the IT Supple- Medicaid patients made hospitals less likely to
Exhibit 3
AK
AR Without telehealth
SD With telehealth
ME
VA
NH
HI
NE
SC
NJ
WA
CO
OR
ID
MA
GA
MT
WI
NM
IA
DE
MN
MI
PA
NC
MO
AZ
IN
KY
OH
MD
ND
CA
FL
NV
LA
OK
WY
CT
KS
WV
TN
VT
DC
NY
TX
IL
MS
AL
UT
RI
Number of hospitals
SOURCE Authors’ analysis. NOTES Limited to hospitals that responded to the Information Technology Supplement to the 2012 Ameri-
can Hospital Association Annual Survey of Hospitals (see Note 10 in text). States are arrayed in order of highest proportion (Alaska) to
lowest proportion (Rhode Island) of telehealth adoption.
The Dorenfest Institute for Health and equity holder in Healthrageous and
Information of the HiMSS Foundation an advisory board member at Qualcomm
(see Note 17 in text) provided data for Life.
this study. Joseph Kvedar is a consultant