A systematic search was conducted targeting the period between January 2008 and April 2020. No protocol has been published. Keywords (diabetes mellitus, telemetry, telemonitoring, and telemedicine) were selected from the MEDLINE Medical Subject Headings and EMBASE Subject Headings databases and searched in titles/abstracts (Multimedia Appendix 1). In general, the steps were as follows: (1) search in five relevant databases, (2) eliminate duplicates, (3) screen titles and abstracts, (4) assess peer-reviewed publications for eligibility, (5) perform additive research via reference lists, (6) select T2DM studies, (7) extract relevant data, and (8) classify the publications.

Publications addressing telemetric interventions targeting T2DM management were included.

Telemetry was defined as “a mode of delivering healthcare services through the use of telecommunications technologies, including but not limited to asynchronous and synchronous technology, and remote patient monitoring technology, by a healthcare practitioner to a patient or a practitioner at a different physical location than the healthcare practitioner” [7]. We included video consultations, telephone counselling, asynchronous communication by email, SMS text messaging, internet/web-based platforms, and mixed forms.

Studies were screened and selected by two independent reviewers. Disagreements were resolved by a consensus-based discussion. We selected studies that met the following inclusion criteria: (1) peer-reviewed articles and studies; (2) written in English or German; (3) study design was an SR, MA, CT, or RCT; and (4) included interventions that involved direct interaction between patients and health care professionals through feedback and data transmission. We also considered quantitative and qualitative studies.

Smartphone/mobile app–based interventions were excluded and analyzed separately in another publication. We also rejected publications that observed mixed populations (eg, pooled patients with T1DM and T2DM), provided pooled data with other digital applications, addressed prevention or diagnosis, or focused on the presentation of technologies. Multimedia Appendix 2 shows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram.

Year of publication, location of the study, duration of the intervention, study design, sample sizes, intervention and control groups used, frequency of contact, feedback methods, outcomes, effects, statistical significance, and conclusions were extracted from each publication (step 7).

For analysis, the interventions were classified a priori (step 8) based on the technologies used, study design, and outcomes (Figure 1).

We conducted a small subgroup MA to assess whether the impact of the four intervention types, as well as the short- and long-term effects on the management of HbA_1c concentrations, differed. To determine the change in HbA_1c, we pooled appropriate RCTs and calculated the differences in means and 95% CIs for the intervention and control groups at the study end points. RCTs in which the changes from baseline to the end of the study were reported as a percentage were included. Studies in which the control group received telemetric support were excluded. Mean deviations and SDs were extracted unchanged.

In addition, the publication bias was assessed visually as a funnel plot using HbA_1c values based on the RCTs and the mean differences (MDs) from our subgroup MA.

We also pooled the number of patients, specifically the number of unique patients as well as the number of patient cases related to the outcomes. In the former scenario, each patient occurred only once, addressing the number of individual patients (without SRs and MAs), and in the latter scenario, with a focus on specific outcomes, patient cases were analyzed based on the respective outcomes and thus may have been included several times (including SRs and MAs).

All SRs and MAs reported clear decreases in HbA_1c values (P<.05) by implementing telemetric interventions [5,11-17]. MAs (5/5, 100%) indicated that telemetry was significantly associated with an obvious improvement between –0.37% and –0.55% in HbA_1c values compared with the usual care (P<.001 [15], P<.001 [13], P<.001 [12], P<.001 [5], and P<.05 [10]).

According to Kim et al [5], telemonitoring was associated with a significantly reduced BMI (weighted MD=–0.25 kg/m², 95% CI –0.49 to –0.01, I²=16.7%) compared with usual care.

Due to the heterogeneous data situation, Zhai et al [13] could not draw any conclusions regarding cost-effectiveness.

Overall, 89% (8/9) of the studies reported a clear reduction in HbA_1c values. More specifically, five RCTs (5/9, 56%) indicated significant positive effects (P=.022 [18], P=.004 [19], P=.023 [20], P<.05 [21], and P=.013 [22]). For example, HbA_1c values declined significantly in an intervention group with weekly video conferences compared with a control group (0.49% versus 0.17%; P=.013) [22].

Overall, definite improvements regarding FBG were documented. Tavsanli et al [22] (weekly video conferences) and Rasmussen et al [20] (average 4.1 video consultations in 6 months) reported clearly lower FBG levels in the intervention group compared with control groups (P>.05 [22] and P<.015 [20]), whereas Hansen et al [18] provided monthly video conferences additional to usual care and reported no substantial changes in FBG levels in relation to the study (significance not reported).

In general, most RCTs observed no essential changes in systolic and diastolic BP measurements (intergroup P>.05 [20] and significance not reported [18]). However, a clear improvement in BP was seen in a 12-month videoconferencing intervention, as reported by Davis et al [19], although the effect was not significant compared with a control group (intervention systolic BP 130.8 mmHg, SD 3.6 mmHg, versus 127.6 mmHg, SD 4.0 mmHg, P=.76; diastolic BP 72.7 mmHg, SD 2.1 mmHg, versus 70.2 mmHg, SD 2.2 mmHg, P=.64).

Rasmussen et al [20] showed a significantly higher weight loss with in-person clinic visits (–1.7 kg) compared with video consultations (–0.6 kg; P=.023).

Hansen et al [18] found no obvious changes in terms of BMI, whereas Davis et al [19] indicated substantial improvements compared to usual care, although the finding was not significant (30.6 kg/m², SD 1.4 kg/m² versus 35.8 kg/m², SD 1.4 kg/m²; P=.73).

Hansen et al [18] noted that significant changes in mental or physical health rankings were not detected.

Gordon et al [23] revealed shorter travel times and less time in waiting rooms according to interviews with participants, although no statistical measurements were performed.

Carlisle and Warren [24] suggested that consumer-friendly technologies and the integration of telemetry into everyday lives are important for the successful implementation of telemetry interventions.

In summary, all studies showed precise improvements in HbA_1c levels with audio interventions in real time. Odnoletkova et al [25] and Walker et al [26] reported significant improvements in their intervention groups compared with matched control groups (intervention group: –0.2%, 95% CI –0.3 to –0.1, P=.003 [25]; and intervention group versus control group: –0.23% versus 0.13%, P=.04 [26]). Sarayani et al [27], Trief et al [28], Maslakpak et al [29], Blackberry et al [30], Benson et al [31], and Vasconcelos et al [32] displayed clear, but not significant, improvements in HbA_1c levels compared with control groups (P>.05). Notably, some control groups [27-29,31] received some forms of telemetric or even educational support. Interestingly, Walker et al [26] found that patients who completed at least six phone calls with a health educator over a 12-month period had significant reductions in their HbA_1c concentrations (P<.05).

The RCT of McMahon et al [10] was classified in two categories (“real-time audio” and “asynchronous” interventions) because it involved the comparison of a telephone-based intervention with two “asynchronous” interventions. HbA_1c decreased at a rate of 0.32% every 3 months for the online arm, 0.36% for the telephone arm, and 0.41% for the web training arm (all P<.001).

The “real-time audio” intervention studies showed obvious improvements in FBG. In the study by Varney et al [33], FBG levels clearly improved in subjects in the intervention group (8.9 mmol/L, 95% CI 8.0 to 9.7, to 8.5 mmol/L, 95% CI 7.7 to 9.4) compared with those in the control group (P=.02), but not in a long-term way, while Maslakpak et al [29] outlined distinct, but not significant, differences between telephone and control groups (P=.766).

In general, all “real-time audio” intervention studies reported clear improvements in BP. Trief et al [28] showed a greater improvement in systolic BP in the “individual calls” group than in the “diabetes education” group at 8 months (P=.021). Vasconcelos et al [32] reported improvements in systolic BP (130.25 mmHg to 125.87 mmHg; P=.171) and diastolic BP (72.12 mmHg to 71.12 mmHg; P=.640). In addition, Varney et al [33] indicated significant improvements in diastolic BP within the telephone group (80 mmHg, 95% CI 76 to 84, to 74 mmHg, 95% CI 71 to 77), but these were not sustained.

According to Odnoletkova et al [25], the difference between the groups in favor of telecoaching was a change in body weight of –1.1 kg (P=.004).

In general, all trials noted slight improvements in BMI. Odnoletkova et al [25] and Trief et al [28] reported significant improvements between groups (P=.003 [25] and P=.021 [28]), whereas Vasconcelos et al [32] indicated a slight decrease (29.99 kg/m² to 29.96 kg/m²) that was not significant (P=.764).

“Real-time audio” interventions appear to be moderate in terms of cost-effectiveness (no statistical significances reported). Schechter et al [34] concluded that the costs were moderate relative to the benefits, whereas Varney et al [35] revealed that the cost of a 10-year intervention was covered by the financial savings, with a tendency for health profits.

Overall, the majority of the studies (all RCTs; 23/34, 96%) reported apparent improvements. Eleven RCTs reported significant improvements in HbA_1c in the intervention groups compared with the control groups (P<.05) [36-46], whereas 3 RCTs showed significant beneficial effects within their intervention groups (P<.05) [47-49]. In addition, 5 RCTs found improvements in the intervention groups compared with matched controls, but the results were not statistically significant (P>.05) [50-54]. Ramadas et al [55], Tildesley et al [38], and Cho et al [56] mentioned significant improvements within their intervention groups (P=.004 [55]; real-time continuous glucose monitoring, P<.001, versus internet blood glucose monitoring system, P<.05 [38]; and P<.01 [56]), but the differences between groups were not significant (P>.05).

The studies showed significant improvements in FBG levels in the intervention groups compared with control groups (8.9 mmol/L, SD 3.9 mmol/L, versus 7.9 mmol/L, SD 2.5 mmol/L, P=.015 [55]; and P=.005 [46]).

In summary, the publications reported apparent improvements in BP. Wild et al [39] found significant improvements in systolic BP (P=.017) and diastolic BP (P=.006) in the intervention group compared with the control group. Wakefield et al [40] also found a significant decrease in systolic BP in the high-intensity arm of the intervention (home telehealth device with algorithm; P=.01). Fang and Deng [44] found improvements in systolic BP (P=.069) and diastolic BP (P=.693) in the treatment group, but they were not significant.

In general, the studies revealed clear beneficial effects of “asynchronous” interventions on both body weight and BMI. For example, Luley et al [41] showed large significant improvements in body weight (–11.8 kg, SD 8.0 kg; both inter- and intragroup comparisons with P=.000) and BMI (–4.1 kg/m²; both intergroup and intragroup comparisons with P=.00).

No clinically important improvements in HRQoL were seen according to Dario et al [51].

A weight-loss telemonitoring intervention from Luley et al 2011 [41] showed an effective decline in medication costs of €83 (US $101) per patient in 6 months.

Cho et al [57] showed a significant time savings for physicians of approximately 55% focusing on patients with HbA_1c levels greater than 6.5% (P<.05).

In general, most publications (21/24, 88%) reported clear significant improvements in HbA_1c (P<.05). Three of these studies were RCTs that achieved significant improvements within their intervention groups (P<.001 [58]; P=.27 [59]; and P value not reported [60]) but not significant differences between the intervention and control groups (P>.05).

Overall, the studies showed mostly positive effects of “combined” interventions on FBG [58,61-64]. For example, Zhou et al [64] and Jeong et al [58] found significant reductions in FBG levels compared with the control groups (8.73 mmol/L to 7.06 mmol/L, P<.001 [64]; and –12.28 mg/dL, SD 41.20 mg/dL, P=.027 (telemedicine group [58]).

Approximately 85% (11/13) of the “combined” intervention studies reported beneficial effects on BP [59-61,63-70]. Kempf et al [70] and Crowley et al [65] found significant improvements compared with control groups (systolic BP, P<.01 [70]; and systolic BP, P=.035, and diastolic BP, P=.013 [65]). However, some RCTs noted improvements in their intervention groups (ie, P<.05) but no significant differences between the groups (P>.05) [60,66-69,71]. Additionally, Kesavadev et al (cohort study with 1000 participants [63]) and Dienstl et al (observational study [61]) indicated similar significant improvements in systolic and diastolic BP (P<.01 [63] and P<.001 [61]).

Most “combination” intervention studies (5/7, 71%) found clear improvements in body weight [60,61,69,70,72]. For example, Kempf et al [70] reported a significant reduction of 6.2 (SD 4.6) kg in the intervention group compared with the control group (–1.0 kg, SD 3.4 kg; P<.01).

The majority of publications (7/9, 78%) showed an apparent reduction of BMI. Significant improvements were outlined by 6 studies (intragroup P=.047 [59], intragroup P<.01 [63], intragroup P<.001 [61], intergroup P=.036 [73], intergroup P<.01 [70], and intergroup P<.05 [74]). For example, Kesavadev et al [63] (n=1000 patients) showed a significant reduction of 0.3 kg/m² (P<.01) and Kempf et al [70] reported –2.1 (SD 1.5) kg/m² in the intervention group versus –0.3 (SD 1.1) kg/m² in the control group (P<.01).

All studies found clear improvements in DRQoL and HRQoL. Kempf et al [70] and Nicolucci et al [69] showed significant intergroup improvements in HRQoL (P<.01 and P<.03, respectively). Jha et al [62] and Dienstl et al [61] (observational studies) reported significant beneficial effects with regard to DRQoL (P=.015 and P<.001, respectively).

Warren et al [75] and Kesavadev et al [63] reported that “combined” interventions are cost-effective. The total costs for the internet-based treatment group were lower than those for the control group (mean US $3781 versus US $4662; P<.001 [75]). According to Kesavadev et al [63], the extra cost was US $9.66/month (significance not reported), but money and time saved in physical visits made up for the extra costs.

Hsu et al [72] reported great time savings with a cloud-based diabetes management program compared with standard face-to-face care (22.5-minute versus 68.8-minute visit time; significance not reported).

All studies reported significant reductions in HbA_1c compared with control subjects (P<.001 [76], P=.005 [77], and P=.013 [78]).

Tang et al [76] detected improvements in BP (systolic BP, P=.306, and diastolic BP, P=.374) but no effects on body weight (P=.232).

A small MA showed that, compared with the control group, 6-month interventions (n=2) were associated with a greater effect size (MD=–1.15%, 95% CI –1.84 to –0.45) than 12-month interventions (n=2) (MD=–0.6%, 95% CI –0.99 to –0.21).

The subgroup analysis revealed an effect size, compared to usual care, of MD=–0.37% (95% CI –0.79 to 0.05) for 6-month interventions (n=3) compared with –0.5% in the 3-month intervention [29] and –0.06% in the 18-month intervention [30].

The greatest effect was seen in 12-month interventions (n=2) (MD=–0.77%, 95% CI –2.25 to 0.72), followed by 6-month interventions (n=8) (MD=–0.57%, 95% CI –0.75 to –0.39), and 3-month interventions (n=3) (MD=–0.38%, 95% CI –0.54 to –0.22).

The 3-month interventions (n=5) had the greatest effect (MD=–0.65%, 95% CI –0.98 to –0.31), whereas 6-month interventions (n=7) had a slightly smaller effect (MD=–0.50%, 95% CI –0.71 to –0.30). In comparison, the effect of 12-month interventions (n=4) was even smaller (MD=–0.25%, 95% CI –0.73 to 0.24). The subgroup “video clip” interventions (n=2) showed a reduction of MD=–0.23 (95% CI –0.23 to –0.23).