Professional Documents
Culture Documents
Annual Review 2012
Annual Review 2012
Annual Review 2012
ANNUAL
REVIEWS Further
Wireless Technology
Click here for quick links to
Annual Reviews content online,
including:
in Disease Management
• Other articles in this volume
• Top cited articles
and Medicine
• Top downloaded articles
• Our comprehensive search
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
479
ME63CH32-Clifford ARI 12 December 2011 14:21
issues such as compliance (4). There is prelim- Before discussing this new area of wireless
inary evidence that telemedicine solutions may monitoring, we first present an overview of the
be able to reduce the rate of emergency ad- types of data transmitted and discuss some of
BP: blood pressure
missions, but this has only been demonstrated the related issues concerning privacy, standards,
BR: breathing rate
in the United Kingdom in studies compris- and evaluation. In particular, we concentrate on
ing small numbers of patients. The best results ambulatory monitoring of patients in the home SpO2 : peripheral
oxygen saturation
have been reported by the Veterans Adminis- (or “biomonitoring”), telemedicine, and recent
tration (VA) in the United States, which showed trends in hospital implementations of wireless
a 20.7% decrease in health service utilization monitoring. We also discuss evidence (or lack
by a cohort of 1,963 patients; however, this was thereof ) concerning the success of such moni-
achieved at a relatively high cost of $1,600 per toring systems.
patient per year (7). The latest systematic re-
view of home telemedicine for COPD showed
that existing telemedicine solutions reduce the BIOMONITORING
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
rate of hospitalization, but patients with tele- A comprehensive review of physiological data
phone support had a higher mortality rate measured by biomonitoring is provided by
than those in the group that received standard Budinger (11), who describes the various meth-
care (8). ods with which each parameter has been ac-
There are around 60 telemedicine projects quired to date, and the wireless standards by
currently running across the United Kingdom’s which they are transmitted (and which are still
National Health Service in England alone. used in contemporary practice). The author
Some have demonstrated positive outcomes, concludes that blood pressure (BP), breathing
although most of these have been small-scale rate (BR), the ECG, and peripheral oxygen
proof-of-concept studies with typically fewer saturation (SpO2 ) are parameters that require
than 100 patients enrolled. There is also con- “more or less continuous monitoring” in the
siderable variation in the magnitude of the home, noting that engineering design contin-
outcomes reported (9). ues to be limited by the battery life of the equip-
The case for wireless technology seems far ment. Other parameters that are typically mon-
more pressing in resource-constrained loca- itored wirelessly include blood glucose, weight,
tions, such as the rural and peri-urban envi- physical activity, and sleep.
ronments of developing countries. Over the In the hospital environment, wireless
past decade there has been a “leapfrogging” of telemetry has been mostly confined to ECG;
wired communication systems in such places, (noninvasive) BP and SpO2 have also been mon-
where mobile phone technology provides low- itored in a similar fashion. The main reason
cost and affordable communication for almost for in-hospital telemetry is to grant the patient
all of society. Recent estimates indicate that up some freedom of movement (enabling a quicker
to 90% of the world’s people are within reach of recovery) and yet allow continuous monitor-
a mobile phone transmitter, and that the num- ing for life-threatening events, such as arrhyth-
ber of mobile phone users is approaching the mias. Typically, a radio-frequency system, such
number of literate people on earth (10). More- as the hospital’s Wi-Fi network, is used to trans-
over, the lack of trained healthcare specialists in mit the data. Recently, several companies have
developing countries, and the migration of such made telemetry systems available in the inten-
personnel to higher-paying economies, means sive care unit (ICU). These “eICU” systems
that there is a pressing need for mHealth, if vary in functionality but are based on transmis-
a financially affordable delivery model can be sion of the standard bedside vital signs (such as
implemented. In fact, mHealth is beginning to the ECG) and on integrating them with other
be adopted rapidly throughout both developed hospital information for display on a remote
and resource-poor economies. “dashboard,” allowing the clinician to interact
remotely with the bedside caregivers. A webcam only four that considered cost-effectiveness.
is therefore often also employed. The integra- Whitten et al. (18) concluded from a study of
tion of so much real-time data has facilitated the 551 articles on telemedicine that “only 38 con-
use of computational predictive alert systems, tained any type of real data. Because many of
which have shown some promise, particularly these 38 studies proved to be inadequately de-
in the lower-acuity setting (12). signed or conducted, we were unable to per-
Several systems for nonhospital wireless form a traditional meta-analysis.” Mair et al.
health monitoring have been implemented. (19) found that only 29 of 246 studies that men-
Alaoui et al. (13) described the monitoring of tioned cost as an outcome were suitably rig-
congestive heart failure, which affects five mil- orous for the purposes of meta-analysis. Agha
lion Americans. In congestive heart failure, the et al. (20) showed from a study of a one-year
heart becomes weak and loses its ability to pump telemedicine program in the United States that
blood throughout the body, resulting in the ac- telemedicine costs (involving teleconsultation,
cumulation of fluid in the lungs. Daily monitor- with patients actively involved in vital-sign ac-
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
ing of weight, BP, and SpO2 is required. The quisition) were 58% that of routine care (in-
authors describe a secure Internet-based system volving outpatient consultations at a treatment
for transferral of patient data from a home mon- center), and 29% that of onsite care in the pa-
itoring measurement device to a central station tient’s home by clinical staff.
used by clinical staff.
Donoghue et al. (14) described the archi-
tecture of a telemedicine system that noninva- mHEALTH
sively measures BP every 30 min, using mo- Although Internet-based systems are increas-
bile phones as a medium between Bluetooth ingly common, a significant growth area is the
sensors and the central station. Roine et al. monitoring of health using a mobile phone.
(15) surveyed the literature for publications on Waves of low-cost or freely downloadable mo-
“telemedicine” between 1990 and 2000. Fifty bile phone applications (“apps”) are beginning
studies were deemed suitable for consideration, to appear, which either process the data on
of which 34 assessed clinical outcomes and 16 the mobile phone itself or process it remotely
assessed economic benefit. The authors found after upload to a central server (in so-called
that “most of the available literature referred “cloud computing” environments). These apps
only to pilot projects. . .and most of the stud- include stethoscopes, sleep structure analyzers,
ies were of low quality,” part of the so-called exercise or physical activity assistants, cardiac
“plague of pilots.” These studies included tele- analysis systems, mental health trackers, and
radiology, in which CT scans are transmitted Parkinson’s disease trackers, among many
remotely; those involving home monitoring re- others (21–23). Mobile phones are being con-
quired patient involvement in order to make structed to be used with integrated health sen-
physiological measurements. The authors con- sors (24), to interact with external sensors (25),
cluded that economic benefits of telemedicine to allow connection to plug-in external sensors
have yet to be proved, agreeing with a criticism (26), or simply to exploit the relatively complex
previously made in a Lancet editorial (16). suite of sensors that a typical mobile phone now
Similarly, Celler et al. (17) noted that “as is contains (such as a microphone, camera, GPS,
common for most telemedicine applications, a and accelerometer) to provide information on
strong evidence base for cost effectiveness and a subject’s health.
improved healthcare outcomes is still not avail- However, it is not sufficient merely to col-
able,” which they attributed to the lack of con- lect data. An integrated approach is needed,
trolled trials described in the literature. In a re- whereby data are stored in a flexible database
view of 175 articles on telemedicine in chronic and both humans and automated algorithms
disease management, these authors found can parse the data to provide warnings of
deterioration and other actionable information. phenomenon known as aliasing, which intro-
Many such architectures have been created, duces “phantom” effects into the signal when
ranging from in-hospital wireless ECG mon- recording data less frequently than twice the
itoring to complex mHealth systems. Celi et al. fastest change in the data. Nyquist’s sampling
(27) implemented an open-source telemedicine theorem states that we need to sample at least
infrastructure, primarily designed for use in de- twice as fast as the highest frequency in the sig-
veloping countries, which allows upload of any nal in order to obtain an accurate representa-
type of medical data to a remote server and syn- tion of the information that we are attempting
chronization of that data set to an open-source to obtain. A typical example of this is when a
medical record system called OpenMRS (28). wheel on a cart in an old western film picks up
The system also allows multiple independent speed and starts to rotate faster than 20 rev-
annotations of the data, feedback to the pa- olutions per second. Because the film is being
tient or healthcare worker, and annotation with shot at 40 Hz (or frames per second), the wheel
a universal medical lexicon (the UMLS). starts to appear to spin in reverse. (We can guess
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
The provision of healthcare through using that the wheel shouldn’t be spinning in that di-
mHealth has several natural advantages over ex- rection, and apply some extrapolation to fill in
isting care pathways: the data and make the wheel appear to spin the
1. Lower cost of capital investment right way, but we can only do so because it is
2. Users’ familiarity with devices and a simple linear system and we are absolutely
interfaces sure that the wheel cannot spin backwards!
3. A generic interface that can be easily cus- However, in general, we will never know if we
tomized for disadvantaged populations have filled in the missing data correctly.) Now
(large icons, voice recognition) set this in the context of current recommenda-
4. Seamless data upload with accurate time- tions for home monitoring of BP in the United
stamps and geospatial markers States by the American Society of Hyperten-
5. Natural security (led by “mBanking”)— sion, which suggest we should record BP twice a
i.e., access requires something you know day (29). BP normally has at least two peaks and
(a password) and something you have (the two troughs during a 24-h period (30), so sam-
device) pling only twice in a 24-h period could acciden-
6. Use of a private, yet individual, iden- tally result in measurement of either the highest
tifier, e.g., International Mobile Equip- or lowest excursions of the BP, falsely indicating
ment Identity (IMEI) number or Media hypertension or hypotension. It is interesting to
Access Control (MAC) address note that previous recommendations suggested
7. Allows construction of a long-term medi- sampling BP once per day, or less frequently,
cal record, allowing detailed personalized which is known to have led to overprescrip-
healthcare tion of hypertensive medication due to “white
8. Automated data upload, no need for user coat syndrome” (a temporary increase in a pa-
intervention tient’s BP due to the stress of being examined
9. Natural route for data feedback to the by a clinician). This syndrome may therefore
user be sometimes compounded by aliasing. A re-
cent study by Hug et al. (31, 32) demonstrated
that hypotensive alarms generated as a result of
THE PROBLEMS OF WIRELESS the monitoring of intra-arterial BP in the ICU
HEALTHCARE MONITORING were systematically being rejected by clinicians
Wireless monitoring systems also allow us even when those alarms were correct, probably
to address a largely ignored problem in because no immediate review procedure was in
medicine—the fact that we often sample be- place. On average, hypotensive events (which
low the “Nyquist frequency,” leading to a have been linked to poorer outcomes in the
Evaluation (CHIME) project measures respi- sion. HL7 is based on XML (41) or on the Eu-
ration from the impedance pneumogram, heart ropean VITAL standard (42), which specifies
rate (HR) and HR variability from an ECG, formats for transferring vital-sign data. Lebak
SpO2 from a pulse oximeter, and sleep position et al. (43) proposed an HL7-based architecture
from an accelerometer (36). In a clinical study whereby vital-sign data were collected contin-
of home infant monitoring, sensitivity to brady- uously from sensors by a local server, which
cardia was reported to be 100%, while positive periodically uploaded batches of patient data
predictivity was 6.5%. The authors concluded to a remote central server, usable by clinicians.
that although the system is sensitive to move- There are several standards for wirelessly
ment artifact, the quality of acquired data (and transmitting data, including the battery-hungry
its breath-detection algorithm) makes it prefer- high-bandwidth Bluetooth, its low-powered
able to conventional monitors. cousin Zigbee, the various forms of Wi-Fi,
Significant recent advances have been made standard mobile phone voice channels, Short
in noncontact wireless health monitoring. Message Service (SMS), Multimedia Messaging
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
These include using cameras to monitor phys- Service (MMS), General Packet Radio Service
ical movement during sleep or to extract HR (GPRS), Universal Mobile Telecommuni-
and respiration rate using video data of a pa- cations System (UMTS), satellite, and even
tient’s face acquired from webcams (37). Such standard radio. Each of these has trade-offs in
work has shown promising noncontact mea- terms of cost, bandwidth, power consumption,
surement of the ECG, which can detect a coarse and reliability. Several of these communication
heartbeat at a distance of several centimeters. methods do not possess error-checking, so
Radar has also been used to measure BR and data can be lost without the sender knowing.
HR in low-power Zigbee-compliant devices Accurate time-stamping of data is a particularly
such as that proposed by Zito et al. (38) and problematic issue in these circumstances. In
Scilingo et al. (39), which can be integrated into medicine, the accuracy of the time at which
the fabric of a shirt. Audio and video analy- an event occurred is often important; without
sis is also being used to identify sleep-related it causality can be lost, timely interventions
diseases (21), Parkinson’s disease (22), and prevented, or erroneous patterns observed that
depression (23). mask events or lead to false alarms. SMS (or
“texting”) is particularly sensitive to such issues.
Texts sometimes take >24 h to arrive at their
DATA TRANSMISSION destinations, with no guarantee of preserving a
AND INTERCHANGE given sequence of messages. SMS is therefore
A survey of remote vital-sign monitoring by only appropriate for a very limited type of trans-
Meystre (40) noted that “telemonitoring is still missions, which are relatively insensitive to
overshadowed by teleconsulting and teledi- missing data and which change on the timescale
agnosis.” Meystre used the term biotelemetry of weeks rather than days (such as transmissions
to distinguish the remote measurement of of a patient’s weight). Time-stamping issues can
physiological data from other home-based care be solved for systems that have a sampling fre-
systems. The author concluded that bioteleme- quency on the order of a few seconds; the time
try systems are typically developed in isolation, can be extracted from the mobile phone (which
and that an equivalent to the DICOM (Digital is regularly synchronized with the mobile
Imaging and Communications in Medicine) network) and then appended to the transmis-
standard used by radiologists to transmit im- sion. GPRS and Wi-Fi connections also allow
ages is required. He suggested that this could be solutions for timeliness and reliability through
based on Health Level 7 (HL7), which was in- additions to the communications protocol that
troduced in 1987 for electronic transmission of explicitly determine quality of transmission
healthcare data and is currently in its third ver- (e.g., 27).
Storage is often a combination of binary meta-data in the HL7 packets that specify de-
data (for data acquired at a high frequency, vice configuration, details of any filters applied
such as ECG) and relational databases (which to the data, and known accuracy. It is also im-
often store lower-frequency data such as daily portant that signal-quality information is sent
weight, or data extracted or aggregated from with each transmission, since all data are cur-
higher-frequency data). Several standards rently treated equally, independent of the actual
exist for signals and images, such as the reliability of the acquisition system used to ob-
European Data Format (EDF) and the Digital tain a given measurement. Most devices have
Imaging and Communications in Medicine built-in signal-quality evaluation, such as the
(DICOM) standard, but no standard exists “movement artifact” in a typical pulse oxime-
for relational databases, although examples ter, but it is extremely rare for devices to provide
have been made available with an open-source anything more than a coarse version of the qual-
license (44). Many monitors and systems now ity index on a visual display. In order to process
exchange data via HL7, as described above. data accurately, it is important to understand
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
However, HL7 only tells devices how to how much each measurement can be trusted.
communicate; it does not encode the semantics It is almost impossible to differentiate artifacts
of the contained data. For example, a typical from real events if only derived parameters are
transmission could include data concerning a provided (for example, only a BR measurement
BP measurement (stored as a number using and not the impedance plethysmography sig-
text), a binary ECG snippet, and a diagnosis. nal used to derive it). When the quality of the
The data packet does not specify from which underlying signal is accurately estimated, it is
monitor the BP reading came, or whether it was possible to use this information to greatly im-
measured invasively or noninvasively, or even prove the sensitivity and specificity of monitor-
from what location on the body the reading was ing algorithms that subsequently use the data
taken (which will change the reading and its (34).
interpretation). The ECG may be represented The issue of ontological encoding of data
in an arbitrary format, and the diagnosis may is far more challenging to solve. Although sev-
appear as free text in any language. eral ontologies exist (e.g., the Unified Medical
The issues surrounding data formats are rel- Language System and its constituent lexi-
atively trivial to solve if a sufficiently flexible and cons such as SnoMed, LOINC, and RxNorm),
open format is used such as the Wave Form which enable a standardized description of
DataBase format (45). Incidentally, the XML medical data, there is a problem of degener-
format specified by the U.S. Food and Drug acy: a given medical condition or intervention
Administration (FDA) is probably not appro- can be encoded in multiple ways, making the
priate as a standard because it becomes rapidly comparison of medical data time-consuming or
cumbersome for the transmission of even mod- impossible. Some progress has been made in
erate recording lengths of ECG data. The this arena by using limited use-specific subsets
issue of unknown sources of data can also be of given ontologies (27, 44). Unfortunately, the
solved easily. An example is the current picture explosion of proprietary systems in the mobile
format taken by digital cameras and cameras monitoring field has led to a proliferation of ad
in mobile phones. Each picture carries meta- hoc and proprietary database formats, prevent-
information concerning the compression, fil- ing data from one manufacturer being com-
ters, camera chassis, lens type, resolution, and pared with data from another. A reasonable so-
shutter speed used to take the photograph, lution to this problem is to make the database
along with the time at which it was taken, and formats available in an open-source manner,
the location (if GPS is enabled). To solve the is- and label the categories with an agreed subset
sue of unknown sources would require a change of the UMLS. However, manufacturers often
in regulation to require manufacturers to send allow the customer (i.e., the clinical team) to
extend the database format during use, seriously systems that require patients’ cooperation in
affecting the ability of automated algorithms to making measurements.
analyze the data. Stewart et al. (48) noted that 1% of the Na-
Two final concerns with wireless patient tional Health Service budget is devoted to heart
monitoring are worth noting. First, wireless failure management, with 60% of this spent on
networks can introduce large delays (sometimes hospital admission. Most patients with heart
several hours) in data transmission, and so pa- failure are managed by primary care physicians.
tient data viewed on multiple devices can eas- The authors emphasized that a ten-minute con-
ily become unsynchronized. Synchronization sultation with a general practitioner is proba-
of databases has long been an issue with dis- bly insufficient for accurate diagnosis and man-
tributed software, and techniques from this field agement of heart failure, and that avoidance of
may be valuable in dealing with such issues. “white coat syndrome” (the anomalously high
Second, wireless transmission of data raises is- BP reading that results from the stress of the
sues of patient privacy—particularly the risk clinical examination) should motivate the use
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
of interception of patient data during wire- of home monitoring. The authors conducted a
less transmission. The definition of “private” 20-patient study in which check-ups with nurs-
medical data or “protected health information” ing staff via video conferencing took place once
varies significantly. Some countries have no pri- per week.
vacy standards at all, and clinicians transmit pa- Woodend et al. (49) conducted a random-
tient data via public email accounts without en- ized controlled trial testing the effect of three
cryption; at the other extreme, even an ECG months of care via home monitoring, which col-
without any identifiers is considered “private lected 12-lead ECG during video consultations
data” in the United Kingdom. The potential for with clinical staff. The authors reported im-
international telemedicine is therefore some- proved patient outcomes. Louis et al. (50) sur-
what complicated. veyed “telemonitoring” methods for patients
who have suffered chronic heart failure. At least
30% of patients diagnosed with heart failure
upon hospital discharge are readmitted within
QUANTIFYING THE IMPACT three months, and up to 54% are readmit-
OF WIRELESS HEALTHCARE ted within six months. Admissions are typi-
Bondmass et al. (46) claimed to have completed cally prolonged and recurrent. Rich et al. (51)
the first study of telemedicine for home mon- and Stewart et al. (48) showed that daily vis-
itoring that reports clinical and economic out- its to patients’ homes by a nurse improved pa-
comes. Patients in this study measured their tient outcomes and reduced readmission rates.
own vital signs during telephone consultations Uncontrolled studies have shown patient ac-
with nurses. Alarms were generated whenever ceptability of home monitoring ranging from
vital signs exceeded predefined upper or lower 80% to 90% and have reported reductions
thresholds, and these alarms were passed to in hospital readmission rates, improvement in
clinical staff in the study hospital. patient quality of life, and a resultant over-
De Lusignan (47) conducted a study of all decrease in annual medical costs due to
20 patients, wirelessly recording HR, BR, BP, the reduction in readmission rate. The authors
ECG, and temperature, with results collected concluded that larger, randomized trials are re-
at a central monitoring station. Vital-sign quired in order to confirm these findings. They
accuracy was compared with periodic man- attributed the benefits of home monitoring to
ual measurements made by a nurse. The au- factors such as detecting new-onset atrial fibril-
thors concluded that the objectivity associ- lation, persistent hypotension, and overdiure-
ated with collecting patient data via continuous sis (by measuring weight as a measure of fluid
home monitoring is one of its advantages over balance).
Koch (52) surveyed 578 studies published tions being monitored are chronic. Little eco-
between 1990 and 2003 on the subject of “home nomic analysis was performed in the 65 studies.
telemedicine,” noting that “most publications Most of the 65 studies did not involve random-
deal with vital-sign measurement and au- ization, and only 8% included a control group.
dio/video consultations.” The author observed The majority of studies collected data once per
that “publications about . . . decision support for day or once per week and included vital-sign
staff are relatively sparse.” The clinical applica- data. The authors also concluded that the qual-
tion domains were mostly in the monitoring of ity and reliability of data acquired in the studies
patients with chronic diseases, the elderly, and were similar to those obtainable by clinical staff,
children. The survey concluded by noting the and that, given the weight of evidence provided,
limitations of home telemedicine, including the future studies need not continue investigating
lack of standards to combine incompatible in- the quality of transferred data. The authors’
formation systems, the lack of a framework by definition of “telemonitoring” included mea-
which such systems could be evaluated, and the surement of physiological parameters with
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
lack of proper guidelines for practical imple- the active involvement of the patient, rather
mentation of home monitoring. than entirely passive monitoring. Although the
A similar survey by Martinez et al. (53) majority of these studies required patients to
considered 13 randomized controlled trials, 10 actively manage their data acquisition, alerts
nonrandomized controlled trials, and 19 un- based on univariate thresholds could be gener-
controlled trials or descriptive studies. The au- ated if measurements were sufficiently “abnor-
thors observed that most of the studies reviewed mal,” with those alerts then automatically sent
did not discuss technical feasibility or technical to clinicians for attention (56). Some studies
problems. However, these trials used patient- (57) have measured parameters continuously
acquired vital signs, and the phrase “continu- using implanted sensors. Outcomes also appear
ous vital-sign measurement” was used to denote to have improved for patients implanted with
daily measurements made by the patient us- a “CardioMEMS” BP-measuring device (58).
ing electronic devices (including the ECG; 54). A survey of publications on home telemoni-
Even among studies based on patient-acquired toring for respiratory conditions between 1966
data, the authors concluded that “there are still and 2007 found 23 relevant studies (59), of
few published data on the feasibility and the im- which 7 pertained to patients undergoing pul-
pact of home monitoring.” monary transplantation, 12 were related to
Pare et al. (55) performed a meta-analysis asthma sufferers, and 4 involved patients with
of 65 studies that took place between 1990 COPD and other severe respiratory illnesses.
and 2006. Of these studies, 18 concerned pul- Eight of the 23 studies were randomized tri-
monary conditions, 17 diabetes, 16 cardiac dis- als. As in other reviews, these studies required
eases, and 14 hypertension. The results of this patient intervention in order to perform physi-
analysis showed that “the magnitude and signif- ological measurements, which were then trans-
icance of the telemonitoring effects on patients’ mitted via various means to clinicians.
conditions still remain inconclusive for all four The case for mHealth in particular has
chronic illnesses.” The authors concluded that yet to be proven, although evidence is begin-
effects on patient outcomes (e.g., decrease in ning to surface in much the same manner as
emergency visits, hospital readmissions, and for more traditional wireless monitoring. Part-
hospital length of stay) were “more consistent ners in Health have demonstrated that drug
in pulmonary and cardiac studies than diabetes stock-outs and overstocking (with consequent
and hypertension.” Importantly, although pa- improvement in mortality and costs) can be
tient compliance was reportedly high in all stud- achieved through increasing the speed of trans-
ies, compliance was seen to decrease through mission of HIV-related data and recording an-
time, which is of concern given that the condi- tiretroviral usage (60). Another recent example
is that of Pésinet (61), which uses mobile phones “apps” developed for the iPhone, many of
to provide a preventive medical diagnostic for them are simply cosmetic and add little or
children between 0 and 5 years of age. Local no additional value above the ability to re-
staff in Senegal and Mali use a simple mobile view data on a smaller, more portable screen.
phone interface to enter and transmit informa- Interestingly, it is the poorer health sectors
tion about the infants’ weights and symptoms (and the associated nongovernmental organi-
(e.g., fever) to pediatricians via SMS. The rapid zations) that have been driving much of the
and consistent transmission of data that indi- revolution in low-cost mHealth. The rapid and
cate the early stages of health problems (such exciting changes that are occurring in emerg-
as malnutrition) has led to a claimed reduction ing economies, necessitated by economic and
in mortality (in one group of 300 subjects in environmental constraints, may even provide a
Senegal) from 120 per 1000 to 8 per 1000. That model for developed economies to lower health
is, in over 90% of cases, if the disease is quickly costs and increase efficiencies. The leapfrog ef-
reported, the child will survive (61). fect in wireless telecommunication may trans-
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.
LITERATURE CITED
1. Rivera-Ruiz M, Cajavilca C, Varon J. 2009. Einthoven’s string galvanometer. Tex. Heart Inst. J. 35(2):174–
78
2. Meyer TJ, Eveloff SE, Bauer MS, et al. 1994. Adverse environmental conditions in the respiratory and
medical ICU settings. Chest 105(4):1211–16
3. Parthasarathy S, Tobin MJ. 2004. Sleep in the intensive care unit. Intensive Care Med. 30(2):197–206
4. GSMA, McKinsey & Co. 2010. m-Health: a new vision for healthcare. http://gsmworld.com/documents/
mHealth_report.pdf. Accessed Apr. 1, 2011
5. Mathers CD, Boerma T, Ma Fat D. 2009. Global and regional causes of death. Br. Med. Bull. 92:7–32
6. Mannino DM, Buist AS. 2007. Global burden of COPD: risk factors, prevalence, and future trends. Lancet
370(9589):765–73
7. Darkins A, Ryan P, Kobb R, et al. 2008. Care coordination/home telehealth: the systematic implemen-
tation of health informatics, home telehealth, and disease management to support the care of veteran
patients with chronic conditions. Telemed. e-Health 14:1118–26
8. Polisena J, Tran K, Cimon K, et al. 2010. Home telehealth for chronic obstructive pulmonary disease: a
systematic review and meta-analysis. J. Telemed. Telecare 16:120–27
9. Cruickshank J, Beer G, Winpenny E, et al. 2010. Healthcare without walls: a framework for delivering telehealth
at scale. 2020health Rep., Nov. 24. http://www.2020health.org/2020health/research/telehealth.html
10. UNESCO Institute for Statistics (UIS). 2002. Literacy and nonformal education sector, world illiter-
acy rate 1970–2000 (prognosis for 2005–2015), age 15 years and over, regional adult illiteracy rate
34. Li Q, Mark RG, Clifford GD. 2008. Robust heart rate estimation from multiple asynchronous noisy
sources using signal quality indices and a Kalman filter. IOP Physiol. Meas. 29:15–32
35. Nemati S, Malhotra A, Clifford GD. 2010. Data fusion for improved respiration rate estimation. EURASIP
J. Adv. Signal Process. 2010:926305
36. Neuman MR, Watson H, Mendenhall RS, et al. 2001. Cardiopulmonary monitoring at home: the CHIME
monitor. Physiol. Meas. 22:267–86
37. Takano C, Ohta Y. 2007. Heart rate measurement based on a time-lapse image. Med. Eng. Phys. 29(8):853–
57
38. Zito D, Pepe D, Neri B, et al. 2008. Feasibility study and design of a wearable system-on-a-chip pulse
radar for contactless cardiopulmonary monitoring. Int. J. Telemed. App. doi:10.1155/2008/328597
39. Scilingo EP, Lanat A, Zito D, et al. 2008. Wearable monitoring of cardiopulmonary activ-
ity through radiant sensing. Proc. 5th Int. Workshop on Wearable Micro and Nanosystems for
Personalised Health. http://www.phealth2008.com/Events/agenda_extendida/ses2_Scilingo_full.pdf.
Accessed July , 2011
40. Meystre S. 2005. The current state of telemonitoring: a comment on the literature. Telemed. e-Health
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
11(1):63–69
41. HL7 Version 3 Standard: Common Message Element Types, Release 2, ANSI/HL7 V3 CMET, R2–2009
42. Anagnostaki A, Pavlopoulos S, Koutsouris D. 2001. XML and the VITAL standard: the document-
oriented approach for open telemedicine applications. Medinfo 10(Pt. 1):77–81
43. Lebak JW, Yao J, Warren S. 2004. HL7-compliant healthcare information system for home monitoring.
Proc. Int. Conf. IEEE EMBS, 26th, San Francisco, Sept. 1–5, pp. 3338–41
44. Saeed M, Villarroel M, Reisner AT, et al. 2011. Multiparameter Intelligent Monitoring in Intensive Care
II (MIMIC-II): a public-access intensive care unit database. Crit. Care Med. 39(5):952–60
45. Moody GB. 2011. WFDB programmer’s guide, 10th ed. http://www.physionet.org/physiotools/wpg.
Accessed July 2011
46. Bondmass M, Bolger N, Castro G, Avitall B. 2000. The effect of physiologic home monitoring and
telemanagement on chronic heart failure outcomes. Internet J. Anaes. 4(2). Online only
47. de Lusignan S. 2000. Commentary. Evid. Based Cardiovasc. Med. 8:92–93
48. Stewart S, Marley JE, Horowitz JD. 1999. Effects of a multi-disciplinary, home-based intervention on
planned readmissions and survival among patients with chronic congestive heart failure: a randomised
controlled study. Lancet 354:1077–83
49. Woodend AK, Sherrard H, Fraser M, et al. 2008. Telehome monitoring in patients with cardiac disease
who are at high risk of readmission. Heart Lung J. Acute Crit. Care 37:36–45
50. Louis AA, Turner T, Gretton M, et al. 2003. A systematic review of telemonitoring for the management
of heart failure. Eur. J. Heart Failure 5:583–90
51. Rich MW, Beckham V, Wittenberg C, et al. 1995. A multidisciplinary intervention to prevent the read-
mission of elderly patients with congestive heart failure. N. Engl. J. Med. 333:1190–95
52. Koch S. 2005. Home telehealth—current state and future trends. Int. J. Med. Informatics 75:565–76
53. Martinez A, Everss E, Rojo-Alvarez JL, et al. 2006. A systematic review of the literature on home moni-
toring for patients with heart failure. J. Telemed. Telecare 12:234–41
54. Nangalia V, Prytherch DR, Smith GB. 2010. Health technology assessment review: remote monitoring
of vital signs—current status and future challenges. Crit. Care 14:233
55. Pare G, Jaana M, Sicotte C. 2007. Systematic review of home telemonitoring for chronic diseases: the
evidence base. J. Am. Med. Informatics Assoc. 14(3):269–77
56. Roth A, Kajiloti I, Elkayam I, et al. 2004. Telecardiology for patients with chronic heart failure: the ‘SHL’
experience in Israel. Int. J. Cardiol. 97:49–55
57. Kjellstrom B, Igel D, Abraham J, et al. 2005. Trans-telephonic monitoring of continuous haemodynamic
measurements in heart failure patients. J. Telemed. Telecare 11:240–44
58. Abraham WT, Adamson PB, Bourge RC, et al. 2011. Wireless pulmonary artery haemodynamic moni-
toring in chronic heart failure: a randomised controlled trial. Lancet 377:658–66
59. Jaana M, Pare G, Sicotte C. 2009. Home telemonitoring for respiratory conditions: a systematic review.
Am. J. Manag. Care 15(5):313–20
60. Yamanija J, Durand R, Bayona J, et al. 2006. Comparing actual medication consumption against the
quantities ordered and a prediction using an information system. Int. J. Tuber. Lung Dis. 10:S69–S70
61. Gaye M, Moreau O. 2008. Pésinet: a health care initiative for the reduction of infant mortality. Growing
Inclusive Markets (GIM) Case Study No. A002. New York: United Nations Development Programme.
http://cases.growinginclusivemarkets.org/documents/99. Accessed July 2011
Annu. Rev. Med. 2012.63:479-492. Downloaded from www.annualreviews.org
by University of Oxford - Bodleian Library on 01/24/12. For personal use only.
Annual Review of
Medicine
v
ME63-frontmatter ARI 20 December 2011 12:3
vi Contents
ME63-frontmatter ARI 20 December 2011 12:3
Indexes
Errata
Contents vii