Type 2 Diabetes: Scope of Literature


This review sought to collect and interpret the evidence base on home telehealth in managing type 2 diabetes. We looked at 3 areas:

  • Patients: How is home telehealth used in care management for people with type 2 diabetes?  Do patients like it?  How clinically effective is it?
  • Providers: What is the impact of home telehealth on health human resources?  What are the roles of nurses, general practitioners, and specialists in its delivery?  How do providers characterize their experiences with home telehealth?
  • System: How is home telehealth incorporated into the care continuum?  What is the economic impact of incorporating home telehealth into the management of type 2 diabetes?  What policies need to be in place for home telehealth to be successful?

In order to answer these questions, we conducted a systematic search for literature published from 2005 to 2010.  This search retrieved 33 studies that met our inclusion criteria.  Studies of particular note included the Informatics for Diabetes Education and Telemedicine (IDEATel) series, the Department of Veteran Affairs Care Coordination Home Telehealth (VA-CCHT) programs, and the South Korean SMS/biodang.com intervention.  The quality of the evidence was mixed, but included a substantial amount of high-quality material.  A final search for publications from 2011 to 2012 found 8 additional studies.  Though findings from these studies were incorporated into the review, be aware that material from the 2011-2012 period was screened and analyzed in a slightly different way than the studies retrieved in our original search.



Number and Location

A search of literature published from 2005-2010 located 59 publications on home telehealth in the management of type 2 diabetes. However, several sets of publications stemmed from multi-part or case series studies. Publications were considered related if they presented outcomes from the same intervention. The total number of distinct studies was placed at 33. As articles were often drafted at different stages of study implementation, sample size, population characteristics, and study duration were not necessarily constant.

Approximately half a dozen ‘publication clusters’ were identified. The largest of these clusters arose from the Informatics for Diabetes Education and Telemedicine (IDEATel) study, the Department of Veteran Affairs Care Coordination Home Telehealth (VA-CCHT) program, and the South Korean SMS/biodang.com intervention.  Our search retrieved 8, 9, and 9 publications, respectively, from these studies. See C.8.1.2: Study Design for more details. Other multi-part studies included Bond et al. (2007, 2010), Forjuoh et al. (2007, 2008),and Istepanian et al. (2009a, b).

Over half of the 33 studies found were located in the United States. Of these, 1 was based on a reservation (Robertson et al., 2007). South Korea was the setting for 5 studies. Only 3 Canadian studies were found: 2 from Ontario and 1 from British Columbia. The majority of the remaining studies were conducted in Europe, including Germany, Spain, and Poland. See chart, below, for more details.

A scan of material from 2011-2012, a time period not covered by our initial searches, found 8 additional studies that met our inclusion criteria and filled gaps left by the first rounds of searching (Arora et al., 2012; Cho et al., 2011; Glasgow et al., 2011; Katz et al., 2012; Logan et al., 2011; O’Reilly et al., 2011; Sarker et al., 2011; Weinstock et al., 2011a). These studies were analyzed in a slightly different way than the studies retrieved in our original search, and their findings were incorporated into this review in a limited fashion. For more details, please see Methods.

And on the qualitative side . . .There were 3 qualitative articles found of relevance to Type 2 diabetes, 2 of which centered on the IDEAtel project (Starren et al., 2005 Trief et al., 2008). The other article focused on a diabetes monitoring and a messaging device program administered by 2 Veterans Affairs medical centres in the Midwestern United States (Hopp et al., 2007). It should be noted that the participants in Hopp et al. (2007) were nurses who were involved in delivering the intervention. The patient experiences reported in this study are derived from provider observations rather than direct measurement.All 3 articles examined patients’ and providers’ experiences with the use of a home telehealth unit. Patients could input various measurements (e.g. blood sugar levels, weight, blood pressure) into this unit. These were then sent to their health provider for review and monitoring. The IDEAtel project also included videoconferencing sessions between the patient and a nurse/dietician. These took place every 4-6 weeks.


Study Design

The Oxford 2011 Levels of Evidence were used to assess the strength of the evidence base.[1] Studies were placed on a scale running from Level 1,[2] considered the highest level of evidence, through to Level 5.[3] Levels are based primarily on study design. Studies were also assigned scores for quality of execution and reporting. Low execution/reporting scores resulted in downgrading.

The Oxford 2011 Levels of Evidence are intended to provide guidance rather than absolute judgments, and do not obviate the need for careful appraisal of local needs and context. The quality of studies within a given level can vary, as can their applicability to select populations. Furthermore, this system is not suitable for all forms of assessment. In the text that follows, the Oxford 2011 Levels of Evidence are used only when discussing clinical outcomes.

The evidence base for home telehealth in the management of type 2 diabetes was extensive, but of mixed strength. Roughly half of the material retrieved qualified as Level 2 evidence[4], and slightly less than that as Level 4 evidence[5]. All save 1[6] of the remaining studies were Level 3 evidence.[7] Although the majority of studies received a ‘Moderate’ rating for execution/reporting, roughly 1 in 10 was downgraded a level for a low score. Particularly strong studies included Glasgow et al. (2010), McMahon et al. (2005), Shea et al. (2006, 2007), and Stone et al. (2010).

The number of participants enrolled ranged from 7 (Watson et al., 2009) to 6,185 (Weppner et al., 2010). Most articles reported on studies with sample sizes in the low hundreds. Notable exceptions were the Informatics for Diabetes Education and Telemedicine (IDEATel) publications, which generally reported on cohorts of roughly 1,500, and the Veteran Affairs Care Coordination Home Telehealth (VA-CCHT) articles, which looked at up to 800 enrollees.

Duration of intervention was under 1 year in most cases. Once again, the IDEATel and VA-CCHT studies were exceptions. Publications from the former spanned a period of 5 years; those from the latter, 4 years.

The most comprehensive reporting of outcomes was found in the multi-part and case series studies mentioned in Section C.8.1.1: Number and Location (IDEATel, VA-CCHT, and the Korean SMS/biodang.com intervention).

The IDEATel program enrolled participants from urban and rural areas of New York State. The intervention group was provided with a home telehealth unit and received bi-monthly video visits from a dietitian. Publications include reports of 1- and 5-year findings. Outcomes examined included blood glucose control, blood pressure, depression, physical activity levels, waist circumference, BMI, and the self-efficacy as a mediating factor in intervention success. Articles on this program included Izquierdo et al. (2007, 2010), Palmas et al. (2007), Shea et al. (2006, 2009), Trief et al. (2006, 2009), and Weinstock et al. (2011a). Sample sizes ranged from approximately 350 to 1,665.

The VA-CCHT program was aimed at veterans with diabetes, particularly those with high health care use. Patients were given a home messaging device and telephone access to caregivers. Outcomes were reported at 1 year, 2 years, and 4 years, and included service use, mortality risk, cost-effectiveness, and survival days. Publications on this program were Barnett et al. (2006, 2007), Chang et al. (2007), Chumbler et al. (2005a, b), and Jia et al. (2009). Sample sizes alternate between approximately 400 and approximately 800, depending on whether or not a control group is present.

Our search also retrieved 3 Veterans Affairs publications on home telehealth for diabetes management that did not appear to be part of the CCHT program (Dang et al., 2007; McMahon et al., 2005; Stone et al., 2010).

A South Korea-based intervention using SMS technology and the disease management website biodang.com gave rise to 9 publications. While the patient population varied from 1 article to the next, the intervention appeared unchanged. Articles in this cluster included Cho et al. (2006, 2009), Kim (2007), Kim et al. (2006), Kim and Jeong (2007), Kim and Kim (2008), Kim and Song (2008), and Yoon and Kim (2008).


Population Characteristics: Demographics

Age and Sex

Mean age for most studies was between 60 and 69. The IDEATel study was notable for its more elderly population. Mean age of IDEATel subjects was approximately 70 years. Participants in the VA-CCHT program fell into the same range, with a mean of 67-68. The studies from South Korea were unusual in that they tended to enroll younger participants, with a majority of publications focused on patients under 50 years of age.

Most studies had equal numbers of male and female participants. The VA-CHHT studies, unsurprisingly, were exceptions; the samples enrolled in these were overwhelmingly male (98-100%).

Race and Ethnicity

Approximately half of the 33 studies reported the race and/or ethnicity of their subjects. Interestingly, with the exception of Istepanian et al. (2009a, b) (London, England), all of the studies reporting race were located in the USA. Researchers on the IDEATel and VA-CCHT programs were particularly thorough in reporting this information. Approximately half of the IDEATel participants were white, slightly over a third identified as Hispanic or Latino, and roughly 15% were black or African-American. Most of the non-white subjects were located in New York City, with 99% of these participants identified as Hispanic or Latino, black, African American, or other. In contrast, 91.5% of subjects located outside of New York City were identified as white.

The VA-CCHT program was carried out at study sites in Florida, Puerto Rico, and Georgia. Nearly 50% of the sample population was identified as Hispanic, 40% were white, and approximately 10% were black or ‘other’. Chumbler et al. (2005b) focused only on VA-CCHT sites in north central Florida and south Georgia; in these locations, approximately 85% of the participants were white.

Only 3 studies reported a high percentage of participants from indigenous populations. Robertson et al. (2007) took place on a Northern Plains Indian reservation. Participants in Levine et al. (2009) were Native Americans recruited from Indian Health Centres in Alabama, Idaho, and Arizona. In Lorig et al. (2010), 14.5% of participants were American Indian or Alaskan Native.

Socioeconomic Status

The socioeconomic status of participants was reported in under half of the studies retrieved. Indicators used included educational attainment, Medicaid eligibility (USA), and individual or neighbourhood income.

Several interventions specifically targeted populations of lower socioeconomic status.  The IDEATel program enrolled underserved Medicare beneficiaries. Other common characteristics of IDEATel subjects were low levels of education, with approximately 10 years schooling on average (Izquierdo et al., 2007; Shea et al., 2009; Trief et al., 2006, 2009; Weinstock et al., 2011a), Medicaid eligibility (Izquierdo et al., 2010; Shea et al., 2006, 2007, 2009), and an income lower than $30,000/year (Izquierdo et al., 2007; Shea et al., 2006, 2007). Patients from the New York City study locations were markedly poorer than those outside the city: nearly 84% had an income less than $10,000.

The VA-CCHT program targeted high priority health care veterans, who were identified by low-income means-testing and high medical service use (Chumbler et al., 2005b, 2009; Jia et al., 2009). Chumbler et al. (2005a) reports that 62% of study subjects had attained less than a high school education.

Other studies focusing on patients with low socioeconomic status were Bond et al. (2007, 2010), Lee et al. (2007), and Glasgow et al. (2010).

Computer Literacy

Participating patients were required to have access to a computer and/or mobile phone in 9 interventions. An internet connection was usually required as well. See Table C.8.1.3: Population Characteristics – Computer Literacy, below, for complete list. Note that the 8 bolded citations are for publications that emerged from the same intervention: South Korea’s SMS/biodang.com intervention, a mobile phone/website-based disease management program. South Korea is noted for having high mobile phone and Internet connectivity. It is also noteworthy that (a) none of the South Korean studies featured low-income or marginalized populations; and (b) patient samples were considerably younger than those in most studies retrieved, with a mean age of under 50 years.

Internet literacy was an explicit eligibility requirement in 4 studies (Cho et al., 2006; Grant et al., 2008; Tjam et al., 2006; Song et al., 2009). Of 3 studies located in Canada, 2 had technology-related eligibility requirements. Tjam et al. (2006), located in Ontario, required patients to be internet literate. Tildesley et al. (2010), which took place in British Columbia, only included patients who had Internet access. The means by which technical literacy was measured are not described.

In the IDEATel intervention, approximately 20% of patients were identified as “know[ing] how to use a computer”. However, this percentage varied by location. Only around 5% of NYC patients knew how to use a computer, compared with about 33% of non-NYC patients.

Watson et al. (2009), Forjuoh et al. (2007, 2008) and Bond et al. (2007, 2010) had study samples with high previous use of computers and/or internet. In McMahon et al. (2005), approximately 30% of patients had internet access before the study began. In Kim and Kang (2006), only 27% of the participants could access the web-based intervention at home.

Population Characteristics: Clinical Characteristics

In most cases, studies excluded patients with physical or cognitive disability, life-threatening illnesses and/or moderate to severe complications. Pregnancy or planned pregnancy was another common disqualifying factor.

The most notable of the few exceptions to the exclusion of patient with co-morbidities are the VA-CCHT programs, many of which targeted veterans with complex medical conditions or at an increased risk for high health care service use. The most common co-morbidities in 1 study included congestive heart failure, peripheral vascular disease, and chronic pulmonary disease (Barnett et al., 2007). Other studies including patients with co-morbidities, and the conditions most frequently reported, can be seen in Table 8.1.3: Population Characteristics – Clinical Characteristics I and Table 8.1.3: Population Characteristics – Clinical Characteristics II, below. The most common co-morbidities were hypertension, obesity, hyperlipidemia/dyslipidemia, and cardiovascular disease.

[1] For a comprehensive overview of this system, please refer to Jeremy Howick, Iain Chalmers, Paul Glasziou, Trish Greenhalgh, Carl Heneghan, Alessandro Liberati, Ivan Moschetti, Bob Phillips, and Hazel Thornton. “Explanation of the 2011 Oxford Centre for Evidence-Based Medicine (OCEBM) Levels of Evidence (Background Document)”.
Oxford Centre for Evidence-Based Medicine.

[2] Systematic reviews of randomized trials; n-of-1 trials.

[3] Mechanism-based reasoning.

[4]Boaz et al., 2009; Bond et al., 2007, 2010; Cho et al., 2006, 2009; Glasgow et al., 2010; Grant et al., 2008; Istepanian et al., 2009a,b; Izquierdo et al., 2007, 2010; Kim, 2006; Kim & Jeong, 2007; Kim & Kang, 2006; Kim & Song, 2008; Lorig et al., 2010; McMahon et al., 2005; Ralston et al., 2009; Rodriguez-Idigoras et al., 2009; Shea, 2007; Shea et al., 2006, 2009; Stone et al., 2010; Timmerberg et al., 2008; Tjam et al., 2006; Weinstock et al., 2011; Yoo et al., 2009; Yoon & Kim, 2008.

[5]Barnett et al., 2006, 2007; Buckley et al., 2008; Bujnowksa-Fedak et al., 2006; Chang et al., 2007; Chumbler et al., 2005a,c; Chumbler et al., 2009; Dang et al., 2007; Forjuoh et al., 2007, 2008; Huanguang et al., 2009; Izquierdo et al., 2007; Jia et al., 2009; Kim et al., 2005; ; Kim et al., 2006; Kim & Kim, 2008; Levine et al., 2009; Luzio et al., 2005; Palmas et al., 2007; Sun et al., 2010; Trief et al., 2006, 2009; Trudel et al., 2007; Watson et al., 2009.

[6] Luzio et al., 2007.

[7] Chumbler et al., 2005b; Grant et al., 2008; Lee et al., 2007; Noh et al., 2010; Quinn et al., 2008; Robertson et al., 2007; Song et al., 2009; Tildesley et al., 2010.



Leave a Reply