Table 4:
Association between population characteristics, intervention characteristics, risk-of-bias items and the effect of telemedicine on HbA1C at 4–12 mo
| Variable | No. of trials and within-trial subgroups | Difference in MD (95% CI) | p value | I2 statistic, % |
|---|---|---|---|---|
| Population characteristics | ||||
| Continent | ||||
| North or South America | 47 | 0 (ref) | 65 | |
| Europe | 26 | −0.08 (−0.27 to 0.11) | 0.4 | |
| Asia | 9 | −0.49 (−0.77 to −0.22) | 0.001 | |
| Oceania | 5 | −0.16 (−0.55 to 0.23) | 0.4 | |
| Age (range 24–75 yr) | 83 | 0.003 per 1 yr (−0.005 to 0.01) | 0.4 | 68 |
| Sex, male (range 20%–100%) | 84 | 0.0002 per 1% (−0.005 to 0.005) | 0.9 | 70 |
| Duration of follow-up (range 2.6–24 yr) | 52 | 0.008 per 1 yr (−0.02 to 0.03) | 0.5 | 69 |
| Baseline HbA1C (range 6.4%–10.7%) | 87 | −0.06 per 1% (−0.16 to 0.04) | 0.3 | 68 |
| BMI score (range 23–38) | 62 | 0.02 per 1 score (−0.01 to 0.05) | 0.2 | 71 |
| % using insulin (0%–100%) | 59 | −0.00008 per 1% (−0.004 to 0.003) | 1.0 | 71 |
| % using OHA (range 44%–100%) | 31 | 0.003 per 1% (−0.006 to 0.01) | 0.5 | 72 |
| Type of diabetes mellitus | ||||
| Type 2 | 58 | 0 (ref) | 69 | |
| Type 1 | 11 | 0.05 (−0.22 to 0.33) | 0.7 | |
| Mixed | 9 | 0.20 (−0.09 to 0.50) | 0.2 | |
| Unknown | 9 | 0.13 (−0.14 to 0.41) | 0.3 | |
| Intervention characteristics | ||||
| Patient-to-provider communication | ||||
| Telephone | 14 | 0 (ref) | 69 | |
| Smartphone application | 7 | −0.25 (−0.71 to 0.21) | 0.3 | |
| Web portal | 23 | −0.16 (−0.44 to 0.12) | 0.3 | |
| Smart device | 23 | 0.06 (−0.23 to 0.36) | 0.7 | |
| Provider-to-patient communication | ||||
| Telephone | 51 | 0 (ref) | 67 | |
| SMS text messaging | 12 | −0.28 (−0.52 to −0.05) | 0.02 | |
| Web portal | 20 | −0.35 (−0.56 to −0.14) | 0.001 | |
| CDSS | 27 | 0.10 (−0.08 to 0.28) | 0.3 | |
| Type of provider | ||||
| Nurse | 33 | 0 (ref) | 69 | |
| CDSS | 27 | 0.07 (−0.12 to 0.27) | 0.5 | |
| Diabetes educator | 11 | 0.10 (−0.21 to 0.40) | 0.5 | |
| Physician | 25 | 0.13 (−0.10 to 0.35) | 0.3 | |
| Allied health | 12 | 0.15 (−0.11 to 0.41) | 0.3 | |
| Care manager | 11 | 0.16 (−0.11 to 0.43) | 0.2 | |
| Nonspecialized support | 19 | 0.17 (−0.05 to 0.40) | 0.1 | |
| Frequency of contact | ||||
| Daily | 5 | 0 (ref) | 68 | |
| Weekly | 19 | −0.09 (−0.49 to 0.30) | 0.6 | |
| Every 2 wk | 11 | −0.05 (−0.48 to 0.38) | 0.8 | |
| Monthly | 15 | 0.05 (−0.36 to 0.45) | 0.8 | |
| Less frequently than monthly | 6 | 0.37 (−0.09 to 0.83) | 0.1 | |
| Not reported | 29 | 0.11 (−0.27 to 0.49) | 0.6 | |
| Additional components | ||||
| Interactive | 82 | 0.03 (−0.34 to 0.40) | 0.9 | 68 |
| Medication adjustment | 40 | −0.23 (−0.42 to −0.05) | 0.01 | |
| Exercise | 41 | −0.11 (−0.39 to 0.18) | 0.5 | |
| General education | 65 | −0.21 (−0.44 to 0.02) | 0.1 | |
| Blood pressure management | 8 | −0.002 (−0.31 to 0.30) | 1.0 | |
| Nutrition | 41 | 0.08 (−0.21 to 0.37) | 0.6 | |
| Risk of bias | ||||
| Randomization not described appropriately | 24 | −0.03 (−0.23 to 0.17) | 0.8 | 69 |
| Inadequate or unclear allocation concealment | 60 | −0.07 (−0.25 to 0.11) | 0.5 | 69 |
| Blinding | ||||
| Yes | 18 | 0 (ref) | 69 | |
| No | 12 | 0.12 (−0.19 to 0.43) | 0.4 | |
| Unclear | 57 | 0.15 (−0.08 to 0.38) | 0.2 | |
| Loss to follow-up | ||||
| Reported | 55 | 0 (ref) | 65 | |
| Not reported | 10 | −0.11 (−0.37 to 0.16) | 0.4 | |
| Partially reported | 22 | 0.30 (0.11 to 0.48) | 0.003 | |
| % loss to follow-up (range 0%–39%) | 76 | 0.005 per 1% (−0.006 to 0.02) | 0.4 | 67 |
| No selective reporting | 71 | −0.06 (−0.30 to 0.17) | 0.6 | 69 |
| Funding | ||||
| Public | 45 | 0 (ref) | 69 | |
| Private | 17 | −0.004 (−0.24 to 0.23) | 1.0 | |
| Neither | 13 | 0.01 (−0.24 to 0.26) | 0.9 | |
| Both | 12 | 0.14 (−0.17 to 0.45) | 0.4 | |
| Not intention-to-treat analysis | 40 | −0.14 (−0.31 to 0.04) | 0.1 | 68 |
| Adjustment for potential confounders | 17 | 0.08 (−0.14 to 0.29) | 0.5 | 69 |
Note: BMI = body mass index, CDSS = computer decision support system, CI = confidence interval, HbA1C = glycated hemoglobin, MD = difference in means, OHA = oral hypoglycemic agents, ref = reference category, SMS = short message service.
Categories with < 5 studies were not included in the meta-regression analyses; heterogeneity in the primary analysis was 69%.