Summary
Home telehealth for patients with heart failure is generally used in the hope of improving clinical outcomes and reducing burden on the health care system. The archetypical intervention is designed to enable ongoing patient monitoring: patients are equipped with home telehealth units, given a schedule for data transmission, and contacted by providers when deterioration is detected.
Details
Basic Model
The standard home telehealth intervention for heart failure was designed to enable ongoing patient monitoring. Patients were given home telehealth units, which generally consisted of multiple measurement devices – scales, sphygmomanometers, etc. – attached to a central hub. This hub transmitted patients’ measurements to an online location, typically through their home telephone lines. In some cases, data were integrated into an electronic medical record (e.g. Hudson et al., 2005; Schmidt et al., 2008). Out-of-range values were often automatically flagged by the system (Chaudhry et al., 2008; Bowles et al., 2009; Soran et al., 2008; Cleland et al., 2005; Dansky et al., 2008; Dar et al., 2009; Hudson et al., 2005; Mortara et al., 2008; Bondmass, 2007; Soran et al., 2010; Weintraub, 2010; Schwarz et al., 200; Kulshreshtha et al., 2010).When measurements fell outside pre-determined target ranges, providers contacted patients – usually via telephone – to discuss results and make referrals or medication changes if appropriate.
Another common feature of home telehealth units was the ‘disease dialogue’: a series of questions, usually centred around symptoms, that the patient completed at fixed intervals. Touchscreens could be used for data entry, although this was not invariably the case. Disease dialogues allowed providers to access patients’ concerns and perceived health status in a standardized format, and, as above, to contact patients when necessary.
Human Resource Requirements
Basic human resource requirements were described in 21 studies. Nurses featured prominently in all; they were the first point of contact for the patient and were generally responsible for follow-up as well. The level of physician involvement varied, and the precise nature and extent of the interaction between care providers was sometimes unclear.
In 11 studies, the intervention appeared to have minimal physician involvement (Bondmass, 2007; Dansky et al., 2008; Dar et al., 2009; Gambetta et al., 2007; Hudson et al., 2005; Kashem et al., 2008; Kulshreshtha et al., 2010; Myers et al., 2006; Scalvini et al., 2005a, b; Whitten et al., 2009; Woodend et al., 2008). While a specialist or a patient’s personal physician sometimes had a part in determining treatment goals and outlining management plans, their involvement in the intervention ceased thereafter (Kulshreshtha et al., 2010; Myers et al., 2006). Nurses were responsible for reviewing patient data, contacting the patient, and making management recommendations. These recommendations might include advice to contact a primary or secondary care provider; however, physicians were not directly involved before this contact was made.
In 8 other studies, physicians were involved throughout the intervention (Chaudhry et al., 2009; Bowles et al., 2009; Balk et al., 2008; Schwarz et al., 2009; Cleland et al., 2005; Wakefield et al., 2008b, 2009; Weintraub et al., 2010). Although patient communication and data screening again fell within the purview of the nurses, any out-of-range values triggered physician consultation.
Noteworthy variants on the care coordination process described above include Mortar et al. (2008), which relied heavily on an automated system, and Schmidt et al. (2008), in which there was extensive ongoing communication between members of each patient’s care team. These are discussed in more detail in C.6.2.1: Technical Requirements and C.6.2.3: Departures from Basic Model, below.
Limited information was available on the clinical and administrative burden of home telehealth interventions for heart failure patients. Of the 23 studies retrieved, only 7 provided some indication of the amount of provider time that interventions required. Limited contextual information makes it difficult to interpret many of the numbers given.
Schmidt et al. (2008) report that 2 or fewer contacts were required for the majority of patients. In Soran et al. (2008), 839 alerts were sent to providers over the course of a 6-month study with 160 participants – or roughly 1 contact per month per patient. Kashem et al. (2008), on the other hand, report that patient contacts were almost 6 times as numerous; providers sent ”1250 generic messages and 637 tailored-text messages” (p.124) in response to patients’ data uploads over the 12-month study. In Schwarz et al. (2008), nurses contacted 95% of participants at least once. Over the 90-day intervention, patients received calls from providers regarding abnormal test results on an average of 27 days (range 0-77 days).
Patients in Gambetta et al. (2007) were asked to call a toll-free number each day to answer a short health assessment. Of these contacts, 15% required return calls. Half of the return calls were suggestions for changes in therapy, 25% consisted of education only, and 25% were deemed ‘false positives’ – that is, no return call was actually indicated. Scalvini et al. (2005b) do not report absolute numbers, but note that 64% of patients’ data transmissions did not require or did not receive management recommendations from the study nurse.
Technical Requirements
The aforementioned telehealth hub was by far the most common technical requirement. Medical peripherals varied, but almost invariably included weight and blood pressure monitors. An intelligent system featured in several interventions; in these, out-of-range data were automatically flagged for the receiving provider’s attention. This was an essential component of Mortara et al. (2008). In this intervention, patient data transmissions went directly to an automated system, which alerted a nurse or physician only if values fell outside of a pre-set range.
In Dansky et al. (2009), the telehome monitoring system used was directly integrated into an electronic health records system. Patients’ data transmissions were automatically uploaded to their personal records. Although the intervention was carried out by nurses, physicians were able to access patient data by consulting this electronic health record. A similar set-up was used in Schmidt et al. (2008).
It is interesting to note that mobile devices feature in very few of these studies, given the relatively advanced state of telehealth in the management of cardiovascular disease. The primary reason may be that the measurements that are most useful for remote monitoring of cardiovascular disease are not easily collected with mobile devices.
One might also speculate that older technologies may tend to be used more in home telehealth for cardiovascular disease because home telehealth has been used in managing cardiovascular disease for longer. The time and money spent in developing older technologies may dissuade stakeholders from investing large sums in recent and untried tools, many of which may be incompatible with existing systems. Finally, patient characteristics – particularly age and mobility restrictions – may be a contributing factor.
Departures from Basic Model
Variations on the basic model were minor. Some studies collected patient information through automated voice response systems rather than home telehealth units (Chaudhry et al., 2009; Gambetta et al., 2007; Mortara et al., 2008), while a few required patients to manually enter data online (e.g. Kashem et al., 2008). In Scalvini et al. (2005a, b) and Piotrowicz et al. (2009), patients used mobile devices to make and transmit measurements. In Schmidt et al. (2008), patients were given a programmable medication dispenser that sent data to an electronic health record. The provider contacted the patient via telephone with a reminder and/or inquiry as needed.
One intervention included a substantial education component (Balk et al., 2008). Several others encouraged synchronous patient-provider consultation through videoconferencing (Dansky et al., 2008; Wakefield et al., 2008a, b; Wakefield et al., 2009; Woodend et al., 2008).
