Type 1 Diabetes: Patient Outcomes

Summary

There is strong evidence that home telehealth has the potential to lower hemoglobin A1c (HbA1c), generally without increasing frequency of hypoglycemia.  There is also strong evidence that this reduction can be significantly greater than that achieved with usual care.  This outcome may be due in part to improvements in disease knowledge and increased frequency of blood glucose testing, both of which have been associated with use of home telehealth.  There is no evidence that home telehealth is inferior to usual care in reducing HbA1c.  However, in a sizable number of interventions, patients experience no significant improvement or a level of improvement that is no better than that seen with usual care.  In short, there is strong evidence that home telehealth can be very effective in some circumstances, but it is not clear precisely what those circumstances are.  Factors may include intervention design, characteristics of the sample, and the environment in which the intervention is delivered.   Length of intervention does not appear to be of great import.

Reductions in HbA1c do not necessarily indicate greater stability in day-to-day glycemic control.  There is moderate evidence that home telehealth can significantly reduce blood glucose variability and rates of hypoglycemia.   However, the literature reviewed is highly inconsistent on these points.  In many cases, no effect is seen.  Furthermore, one strongly designed and well executed study found a significant increase in rates of hypoglycemia.  Caution is therefore warranted.

There is strong evidence supporting an association between home telehealth and significant improvements in quality of life.

There is currently no evidence supporting an association between home telehealth and significant improvements in body mass, cholesterol, or blood pressure.  However, few studies have targeted these outcomes.

Patient uptake of home telehealth for type 1 diabetes management is highly variable. There is some indication that interventions in which patients receive little or no feedback tend to have lower uptake. Concerns about ease of use and acceptability of technology appear largely unwarranted at present; we found no reports of negative patient experiences, although it should be acknowledged that those who agree to participate in home telehealth programs may be predisposed to enjoy the mobile devices and data analysis software that often accompanies these interventions.

Patients can derive significant time savings when home telehealth is used for remote appointments.  The majority of the evidence indicates that home telehealth is comparable to usual care, if not superior.  A final scan of literature published from 2011-2012 uncovered one study in which patients used smartphones equipped with diabetes management software.  Some patients were supported by bi-weekly teleconsultations with providers, while others carried on with regular clinic appointments.  Total provider time did not differ (nor did clinical outcomes), but patients saved considerable time when teleconsultations were used.  See Charpentier et al. (2011) for more details.

 

Details

Uptake and Use of Technology

Summary: Uptake and use of technology outcomes were reported in 6 studies (Benhamou et al., 2007; Everett & Kerr, 2010; Farmer et al., 2005; Rigla et al., 2008; Rossi et al., 2010; Wangberg, 2008). Due to variation in patient populations and study design, it is difficult to draw broad conclusions on factors affecting patient uptake and ease of use. Uptake was highly variable across studies, although there is some indication that interventions in which patients receive little or no feedback tend to have lower uptake. Concerns about ease of use and acceptability of technology, on the other hand, appear largely unwarranted at present, as negative perceptions of telehealth technology were not reported in any studies. These results should, however, be interpreted with caution, as patients willing to consent to participate in telehealth studies may have unusually positive attitudes toward technology at the outset. Another possible confounding factor is reporting bias.

Study Details: In studies reporting views on the telehealth system’s ease of use, effects on quality of care, and overall benefit, patients’ perceptions were generally positive. In Benhamou et al. (2007), 81% of patients found the device ‘very easy’ or ‘moderately easy’ to use, and 76% attributed improvements in the quality of their medical care to its use. Gomez et al. (2008) reported that 78% of patients ‘would recommend [the system] in diabetes care’ (p. 475). The average rating of the program used in Cox et al. (2008) was as ‘beneficial, easy to use, and enjoyable’ (p. 1528).

Rates of use were reported in 6 studies. In Benhamou et al. (2007), 29 of 30 participants adhered to a once-a-week data transmission schedule without reminders from investigators. This was the case during both the intervention period, in which providers acknowledged receipt of data and sent treatment recommendations, and the control period, in which no feedback was given. In contrast, Farmer et al. (2005) found that control group patients uploaded blood glucose values significantly less frequently than those in the intervention group: 0 median uploads at 9 months in the control group vs. 11 in the intervention group (p<.0001). Patient uptake varied widely in Rossi et al. (2010), in which the median number of text messages sent by patient to physician over the course of the 6-month study ranged from 6-75 (median 52).

And on the qualitative side . . .

In Armstrong and Powell (2009), participants expressed a high degree of satisfaction with the peer support offered through the board. Participants viewed using the Internet as an enjoyable, empowering, and educational process. Online information seeking was used in conjunction with other approaches to seeking help and advice. Authors suggest that online health information may complement the advice and support provided in formal health care settings rather than posing a challenge to it.

In Rigla et al. (2008), participants had access to an online diabetes management system in both phases of a randomized crossover trial. Patients logged into the system significantly more frequently during the intervention phase, when they also had use of a continuous glucose monitor (19.5+/-9.4 log-ins per patient per week v. 8.5+/-4.8, p<.01). Participants in Everett and Kerr (2010) were advised to use an online disease management system as frequently or infrequently as desired, though preferably every few days. Although participants were able to exchange messages or videoconference with a diabetes nurse specialist, “the emphasis on using [the] the system was to enable self-management . . . it was expected that the patients would make their own changes to management” (p. 10). Of 16 study participants, 12 used the system. Frequency of user log-ins is not reported. Wangberg (2008), while not providing exact numbers, reports that use of an educational website declined rapidly over the course of a 1-month trial, with the majority of those analyzed spending little time on the site. In addition, a large number of participants were lost to follow-up; of 64 enrolled, 27 were excluded from the analysis.

 

Self-Management, Self-Efficacy, and Behaviour Change

Self-management, self-efficacy, and behaviour change outcomes were measured in 9 studies, 5 of them Level 2 evidence (Benhamou et al., 2007; Farmer et al., 2005; Jansa et al., 2006; Vespasiani et al., 2009; Wangberg, 2008) and 4 Level 3 (Albisser et al., 2007; Cox et al., 2008; Rigla et al., 2008; Rossi et al., 2009).

 

Medication Use

Summary: 5 studies examined medication use (Albisser et al., 2007; Jansa et al., 2006; Rigla et al., 2008; Rossi et al., 2009; Vespasiani et al., 2006). On balance, home telehealth does not appear to have a significant effect on medication use. However, findings are not unmixed; there is moderate evidence – 2 Level 3 studies – that telehealth interventions are sometimes accompanied by medication change.

Study Details: Albisser et al. (2007) reported a significant decrease in total daily insulin dosage in the intervention, while Rigla et al. (2008) found a significantly higher number of daily bolus insulin doses in the intervention phase of a crossover study, but no change in the total number of daily doses or in basal doses. Both studies were limited by small sample sizes. A larger number of studies did not find any association between home telehealth and changes in use of long- and short-acting insulin (Jansa et al., 2006; Rossi et al., 2009; Vespasiani et al., 2009).

 

Self-Care Habits

Summary: Self-care, typically measured through frequency of blood glucose testing, was measured in 3 Level 2 studies (Benhamou et al., 2007; Farmer et al., 2005; Wangberg et al., 2008). There is strong evidence that telehealth has the potential to significantly improve self-care, but the extent to which this benefit is generalizable merits further scrutiny. The absence of effect reported in a 12-month intervention, despite high uptake, indicates that some caution is warranted.

Study Details: Farmer et al. (2005)found a significant difference between groups in the number of weeks of the 9-month intervention in which patients tested their blood glucose once or more ((18.8+/-11.1 in control group vs. 27.3+/-11.8 in intervention group; p<.001). Discouragingly, there was no correlation between testing frequency and reduction in HbA1c.

Wangberg (2008), who conducted a 1-month educational and behavioural intervention, also reported a significant main effect on self-care. However, Benhamou et al. (2007) found no intervention effects on adherence (measured by number of daily blood glucose tests), despite high uptake.

 

Disease Knowledge

Summary: Diabetes knowledge was measured in 2 studies: 1 Level 2 (Jansa et al., 2006) and 1 Level 3 (Cox et al., 2008). There is moderate evidence of association between home telehealth interventions and improved disease knowledge. Whether this improvement is significantly greater than that seen in usual care is questionable.

Study Details: Jansa et al. (2006) found significant improvement in disease knowledge in both control and intervention groups. The difference between groups was not significant. Cox et al. (2008) also reported improvements in disease knowledge. The size of this improvement was significantly associated with frequency of log-ons to an educational website.

 

Clinical Outcomes, Symptoms, and Health Status

Clinical outcomes and symptoms and health status outcomes were measured in 11 studies (Albisser, 2007; Benhamou, 2007, 2010; Cox, 2008; Everett & Kerr, 2010; Farmer, 2005; Gomez et al., 2008; Jansa et al., 2006; Rigla et al., 2008; Rossi, 2009, 2010; Vespasiani, 2009). Indicators of glycemic control, particularly hemoglobin A1c, blood glucose variability, and incidence of hypoglycemia, were the most frequently reported measures. Details of these findings are presented in Table C.7.3.1: Patient Outcomes – Clinical Outcomes, Symptoms, and Health Status (below).

 

Measures of Mean Blood Glucose

Summary: Measures of mean blood glucose, most commonly hemoglobin A1c, were made in 4 Level 2 studies (Benhamou, 2007; Farmer, 2005; Jansa et al., 2006; Rossi et al., 2010), 5 Level 3 studies (Albisser, 2007; Benhamou et al., 2010; Gomez et al., 2008; Rigla et al., 2008; Rossi et al., 2009), and 1 Level 5 study (Everett & Kerr, 2010). Findings on changes in mean blood glucose were mixed. Although there is strong evidence associating home telehealth with significant improvements in mean blood glucose[1], several studies did not find significant improvement[2]. Furthermore, while there was also strong evidence of significantly greater improvement than that seen in a control group[3] , there were a number of strongly designed studies in which this was not the case.[4]

Studies in which no benefit was found do not cancel out those in which outcomes were significantly improved. It appears that home telehealth interventions can be effective in bringing about significant improvements in measures of mean blood glucose, but it is also clear that they are ineffective in a sizeable proportion of cases. This may be due to variation in intervention design, characteristics of the sample, or the environment in which the intervention was delivered. Readers are advised to give careful consideration to the generalizability of these studies to the context in which they are operating.

One factor that does not appear to be of great import in the studies examined here is length of intervention. In studies that found significant improvement, length varied from 1 to 12 months. Those that did not find improvement examined similar time frames. There appears to be a slight trend toward lower baseline HbA1c in studies that did not find significant improvement, suggesting that interventions may be more effective for patients with poorer control. However, evidence of this is far from conclusive.

Study Details: Significant improvements in measures of mean blood glucose were found in intervention group participants without corresponding improvement in control group participants in Level 3 studies by Benhamou et al. (2010) and Rigla et al. (2008). Rigla et al. (2008) placed the percentage of in-range glucose values at 62% for the intervention group vs. 52% in the control group (p<.05). The intervention group in Benhamou et al. (2010) was further divided into 2 subgroups, which received different levels of feedback intensity. At the 6-month endpoint, both intervention groups had significantly lower HbA1c values (IG1:-0.46% +/- 0.89, p<.001; IG2: -0.73% +/- 0.84, p<.001). The difference between intervention groups was not significant. Also of note is Cox et al. (2008), in which intervention group participants showed significantly greater improvements in Improved Functioning Score (p=.048).

A Level 2 study by Farmer (2005) reports significant reductions in HbA1c in intervention and control groups (9.2%+/-1.1 to 8.6% +/-1.4, p=.001; and 9.3% +/-1.5 to 8/9% +/-1.4, p<.04, respectively). The difference between groups was not significant. However, median blood glucose was significantly lower in the intervention group throughout the 9-month trial (8.9 mmol/l vs. 10.3 mmol/l, p<.0001).

Another Level 2 study by Benhamou et al. (2007) reports non-significant trends towards improvements in blood glucose control in the intervention group only. Level 2 studies by Rossi et al., (2010) and Jansa et al. (2006) find improvements in both groups, while a Level 3 study by Albisser et al., (2007) reports no change in either. Findings from single-arm studies are consistent in finding improvements in glycemic control (Everett & Kerr, 2010; Gomez et al., 2008; Rossi et al., 2009), although these reach significance only in Gomez et al., 2008.

Limitations of the studies described above include sample size (Rigla et al., 2008; Gomez et al., 2008), incomplete reporting of eligibility criteria and patient characteristics (Rigla et al., 2008; Benhamou et al., 2010), and study design (Everett & Kerr, 2010; Gomez et al., 2008; Rossi et al., 2009). Despite these limitations, the collective findings of these studies constitute strong evidence that home telehealth interventions do not have a negative effect on glycemic control, and may have a positive effect. In addition, Farmer (2005) found that sex had a significant moderating effect. Males experienced a mean HbA1c decrease of .02% (+/-1.0) over the course of the study, whereas that of females averaged 1.0% (+/-1.2) lower (p<.001). These findings were not replicated in any other studies in this review. However, given the strong design and execution of Farmer (2005), further research into this area may be warranted.

For more details, see Table C.7.3.1: Patient Outcomes – Clinical Outcomes, Symptoms, and Health Status (above).

 

Blood Glucose Variability

Summary: Blood glucose variability was measured in 3 studies: 1 Level 2 study (Benhamou et al., 2007) and 2 Level 3 studies (Rigla et al., 2008; Rossi et al., 2009). There is moderate evidence associating home telehealth interventions with reductions in blood glucose variability, although 1 study found no significant intervention effect (Benhamou et al., 2007).

Study Details: Rigla et al. (2008) reported significantly higher scores on the Glucose Risk Index (GRI) during the control phase of a crossover study. While not a direct measure of blood glucose variability, the authors suggest that this difference (9.6 vs. 6.25, p<.05) was indicative of improvement in glucose stability during the intervention phase. Study participants also had a significantly higher percentage of in-range glucose values during the intervention phase (62% vs. 52%, p<.05). In Rossi et al. (2009), participants experienced significantly reduced coefficient of variation values for both fasting glucose (40.8% to 34.1%; p=.01) and post-prandial glucose (39.5% to 28%; p=.01). Benhamou et al. (2007) found no significant differences in blood glucose variability between the control and intervention phases of a crossover study. For more details, see Table C.7.3.1: Patient Outcomes – Clinical Outcomes, Symptoms, and Health Status (above).

 

Frequency of Hypoglycemia

Summary: Frequency of hypoglycemia was measured in 7 studies. Of these, 3 qualified as Level 2 evidence (Benhamou et al., 2007; Farmer et al., 2005; Jansa et al., 2006), while 4 were Level 3 evidence (Benhamou et al., 2010; Gomez et al., 2008; Rigla et al., 2008; Albisser et al., 2007). There is weak evidence associating telehealth with a significantly greater reduction in hypoglycemia than that seen in usual care (Albisser et al., 2007), moderate evidence associating it with a significant increase (Farmer, 2005), and a large number of studies in which no effect was seen. It appears that home-based telehealth cannot be consistently relied upon to alter rates of hypoglycemia, although 1 Level 2 study and 2 Level 3 studies[5] provide strong evidence that it has the potential to do so under some circumstances. However, there is a high level of inconsistency within the group of studies retrieved. Roughly half of the studies found no intervention effect, and those that did were not in agreement on the direction of the change.

Study Details: No intervention effect was found in Benhamou et al. (2007, 2010), Gomez et al. (2008), or Rigla et al. (2008). Farmer (2005) reported a significant increase in incidence hypoglycemia in patients using home-based telehealth. However, Albisser et al. (2007) and Jansa et al. (2006) found significant decreases. In the former, this reduction was also significantly greater than that seen in the control group. For more details, see Table C.7.3.1: Patient Outcomes – Clinical Outcomes, Symptoms, and Health Status (above).

 

Other

Summary: Studies meeting inclusion criteria for type 1 diabetes focused almost exclusively on measures of glycemic control when reporting clinical outcomes. Non-glycemic measures that were studied include body mass (Rossi et al., 2010; Jansa et al., 2006), cholesterol (Rossi et al., 2010), and blood pressure (Rossi et al., 2010).  No studies found significant reductions in non-glycemic measures or significant differences between intervention and control groups. There is currently no evidence supporting an association between home telehealth and significant improvements in body mass, cholesterol, or blood pressure. However, few studies have targeted these outcomes.

Study Details: See Table C.7.3.1: Patient Outcomes – Clinical Outcomes, Symptoms, and Health Status (above).

 

Quality of Life

Summary: Quality of life was measured in 3 studies (Benhamou et al., 2007; Jansa et al., 2006; Rossi, 2010). All were Level 2 evidence and administered validated diabetes-related quality of life surveys to study participants. There is strong evidence supporting an association between telehealth interventions and significant improvements in quality of life, although the evidence does not uniformly support this conclusion: 1 study found no significant improvements in intervention or control group patients (Jansa et al., 2006). Results from Benhamou et al. (2007) suggest that the long-term effects of home telehealth interventions on quality of life merit further study.

Study Details: Rossi et al. (2010) found significant improvement in 4 subscales[6] of the SF-36 among a subgroup of study participants. Benhamou et al. (2007) also reported significant improvements in quality of life during the intervention phase of a 6-month crossover study. Furthermore, these improvements persisted throughout the 6-month follow-up phase. Jansa et al. (2006) found no significant changes in either control or intervention participants.

 

Cost and Time Savings

Summary: There is little information available on the effect of home telehealth on patient cost and time savings. This gap in the literature limits attempts to calculate the overall economic value of home telehealth interventions. While patient time commitments may not have a direct impact on the economic value of telehealth within a given health care system, less patient travel time could be expected to translate into societal benefits through time saved in lost productivity.

Study Details: In 2 studies that did report patient time savings, outcomes were favourable to the intervention group. Jansa et al. (2006) found substantial time savings in the intervention group compared with the control group, with the former spending an average of 23 hours less (15 hours vs. 48 hours) in appointment and appointment travel time over the course of the 6 month study. Benhamou et al. (2010) reported average time savings of 5 hours in the intervention group over the same time period.

Recent Developments: A scan of material from 2011-2012, a time period not covered by our initial searches, found 1 article that addressed cost and time savings (Charpentier et al., 2011). In this study, patients used smartphones equipped with diabetes management software and were supported through either teleconsultations or clinic visits. Patients using teleconsultations had reduced travel time. Reductions in HbA1c did not differ between the 2 groups. Total provider time was also comparable. As this study was not subject to the same level of analysis as the other studies included in this review, we will not attempt further analysis. See Charpentier et al. (2011) for more details.

 


[1] Three Level 2 studies and two Level 3 studies: Farmer, 2005; Jansa et al., 2006; Rossi et al., 2010; Gomez et al., 2008 and Benhamou et al., 2010, respectively.

[2] One Level 2 and twoLevel 3 studies: Benhamou et al., 2007; and Albisser, 2007; Rossi, 2009, respectively

[3] Two Level 2 studies: Farmer, 2005 (median blood glucose) and Benhamou et al., 2010

[4]  Three Level 2 studies and one Level 3 study: Farmer, 2005 (HbA1c); Jansa et al., 2006; Rossi et al., 2010; and Albisser et al., 2007, respectively.Level 2 studies: Farmer, 2005 (HbA1c); Jansa et al., 2006; Rossi et al., 2010

[5] Albisser et al., 2007; Farmer et al., 2005; Jansa et al., 2006

[6] Role physical, role emotional, general health, and vitality.

Leave a Reply