The Effectiveness of Telemedicine Solutions in Type 1 Diabetes Management: A Systematic Review and Meta-analysis

Abstract

Background:

Telemedicine holds a potential to strengthen self-management support outside health care settings in the everyday management of type 1 diabetes (T1D). However, existing effectiveness reviews are older or include a relatively narrow focus on specific definitions of telemedicine or included databases.

Objective:

To conduct a systematic review of the effectiveness of telemedicine solutions versus any comparator on diabetes-related outcomes among people with T1D.

Methods:

Studies including adults (≥18 years) with T1D published before October 14, 2020, were eligible. Primary outcome was glycated hemoglobin (HbA1c, %). The Cochrane Library, PubMed, EMBASE, and CINAHL were searched. Meta-analysis based on the mean difference in HbA1c% was used to pool effects. The Cochrane tool was applied to assess risk-of-bias, and the certainty of evidence was graded using the GRADE approach.

Results:

A total of 22 studies were included (with 1615 participants). Treatment effect for HbA1c% favored telemedicine (mean difference of −0.26% [95% confidence interval:−0.37% to −0.15%]) with moderate effect certainty. Heterogeneity was moderate (I² = 33.30%). Although not significant, secondary outcomes were all in favor of telemedicine except number of severe hypoglycemic events and diabetes knowledge, but the certainty of the evidence for those outcomes was all low or very low.

Discussion:

Reducing average HbA1c% levels are important to combat the risk of diabetic complications and premature death. However, the evidence mostly consist of small studies with a relative short duration and the estimated pooled effect is smaller than could be expected from quality improvement strategies in general for diabetes management.

PROSPERO number:

CRD42020123565.

Keywords

blood glucose diabetes mellitus type 1 meta-analysis systematic review telemedicine

Introduction

Type 1 diabetes (T1D) is a chronic disease that requires constant decision-making in terms of self-management and monitoring, which often results in poor adherence to recommended guidelines.¹ Self-management is often highlighted as the cornerstone in the management of T1D and has previously been associated with improvements in different health-related outcomes.¹ Self-management consists of different modalities such as educational, psychological, behavioral, and clinical assistance activities to help the patient navigate through his or her condition.² Thus, support for self-management is highly encouraged as part of the management of T1D.¹

Application of telemedicine is emerging as a way to strengthen self-management support outside health care settings in the everyday management of diabetes.^3,4 Different conceptions of what constitutes the concept of “telemedicine” are common. Sood and colleagues reported 104 different definitions in a review of the term.⁵ In this review, we apply the World Health Organization (WHO) definition of telemedicine as

the delivery of health care services, where distance is a critical factor, by all health care professionals using information and communication technologies for the exchange of valid information for diagnosis, treatment and prevention of disease and injuries, research and evaluation, and for the continuing education of health care providers, all in the interests of advancing the health of individuals and their communities.⁶

What constitutes the “delivery of health care services” and “valid information for diagnosis, treatment and prevention” is somewhat open for interpretation. Therefore, the definition is further elaborated to include four elements that have been used previously^7,8: (1) information are obtained from a patient (whether audio, video, measured physiological data, or written responses) and then (2) electronically transferred over a geographical distance to (3) a health professional who exercise skill and judgment (4) in the provision of tailored feedback to the patient. Both synchronous and asynchronous technologies apply. Asynchronous telemedicine is defined as store-and-forward transmission of patient information with non-real-time communication (eg, email, text messaging, Internet platform), whereas synchronous telemedicine refer to real-time communication and transfer of data. A combination is also applied to include setups where data whereas collected asynchronously, while follow-up communication was conducted synchronously. Wearables and sensors are encompassed within the telemedicine definition, when automated patient-data are transferred over a distance with a health professional at the receiving end, who evaluates the data and gives feedback to patients. Similar conceptualizations of telemedicine that includes sensors can be found, for example, Klonoff and colleagues, who applied a framework that distinguishes between “sensor-based” and “non-sensor-based” telemedicine.⁹

A recent systematic review and meta-analysis reported that telemedicine can be considered a safe option for providing self-management support for people with diabetes.¹⁰ However, existing T1D effectiveness reviews of telemedicine lack a meta-analysis or pool results from diverse study designs,¹¹ include a narrow definition of telemedicine or relatively few data sources,¹² and do not include an assessment of the certainty of the evidence for included outcomes.^11-13 The same can be said for reviews pooling both T2D and T1D.^14-16 Moreover, because of the rapid development in the field of telemedicine, a new systematic review and meta-analysis within the field is indicated.

The objective of this study is therefore to systematically evaluate the effect on HbA1c% levels and other secondary outcomes found in existing telemedicine solutions versus any comparator without the use of telemedicine among adults with T1D.

Methods

Study Design

This study is a systematic review with embedded meta-analysis, which is conducted and reported according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines.¹⁷ The study is PROSPERO registered with identification CRD42020123565 on April 2020. A study protocol with comprehensive search protocol covering T1D, T2D, and gestational diabetes mellitus (GDM) has previously been published¹⁸ and summarized below.

Eligibility Criteria

This review considered all studies published prior to October 14, 2020 that included adults (≥18 years), who had a diagnosis with T1D. Studies that reported mixed disease populations (eg, diabetes and chronic obstructive pulmonary disease) and mixed diabetes types were also included if the results related to the population with T1D were reported separately. Studies which only included participants at risk of diabetes or having prediabetes were excluded. Studies were eligible if they included interventions with remote feedback or communication between a patient and health care professionals, either as substitutes or alternatives to usual care. Fully automatic interventions were also considered. The primary outcome was glycemic control measured as change in glycated hemoglobin (HbA1c%). Only randomized controlled trials (RCTs) were considered, both parallel and cross-over designs. Studies published in English, Danish, Norwegian, or Swedish as peer-reviewed full-text papers were included.

Information Sources

The search was performed on October 14 in the Cochrane Library Central Register of Controlled Trials (CENTRAL), PubMed, EMBASE, and CINAHL by two authors (SHL, SH) with the help of a research librarian. Web of Science, SCOPUS, and Google Scholar were applied for additional citation searches.

Search Strategy

The search protocol covering T1D, T2D, and gestational diabetes is presented in Appendix 1. To qualify the systematic search, an initial unstructured search to identify relevant terms was performed in PubMed and CINAHL. The search strategy in the following systematic search was thereafter adapted for each database. Search terms included different synonyms, near-synonyms, acronyms, and spellings for all index terms and keywords. Thesaurus, Boolean operators, abstract/title/keywords, phrase, truncation, free text, and advanced search were utilized. With an eye to identifying additional studies, citation searches were applied.

Study Selection

All eligible studies were uploaded into RefWorks (Refworks, RefWorks-COS, ProQuest RefWorks 2.0, 2010) where duplicates were removed using the functions Exact duplicates and Close duplicates. Two authors (SH, JA) then screened title and abstracts with respect to adherence to the eligibility criteria of the review. Three authors (SH, SHL, JA) then conducted a full review of the remaining studies. Any disagreement between the three authors was resolved through discussion or by inclusion of other authors resulting in a final sample of studies. Reasons for exclusion of studies in the full review were recorded.

Data Extraction

A standardized worksheet in Microsoft Excel (2016) was applied by three authors (SHL, SH, JA) to extract data from studies. These included trial characteristics (author, publication year, country, sample size, study duration), patient characteristics (age, gender, body mass index [BMI]), and HbA1c outcomes. Furthermore, the characteristics of the telemedicine interventions were extracted including the purpose of the different components (general glycemic improvements, diet, exercise, medication, fatigue, diabetes education, and prevention of hypo- or hyperglycemic events), the type of peripherals (eg, blood glucose monitors), involved health professions in the delivery of telemedicine, synchronous or asynchronous implementation, and implementation context (eg, hospital, specialized diabetes clinic, university, community, or cross-sectional). Authors resolved potential disagreements through discussion with an option to include additional authors. If studies reported other measures than means and standard deviations (eg, medians and confidence intervals [CIs]), these were transformed by traditional methods^19,20 and scaled to HbA1c% when relevant (eg, if outcomes was reported as mmol/mol).

Assessment of Methodological Quality

Two reviewers (JA, FWU) assessed the risk of bias of all included studies independently using the revised Cochrane tool²¹ and potential disagreements were resolved by discussion.

Data Synthesis

An overall treatment effect was estimated with a meta-analysis based on the mean difference (MD) from each study. I² tests were used to assess heterogeneity. Results were combined with either a random or fixed effects model (depending on if the I² statistic was higher or lower than 50%, respectively). Results were subject to post hoc subgroup analysis of all extracted study and participant characteristics (continuous variables were binary divided into groups based on variable means). Sensitivity analyses were applied to investigate the impact on results by excluding potential study outliers. Visual inspection of the funnel plot and Egger’s test were used to evaluate publication bias. Analyses were all performed in Stata 16 (StataCorp).

Certainty Assessment

A Summary of Findings (SoF) Table was created using GRADEPro GDT 2015 (McMaster University, ON),^22,23 which presents the absolute effect for the groups (treatment and control) and ranking of the quality of the evidence for each included outcome. In the handbook for GRADE,²⁴ these are based on overall judgments of the risk of bias,²⁵ indirectness,²⁶ imprecision,²⁷ inconsistency,²⁸ and risk of publication bias.²⁹

Results

Figure 1 is a flowchart overview of the selection of studies. From the systematic search, 16 309 studies were identified (11 062 after duplicate removal) of which 1125 studies were full-text assessed after screening titles and abstracts. Finally, 22 articles met the inclusion criteria.

Figure 1.

Selection of studies.

Study references and characteristics are shown in Table 1. Two studies were published prior to 2000 (9%), eight in the 2000s (36%), and 12 in the 2010s (55%). One study was multinational (5%), whereas 14 studies were from Europe (64%), four from North America (18%), two from Asia/Middle East (9%), and one from Australia/New Zealand (5%). Size of studies varied from 10 to 158 participants averaging 73 per study. Mean study duration was six months ranging from one to 12 months. Mean proportion of men was 26% (range 0%-67%), average age 35 (range 24-44), mean baseline BMI was 24.9 (range 22.78-26.10), and average baseline HbA1c% was 8.7% (range 7.35%-10.89%).

Table 1.

Study and Participant Characteristics.

Study	Publication year	Country	Sample size	Duration, mo	Mean age, y	Male, %	Baseline HbA1c%	Baseline BMI
Ahring et al.³⁰	1992	Canada	42	3	41.43	30.48	10.89	N/A
Benhamou et al.³¹	2007	France	30	6	41.25	7.00	8.27	24.10
Bertuzzi et al.³²	2018	Italy	77	12	34.96	12.56	N/A	N/A
Charpentier et al.³³	2011	France	120	6	34.24	21.49	9.01	25.44
Drion et al.³⁴	2015	Netherlands	63	3	34.02	20.00	N/A	25.49
Esmatjes et al.³⁵	2014	Spain	154	6	31.85	34.48	9.25	25.25
Farmer et al.³⁶	2005	UK	93	9	23.86	27.51	9.25	25.21
Gómez et al.³⁷	2002	Spain	20	6	N/A	N/A	8.25	N/A
Jansà et al.³⁸	2006	Spain	40	6	25.00	10.50	8.65	23.40
Kirwan et al.³⁹	2013	Australia	72	9	35.20	14.00	8.78	N/A
Kotsani et al.⁴⁰	2018	Greece	94	3	26.98	23.53	8.20	N/A
McCarrier et al.⁴¹	2009	USA	78	12	37.00	67.00	8.00	N/A
Menting et al.⁴²	2017	Netherlands	120	5	43.65	23.00	8.05	26.10
Montori et al.⁴³	2004	USA	31	6	42.94	32.00	8.90	25.95
Rigla et al.⁴⁴	2007	Spain	20	1	40.60	50.00	8.00	N/A
Rossi et al.⁴⁵	2010	Multinational	130	6	35.74	43.00	8.30	N/A
Rossi et al.⁴⁶	2013	Italy	127	6	36.33	48.00	8.45	24.40
Sacco et al.⁴⁷	2004	USA	10	3	38.20	30.00	9.11	N/A
Suh et al.⁴⁸	2014	South Korea	57	3	33.00	37.00	9.50	22.78
Hejlesen et al.⁴⁹	1998	Denmark	12	2	26.50	3.00	9.40	N/A
Linden et al.⁵⁰	2017	Sweden	158	6	30.79	0.00	7.35	N/A
Yaron et al.⁵¹	2019	Israel	67	12	44.07	15.78	7.77	26.01

Abbreviations: HbA1c, glycated hemoglobin; BMI, body mass index, N/A = not available.

Characteristics of the telemedicine technologies are illustrated in Table 2. One study had daily contacts (5%), eight biweekly (36%), six weekly (27%), three studies a monthly contact (14%), and one study (5%) less frequent (twice in the study period). Across studies, 13 used a web portal (59%), nine included a blood glucose meters (BGMs) (41%) (one study also supplemented with a continuous glucose monitor, (CGM) [5%]), eight a telephone (36%), six a smartphone or tablet application (27%), six mail correspondence (27%), and four applied SMS communication (18%). Two studies applied a synchronous setup (9%), nine an asynchronous (41%), and 10 studies a combination of synchronous follow-up communication with asynchronous monitoring data transfer (45%). The purpose of intervention components were glycemic improvement (19 studies, 86%), dietary support (17 studies, 77%), medication support (17 studies, 77%), exercise support (13 studies, 59%), educational (11 studies, 50%), prevention of glycemic events (hypo- or hyperglycemia) (five studies, 23%), and fatigue improvement (one study, 5%). The telemedicine interventions were staffed by physicians in nine studies (41%), diabetes nurses in five studies (23%), coaches or psychologists in four studies (18%), a diabetes team in three studies (14%), and a research team in one study (5%). One study used the same health care professions to staff the control group (5%), three did not (14%), whereas 18 studies did not report sufficient details (82%).

Table 2.

Telemedicine Intervention Characteristics.

Study	Publication year	Setting	Frequency of contact	Included technology								Purpose of intervention components
Study	Publication year	Setting	Frequency of contact	CGM	Blood glucose meter	Synchronous or Asynchronous	Telephone	Email	App	SMS	Pump	Glycemic improvement	Dietary support	Exercise support	Medication support	Fatigue improvement	Educational	Prevention of glycemic event
Ahring et al.³⁰	1992	Hospital	Biweekly		Yes	Combined	Yes					Yes	Yes		Yes
Benhamou et al.³¹	2007	Specialized	Weekly		Yes	Asynchronous				Yes		Yes
Bertuzzi et al.³²	2018	Specialized	Monthly	Yes	Yes	Combined		Yes				Yes	Yes	Yes	Yes		Yes
Charpentier et al.³³	2011	Hospital	Biweekly			Combined	Yes		Yes			Yes	Yes	Yes	Yes
Drion et al.³⁴	2015	Outpatient clinic	N/A			Combined			Yes			Yes	Yes	Yes	Yes
Esmatjes et al.³⁵	2014	Hospital	Monthly		Yes	Combined						Yes	Yes	Yes	Yes
Farmer et al.³⁶	2005	Specialized	Biweekly		Yes	Synchronous						Yes	Yes	Yes	Yes
Gómez et al.³⁷	2002	Specialized	Biweekly			Asynchronous	Yes					Yes	Yes		Yes
Jansà et al.³⁸	2006	Specialized	N/A		Yes	Combined	Yes	Yes				Yes	Yes	Yes	Yes		Yes	Yes
Kirwan et al.³⁹	2013	N/A	Weekly			Asynchronous		Yes	Yes	Yes		Yes	Yes	Yes	Yes
Kotsani et al.⁴⁰	2018	University	Weekly		Yes	Combined	Yes	Yes				Yes	Yes	Yes	Yes		Yes	Yes
McCarrier et al.⁴¹	2009	Specialized	Weekly			Asynchronous		Yes				Yes	Yes	Yes	Yes		Yes
Menting et al.⁴²	2017	University	Biweekly			Asynchronous		Yes						Yes		Yes	Yes	Yes
Montori et al.⁴³	2004	N/A	Biweekly			Combined	Yes					Yes			Yes
Rigla et al.⁴⁴	2007	Medical center	Weekly		Yes	Asynchronous						Yes	Yes		Yes		Yes
Rossi et al.⁴⁵	2010	Outpatient clinic	N/A			Asynchronous			Yes	Yes		Yes	Yes		Yes		Yes
Rossi et al.⁴⁶	2013	Outpatient clinic	Biweekly			Asynchronous			Yes	Yes		Yes	Yes	Yes	Yes		Yes
Sacco et al.⁴⁷	2004	University	Weekly			Synchronous	Yes						Yes	Yes	Yes		Yes
Suh et al.⁴⁸	2014	Medical center	Biweekly			Combined	Yes		Yes			Yes	Yes	Yes	Yes		Yes
Hejlesen et al.⁴⁹	1998	Hospital	Daily			Asynchronous						Yes						Yes
Linden et al.⁵⁰	2017	Hospital	Twice			Unknown							Yes				Yes
Yaron et al.⁵¹	2019	Specialized	Monthly		Yes	Combined					Yes	Yes						Yes

Abbreviation: CGM, continuous glucose monitors; SMS, short message service

Figure 2 presents the result of the risk of bias assessments across studies (Appendix 2 contains individual study assessments). Overall, a high risk of bias was evaluated in 50% of the included studies, moderate in 41%, and low in 9%. This is primarily due to missing outcomes at follow-up. A high proportion of studies (59%) reported results of complete case or per-protocol analyses without appropriate consideration of the potential impact of missing data when loss-to-follow-up are unbalanced or drop-out reasons are different across allocation (ie, testing the missing data mechanism or assuming missing at random followed by imputation and/or tests for association with baseline variables and missingness and/or performing sensitivity and/or adjusted analyses).

Figure 2.

Overview of the risk of bias assessments.

Effect on HbA1c%

The results from the meta-analysis and individual studies are shown in Figure 3. The MD for HbA1c% was −0.26 (95% CI: −0.37 to −0.15) favoring telemedicine. Heterogeneity was moderate (I² = 33.30%). Studies published in North America (MD −0.71 [95% CI: −1.12 to −0.30]) or Australia/New Zealand (MD −0.78 [95% CI: −1.23 to −0.33]) and studies with no educational component (MD −0.39 [95% CI: −0.54 to −0.23]) or a proportion of man lower than 26% (MD −0.38 [95% CI: −0.54 to −0.21]) exhibited significantly larger HbA1c% treatment effects as shown in Appendix 3. Removing study outliers (Sacco and colleagues and Ahring and colleagues), because of a combination of an old publication date and relatively large treatment effects, decreased the MD to −0.25 (95% CI: −0.40 to −0.10) as seen in Appendix 4.

Figure 3.

Results from the meta-analysis.

Secondary Outcomes

Included secondary outcomes were weight loss, fasting blood glucose, the number of severe hypoglycemic events, diabetes knowledge, systolic and diastolic blood pressure, total cholesterol, high- and low-density lipoprotein, and triglycerides (Appendix 5). All outcomes favored telemedicine except the number of severe hypoglycemic events and diabetes knowledge, which was slightly in favor of the comparator, but none are statistically significant (at a 5% level).

Certainty of Findings

The certainty of evidence of the calculated treatment effect of HbA1c% was judged as moderate due to serious problems with risk of bias (Table 3). Inclusion criteria for this review were broad and evidence were therefore not downgraded due to indirectness, although studies have a relatively short follow-up. The overall effect size was statistically significant and passed the optimal information size criteria. Publication bias was not strongly suspected. The funnel plot (Figure 4) was somewhat asymmetrical favoring telemedicine and Egger’s test was insignificant (P = .1499). Industry sponsorship was declared or unknown in 14/22 studies (in 12 of these cases private companies supplied the telemedicine equipment only) and there was no relation between presence of industry sponsorship, study size, or estimated average treatment effect. No upgrades to the evidence were made. The certainty of the secondary outcomes was low for severe hypoglycemic events and weight loss and very low for fasting blood glucose, diastolic and systolic blood pressure, total cholesterol, low- and high-density lipoprotein, triglycerides, and diabetes knowledge.

Table 3.

Summary of Findings (GRADE Profile).

Certainty assessment							No. of patients		Effect		Certainty	Importance
No of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Telemedicine	Usual Care	Relative (95% CI)	Absolute (95% CI)	Certainty	Importance
HbA1c%
20	Randomized trials	Serious^a	Not serious	Not serious	Not serious	None	736	739	—	MD 0.26 % lower(0.37 lower to 0.15 lower)	⨁⨁⨁◯Moderate	IMPORTANT
Severe hypoglycemic events
7	Randomized trials	Serious^b	Not serious	Not serious	Serious^c	None	18/283 (6.4%)	6.1%	logodds 0.05(−0.63 to 0.74)		⨁⨁◯◯Low	IMPORTANT
Fasting blood glucose
4	Randomized trials	Serious^d	Serious^e	Not serious	Serious^c	None	204	204	Abbreviation:	5.42 mg/dL lower(34.69 lower to 23.84 higher)	⨁◯◯◯Very low	IMPORTANT
Diastolic blood pressure
2	Randomized trials	Serious^f	Not serious	Not serious	Very serious^g	None	130	127	—	1.12 mmHg lower(2.89 lower to 0.64 higher)	⨁◯◯◯Very low	IMPORTANT
Systolic blood pressure
2	Randomized trials	Serious^f	Not serious	Not serious	Very serious^g	None	130	127	—	MD 0.34 mmHg lower(2.99 lower to 2.31 higher)	⨁◯◯◯Very low	IMPORTANT
Total cholesterol
2	Randomized trials	Serious^f	Not serious	Not serious	Very serious^g	None	130	127	—	1.59 mg/dL lower(9.48 lower to 6.3 higher)	⨁◯◯◯Very low	IMPORTANT
High-density lipoprotein
2	Randomized trials	Serious^f	Serious^e	Not serious	Very serious^g	None	130	127	—	1.46 mg/dL lower(5.38 lower to 2.46 higher)	⨁◯◯◯Very low	IMPORTANT
Low-density lipoprotein
2	Randomized trials	Serious^f	Not serious	Not serious	Very serious^g	None	130	127	—	0.99 mg/dL lower(8.6 lower to 6.61 higher)	⨁◯◯◯Very low	IMPORTANT
Diabetes knowledge
2	Randomized trials	Serious^f	Not serious	Not serious	Very serious^g	None	98	96	—	0.03 points lower(1.2 lower to 1.14 higher)	⨁◯◯◯Very low	IMPORTANT
Triglycerides
2	Randomized trials	Serious^f	Serious^e	Not serious	Very serious^g	None	130	127	—	5.17 mg/dL lower(30.11 lower to 19.76 higher)	⨁◯◯◯Very low	IMPORTANT
Weightloss
4	Randomized trials	Serious^h	Not serious	Not serious	Serious^c	None	230	223	—	0.67 kg lower(1.67 lower to 0.33 higher)	⨁⨁◯◯Low	NOT IMPORTANT

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

Explanations.

Only 9% of studies have a low risk of bias. The correlation between the extent of risk of bias and treatment effect is statistically insignificant (Spearman rho = 0.2434, P = .3012) as are the relationship between risk of bias and study weights (Spearman rho = 0.2205, P = .3501). Evidence is downgraded one level.

Six of seven studies exhibit high risk of bias.

Very wide confidence intervals in studies, uncertain overall effect, few studies.

Three of four studies exhibit high risk of bias.

Large variation in effect sizes, little overlap in confidence intervals for some studies and large heterogeneity.

Both studies exhibit high risk of bias.

Very wide confidence intervals, uncertain effect and very few studies.

All four studies exhibit high risk of bias.

Figure 4.

Funnel plot. Abbreviation: CI, confidence interval.

Discussion

The pooled results of this review favored telemedicine (MD −0.26 [95% CI: −0.37 to −0.15]) and were robust to outliers. A moderate heterogeneity was found (I²=33.30%). Subgroup analyses revealed that studies with a relatively low proportion of men or with no educational component as well as publications from North America and Australia/New Zealand reported significantly larger effects of telemedicine. Secondary outcomes mostly favored telemedicine, but none were significant. Current evidence certainty on the primary effect on HbA1c% was moderate and lower for all secondary outcomes.

Comparison With Other Reviews

Results are comparable with the two most recent systematic reviews of telemedicine for patients with T1D. Operating with a much narrower definition of telemedicine and without certainty assessment, Wang and colleagues¹² found a similar effect in a 2019 study (MD −0.25 [95% CI: −0.41 to −0.09]) including eight RCTs from three databases. Lee and colleagues¹³ have estimated a lower treatment effect (MD −0.18 [95% CI: −0.33 to −0.04]) with a different target T1D population. They included 28 RCTs published up to 2016 that, in addition to our study, also included studies from PsycInfo. Furthermore, they concluded that the evidence of effect on HbA1c% was more uncertain (very low) because of serious problems with risk of bias, inconsistency, and indirectness. We did not find critical problems with heterogeneity and, on the balance, we conservatively did not downgrade HbA1c% for indirectness, as we found no statistically significant differences across studies with different lengths of follow-up and intervention characteristics.

Implications

The treatment effect demonstrated in this study is slightly less than could be expected of other quality improvement strategies for diabetes management (MD −0.37, derived from 120 studies).⁵² Thus, there might be other initiatives that will lead to a higher effect on participant blood glucose levels. In addition, published studies are small with a relatively short duration. As telemedicine studies are unblinded, there is a risk that treatment effects are subject to a Hawthorne effect,⁵³ which might explain a higher effect in the telemedicine group. However, if a Hawthorne effect was present, it could be expected that treatment effects would diminish over time, which cannot be confirmed (although not statistically significant, the subgroup of studies in Appendix 3 with a follow of 12 months or more report higher average effects than studies with shorter duration [effect on HbA1c% levels are −0.43 vs −0.23, respectively]). Should the average effect be maintained, important gains could be made by the estimated effect of telemedicine. A large prospective study has estimated that lowering average HbA1c levels by 1% would lead to a projected 37% reduction in long-term microvascular complications and a 21% lower mortality in people with type 2 diabetes (T2D).⁵⁴ Could these effects be transferred to T1D, telemedicine technologies could hold a potential to reduce microvascular complications by 9.6% (95% CI: 5.6%-13.7%) and mortality by 5.5% (95% CI: 3.3%-7.8%).

A plausible explanation for the higher effect in studies with a lower proportion of men could be that women have been shown to prefer more interactive diabetes resources, because they are more prone to disclose disease to their surroundings.⁵⁵ Lower effect of telemedicine with an educational component has previously been linked to the presence of various barriers.⁵⁶ However, there is no simple explanation for higher treatment effects in studies published in North America compared with Europe, other than it might just be because of the limited number of studies. Alternative causes could be variations in health care design and structure, technology maturity across markets, participant motivation not reported in the studies, and so on.

Limitations

This study has limitations. First, the limited size of the evidence base combined with a broad definition of telemedicine made it difficult to make firm conclusions about treatment effects especially in subgroups (nonsignificance) and in grading indirectness (studies implemented a variety of concrete technologies under the umbrella-term telemedicine). Second, we have included studies reported before the COVID pandemic. We acknowledge that the pandemic has led to an increased use of video technology and CGMs and that recent experience of telemedicine effectiveness during the pandemic has been published. However, including studies from the pandemic would change the scope for the review. More ad hoc implementation of telemedicine would be expected, when organizations were forced to find new ways of delivering health care leading to results (high or low treatment effects) that might not be generalizable to a context post-COVID (eg, if extraordinary resources were assigned to telemedicine operation, broader target populations were included, assigned health professionals were different). Third, data extraction could have included more baseline data from studies such as time since diagnosis, blood pressure, cholesterol levels, and medication. However, reported baseline data varied considerably and would have led to a relatively large proportion of missing study data. Fourth, for two studies, we did not have data to calculate an MD in HbA1c% because of the lack of outcome data.

Future Research

Telemedicine to people with diabetes is a complex intervention with different purposes of included components, different peripherals, mode, and frequency of contact and patient group characteristics. Moreover, telemedicine is implemented in various sectors and organizational settings. Consequently, experimental evaluation has been criticized for not being able to understand or sufficiently describe which mechanisms and contexts that would make these complex interventions lead to desired outcomes for patients—both average population effects and individually.⁵⁷ To accommodate this criticism, a qualitative strategy to increase certainty of T1D outcomes is to conduct more RCT studies that explicitly describe the background for the chosen technology configuration and a theory for how context and mechanisms could lead to which outcomes.

Future reviews should utilize that more RCTs testing CGMs compared with BGMs have been published from 2021 and onward. CGMs provide both people with T1D and health care professionals’ real-time access to glucose information, which recent meta-analyses have associated with some improvement in HbA1c% compared with BGMs.^58,59 As previously discussed, careful preplanned subgroups of studies implemented during the pandemic are suggested. Evidence grading is important, because existing CGM evidence within diabetes management in general reported a high heterogeneity, a short study duration, and comprehensive industry sponsorship.⁵⁹

Conclusions

There is moderate evidence that telemedicine lead to reduced HbA1c% levels for adults with T1D. This is important to reduce the risk of diabetic complications and premature death. However, the estimated pooled effect is smaller than could be expected from quality improvement strategies in general for diabetes management. Some additional benefits of telemedicine might exist, but the evidence for all secondary outcomes were low or very low.

Supplemental Material

sj-docx-1-dst-10.1177_19322968221076874 – Supplemental material for The Effectiveness of Telemedicine Solutions in Type 1 Diabetes Management: A Systematic Review and Meta-analysis

Supplemental material, sj-docx-1-dst-10.1177_19322968221076874 for The Effectiveness of Telemedicine Solutions in Type 1 Diabetes Management: A Systematic Review and Meta-analysis by Flemming Witt Udsen, Stine Hangaard, Clara Bender, Jonas Andersen, Thomas Kronborg, Peter Vestergaard, Ole Hejlesen and Sisse Laursen in Journal of Diabetes Science and Technology

Footnotes

Acknowledgements

The authors would like to thank research librarian, Connie Skrubbeltrang, who assisted in search of literature.

BGM, blood glucose meter; BMI, body mass index; CENTRAL, Cochrane Library Central Register of Controlled Trials; CGM, continuous glucose monitor; CI, confidence interval; GDM, gestational diabetes mellitus; HbA1c, glycated hemoglobin; MD, mean difference; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses; RCT, randomized controlled trial; SMS, short message service; SoF, summary of findings; T1D, type 1 diabetes; T2D, type 2 diabetes; WHO, World Health Organization.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This is an independent manuscript commissioned and jointly funded by the Steno Diabetes Center North Denmark and Aalborg University.

Data Sharing

No additional data.

ORCID iDs

Flemming Witt Udsen

Stine Hangaard

Clara Bender

Jonas Andersen

Sisse Laursen

Supplemental Material

Supplemental material for this article is available online.

References

Powers

Bardsley

Cypress

, et al. Diabetes self-management education and support in type 2 diabetes: a joint position statement of the American Diabetes Association, the American Association of Diabetes Educators, and the Academy of Nutrition and Dietetics. Clin Diabetes. 2016;34(2):70-80. doi:10.1016/j.jand.2015.05.012.

Beck

Greenwood

Blanton

, et al. 2017 national standards for diabetes self-management education and support. Diabetes Educ. 2017;43(5):449-464.

Borries

Dunbar

Bhukhen

, et al. The impact of telemedicine on patient self-management processes and clinical outcomes for patients with Types I or II Diabetes Mellitus in the United States: a scoping review. Diabetes Metab Syndr Clin Res Rev. 2019;13(2):1353-1357. doi:10.1016/j.dsx.2019.02.014

Whitehead

Seaton

. The effectiveness of self-management mobile phone and tablet apps in long-term condition management: a systematic review. J Med Internet Res. 2016;18(5):e97.

Sood

Mbarika

Jugoo

, et al. What is telemedicine? a collection of 104 peer-reviewed perspectives and theoretical underpinnings. Telemed J E Health. 2007;13(5):573-590. http://www.liebertonline.com/doi/pdfplus/10.1089/tmj.2006.0073%5Cn http://www.ncbi.nlm.nih.gov/pubmed/17999619. Accessed January 22, 2022.

World Health Organization Global Observatory for eHealth. Telemedicine: opportunities and developments in Member States. Observatory. 2010;2:96.

McLean

Liu

Pagliari

Car

Sheikh

. Telehealthcare for asthma. Cochrane Database Syst Rev. 2009;2. http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L355267549. Accessed January 22, 2022.

McLean

Nurmatov

Liu

JLY

Pagliari

Car

Sheikh

. Telehealthcare for chronic obstructive pulmonary disease. Cochrane Database Syst Rev. 2011;(7):1-52. http://www.ncbi.nlm.nih.gov/pubmed/21735417. Accessed January 22, 2022.

Klonoff

. Telemedicine for diabetes : current and future trends. J Diabetes Sci Technol. 2016;10(1):3-5.

10.

Hanlon

Daines

Campbell

McKinstry

Weller

Pinnock

. Telehealth interventions to support self-management of long-term conditions: a systematic metareview of diabetes, heart failure, asthma, chronic obstructive pulmonary disease, and cancer. J Med Internet Res. 2017;19(5):e172.

11.

Baron

Mcbain

Newman

Psych

Hons

. The impact of mobile monitoring technologies on glycosylated hemoglobin in diabetes: a systematic review. J Diabetes Sci Technol. 2012;6(5):1185-1196.

12.

Wang

Shu

, et al. Mobile health in the management of type 1 diabetes: a systematic review and meta-analysis. BMC Endocr Disord. 2019;19(1):1-10.

13.

Lee

SWH

Ooi

Lai

. Telemedicine for the management of glycemic control and clinical outcomes of type 1 diabetes mellitus: a systematic review and meta-analysis of randomized controlled studies. Front Pharmacol. 2017;8:1-8.

14.

Jaana

Paré

. Home telemonitoring of patients with diabetes: a systematic assessment of observed effects. J Eval Clin Pract. 2007;13(2):242-253.

15.

Faruque

Wiebe

Ehteshami-Afshar

, et al. Effect of telemedicine on glycated hemoglobin in diabetes: a systematic review and meta-analysis of randomized trials. CMAJ. 2017;189(9):E341-E364.

16.

Wen

Wang

, et al. Effect of telemedicine intervention on hypoglycaemia in diabetes patients: a systematic review and meta-analysis of randomised controlled trials. J Telemed Telecare. 2019;25(7):402-413.

17.

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372.

18.

Laursen

Hangaard

Udsen

Vestergaard

Hejlesen

. Effectiveness of telemedicine solutions for the management of patients with diabetes: protocol for a systematic review and meta-analysis. JMIR Res Protoc. 2020;9(11):1-7.

19.

Wan

Wang

Liu

Tong

. Estimating the sample mean and standard deviation from the sample size, median, range and/or interquartile range. BMC Med Res Methodol. 2014;14(1):1-13.

20.

Hozo

Djulbegovic

Hozo

. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol. 2005;5:1-10.

21.

Sterne

JAC

Savović

Page

, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:1-8.

22.

Guyatt

Thorlund

Oxman

, et al. GRADE guidelines: 13. Preparing Summary of Findings tables and evidence profiles—continuous outcomes. J Clin Epidemiol. 2013;66(2):173-183. doi:10.1016/j.jclinepi.2012.08.001

23.

Guyatt

Oxman

Sultan

, et al. GRADE guidelines: 11. Making an overall rating of confidence in effect estimates for a single outcome and for all outcomes. J Clin Epidemiol. 2013;66(2):151-157.

24.

Schünemann

Brozek

Guyatt

Oxman

. Handbook for grading the quality of evidence and the strength of recommendations using the GRADE approach. https://gdt.gradepro.org/app/handbook/handbook.html. Published 2013. Accessed January 22, 2022.

25.

Guyatt

Oxman

Vist

, et al. GRADE guidelines: 4. Rating the quality of evidence—study limitations (risk of bias). J Clin Epidemiol. 2011;64(4):407-415.

26.

Guyatt

Oxman

Kunz

, et al. GRADE guidelines: 8. Rating the quality of evidence—indirectness. J Clin Epidemiol. 2011;64(12):1303-1310.

27.

Guyatt

Oxman

Kunz

, et al. GRADE guidelines 6. Rating the quality of evidence—imprecision. J Clin Epidemiol. 2011;64(12):1283-1293.

28.

Guyatt

Oxman

Kunz

, et al. GRADE guidelines: 7. Rating the quality of evidence—inconsistency. J Clin Epidemiol. 2011;64(12):1294-1302.

29.

Guyatt

Oxman

Montori

, et al. GRADE guidelines: 5. Rating the quality of evidence—publication bias. J Clin Epidemiol. 2011;64(12):1277-1282.

30.

Ahring

JPK

Joyce

Farid

. Telephone modem access improves diabetes control in those with insulin- requiring diabetes. Diabetes Care. 1992;15(8):971-975.

31.

Benhamou

Melki

Boizel

, et al. One-year efficacy and safety of Web-based follow-up using cellular phone in type 1 diabetic patients under insulin pump therapy: the PumpNet study. Diabetes Metab. 2007;33(3):220-226.

32.

Bertuzzi

Stefani

Rivolta

, et al. Teleconsultation in type 1 diabetes mellitus (TELEDIABE). Acta Diabetol. 2018;55(2):185-192. https://doi.org/10.1007/s00592-017-1084-9

33.

Charpentier

Benhamou

Dardari

, et al. The Diabeo software enabling individualized insulin dose adjustments combined with telemedicine support improves HbA1cin poorly controlled type 1 diabetic patients: a 6-month, randomized, open-label, parallel-group, multicenter trial (TeleDiab 1 study). Diabetes Care. 2011;34(3):233-239.

34.

Drion

Pameijer

Van Dijk

Groenier

Kleefstra

Bilo

HJG

. The effects of a mobile phone application on quality of life in patients with type 1 diabetes mellitus: a randomized controlled trial. J Diabetes Sci Technol. 2015;9(5):1086-1091.

35.

Esmatjes

Jansà

Roca

, et al. The efficiency of telemedicine to optimize metabolic control in patients with type 1 diabetes mellitus: telemed study. Diabetes Technol Ther. 2014;16(7):435-441. http://online.liebertpub.com/doi/abs/10.1089/dia.2013.0313. Accessed January 22, 2022

36.

Farmer

Gibson

Dudley

, et al. A randomized controlled trial of the effect of real-time telemedicine support on glycemic control in young adults with type 1 diabetes ({ISRCTN} 46889446). Diabetes Care. 2005;28(11):2697-702. http://ezproxy.library.uvic.ca/login?url=http://search.ebscohost.com/login.aspx?direct=true&db=ccm&AN=106351263&site=ehost-live&scope=site. Accessed January 22, 2022.

37.

Gómez

Hernando

García

, et al. Telemedicine as a tool for intensive management of diabetes: the DIABTel experience. Comput Methods Programs Biomed. 2002;69(2):163-177.

38.

Jansà

Vidal

Viaplana

, et al. Telecare in a structured therapeutic education programme addressed to patients with type 1 diabetes and poor metabolic control. Diabetes Res Clin Pract. 2006;74(1):26-32.

39.

Kirwan

Vandelanotte

Fenning

Duncan

. Diabetes self-management smartphone application for adults with type 1 diabetes: randomized controlled trial. J Med Internet Res. 2013;15(11):e235.

40.

Kotsani

Antonopoulou

Kountouri

, et al. The role of telenursing in the management of Diabetes Type 1: a randomized controlled trial. Int J Nurs Stud. 2018;80(April 2017):29-35. doi:10.1016/j.ijnurstu.2018.01.003

41.

McCarrier

Ralston

Hirsch

, et al. Web-based collaborative care for type 1 diabetes: a pilot randomized trial. Diabetes Technol Ther. 2009;11(4):211-217.

42.

Menting

Tack

van Bon

, et al. Web-based cognitive behavioural therapy blended with face-to-face sessions for chronic fatigue in type 2 diabetes: a multicentre randomised controlled trial. Lancet Diabetes Endocrinal. 2017;5:448-456.

43.

Montori

Helgemoe

Guyatt

, et al. Telecare for patients with type 1 diabetes and inadequate glycemic. Diabetes Care. 2004;27:1088-1094.

44.

Rigla

Elena Hernando

Gómez

, et al. A telemedicine system that includes a personal assistant improves glycemic control in pump-treated patients with type 1 diabetes. J Diabetes Sci Technol. 2007;1(4):505-510.

45.

Rossi

MCE

Nicolucci

Di Bartolo

, et al. Diabetes interactive diary: a new telemedicine system enabling flexible diet and insulin therapy while improving quality of life: an open-label, international, multicenter, randomized study. Diabetes Care. 2010;33(1):109-115.

46.

Rossi

Nicolucci

Lucisano

, et al. Impact of the “Diabetes interactive diary” telemedicine system on metabolic control, risk of hypoglycemia, and quality of life: a randomized clinical trial in type 1 diabetes. Diabetes Technol Ther. 2013;15(8):670-679. http://online.liebertpub.com/doi/abs/10.1089/dia.2013.0021. Accessed January 22, 2022.

47.

Sacco

Morrison

Malone

. A brief, regular, proactive telephone “coaching” intervention for diabetes: rationale, description, and preliminary results. J Diabetes Complications. 2004;18(2):113-118.

48.

Suh

Park

Cho

, et al. A Randomized controlled trial of the effect of internet based mentoring program for type 1 diabetes patients with inadequate glycemic control. Diabetes Metab J [Internet]. 2014;38:134-142. http://www.endocrine-abstracts.org/ea/0029/ea0029p685.htm. Accessed January 22, 2022.

49.

Hejlesen

Andreassen

Frandsen

, et al. Using a double blind controlled clinical trial to evaluate the function of a Diabetes Advisory System: a feasible approach? Comput Methods Programs Biomed. 1998;56(2):165-173.

50.

Linden

Berg

Adolfsson

Sparud-Lundin

. Person-centred, web-based support in pregnancy and early motherhood for women with Type 1 diabetes mellitus: a randomized controlled trial. Diabet Med. 2018;35(2):232-241.

51.

Yaron

Sher

Sorek

, et al. A randomized controlled trial comparing a telemedicine therapeutic intervention with routine care in adults with type 1 diabetes mellitus treated by insulin pumps. Acta Diabetol. 2019;56(6):667-673. doi:10.1007/s00592-019-01300-1

52.

Tricco

Ivers

Grimshaw

, et al. Effectiveness of quality improvement strategies on the management of diabetes: a systematic review and meta-analysis. Lancet. 2012;379(9833):2252-2261.

53.

Adair

. The Hawthorne effect: a reconsideration of the methodological artifact. J Appl Psychol. 1984;69:334-345.

54.

Stratton

Adler

Neil

, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321. doi:10.1016/j.resuscitation.2010.10.016.

55.

Mathew

Gucciardi

De Melo

Barata

. Self-management experiences among men and women with type 2 diabetes mellitus: a qualitative analysis. BMC Fam Pract. 2012;13(1):1.

56.

Harris

Miller

Amiel

Mulnier

. Characterization of adults with type 1 diabetes not attending self-management education courses: the Barriers to Uptake of Type 1 Diabetes Education (BUD1E) Study. Qual Health Res. 2019;29(8):1174-1185.

57.

Pawson

Tilley

. Realistic Evaluation. London: Sage; 1997.

58.

Kłak

Mańczak

Owoc

Olszewski

. Impact of continuous glucose monitoring on improving emotional well-being among adults with type 1 diabetes mellitus: a systematic review and meta-analysis. Polish Arch Intern Med. 2021;131(9):808-818.

59.

Maiorino

Signoriello

Maio

, et al. Effects of continuous glucose monitoring on metrics of glycemic control in diabetes: a systematic review with meta-analysis of randomized controlled trials. Diabetes Care. 2020;43(5):1146-1156.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

3.20 MB