The Effectiveness of Telemedicine Solutions for the Management of Type 2 Diabetes: A Systematic Review,Meta-Analysis,and Meta-Regression

Abstract

Background:

Previous systematic reviews have aimed to clarify the effect of telemedicine on diabetes. However, such reviews often have a narrow focus, which calls for a more comprehensive systematic review within the field. Hence, the objective of the present systematic review, meta-analysis, and meta-regression is to evaluate the effectiveness of telemedicine solutions versus any comparator without the use of telemedicine on diabetes-related outcomes among adult patients with type 2 diabetes (T2D).

Methods:

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. We considered telemedicine randomized controlled trials (RCT) including adults (≥18 years) diagnosed with T2D. Change in glycated hemoglobin (HbA1c, %) was the primary outcome. PubMed, EMBASE, and the Cochrane Library Central Register of Controlled Trials (CENTRAL) were searched on October 14, 2020. An overall treatment effect was estimated using a meta-analysis performed on the pool of included studies based on the mean difference (MD). The revised Cochrane risk-of-bias tool was applied and the certainty of evidence was graded using the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach.

Results:

The final sample of papers included a total of 246, of which 168 had sufficient information to calculate the effect of HbA1c%. The results favored telemedicine, with an MD of −0.415% (95% confidence interval [CI] = −0.482% to −0.348%). The heterogeneity was great (I² = 93.05%). A monitoring component gave rise to the higher effects of telemedicine.

Conclusions:

In conclusion, telemedicine may serve as a valuable supplement to usual care for patients with T2D. The inclusion of a telemonitoring component seems to increase the effect of telemedicine.

Introduction

Diabetes is one of the most common chronic diseases and a major health care problem worldwide.^1,2 In 2017, an estimated 8.4% of the adult global population had diabetes, which is expected to increase to approximately 9.9% (700 million) in 2045, primarily due to an increase in unhealthy dietary habits, obesity, and sedentary lifestyle.^1,2 The global prevalence is predicted to increase mostly in low- and middle-income countries. ³ The economic impact of diabetes is considerable, spans health care services, and affects individuals, families, and national productivity.^1,4

Type 2 diabetes (T2D) constitutes approximately 90% to 95% of diabetes cases.^1,5 The T2D is a progressive disease associated with significant premature mortality, morbidity, and several complications, such as cardiovascular disease, nephropathy, neuropathy, and retinopathy.^6,7 Diabetes patient care is a multifaceted and complex process, mainly aimed at attaining optimal glycemic control to prevent and control diabetes-related complications. ⁸ However, sustaining optimal glycemic control for people with diabetes is both demanding and challenging because it requires numerous daily self-management decisions and care activities. ⁹ These challenges include estimating the appropriate diabetes medication dosage to avoid hypoglycemic and hyperglycemic events and adherence to obstacles in terms of following the recommended guidelines.^8,9 Adherence to the complex T2D treatment regimen is difficult to maintain ¹⁰ and medial adherence rates have been reported to range from 36% to 93%. ¹¹

Self-management strategies are considered an essential part of diabetes treatment and are associated with improvements in health-related outcomes. ⁹ A potential solution to support ongoing diabetes self-management support is the use of telemedicine, ¹² which has been suggested as a promising but unproven approach to support people with diabetes in the management of their disease. ¹³ Telemedicine can be defined as the delivery of health care services over a distance using information and communication technologies. ¹⁴ However, no definitive definition exists. ¹⁴ Telemedicine solutions may include a variety of different technologies and various delivery forms, including monitoring, education, consultative services, coaching, and counseling tasks.^13,15-17 Telemedicine interventions constitute different constellations, such as simple reminders via text messaging, video consultation, and transmission of patient data (eg, blood glucose, blood pressure, dietary and medication intake, and physical activity) with feedback from health care professionals via web portals or via telephone.^13,15,16 As diabetes predominantly needs to be managed outside health care facilities and to a large extent requires self-management, telemedicine holds the potential to provide sufficient self-management support to people with T2D.^18,19

Previous systematic reviews have aimed to clarify the effect of telemedicine on diabetes.^15-17,20-28 However, these previous reviews have often focused on a specific type of telemedicine, a specific outcome, and/or a specific comparator, which calls for a more comprehensive and inclusive systematic review seeking to compare and synthesize findings for treatment outcomes while adjusting for different study characteristics. In addition, the field of telemedicine is developing rapidly; thus, a large number of studies likely have been published recently, calling for an updated review. ²⁹ Hence, the objective of the present systematic review, meta-analysis, and meta-regression was to evaluate the effectiveness of telemedicine solutions versus any comparator without the use of telemedicine on diabetes-related outcomes among adult patients with T2D.

Methods

Study Design

This systematic review, meta-analysis, and meta-regression was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. ³⁰ A comprehensive search protocol was published elsewhere ³¹ and PROSPERO-registered with identification CRD42020123565 on April 2020. The search was part of a comprehensive search that included T2D as well as type 1 diabetes (T1D) and gestational diabetes ³¹ , which will be reported separately.

Eligibility Criteria

Studies were considered if they included adults (≥18 years) diagnosed with T2D. Studies that included mixed populations (eg, T1D and T2D) were only included if the data for the T2D population were reported separately. Studies were excluded if only participants at risk of diabetes or participants with prediabetes were included. Furthermore, the studies must have included telemedicine interventions that were substituted for usual practice or served as an alternative to usual practice. The telemedicine interventions had to include remote feedback/communication between a patient and health care professional(s). Alternatively, feedback/communication could occur between the patient and a trained peer. Telemedicine interventions that were wholly automatic were also considered.

Studies that reported on any relevant patient diabetes-related outcome were included. The primary outcome was change in glycated hemoglobin (HbA1c, %).

We only considered randomized controlled trials (RCTs)—both parallel and crossover designs. Studies published in English, Danish, Norwegian, and Swedish as peer-reviewed full-text papers were included. All studies published before October 14, 2020, were considered.

Information Sources

The search was performed in PubMed, EMBASE, the Cochrane Library Central Register of Controlled Trials (CENTRAL), and CINAHL. Two authors performed the database searches (S.H.L. and S.H.). A research librarian assisted. Additional citation searches were performed in the Web of Science, SCOPUS, and Google Scholar.

Search Strategy

Initially, an unstructured search was performed in PubMed, CINAHL, and Google Scholar to identify relevant search terms and thus qualify the systematic search. The systematic search followed the initial search. The search strategy was adapted for each database. The search terms included various synonyms, near-synonyms, acronyms, and spellings for all keywords and index terms. A variety of search functions were applied, including thesaurus, Boolean operators, abstract/title/keywords, phrase, truncation, free text, and advanced search. Citation searches were applied to identify additional studies.

Selection Process

First, all of the identified papers were uploaded into RefWorks (ProQuest RefWorks 2.0, 2010). Second, duplicates were removed using the functions Exact duplicates and Close duplicates. Third, titles and abstracts were screened by two authors with respect to the eligibility criteria of the review (S.H. and S.H.L.). Fourth, the remaining studies underwent full review by three authors with respect to the eligibility criteria of the review (S.H., S.H.L., and J.D.A.). Disagreement between the authors was resolved through discussion by the three authors alone or by inclusion of other authors. During the full review, the reasons for exclusion of studies were recorded, and afterward, a final sample of papers was identified.

Data Extraction

Data were extracted by three independent authors (S.H., S.H.L., and J.D.A.) using a standardized sheet in Microsoft Excel (2016). Extracted data included trial characteristics (author, publication year, country, sample size, and study duration), patient characteristics (age, sex, and body mass index [BMI]), and HbA1c outcomes. In addition, the characteristics of the telemedicine interventions were extracted, including the frequency of contact, implementation setting (primary care, hospital, specialized outpatient clinic, university, community or cross-sectional), peripherals (eg, glucometers, pedometers, blood pressure monitors, and scales), and the general purpose of intervention components (monitoring, consultation, counseling, coaching, education, mentoring, and reminding). Disagreements between the authors were resolved through discussion. Additional authors were included in the discussions when necessary.

Risk of Bias Assessment

The revised Cochrane risk-of-bias tool was applied. ³² Four reviewers (J.D.A., S.H., T.K., and F.W.U.) assessed the included studies independently and resolved potential disagreements by discussion.

Data Synthesis

All statistical analyses were performed in Stata 16 (Stata Statistical Software: Release 16, StataCorp 2019.; StataCorp LLC, College Station, Texas). Reported medians, interquartile ranges, ranges, and confidence intervals were transformed to means and standard deviations by traditional methods^33,34 and scaled to HbA1c% when relevant (eg, if outcome was reported as mmol/mol). An overall treatment effect was estimated with a meta-analysis of the pool of included studies based on the mean difference (MD). Heterogeneity was assessed statistically using I² tests. The results were combined with a random-effects model (due to heterogeneity, ie, an I² statistic > 50%). Univariate a priori subgroup analyses based on meta-regression of the telemedicine characteristics were conducted and combined with post hoc analyses of the association of study and patient characteristics with the treatment effect of telemedicine. Publication bias was evaluated using visual inspection of the funnel plot and Egger test.

Certainty Assessment

The Grading of Recommendations, Assessment, Development and Evaluations (GRADE) approach was applied. A summary of findings (SoF) table was created using GRADEPro GDT 2015 (McMaster University, Ontario, Canada),^35,36 which presents the absolute risks for the groups (treatment and control), estimates of relative risk, and ranking of the quality of the evidence, which is based on the risk of bias, ³⁷ indirectness, ³⁸ imprecision, ³⁹ inconsistency, ⁴⁰ and risk of publication bias. ⁴¹

Results

The flowchart in Figure 1 describes our selection of studies. In the literature search, 16 309 studies were found and 1125 eligible studies were assessed by full-text reading after screening titles and abstracts. After full-text assessment, 246 articles met our inclusion criteria.

Figure 1.

Flowchart.

The characteristics of the individual studies are presented in Table 1. One study was multinational, 88 articles were conducted in North America (36%), 84 in Asia (35%), 44 in Europe (18%), 16 in Australia/New Zealand (7%), five in Africa (2%), and five in South America (5%). Four studies were published before 2000 (2%), 50 in the 2000s (21%), 165 in the 2010s (68%), and 24 (10%) in 2020. Sample sizes ranged from 17 to 4078, with an average of 251 participants per study. The study duration ranged from one to 96 months, with a study average of 8.5 months. Across studies, the mean proportion of men was 50.4% (range = 15%-100%), the average age at baseline was 57 (range = 37-73) years, and the baseline mean BMI was 30 (range = 22-40). The average baseline HbA1c% levels ranged from 5.70% to 11.05%, with an average of 8.33%.

Table 1.

Study and Participant Characteristics.

Study	Publication year	Country	Sample size	Duration (months)	Mean age years	Male %	Baseline BMI	Baseline HbA1c
Abaza and Marschollek 42	2017	Egypt	90	3	52	44	NA	9.66
Agarwal et al 43	2019	Canada	50	4	64	55	NA	7.44
Agarwal et al 44	2019	Canada	240	3	52	52	NA	8.96
Aguiar et al 45	2018	Brazil	80	12	62	67	NA	8.95
Akinci et al 46	2018	Turkey	66	2	52	36	32	8.34
Al Omar et al 47	2020	United Arab Emirates	218	6	42	42	NA	8.45
Alanzi et al 48	2018	Saudi Arabia	20	6	NA	75	NA	8.53
Albikawi et al 49	2016	Jordan	168	3	51	46	NA	NA
Alghafri et al 50	2018	Oman	232	3	44	43	33	7.96
Ali et al 51	2016	India & Pakistan	1146	30	54	46	27	9.90
Ali et al 52	2020	India	404	24	53	41	27	9.15
Aliha et al 53	2013	Iran	61	3	53	50	28	9.70
Alonso-Domínguez et al 54	2019	Spain	204	3	61	54	30	6.85
Alotaibi et al 55	2016	Saudi Arabia	20	6	45	NA	NA	8.55
Anderson et al 56	2009	USA	310	24	56	41	34	7.48
Anderson et al 57	2010	USA	295	12	NA	42	35	8.00
Anderson-Loftin et al 58	2005	USA	97	6	57	23	35	7.90
Andreae et al 59	2020	USA	230	3	59	20	NA	8.29
Anzaldo-Campos et al 60	2016	Mexico	301	10	52	38	31	11.05
Arora et al 61	2014	USA	128	6	51	36	NA	10.10
Asante et al 62	2020	Ghana	60	3	56	22	28	8.95
Avdal et al 63	2011	Turkey	122	6	52	49	NA	8.07
Azizi et al 64	2016	Iran	72	4	55	50	NA	NA
Benson et al 65	2018	USA	120	12	60	55	37	8.20
Blackberry et al 66	2013	Australia	473	18	63	57	NA	8.06
Bluml et al 67	2019	USA	446	3	54	41	37	10.25
Boels et al 12	2019	Netherlands	230	6	59	60	32	8.20
Bogner et al 68	2012	USA	182	3	57	32	34	7.10
Bohingamu Mudiyanselage et al 69	2018	Australia	177	12	70	53	NA	5.70
Lashkari et al 70	2013	Iran	50	3	NA	NA	29	9.68
Browning et al 71	2016	China	711	12	64	52	26	10.45
Bujnowska-Fedak et al 72	2011	Poland	100	6	55	54	25	7.65
Buysse et al 73	2019	Belgium	153	24	37	50	NA	8.30
Capozza et al 74	2015	USA	156	6	53	37	NA	9.11
Carter et al 75	2011	USA	47	9	51	36	36	8.91
Castelnuovo et al 76	2011	Italy	72	12	52	52	NA	NA
Chao et al 77	2019	Taiwan	121	3	NA	NA	25	8.70
Chen et al 78	2008	Taiwan	78	6	59	44	26	9.61
Chen et al 79	2018	China	233	NA	59	51	NA	NA
Chiu et al 80	2016	Taiwan	182	8	65	52	26	7.65
Cho et al 81	2006	Korea	80	30	53	54	23	7.60
Cho et al 82	2017	Korea	484	6	53	64	26	7.84
Choe et al 83	2005	USA	80	14	52	47	NA	10.15
Choudhry et al 84	2018	USA	4078	12	60	55	NA	9.65
Clark et al 85	2004	UK	100	12	60	58	31	8.42
Crowley et al 86	2013	USA	369	12	61	28	NA	8.00
Crowley et al 87	2016	USA	50	6	60	96	NA	10.45
Dale et al 88	2009	UK	231	6	NA	60	NA	8.56
Dario et al 89	2017	Italy	299	12	73	56	NA	7.94
Davis et al 90	2010	USA	165	12	60	25	37	8.96
Del Prato et al 91	2012	Italy	291	5.5	58	52	30	8.86
Delahanty et al 92	2019	USA	211	12	62	45	35	7.70
Doupis et al 7	2019	Greece	457	8	63	51	31	7.85
Dugas et al 93	2018	USA	27	3	NA	89	NA	9.23
Duruturk and Özköslü 94	2019	Turkey	50	1.5	53	NA	31	7.36
Döbler et al 95	2018	Germany	249	12	52	70	36	7.70
Eakin et al 96	2013	Australia	302	6	58	56	33	7.45
Eakin et al 97	2014	Australia	302	24	58	56	33	7.45
Egede et al 98	2017	USA	113	6	54	19	36	10.10
Egede et al 99	2018	USA	90	12	63	98	NA	7.11
Estey et al 100	1990	Canada	60	4	NA	46	NA	6.21
Faridi et al 101	2008	USA	30	3	56	37	36	6.45
Farsaei et al 102	2011	Iran	172	3	53	34	NA	9.10
Fernandes et al 103	2016	Brazil	219	6	63	31	29	8.07
Fortmann et al 104	2017	USA	126	6	48	25	32	9.55
Fottrell et al 105	2019	Bangladesh	2470	18	NA	47	22	NA
Fountoulakis et al 106	2015	Greece	80	12	62	69	30	9.80
Franc et al 107	2020	France	665	12	39	48	26	9.10
Franciosi et al 108	2011	Italy	62	6	49	74	31	7.90
Frosch et al 109	2011	USA	201	6	55	52	33	9.60
García et al 110	2015	USA	72	6	50	33	36	8.60
Garg et al 111	2017	USA	184	12	64	60	35	9.05
Gagliardino et al 112	2013	Argentina	198	12	61	49	33	7.21
Wasif Gillani 113	2016	Malaysia	150	6	53	55	28	9.89
Gimbel et al 114	2020	USA	240	12	63	62	33	7.55
Glasgow et al 115	2006	USA	327	2	62	50	32	7.45
Glasgow et al 116	2006	USA	327	2	61	55	33	7.25
Glasgow and Toobert 117	2000	USA	320	6	59	39	NA	7.60
Glasgow et al 118	2002	USA	320	12	59	39	NA	7.59
Glasgow et al 119	2005	USA	886	12	63	49	NA	7.32
Goodarzi et al 120	2012	Iran	100	3	54	22	28	7.87
Goodarzi et al 121	2018	Iran	76	3	57	100	NA	NA
Graziano and Gross 122	2009	USA	120	3	62	55	NA	8.65
Griffin et al 123	2014	UK	478	12	60	62	33	7.12
Gupta et al 124	2020	India	81	4	50	52	28	8.45
Haider et al 125	2019	Australia	229	6	59	83	31	NA
Hansen et al 126	2017	Denmark	165	8	58	64	34	9.30
Hare et al 127	2011	Australia	223	33	55	56	32	7.55
Hee-Sung 128	2007	South Korea	60	3	50	42	24	6.82
Heisler et al 129	2019	USA	290	6	63	98	NA	9.10
Hidrus et al 130	2020	Malaysia	100	3	NA	50	NA	NA
Hokanson et al 131	2006	USA	114	6	54	57	33	8.60
Holbrook et al 132	2009	Canada	511	6	61	51	32	7.05
Holmen et al 133	2014	Norway	151	12	57	55	31	8.25
Hordern et al 134	2009	Australia	223	12	56	55	32	7.55
Huang et al 135	2019	Singapore	50	3	52	49	29	8.65
Huizinga et al 136	2010	USA	165	24	56	60	34	6.64
Hunt et al 137	2014	USA	17	3	NA	41	NA	6.59
Iljaž et al 138	2017	Slovenia	120	12	55	30	32	6.95
Islam et al 139	2019	Bangladesh	239	6	48	46	NA	NA
Jahangard-Rafsanjani et al 140	2015	Iran	101	5	57	50	29	7.56
Jain et al 141	2018	India	299	6	57	57	24	8.16
Jarab et al 142	2012	Jordan	171	6	64	57	33	8.45
Jennings et al 143	2014	Australia	436	8	58	52	33	NA
Jeong et al 144	2018	Korea	338	6	53	67	25	8.30
Jiwani et al 145	2020	USA	26	6	58	30	39	9.30
Kardas et al 146	2016	Poland	62	1.5	59	60	31	6.81
Kassavou et al 147	2020	UK	135	3	NA	54	NA	NA
Kempf et al 148	2017	Germany	202	12	59	54	36	8.30
Keogh et al 149	2011	Ireland	121	6	59	63	32	9.18
Kim and Utz 150	2019	South Korea	155	3	51	48	NA	9.14
Kim 151	2007	Korea	60	3	47	43	24	7.84
Kim and Jeong 152	2007	Korea	60	6	47	43	24	7.84
Kim and Song 153	2008	Korea	40	12	47	47	25	7.85
Kim and Kim 154	2008	Korea	40	6	47	47	25	7.85
Kim and Oh 155	2003	Korea	50	3	60	30	24	8.50
Kim et al 156	2005	Korea	35	3	61	36	24	8.60
Kim and Kang 157	2006	Korea	73	3	55	53	NA	7.94
Kim et al 158	2009	USA	83	6.5	56	56	26	9.25
Kim et al 159	2010	Korea	100	3	48	50	24	9.80
Kim et al 160	2015	Korea	70	6	66	50	25	8.55
Kim et al 161	2016	China	220	6	54	48	26	7.95
King et al 162	2006	USA	335	2	61	49	32	NA
Kirkman et al 163	1994	USA	275	12	64	99	NA	10.70
Kleinman et al 164	2017	India	91	6	48	70	29	9.25
Krein et al 165	2004	USA	246	18	61	97	NA	9.25
Ku et al 166	2020	Australia	40	3	50	35	28	8.95
Kusnanto et al 167	2019	Indonesia	30	3	NA	43	NA	8.46
Kwon et al 168	2004	Korea	110	3	54	61	24	7.39
Lauffenburger et al 169	2019	USA	1400	12	55	63	NA	9.35
Lazo-Porras et al 170	2020	Peru	172	18	61	37	28	8.55
Lee et al 171	2017	Malaysia	85	3	53	51	30	8.74
Lee et al 172	2020	South Korea	72	6	NA	NA	26	7.44
Lee et al 173	2019	Malaysia	240	12	56	54	NA	9.00
Li et al 174	2016	China	53	6	62	53	24	7.73
Liebreich et al 175	2009	Canada	49	3	54	41	34	NA
Lim et al 176	2011	Korea	154	6	68	21	25	7.85
Lim et al 177	2016	Korea	100	6	65	75	26	8.00
Lorig et al 178	2010	USA	761	6	54	29	NA	6.41
Lujan et al 179	2007	USA	150	6	58	22	NA	7.96
Luley et al 180	2011	Germany	70	6	58	49	35	7.55
Lutes et al 181	2017	USA	200	12	53	NA	38	9.09
Lynch et al 182	2014	USA	61	6	54	33	36	7.65
MacPhail et al 183	2014	Australia	87	4	68	NA	31	NA
Marios et al 184	2012	Australia	39	6	63	53	33	7.73
Maslakpak et al 185	2017	Iran	90	3	50	60	29	8.00
Mayberry et al 186	2020	USA	379	6	57	46	NA	8.13
McEwen et al 187	2017	USA	157	9	54	35	33	9.93
McKay et al 188	2001	USA	78	2	52	47	NA	NA
McKay et al 189	2002	USA	160	3	59	45	NA	7.48
McKee et al 190	2011	USA	55	6	60	33	33	8.22
McMahon et al 191	2012	USA	152	12	61	93	34	9.85
Mease 192	2000	USA	28	3	63	39	NA	9.50
Mons et al 193	2013	Germany	204	18	68	61	NA	8.10
Moriyama et al 194	2009	Japan	75	12	66	46	NA	7.47
Mwangi et al 195	2020	Kenya	104	3	62	32	25	NA
Namjoo Nasab et al 196	2017	Iran	64	3	52	48	27	NA
Nesari et al 197	2010	Iran	61	3	51	28	28	9.26
Nicolucci et al 198	2015	Italy	302	12	58	65	29	7.95
Niswender et al 199	2014	Several	611	6	57	51	34	7.95
O’Neil et al 200	2016	USA	563	12	NA	29	NA	8.32
Odegard and Christensen 201	2012	USA	165	12	63	48	NA	NA
Odnoletkova et al 202	2016	Belgium	574	18	63	62	30	7.00
Oh et al 203	2003	Korea	50	3	61	36	25	8.55
Orsama et al 204	2013	Finland	56	10	62	54	32	6.98
Owolabi et al 205	2019	South Africa	216	6	NA	16	32	NA
Parsons et al 206	2019	UK	446	12	62	57	33	8.60
Patja et al 207	2012	Finland	1535	12	65	57	32	7.57
Peasah et al 208	2019	USA	78	3	62	53	35	8.20
Peimani et al 209	2016	Iran	150	3	52	53	28	7.41
Piette et al 210	2011	USA	339	12	56	49	38	7.60
Plotnikoff et al 211	2013	Canada	287	12	62	51	30	7.16
Presley et al 212	2020	USA	120	6	55	28	35	9.98
Quinn et al 213	2017	USA	142	12	52	51	35	9.59
Quinn et al 214	2011	USA	213	12	53	51	35	9.24
Raj and Mathews 215	2020	India	50	6	69	54	NA	10.26
Ralston et al 216	2009	USA	83	12	57	51	NA	8.05
Ramadas et al 217	2018	Malaysia	132	12	51	69	NA	9.00
Rasmussen et al 218	2016	Denmark	40	6	63	68	32	8.55
Rodríguez-Idígoras et al 219	2009	Spain	328	12	64	52	NA	7.51
Ruggiero et al 220	2014	USA	270	12	53	32	33	8.63
Sacco et al 221	2009	USA	62	6	52	42	36	8.50
Sacco et al 222	2012	USA	62	6	52	42	36	8.50
Samuel-Hodge et al 223	2009	USA	201	12	59	36	35	7.78
Sarayani et al 224	2018	Iran	100	9	55	58	30	7.95
Saslow et al 225	2020	USA	64	12	NA	NA	NA	NA
Sazlina et al 226	2015	Malaysia	69	8	64	57	27	8.20
Schillinger et al 227	2009	USA	339	12	56	39	31	9.35
Shahid et al 228	2015	Pakistan	440	4	49	61	27	9.97
Shahsavari and Bakhshandeh Bavarsad 229	2020	Iran	60	3	NA	15	29	8.93
Shetty et al 230	2011	India	215	12	50	NA	27	9.00
Shreck et al 231	2014	USA	526	12	56	33	NA	8.65
Sigurdardottir et al 232	2009	Iceland	58	6	61	68	32	7.99
Skelly et al 233	2009	USA	180	9	67	NA	NA	8.37
Sone et al 234	2002	Japan	2205	36	59	55	23	7.74
Sone et al 235	2010	Japan	2033	96	59	46	23	7.85
Song and Kim 236	2009	Korea	49	3	50	43	25	9.20
Spencer et al 237	2018	USA	222	6	49	39	33	7.93
Steventon et al 238	2014	England	513	12	65	58	31	8.42
Sun et al 239	2019	China	91	6	68	46	23	7.86
Sunil Kumar et al 240	2020	India	300	6	65	60	NA	7.60
Takenga et al 241	2014	Democratic Republic of Congo	40	2	NA	73	NA	8.63
Tamban et al 242	2013	Philippines	104	6	50	27	28	7.84
Tan et al 243	2018	Singapore	142	2	62	NA	NA	9.72
Tang et al 244	2013	USA	415	12	54	60	NA	9.26
Teston et al 245	2017	Brazil	134	5	NA	32	NA	NA
Thom et al 246	2013	USA	299	6	55	48	34	9.99
Torbjørnsen et al 247	2014	Norway	164	4	58	59	32	7.85
Tu et al 248	1993	USA	31	3	65	33	NA	NA
Van Dyck et al 249	2013	Belgium	92	12	62	69	30	7.30
Van Dyck et al 250	2011	Belgium	92	12	62	NA	30	NA
Van Vugt et al 251	2016	The Netherlands	132	6	68	53	30	6.59
Varney et al 252	2014	Australia	94	12	62	68	32	8.35
Vaughan et al 253	2020	USA	89	6	55	28	34	8.86
Vervloet et al 254	2014	The Netherlands	604	24	55	55	NA	NA
Vinithia et al 255	2019	India	248	24	43	68	27	9.50
Von Storch et al 256	2019	Germany	2441	3	59	81	31	6.99
Wakefield et al 257	2011	USA	302	12	68	94	33	7.15
Wakefield et al 258	2012	USA	302	12	68	94	33	7.15
Wakefield et al 259	2014	USA	108	6	60	44	NA	7.30
Waki et al 260	2014	Japan	54	3	57	76	27	7.05
Walker et al 261	2011	USA	527	12	56	33	31	8.65
Wang et al 262	2020	Mongolia	171	12	55	48	26	9.15
Wang et al 263	2019	China	120	6	45	32	NA	8.65
Wang et al 264	2017	China	212	6	54	55	25	7.95
Warren et al 265	2018	Australia	157	6	61	55	34	8.27
Weinberger et al 266	1995	USA	275	12	64	99	NA	10.70
Whittemore et al 267	2004	USA	53	6	58	NA	35	7.65
Wichit et al 268	2017	Thailand	140	3	58	27	27	6.65
Wild et al 269	2016	UK	321	9	61	67	33	8.85
Williams et al 270	2012	Australia	120	6	57	65	33	8.55
Williams et al 271	2017	New Zealand	138	6	55	38	40	8.15
Wolever et al 272	2010	USA	56	6	53	23	NA	7.93
Wolf et al 273	2004	USA	147	12	53	40	38	7.70
Wu et al 274	2017	Australia & Taiwan	181	1	66	61	NA	NA
Yang et al 275	2020	South Korea	401	3	56	51	26	7.96
Yasmin et al 276	2020	Bangladesh	320	12	52	23	NA	NA
Yoo et al 277	2009	Korea	123	3	58	59	26	7.50
Yoon and Kim 278	2008	South Korea	60	12	47	43	24	7.83
Young et al 279	2005	UK	591	12	67	58	30	7.93
Yu et al 280	2019	China	185	6	52	28	26	8.60
Zamanzadeh et al 281	2017	Iran	66	3	49	41	NA	NA
Zhou et al 282	2014	China	114	3	NA	NA	24	8.33

Abbreviations: BMI, body mass index; HbA1c, glycated hemoglobin; NA, not available.

Table 2 illustrates the telemedicine technologies implemented across studies. Seventy-one studies (29%) were conducted in a hospital setting, 58 studies (24%) in the primary care sector, 37 in communities (15%), 34 in specialized outpatient clinics (14%), 14 in a university setting (6%), and four in cross-sectorial implementations (2%). The frequency of contact with patients across studies was daily (30 studies, 12%), weekly (83 studies, 34%), every two weeks (26 studies, 11%), monthly (40 studies, 16%), and more rarely (12 studies, 5%). Twenty-four (10%) of the studies reported a “tailored” number of contacts with participants (10%). Across studies, the explicitly reported peripherals were scales (13 studies, 5%), glucometers (45 studies, 19%), blood pressure monitors (19 studies, 8%), and pedometers (16 studies, 7%).

Table 2.

Telemedicine Intervention Characteristics.

Study	Publication year	Setting	Frequency of contact	Included peripherals				Intervention components
				Glucometer	Pedometer	BP monitor	Scale	Monitoring	Consultation	Counseling	Coaching	Education	Mentoring	Reminding
Abaza and Marschollek 42	2017	Hospital	Daily	x			x					x		x
Agarwal et al 43	2019	Community	Weekly									x
Agarwal et al 44	2019	Hospital	Daily									x		x
Aguiar et al 45	2018	Hospital	Once						x
Akinci et al 46	2018	University												x
Al Omar et al 47	2020	Primary	Daily									x
Alanzi et al 48	2018	Community		x				x
Albikawi et al 49	2016	Specialized	Once						x
Alghafri et al 50	2018	Primary	Monthly									x
Ali et al 51	2016	Specialized	Monthly						x
Ali et al 52	2020	Hospital	Weekly					x
Aliha et al 53	2013	Specialized	Weekly						x
Alonso-Domínguez et al 54	2019	Primary	Daily					x
Alotaibi et al 55	2016	Hospital	Weekly	x				x				x
Anderson et al 56	2009	Community	Monthly								x
Anderson et al 57	2010	Community	Tailored						x		x
Anderson-Loftin et al 58	2005	Primary	Weekly								x	x
Andreae et al 59	2020	Community	Weekly									x
Anzaldo-Campos et al 60	2016	Primary		x				x				x
Arora et al 61	2014	Hospital	Daily									x		x
Asante et al 62	2020	Specialized	Weekly									x
Avdal et al 63	2011	University		x				x
Azizi et al 64	2016	Specialized						x						x
Benson et al 65	2018	Primary	Monthly								x	x
Blackberry et al 66	2013	Primary	Monthly								x
Bluml et al 67	2019	Primary	Weekly						x			x
Boels et al 12	2019	Hospital	Daily									x
Bogner et al 68	2012	Primary	Twice					x				x
Bohingamu Mudiyanselage et al 69	2018	Community	Daily					x
Lashkari et al 70	2013		Weekly							x
Browning et al 71	2016	Community	Monthly								x
Bujnowska-Fedak et al 72	2011	Primary	Weekly	x				x
Buysse et al 73	2019	Hospital	Monthly					x
Capozza et al 74	2015	Primary	Daily									x		x
Carter et al 75	2011	Primary	Weekly	x		x	x	x	x
Castelnuovo et al 76	2011	Hospital	Fortnight		x			x		x
Chao et al 77	2019	Hospital										x
Chen et al 78	2008	Hospital	Weekly											x
Chen et al 79	2018	Hospital	Weekly									x
Chiu et al 80	2016	Community	Weekly								x
Cho et al 81	2006	Hospital	Weekly					x
Cho et al 82	2017	Specialized	Weekly	x		x		x
Choe et al 83	2005	University	Monthly									x
Choudhry et al 84	2018	Primary							x
Clark et al 85	2004	Specialized	Fortnight								x
Crowley et al 86	2013	Primary	Monthly									x
Crowley et al 87	2016	Hospital	Fortnight					x	x
Dale et al 88	2009	Primary	Tailored						x				x
Dario et al 89	2017		Tailored	x				x
Davis et al 90	2010	Community	Monthly		x							x
Del Prato et al 91	2012			x				x
Delahanty et al 92	2019	Community	Weekly									x
Doupis et al 7	2019	Specialized	Weekly									x
Dugas et al 93	2018	Specialized	Daily		x			x					x
Duruturk and Özköslü 94	2019	Hospital	Weekly					x
Döbler et al 95	2018	Specialized	Monthly							x
Eakin et al 96	2013	Primary	Fortnight		x		x			x
Eakin et al 97	2014	Primary	Fortnight		x		x			x
Egede et al 98	2017	Community	Weekly	x		x		x
Egede et al 99	2018	Community
Estey et al 100	1990	University	Fortnight							x		x
Faridi et al 101	2008	Community	Daily					x						x
Farsaei et al 102	2011	Specialized	Weekly									x
Fernandes et al 103	2016	Primary	Monthly								x
Fortmann et al 104	2017	Community	Daily					x				x		x
Fottrell et al 105	2019	Community	Weekly									x
Fountoulakis et al 106	2015	Hospital	Tailored	x				x
Franc et al 107	2020	Primary	Daily					x
Franciosi et al 108	2011	Specialized	Monthly	x								x
Frosch et al 109	2011	Primary	Tailored								x
García et al 110	2015	Community	Fortnight	x							x
Garg et al 111	2017	Hospital	Weekly	x						x
Gagliardino et al 112	2013	Specialized	Tailored										x
Wasif Gillani 113	2016	Hospital	Twice	x				x
Gimbel et al 114	2020	Primary	Daily			x		x
Glasgow et al 115	2006	Primary care	Twice								x
Glasgow et al 116	2006		Twice								x
Glasgow and Toobert 117	2000	Specialized	Fortnight							x
Glasgow et al 118	2002	Specialized	Fortnight						x
Glasgow et al 119	2005	Primary care								x
Goodarzi et al 120	2012		Tailored									x
Goodarzi et al 121	2018	University	Weekly									x
Graziano and Gross 122	2009	Cross-sectional	Daily					x		x
Griffin et al 123	2014	Primary	Monthly								x
Gupta et al 124	2020	Hospital	Weekly								x	x
Haider et al 125	2019	Hospital	Weekly									x
Hansen et al 126	2017	Hospital	Monthly					x	x
Hare et al 127	2011	Hospital	Tailored							x
Hee-Sung 128	2007	Hospital	Monthly					x				x		x
Heisler et al 129	2019	Hospital	Monthly									x
Hidrus et al 130	2020	Hospital	Daily								x
Hokanson et al 131	2006	Specialized	Tailored							x
Holbrook et al 132	2009	Primary												x
Holmen et al 133	2014	Primary	Monthly					x		x
Hordern et al 134	2009	Hospital	Tailored					x		x
Huang et al 135	2019	Specialized	Daily									x
Huizinga et al 136	2010	University	Monthly							x
Hunt et al 137	2014	Specialized	Weekly					x
Iljaž et al 138	2017	Primary	Tailored					x		x				x
Islam et al 139	2019	Hospital	Daily								x
Jahangard-Rafsanjani et al 140	2015	Community	Monthly							x
Jain et al 141	2018	Community	Weekly								x	x
Jarab et al 142	2012	Specialized	Weekly							x
Jennings et al 143	2014									x
Jeong et al 144	2018	Hospital	Twice	x				x	x
Jiwani et al 145	2020	Community	Weekly									x
Kardas et al 146	2016	Primary		x	x	x	x	x
Kassavou et al 147	2020	Primary	Weekly									x
Kempf et al 148	2017	Specialized	Weekly								x
Keogh et al 149	2011	Specialized	Once								x
Kim and Utz 150	2019	Hospital	Weekly								x	x
Kim 151	2007	Hospital	Weekly					x						x
Kim and Jeong 152	2007	Hospital	Weekly					x						x
Kim and Song 153	2008	Hospital	Weekly					x						x
Kim and Kim 154	2008	Hospital	Weekly					x						x
Kim and Oh 155	2003	Hospital	Weekly							x		x
Kim et al 156	2005	Hospital	Weekly							x		x
Kim and Kang 157	2006	Hospital	Tailored							x
Kim et al 158	2009	Community	Monthly	x		x		x		x
Kim et al 159	2010	Hospital	Daily	x				x						x
Kim et al 160	2015	Hospital						x
Kim et al 161	2016	Hospital	Tailored					x
King et al 162	2006	Primary	Twice								x
Kirkman et al 163	1994	Primary	Monthly							x
Kleinman et al 164	2017	Specialized						x						x
Krein et al 165	2004	Specialized								x
Ku et al 166	2020	Hospital	Weekly	x				x			x
Kusnanto et al 167	2019	Primary	Daily								x
Kwon et al 168	2004	Hospital	Tailored					x						x
Lauffenburger et al 169	2019		Weekly								x	x
Lazo-Porras et al 170	2020	Hospital	Weekly					x			x
Lee et al 171	2017	Community		x				x		x				x
Lee et al 172	2020	Hospital	Weekly					x				x
Lee et al 173	2019	Primary	Weekly	x				x
Li et al 174	2016	Hospital	Monthly								x	x
Liebreich et al 175	2009		Weekly							x		x
Lim et al 176	2011	Hospital		x				x
Lim et al 177	2016	Hospital		x	x			x
Lorig et al 178	2010		Weekly										x
Lujan et al 179	2007	Community	Fortnight							x		x
Luley et al 180	2011	Hospital	Weekly		x		x	x
Lutes et al 181	2017	Primary	Monthly	x	x		x			x
Lynch et al 182	2014	Community	Weekly										x
MacPhail et al 183	2014	Primary	Twice								x
Marios et al 184	2012	Primary	Weekly					x		x
Maslakpak et al 185	2017		Weekly									x
Mayberry et al 186	2020	Primary	Monthly								x
McEwen et al 187	2017	Community	Tailored								x
McKay et al 188	2001		Weekly					x			x
McKay et al 189	2002	Primary	Weekly								x
McKee et al 190	2011	Community		x		x		x			x
McMahon et al 191	2012		Tailored					x		x				x
Mease 192	2000	Primary	Weekly			x		x
Mons et al 193	2013	Primary	Monthly							x
Moriyama et al 194	2009	Hospital	Fortnight								x	x
Mwangi et al 195	2020	Hospital	Monthly									x		x
Namjoo Nasab et al 196	2017	Community	Weekly							x
Nesari et al 197	2010	Specialized	Weekly							x		x
Nicolucci et al 198	2015	Primary	Monthly	x		x	x	x						x
Niswender et al 199	2014	Cross-sectional	Fortnight						x
O’Neil et al 200	2016	University	Weekly							x		x
Odegard and Christensen 201	2012	Community	Monthly							x		x
Odnoletkova et al 202	2016		Fortnight								x
Oh et al 203	2003	Hospital	Weekly							x		x
Orsama et al 204	2013	Community	Tailored	x	x	x	x	x
Owolabi et al 205	2019	Primary	Daily								x	x		x
Parsons et al 206	2019	Primary	Weekly						x			x
Patja et al 207	2012	Primary	Monthly								x
Peasah et al 208	2019	Primary	Weekly								x			x
Peimani et al 209	2016	Hospital	Weekly									x
Piette et al 210	2011	Community	Weekly							x
Plotnikoff et al 211	2013	Community	Tailored								x
Presley et al 212	2020	Community	Weekly					x			x
Quinn et al 213	2017	Community	Tailored					x				x
Quinn et al 214	2011	Community	Tailored					x				x
Raj and Mathews 215	2020	Hospital	Weekly								x			x
Ralston et al 216	2009	Hospital	Weekly					x
Ramadas et al 217	2018	Hospital								x				x
Rasmussen et al 218	2016	Specialized							x
Rodríguez-Idígoras et al 219	2009	Community	Tailored	x				x
Ruggiero et al 220	2014	Primary	Monthly								x
Sacco et al 221	2009	University	Weekly								x
Sacco et al 222	2012	University	Weekly								x
Samuel-Hodge et al 223	2009	Community	Monthly							x
Sarayani et al 224	2018		Weekly									x
Saslow et al 225	2020	University	Daily									x
Sazlina et al 226	2015	Primary			x								x
Schillinger et al 227	2009		Weekly									x
Shahid et al 228	2015	Specialized	Weekly	x				x		x
Shahsavari and Bakhshandeh Bavarsad 229	2020	Specialized	Weekly								x
Shetty et al 230	2011	Specialized	Weekly											x
Shreck et al 231	2014		Fortnight							x
Sigurdardottir et al 232	2009	Specialized	Weekly								x
Skelly et al 233	2009	Cross-sectional	Fortnight							x		x
Sone et al 234	2002	Specialized	Fortnight		x					x
Sone et al 235	2010	Specialized	Fortnight		x					x		x
Song and Kim 236	2009	Specialized	Weekly							x		x
Spencer et al 237	2018	Community	Fortnight								x		x
Steventon et al 238	2014	Cross-sectional	Daily	x		x		x				x
Sun et al 239	2019	Hospital	Daily	x				x	x					x
Sunil Kumar et al 240	2020	Hospital	Daily					x				x		x
Takenga et al 241	2014	Hospital						x
Tamban et al 242	2013		Weekly											x
Tan et al 243	2018	Primary	Fortnight									x
Tang et al 244	2013			x				x				x
Teston et al 245	2017		Fortnight								x
Thom et al 246	2013	Community	Fortnight								x		x
Torbjørnsen et al 247	2014		Monthly	x						x
Tu et al 248	1993	Hospital	Weekly							x		x
Van Dyck et al 249	2013	Hospital	Fortnight		x					x	x
Van Dyck et al 250	2011	Hospital	Fortnight		x					x	x
Van Vugt et al 251	2016	Primary									x			x
Varney et al 252	2014	Hospital	Monthly								x
Vaughan et al 253	2020	Primary	Weekly						x		x
Vervloet et al 254	2014													x
Vinithia et al 255	2019	Hospital	Weekly								x	x
Von Storch et al 256	2019		Weekly	x				x			x
Wakefield et al 257	2011	Primary	Daily	x		x		x				x
Wakefield et al 258	2012	Primary	Daily	x		x		x				x
Wakefield et al 259	2014	University	Tailored	x		x		x
Waki et al 260	2014	University	Tailored	x	x	x	x	x
Walker et al 261	2011	Specialized	Monthly								x
Wang et al 262	2020	Hospital	Weekly									x
Wang et al 263	2019	Hospital	Weekly					x	x
Wang et al 264	2017	Hospital	Fortnight	x				x						x
Warren et al 265	2018	Primary	Daily	x		x		x	x
Weinberger et al 266	1995	Primary	Monthly							x		x
Whittemore et al 267	2004	Specialized	Monthly								x
Wichit et al 268	2017	Hospital	Once								x	x
Wild et al 269	2016	Primary		x		x	x	x
Williams et al 270	2012	Hospital	Weekly					x
Williams et al 271	2017	Primary	Monthly							x
Wolever et al 272	2010		Weekly								x
Wolf et al 273	2004	Primary	Monthly								x	x
Wu et al 274	2017	University	Once								x			x
Yang et al 275	2020	Primary	Daily					x	x
Yasmin et al 276	2020	Hospital	Fortnight								x	x		x
Yoo et al 277	2009	University	Tailored	x		x	x	x				x
Yoon and Kim 278	2008	Hospital	Weekly					x						x
Young et al 279	2005	Primary	Monthly							x		x
Yu et al 280	2019	Hospital	Daily						x			x
Zamanzadeh et al 281	2017		Daily									x
Zhou et al 282	2014	Hospital	Fortnight	x		x	x	x						x

Abbreviation: BP, blood pressure.

The final sample included 86 studies (35%) with a monitoring component in the telemedicine intervention, 22 studies with a consultation opportunity (9%), 53 studies with a counseling purpose (22%), 63 studies with an opportunity for patients to receive coaching (26%), 81 studies with a patient education component (33%), and eight studies with the possibility of mentoring (3%). Furthermore, 38 studies were able to send reminders (16%).

The evaluation of risk of bias is described for each study in Online Appendix 1 and across studies in Figure 2. Overall, there was a high risk of bias in 45% of the included studies, a moderate risk in 38% and a low risk in 17%. This result was largely attributed to two factors. First were missing outcomes at follow-up, where a high proportion of studies (43%) only reported results of a complete case or per-protocol analysis without appropriate consideration of the relationship with covariates and missingness (ie, tests for missing completely at random or an assumption of missing at random with imputation and/or tests for association with baseline variables and missingness and/or adjusted analyses). Second, there were some concerns regarding the risk of selecting published results, especially due to a lack of registered or published research protocols in the majority of studies (72%).

Figure 2.

Summary of risk of bias assessment.

Effect on HbA1c%

Of the 243 studies, 168 had sufficient information to calculate an MD with standard errors for the effect on HbA1c% and reported treatment effects of telemedicine from one to 96 months. Figure 3 and Table 3 present the results from the individual studies and the meta-analysis. Overall, the results favored telemedicine, with an MD of −0.415%, which was statistically significant (95% confidence interval [CI] = −0.482% to −0.348%). The heterogeneity was great (I² = 93.05%).

Figure 3.

Forest plot of the meta-analysis (magenta line is effect size difference of 0).

Table 3.

Meta-Analysis Summary.

Study	MD	Lower 95% CI	Higher 95% CI	Weight
Abaza and Marschollek 42	−0.11	−0.72	0.50	0.47
Aguiar et al 45	−0.63	−1.08	−0.18	0.57
Akinci et al 46	−0.57	−1.21	0.07	0.45
Alanzi et al 48	−1.25	−2.14	−0.36	0.33
Ali et al 51	−0.40	−0.59	−0.22	0.73
Aliha et al 53	−1.30	−1.88	−0.72	0.49
Alotaibi et al 55	−0.74	−1.56	0.08	0.36
Anderson et al 56	−0.29	−0.60	0.02	0.66
Anderson-Loftin et al 58	−1.00	−1.56	−0.44	0.50
Anzaldo-Campos et al 60	−1.37	−1.81	−0.94	0.58
Arora et al 61	−0.13	−0.56	0.31	0.58
Avdal et al 63	−0.69	−1.04	−0.35	0.64
Azizi et al 64	−0.40	−0.55	−0.25	0.74
Blackberry et al 66	−0.06	−0.27	0.15	0.72
Bogner et al 68	−1.20	−1.52	−0.88	0.65
Lashkari et al 70	−0.99	−1.62	−0.36	0.46
Browning et al 71	0.03	−0.16	0.22	0.73
Bujnowska-Fedak et al 72	−0.06	−0.71	0.59	0.45
Chen et al 78	−0.76	−1.33	−0.19	0.49
Cho et al 81	−0.70	−1.16	−0.24	0.56
Cho et al 82	−0.15	−0.32	0.02	0.74
Choe et al 83	−1.30	−1.88	−0.72	0.49
Crowley et al 86	−0.10	−0.17	−0.04	0.76
Crowley et al 87	−1.00	−1.35	−0.65	0.64
Dale et al 88	0.10	−0.23	0.43	0.65
Dario et al 89	0.01	−0.23	0.25	0.70
Davis et al 90	−0.70	−0.90	−0.50	0.72
Del Prato et al 91	0.00	−0.06	0.06	0. 77
Dugas et al 93	0.18	−0.87	1.23	0.27
Döbler et al 95	−0.80	−1.11	−0.49	0.66
Eakin et al 96	0.00	−0.29	0.29	0.67
Estey et al 100	−0.20	−0.77	0.37	0.49
Faridi et al 101	−0.40	−0.98	0.18	0.49
Farsaei et al 102	−1.50	−1.85	−1.15	0.64
Fortmann et al 104	−0.90	−1.34	−0.46	0.58
Fountoulakis et al 106	−0.70	−1.16	−0.24	0.56
Franciosi et al 108	−0.50	−0.74	−0.26	0.70
Frosch et al 109	−0.30	−0.42	−0.18	0.75
García et al 110	−1.20	−1.45	−0.95	0.69
Garg et al 111	−0.30	−0.65	0.05	0.64
Gagliardino et al 112	−0.20	−0.51	0.11	0.66
Glasgow and Toobert 117	0.10	−0.27	0.47	0.62
Glasgow et al 118	−0.20	−0.55	0.15	0.64
Glasgow et al 119	0.01	−0.13	0.15	0.74
Glasgow et al 115	−0.20	−0.48	0.08	0.68
Glasgow et al 116	0.00	−0.27	0.27	0.68
Goodarzi et al 120	−0.46	−0.88	−0.04	0.59
Graziano and Gross 122	−0.07	−0.45	0.31	0.62
Griffin et al 123	−0.01	−0.18	0.16	0.73
Hare et al 127	−0.10	−0.41	0.21	0.66
Hee-Sung 128	−0.43	−0.81	−0.05	0.62
Holbrook et al 132	−0.50	−0.71	−0.30	0.72
Holmen et al 133	−0.20	−0.68	0.28	0.55
Hordern et al 134	−0.70	−0.99	−0.41	0.67
Iljaž et al 138	−0.30	−0.69	0.09	0.61
Jahangard-Rafsanjani et al 140	−0.40	−0.89	0.09	0.54
Jarab et al 142	−0.90	−1.46	−0.34	0.50
Jeong et al 144	−0.12	−0.40	0.16	0.68
Kardas et al 146	−0.03	−0.51	0.45	0.55
Kempf et al 148	−0.60	−0.91	−0.29	0.66
Keogh et al 149	−0.39	−0.78	0.00	0.61
Kim and Oh 155	−1.20	−1.74	−0.66	0.51
Kim et al 156	−1.00	−1.63	−0.38	0.46
Kim and Kang 157	−0.90	−1.58	−0.22	0.43
Kim 151	−0.72	−1.22	−0.22	0.54
Kim and Jeong 152	−0.66	−1.20	−0.12	0.51
Kim and Song 153	−1.52	−2.02	−1.02	0.54
Kim and Kim 154	−0.59	−1.21	0.03	0.46
Kim et al 158	−0.90	−1.40	−0.40	0.54
Kim et al 159	−0.40	−0.74	−0.06	0.64
Kim et al 160	−0.70	−1.13	−0.27	0.58
Kim et al 161	−0.70	−0.96	−0.44	0.69
Kleinman et al 164	−0.30	−0.77	0.17	0.56
Krein et al 165	0.10	−0.28	0.48	0.62
Kwon et al 168	−0.68	−0.82	−0.55	0.75
Lee et al 171	−0.93	−1.49	−0.37	0.50
Li et al 174	−0.35	−0.95	0.25	0.48
Lim et al 176	−0.40	−0.79	−0.01	0.61
Lim et al 177	−0.60	−1.00	−0.20	0.60
Lorig et al 178	−0.11	−0.26	0.05	0.74
Lujan et al 179	−0.25	−0.68	0.18	0.58
Luley et al 180	−1.00	−1.33	−0.67	0.65
Lutes et al 181	−0.26	−0.66	0.14	0.60
Marios et al 184	0.49	−0.25	1.23	0.40
Maslakpak et al 185	−0.50	−1.09	0.09	0.48
McEwen et al 187	−0.01	−0.46	0.44	0.57
McKay et al 189	0.36	−0.17	0.89	0.52
McKee et al 190	−0.60	−1.39	0.19	0.37
McMahon et al 191	−0.10	−0.56	0.36	0.56
Nesari et al 197	−1.56	−2.18	−0.94	0.46
Nicolucci et al 198	−0.34	−0.57	−0.11	0.71
Niswender et al 199	−0.13	−0.17	−0.10	0.77
Odnoletkova et al 202	−0.10	−0.27	0.07	0.73
Oh et al 203	−1.30	−1.88	−0.72	0.49
O’Neil et al 200	−0.39	−0.59	−0.19	0.72
Orsama et al 204	−0.44	−0.88	0.01	0.58
Peimani et al 209	−0.49	−0.95	−0.03	0.56
Piette et al 210	0.00	−0.28	0.28	0.68
Plotnikoff et al 211	0.21	0.13	0.29	0.76
Quinn et al 214	−0.80	−1.26	−0.34	0.56
Ramadas et al 217	0.10	−0.39	0.59	0.55
Rodríguez-Idígoras et al 219	0.05	−0.18	0.28	0.70
Sacco et al 221	−0.40	−0.95	0.15	0.51
Samuel-Hodge et al 223	−0.10	−0.19	−0.01	0.76
Sarayani et al 224	−0.30	−0.80	0.20	0.54
Schillinger et al 227	−0.30	−0.66	0.06	0.63
Shahid et al 228	−0.73	−0.94	−0.52	0.72
Sigurdardottir et al 232	0.25	−0.26	0.76	0.53
Sone et al 234	−0.17	−0.26	−0.08	0.76
Sone et al 235	0.10	0.01	0.20	0.76
Song and Kim 236	−1.50	−2.13	−0.87	0.46
Tamban et al 242	−0.35	−0.71	0.01	0.63
Tan et al 243	−0.38	−0.77	0.01	0.61
Tang et al 244	−0.23	−0.48	0.02	0.69
Thom et al 246	−0.57	−0.90	−0.24	0.65
Torbjørnsen et al 247	0.00	−0.41	0.41	0.60
Tu et al 248	0.43	−0.65	1.51	0.26
Van Dyck et al 249	−0.30	−0.79	0.19	0.55
Varney et al 252	−0.20	−0.65	0.25	0.57
Wakefield et al 259	−0.10	−0.26	0.06	0.74
Waki et al 260	−0.40	−0.91	0.11	0.53
Walker et al 261	0.10	0.04	0.16	0.77
Wang et al 264	−0.60	−0.87	−0.33	0.68
Warren et al 265	−0.57	−0.89	−0.24	0.65
Weinberger et al 266	−0.60	−0.74	−0.46	0.74
Whittemore et al 267	0.00	−0.56	0.56	0.50
Wichit et al 268	−0.30	−0.68	0.08	0.62
Wild et al 269	−0.50	−0.75	−0.25	0.69
Williams et al 270	−0.80	−1.22	−0.38	0.59
Wolever et al 272	−0.50	−1.22	0.22	0.41
Yoo et al 277	−0.50	−0.84	−0.17	0.65
Yoon and Kim 278	−1.63	−2.11	−1.15	0.55
Zhou et al 282	−0.76	−1.20	−0.32	0.58
Agarwal et al 43	0.80	0.13	1.47	0.44
Alghafri et al 50	0.30	−0.01	0.61	0.66
Al Omar et al 47	−0.70	−1.05	−0.35	0.63
Andreae et al 59	−0.10	−0.46	0.26	0.63
Asante et al 62	−1.30	−2.02	−0.58	0.41
Bluml et al 67	0.00	−0.25	0.25	0.69
Boels et al 12	−0.20	−0.52	0.12	0.66
Bohingamu Mudiyanselage et al 69	−0.21	−0.41	−0.01	0.72
Buysse et al 73	−0.10	−0.67	0.47	0.49
Chao et al 77	−0.90	−1.35	−0.45	0.57
Choudhry et al 84	0.10	0.02	0.19	0.76
Delahanty et al 92	0.00	−0.37	0.37	0.62
Doupis et al 7	0.10	−0.04	0.24	0.74
Duruturk and Özköslü 94	−1.99	−2.80	−1.18	0.36
Gupta et al 124	−0.52	−0.98	−0.06	0.56
Haider et al 125	−0.40	−0.75	−0.05	0.64
Huang et al 135	−0.40	−1.18	0.38	0.37
Jain et al 141	0.26	−0.08	0.60	0.64
Jiwani et al 145	−2.20	−3.26	−1.14	0.26
Kim and Utz 150	0.24	−0.18	0.66	0.59
Ku et al 166	−1.20	−1.91	−0.49	0.41
Kusnanto et al 167	−0.27	−1.02	0.48	0.39
Lauffenburger et al 169	−0.06	−0.21	0.09	0.74
Lee et al 172	−0.50	−0.94	−0.06	0.58
Parsons et al 206	−0.97	−1.20	−0.74	0.71
Peasah et al 208	0.00	−0.52	0.52	0.53
Presley et al 212	0.50	0.00	1.00	0.54
Shahsavari and Bakhshandeh Bavarsad 229	−1.38	−1.89	−0.87	0.53
Sun et al 239	−0.38	−0.75	−0.01	0.62
Sunil Kumar et al 240	−0.87	−1.11	−0.63	0.70
Vaughan et al 253	−0.67	−1.24	−0.10	0.49
Vinithia et al 255	−0.40	−0.71	−0.10	0.66
Von Storch et al 256	−0.37	−0.45	−0.29	0.76
Wang et al 263	−0.80	−1.32	−0.28	0.53
Yu et al 280	−0.40	−0.92	0.12	0.53
Overall	−0.42	−0.48	−0.35	100.00

Abbreviations: CI, confidence interval; MD, mean difference.

A series of univariate meta-regressions are illustrated in Table 4 and were assessed with a 5% significance level. Compared with North American studies (the reference), Asian studies reported larger effects of telemedicine (difference in MD = −0.287, P = .000), as did studies with higher baseline HbA1c% levels (difference in MD = −0.086 per %, P = .008). Compared with primary care settings (reference), hospital settings were also associated with an increase in the effect of telemedicine (difference in MD = −0.290, P = .004). Furthermore, a monitoring component gave rise to greater effects of telemedicine (difference in MD = −0.195, P = .004). Treatment effects of telemedicine were lesser for studies with longer duration (difference in MD = 0.008 per month, P = .015) and for studies with higher proportions of men (difference in MD = 0.005 per %, P = .035) and higher age (difference in MD = 0.022 per year, P = .000). A coaching component led to a lower effect (difference in MD = 0.215, P = .007). There were no statistically significant associations between the effect of telemedicine on HbA1c% levels and publication date, baseline BMI, contact frequency, the included peripherals, or risk of bias.

Table 4.

Association Between Study Covariates and Effect of Telemedicine on HbA1c% (Meta-Regression).

Covariate	N	Difference in MD (SE)	P value	I2 (%)
Study characteristics
Publication decade
Before 2020	4	Reference		92.86
2020s	50	−0.175 (0.287)	.542
2010s	165	−0.082 (0.281)	.769
2020	24	−0.418 (0.312)	.180
Continent
North America	88	Reference		91.15
Europe	44	−0.037 (0.095)	.696
South America	5	−0.104 (0.303)	.731
Africa	5	−0.366 (0.362)	.312
Asia	84	−0.287 (0.078)	.000*
Australia/New Zealand	16	−0.040 (0.140)	.773
Study duration (range = 1-96 months)	168	0.008 per month (0.003)	.015*	92.60
Proportion of men (range = 15%-100%)	159	0.005 per % (0.002)	.035*	93.05
Age (range = 37-73 years)	157	0.022 per year (0.006)	.000*	92.64
Baseline BMI (range = 22-40)	123	0.018 per score (0.010)	.066	92.51
Baseline HbA1c% (range = 5.70%-11.05%)	165	−0.093 per % (0.035)	.007*	92.80
Telemedicine characteristics
Setting
Primary care	58	Reference		90.25
Community	37	−0.019 (0.116)	.868
Hospital	71	−0.290 (0.100)	.004*
Specialized outpatient clinic	34	−0.148 (0.114)	.194
University	14	−0.180 (0.169)	.287
Cross-sectorial	4	0.203 (0.301)	.500
Contact frequency
Daily	30	Reference		91.42
Weekly	83	−0.150 (0.129)	.246
Every two weeks	26	0.045 (0.157)	.776
Monthly	40	0.101 (0.144)	.481
More seldom	12	−0.006 (0.199)	.977
Tailored	24	0.076 (0.148)	.611
Included peripherals
Glucometer	45	−0.052 (0.081)	.523	92.81
Pedometer	16	0.102 (0.132)	.440	92.96
BP monitor	19	−0.008 (0.132)	.953	93.09
Scale	13	0.014 (0.137)	.919	93.12
Intervention components
Monitoring	86	−0.195 (0.068)	.004*	92.30
Consultation	22	−0.015 (0.114)	.895	92.25
Counseling	53	−0.030 (0.083)	.720	92.95
Coaching	63	0.215 (0.080)	.007*	92.44
Education	81	−0.125 (0.072)	.085	92.76
Mentoring	8	0.265 (0.201)	.188	93.03
Reminding	38	−0.151 (0.092)	.100	92.87
Risk of bias
Low	33	Reference		92.82
Some concerns	76	−0.101 (0.100)	.310
High	88	−0.067 (0.100)	.492

Abbreviations: HbA1c%, glycated hemoglobin; MD, mean difference; BMI, body mass index; BP, blood pressure.

*

Statistically significant at a 5% level.

Certainty of the Evidence

Table 5 summarizes the findings. Overall, the certainty of evidence of the calculated effect on HbA1c% was judged as low due to serious problems with the risk of bias and inconsistency.

Table 5.

Summary of Findings Table.

Certainty assessment							No of patients		Effect		Certainty	Importance
No of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Telemedicine	Usual practice	Relative (95% CI)	Absolute (95% CI)
HbA1c%
243	Randomized trials	Serious a	Serious b	Not serious	Not serious	None	29 671	27 465	−	MD 0.415 lower (0.482 lower to 0.348 lower)	⨁⨁◯◯ LOW	IMPORTANT

Abbreviations: CI, confidence interval; HbA1c%, glycated hemoglobin; MD, mean difference.

a

The size of the review implies that no single study contributes dominant weights in the meta-analysis. Indeed, study weights range from 0.26 to 0.77. The correlation between effect size, that is, the MD, and risk of bias across studies was low and insignificant (Spearman ρ = −0.06, P = .3961). However, only 17% of the studies were evaluated to have a low risk of bias. Consequently, the risk of bias was downgraded to one level and assessed as serious.

b

Effect size point estimates vary widely between studies, with significant effects favoring both alternatives. I² was high, both with and without controlling for statistically significant study covariates simultaneously in the meta-regression (I² = 87.8% and I² = 93%).

Discussion

This review aimed to evaluate the effectiveness of telemedicine solutions among adult patients with T2D. Overall, the results favored telemedicine. Asian studies, studies with higher baseline HbA1c% levels, and studies in hospital settings reported larger effects of telemedicine. Moreover, inclusion of a monitoring component in the telemedicine solution gave rise to a higher effect of telemedicine, whereas inclusion of a coaching component led to a lower effect. The results reflect the findings from previous systematic reviews, which have found significant reductions in HbA1c favoring telemedicine.^{13,22,283,284} In line with our findings, Faruque et al and Wu et al found a larger effect among studies with a higher baseline HbA1c.^13,283 In contrast to our findings, Faruque et al found a larger effect in studies that used web portals or text messaging. ¹³ In terms of telemonitoring, Hanlon et al found improved glycemic control in studies that included telemonitoring with feedback ¹⁷ and Jaana et al found significantly reduced HbA1 through telemonitoring. ²² Hence, the ideal telemedicine setup remains to be determined, although the inclusion of a telemonitoring component seems advisable.

The heterogeneity of the included studies was large (I² = 93.05%), which is often seen in reviews of telemedicine.^{13,15,284-286} In the before-mentioned systematic review by Faruque et al, the heterogeneity (I²) ranged from 58% to 75% depending on the time point Hba1c was measured. ¹³ A systematic review and meta-analysis by Tchero found a heterogeneity (I²) of 99% in studies comparing telemedicine with usual care in T2D. ²⁸⁴ Hence, the large heterogeneity found in the present review is not unusually large when compared with similar reviews. The large heterogeneity found in the present review may be explained by differences in the inclusion criteria and context of studies that vary greatly in terms of patient subgroup, provider, technology, organization, communication frequency, outcome, and so on. Moreover, the inclusion of studies from the 1990s and 2000s may have added to the heterogeneity. Telemedicine interventions in diabetes have evolved significantly during the last decades due to technological advances and differences in the interventions are thus expected to have affected heterogeneity. However, the inclusion of studies from the 1990s and 2000s has maintained the broad and inclusive scope that was intended for the present review.

The certainty of the evidence was judged as poor. First, most of the evidence (45%) stems from studies with a high risk of bias and only 17% stems from studies with a low risk of bias. However, the size of the review implies that no single study contributed dominant weights in the meta-analysis, where study weights ranged from 0.26 to 0.77. The correlation between effect size, that is, the MD, and risk of bias across studies was low and nonsignificant (Spearman ρ = −0.06, P = .3961). Taken together, the certainty of evidence was downgraded one level due to risk of bias. Second, imprecision was assessed as not serious, as the effect size was statistically significant, and the MD and lower confidence limit (MD = −0.415, 95% CI = −0.482% to −0.348%) were both clinically relevant, and all were comparable with expected effects for other quality improvement strategies, ²⁸⁷ which is also why no upgrading due to large effects was conducted. The total number of patients (N = 57 136) included in the review was also much higher than the optimal information size threshold (the OIS criterion). Third, inconsistency was assessed as serious. Although confidence intervals overlap considerably, effect size point estimates vary widely between studies, with significant effects favoring both alternatives. I² was high both with and without controlling for statistically significant study covariates simultaneously in the meta-regression (I² = 87.8% and I² = 93%). Fourth, indirectness was not serious in this review, as all included studies were head-to-head experimental trials assessing the same outcome, and the inclusion criteria for both the population and type of intervention were broad. Fifth, although publication bias cannot be rejected with high confidence, the risk of publication bias was evaluated as undetected. Less than 100 participants were included in 35% of the studies. The Egger test for small-study effects was statistically significant (P < .000) and the funnel plot in Figure 4 reveals studies reporting both significant and insignificant positive and negative effects. The funnel plot is somewhat asymmetrical, with more studies demonstrating significant effects favoring telemedicine than negative effects. Of the 243 included studies, 32 reported industry sponsorship and 44 studies did not report whether there were sponsorships or other relevant conflicts of interest. However, the relationship between disclosed industry sponsorship and/or undisclosed relationships with both study size (above/below 100 participants) and MD (above/below mean effect across studies) was statistically nonsignificant (Fisher exact test = 0.31 and 0.11, respectively). Finally, no dose-response gradient was detected, as the contact frequency was insignificant in the meta-regression (Table 4).

Figure 4.

Funnel plot.

The present systematic review has some limitations. First, more baseline data, such as diabetes years, blood pressure, cholesterol levels, and medication, could have been extracted. However, due to differences in reporting style, such an extraction would have resulted in a high proportion of missing data in the analysis. Second, we could have contacted the authors of the individual studies to minimize the amount of missing HbA1c data. However, due to the large sample of included papers, such a search for data was deemed too time-consuming. Third, relevant studies may have been overlooked. Although a very broad search was conducted, the search was still limited to English and Scandinavian.

Conclusion

Telemedicine may serve as an effective supplement to usual care for patients with T2D. The inclusion of a telemonitoring component seems to increase the effect of telemedicine. It seems that those with a higher HbA1c are more likely to benefit from telemedicine. Patients with poor glycemic control may benefit more from telemedicine interventions, as a high HbA1c level leaves further room for improvement. However, the ideal glycemic target group for telemedicine in T2D remains to be determined.

Supplemental Material

sj-docx-1-dst-10.1177_19322968211064633 – Supplemental material for The Effectiveness of Telemedicine Solutions for the Management of Type 2 Diabetes: A Systematic Review, Meta-Analysis, and Meta-Regression

Supplemental material, sj-docx-1-dst-10.1177_19322968211064633 for The Effectiveness of Telemedicine Solutions for the Management of Type 2 Diabetes: A Systematic Review, Meta-Analysis, and Meta-Regression by Stine Hangaard, Sisse H. Laursen, Jonas D. Andersen, Thomas Kronborg, Peter Vestergaard, Ole Hejlesen and Flemming W. Udsen in Journal of Diabetes Science and Technology