OBJECTIVE

Young adults with type 1 diabetes (T1D) often struggle to achieve glycemic control and maintain routine clinic visits. We aimed to evaluate the societal cost-effectiveness of the Colorado young adults with T1D (CoYoT1) Clinic, an innovative care model of shared medical appointments through home telehealth.

RESEARCH DESIGN AND METHODS

Patients self-selected into the CoYoT1 (N = 42) or usual care (N = 39) groups.

RESULTS

Within the trial, we found no significant differences in 9-month quality-adjusted life; however, the control group had a larger decline from baseline in utility than the CoYoT1 group, indicating a quality of life (QoL) benefit of the intervention (difference in difference mean ± SD: 0.04 ± 0.09; P = 0.03). There was no significant difference in total costs. The CoYoT1 group had more study-related visits but fewer nonstudy office visits and hospitalizations.

CONCLUSIONS

The CoYoT1 care model may help young adults with T1D maintain a higher QoL with no increase in costs.

The absolute numbers of young adults with type 1 diabetes (T1D) are on the rise (1). The transition period from pediatric to adult care is challenging and frequently accompanied by missed clinic visits and suboptimal glycemic control (26). An innovative care model—shared medical appointments delivered through home telehealth—was evaluated by the recent Colorado Young Adults with Type 1 Diabetes (CoYoT1) trial. The trial demonstrated that the care model improved patient attendance and diabetes care engagement (3,7). We aimed to evaluate the societal cost-effectiveness of the CoYoT1 model versus usual care (control).

In this prospective pragmatic trial, patients with T1D aged 18–25 years self-selected into either the CoYoT1 or control groups at the Barbara Davis Center for Diabetes. During the trial, we collected patients’ quality of life (QoL) assessed by the EuroQol five-level five-dimension questionnaire, self-reported health care utilization, and clinical staff time related to group and/or individual visits at baseline, 3, 6, and 9 months. Main outcomes included health-related utility, quality-adjusted life years (QALYs), and total costs. Details on the intervention and the clinical findings have been previously published (3,7). We have provided an impact inventory table (8) and reporting checklist (9) in Supplementary Tables 1 and 2.

The 9-month total costs included 1) all direct costs associated with trial staff time as part of the study, health care utilization that occurred outside of the study, device use (continuous glucose monitoring [CGM] and/or pump), and test strip use and 2) all indirect costs associated with reduced work productivity and commute time for an in-person clinic visit, if employed. We calculated costs by multiplying the U.S. Bureau of Labor Statistics median hourly wages (or prices per service) by hours spent (or number of services used) in the 9-month time period. All cost assumptions are provided in Supplementary Table 3. All costs are expressed in 2015 U.S. dollars.

We applied the intent-to-treat principle to all analyses. The Wilcoxon test and the Fisher exact test were used for group comparison as appropriate. We used the ANCOVA method to compare QALYs, adjusting for baseline utility (10). We used linear mixed models to model repeated-measures outcomes and to test effects of treatment, time, and their interaction, respectively. To account for baseline imbalanced costs (11), we used the bootstrap method to calculate mean difference in difference and its 95% CI. We also conducted subgroup analyses per baseline HbA1c level >8.0% and <8.0%.

Eighty-one patients participated in the study, 42 in the CoYoT1 group and 39 in the control group. The CoYoT1 group had a shorter duration of diabetes than the control group, but all other major baseline characteristics were balanced (Supplementary Table 6).

Compared with the control group, the CoYoT1 group had a smaller decline in utility from baseline (mean ± SD: −0.03 ± 0.06 vs. −0.07 ± 0.10; P = 0.03) and less diabetes-related distress (P < 0.01) (Table 1). Nine-month QALYs were similar: 0.70 ± 0.05 years (CoYoT1) vs. 0.68 ± 0.08 years (control) (P = 0.86).

Table 1

Within-trial cost-effectiveness analysis results

CoYoT1 (n = 42)
Control (n = 39)
Pa
Utility and QALYs Mean (SD) Median (range) Mean (SD) Median (range)
Utility at 9 months 0.87 (0.11) 0.90 (0.55, 1.0) 0.82 (0.17) 0.84 (0.39, 1.0) 0.03b
QALYs 0.70 (0.05) 0.70 (0.56, 0.75) 0.68 (0.08) 0.69 (0.46, 0.75) 0.86c
Diabetes distress scale at 9 months 1.78 (0.72) 1.65 (1.0, 3.65) 2.18 (0.69) 2.15 (1.12, 3.65) <0.01b
Per-patient costs ($) Mean (SD) Median (IQR) Mean (SD) Median (IQR) Total direct costs 4,024 (2,471) 3,930 (1,973, 5,545) 8,625 (18,442) 3,996 (1,072, 4,903) 0.68 Trial staff for intervention/control 198 (55) 220 (161, 238) 54 (69) 52 (0, 77) <0.01 Other medical care 201 (394) 58 (0, 199) 3,488 (14,185) 241 (0, 498) 0.02 Strip test use 1,033 (958) 680 (472, 1,070) 975 (529) 816 (544, 1,361) 0.38 Pump use 1,365 (1,269) 1,063 (0, 2,127) 741 (1,264) 0 (0, 1,595) 0.03 CGM use 1,018 (1,391) 0 (0, 1,277) 1,111 (1,695) 0 (0, 3,830) 0.80 Total indirect costs 248 (419) 22 (10, 326) 694 (2,303) 19 (0, 325) 0.43 Missed work 119 (301) 0 (0, 0) 278 (767) 0 (0, 242) 0.58 Poor performance 91 (219) 0 (0, 121) 406 (1,559) 0 (0, 182) 0.30 Total commute time for in-person clinic visits 17 (11) 15 (9, 20) 11 (16) 5 (0, 15) 0.01 Total costs 4,257 (2,590) 4,228 (2,139, 6,061) 8,929 (18,348) 4,271 (2,035, 5,497) 0.79 Clinical variables at 9 months Mean (SD) Median (range) Mean (SD) Median (range) HbA1c 8.40 (1.54) 8.10 (5.8, 11.4) 8.08 (0.95) 7.8 (6.9, 10.3) 0.63b BMI 25.16 (4.54) 25.2 (18.4, 39.0) 25.37 (4.62) 23.7 (19.5, 33.6) 0.18b Number of patients having severe hyperglycemia events (%) 3 (9) 0.11d Number of patients having severe hypoglycemia events (%) 1 (3) 2 (6) 0.61d Number of study visits 3.45 (1.04) 4 (1, 4) 0.64 (0.71) 1 (0, 2) <0.01 Daily strip tests 5.11 (6.89) 3.65 (0.9, 32.7) 3.35 (1.81) 3.2 (0.9, 6.0) 0.61b Pump use: yes (%) 14 (47) 4 (36) 0.73d CGM use: yes (%) 11 (37) 3 (30) 1.00d Subgroup analyses Mean (SD) Median (range) Mean (SD) Median (range) In the subgroup with high baseline HbA1c (≥8.0%) (n = 43) Utility at 9 months 0.88 (0.12) 0.90 (0.59, 1.0) 0.82 (0.15) 0.84 (0.45, 1.0) 0.016b HbA1c at 9 months 9.3 (1.41) 9.25 (7.4, 11.4) 8.5 (1.09) 8.25 (7.5, 10.3) 0.41b Number of clinical visits 3.38 (1.10) 4 (1, 4) 0.53 (0.61) 0 (0, 2) <0.01 Diabetes distress scale at 9 months 1.96 (0.83) 1.76 (1.06, 3.65) 2.07 (0.57) 1.94 (1.23, 3.18) 0.046b In the subgroup with low baseline HbA1c (<8.0%) (n = 34) Utility at 9 months 0.87 (0.12) 0.87 (0.55, 1.0) 0.81 (0.20) 0.86 (0.39, 1.0) 0.71b HbA1c at 9 months 7.41 (0.99) 7.6 (5.8, 9.0) 7.58 (0.44) 7.6 (6.9, 8.1) 0.37b Number of clinical visits 4 (0) 4 (4, 4) 0.75 (0.79) 1 (0, 2) <0.01 Diabetes distress scale at 9 months 1.47 (0.33) 1.47 (1.0, 2.0) 2.29 (0.81) 2.21 (1.12, 3.65) <0.01b CoYoT1 (n = 42) Control (n = 39) Pa Utility and QALYs Mean (SD) Median (range) Mean (SD) Median (range) Utility at 9 months 0.87 (0.11) 0.90 (0.55, 1.0) 0.82 (0.17) 0.84 (0.39, 1.0) 0.03b QALYs 0.70 (0.05) 0.70 (0.56, 0.75) 0.68 (0.08) 0.69 (0.46, 0.75) 0.86c Diabetes distress scale at 9 months 1.78 (0.72) 1.65 (1.0, 3.65) 2.18 (0.69) 2.15 (1.12, 3.65) <0.01b Per-patient costs ($) Mean (SD) Median (IQR) Mean (SD) Median (IQR)
Total direct costs 4,024 (2,471) 3,930 (1,973, 5,545) 8,625 (18,442) 3,996 (1,072, 4,903) 0.68
Trial staff for intervention/control 198 (55) 220 (161, 238) 54 (69) 52 (0, 77) <0.01
Other medical care 201 (394) 58 (0, 199) 3,488 (14,185) 241 (0, 498) 0.02
Strip test use 1,033 (958) 680 (472, 1,070) 975 (529) 816 (544, 1,361) 0.38
Pump use 1,365 (1,269) 1,063 (0, 2,127) 741 (1,264) 0 (0, 1,595) 0.03
CGM use 1,018 (1,391) 0 (0, 1,277) 1,111 (1,695) 0 (0, 3,830) 0.80
Total indirect costs 248 (419) 22 (10, 326) 694 (2,303) 19 (0, 325) 0.43
Missed work 119 (301) 0 (0, 0) 278 (767) 0 (0, 242) 0.58
Poor performance 91 (219) 0 (0, 121) 406 (1,559) 0 (0, 182) 0.30
Total commute time for in-person clinic visits 17 (11) 15 (9, 20) 11 (16) 5 (0, 15) 0.01
Total costs 4,257 (2,590) 4,228 (2,139, 6,061) 8,929 (18,348) 4,271 (2,035, 5,497) 0.79
Clinical variables at 9 months Mean (SD) Median (range) Mean (SD) Median (range)
HbA1c 8.40 (1.54) 8.10 (5.8, 11.4) 8.08 (0.95) 7.8 (6.9, 10.3) 0.63b
BMI 25.16 (4.54) 25.2 (18.4, 39.0) 25.37 (4.62) 23.7 (19.5, 33.6) 0.18b
Number of patients having severe hyperglycemia events (%)  3 (9)  0.11d
Number of patients having severe hypoglycemia events (%) 1 (3)  2 (6)  0.61d
Number of study visits 3.45 (1.04) 4 (1, 4) 0.64 (0.71) 1 (0, 2) <0.01
Daily strip tests 5.11 (6.89) 3.65 (0.9, 32.7) 3.35 (1.81) 3.2 (0.9, 6.0) 0.61b
Pump use: yes (%) 14 (47)  4 (36)  0.73d
CGM use: yes (%) 11 (37)  3 (30)  1.00d
Subgroup analyses Mean (SD) Median (range) Mean (SD) Median (range)
In the subgroup with high baseline HbA1c (≥8.0%) (n = 43)
Utility at 9 months 0.88 (0.12) 0.90 (0.59, 1.0) 0.82 (0.15) 0.84 (0.45, 1.0) 0.016b
HbA1c at 9 months 9.3 (1.41) 9.25 (7.4, 11.4) 8.5 (1.09) 8.25 (7.5, 10.3) 0.41b
Number of clinical visits 3.38 (1.10) 4 (1, 4) 0.53 (0.61) 0 (0, 2) <0.01
Diabetes distress scale at 9 months 1.96 (0.83) 1.76 (1.06, 3.65) 2.07 (0.57) 1.94 (1.23, 3.18) 0.046b
In the subgroup with low baseline HbA1c (<8.0%) (n = 34)
Utility at 9 months 0.87 (0.12) 0.87 (0.55, 1.0) 0.81 (0.20) 0.86 (0.39, 1.0) 0.71b
HbA1c at 9 months 7.41 (0.99) 7.6 (5.8, 9.0) 7.58 (0.44) 7.6 (6.9, 8.1) 0.37b
Number of clinical visits 4 (0) 4 (4, 4) 0.75 (0.79) 1 (0, 2) <0.01
Diabetes distress scale at 9 months 1.47 (0.33) 1.47 (1.0, 2.0) 2.29 (0.81) 2.21 (1.12, 3.65) <0.01b

Statistically significant P values appear in boldface type (P < 0.05). IQR, interquartile range.

aThe default statistical method was Wilcoxon test.

bA linear mixed model was used to compare the groups, adjusting its baseline outcome. The P value is for group comparison across all visits.

cAn ANCOVA was used to test the treatment effect, adjusting its baseline utility.

dA Fisher exact test was used to compare the groups.

The per-person 9-month mean total costs were $4,257 ± 2,590 for the CoYoT1 group and$8,929 ± 18,348 for the control group (P < 0.79) (Table 1). The difference in difference for total costs was −$2,965 (95% CI −$12,199, \$2,777) (Supplementary Tables 7 and 8) and not statistically significant. The CoYoT1 group had more study-related visits but fewer nonstudy office visits (means: 1.27 vs. 3.0; P = 0.01) and hospitalizations (mean frequencies: 0.0 vs. 0.23; two-sided P = 0.15) than the control group (Supplementary Table 9). For key clinical outcomes, including HbA1c, BMI, and number of severe hyperglycemia (and hypoglycemia) events, we found no significant differences. No within-trial incremental cost-effectiveness ratio was calculated due to the lack of significant difference in 9-month total costs or QALYs.

In the subgroup analyses, among patients with high baseline HbA1c (≥8.0%), the CoYoT1 group experienced a small reduction in utility from baseline and maintained diabetes distress scores over time, while control subjects had a greater reduction in utility (P = 0.016) and an increase in diabetes distress (P = 0.046). Among patients with low baseline HbA1c (<8.0%), the CoYoT1 had a reduction in their diabetes distress score by 0.5, whereas control subjects had an increase in their distress score by 0.4 (P < 0.01). In both subgroup analyses, HbA1c were not different for intervention and control (P = 0.41 and 0.37).

Young adults with T1D suffer from poor health outcomes, with only 14% of this population meeting the American Diabetes Association’s HbA1c goal of <7.0% (6). Efforts to improve health outcomes in this population have focused on developing new systems of care that may improve the transition between pediatric and adult medicine (5). Our study is the first to evaluate the societal cost-effectiveness of the CoYoT1 care model, a combination of telemedicine and shared medical appointments, compared with usual care in transition-age young adults with T1D.

During the trial, the CoYoT1 group maintained a higher QoL over time than the control group. In addition, the CoYoT1 group tended to have lower (nonsignificant) health care costs with fewer nonstudy office visits (i.e., urgent care visits) and hospitalizations (nonsignificant). To forecast the long-term implications of the QoL findings, we used the Sheffield model (12) to simulate the patient-level natural history of T1D over the projected lifetime of patients. We found that if the QoL benefits were to persist over a lifetime, there would be a gain of 0.95 QALYs. The lifetime base-case, subgroup, and sensitivity cost-effectiveness analyses were all consistent with each other (Supplementary Tables 1114).

The clinical findings from our trial suggest that the combination of home telemedicine and shared medical appointments is a safe and efficient method for delivering care to young adults with T1D. The model improved clinic follow-up and patient appointment satisfaction, resulting in increased young adult engagement in care (3,7). These features of CoYoT1 likely reduced patients’ diabetes-related distress and helped maintain higher QoL (13). While CoYoT1 enhanced patients’ QoL and increased CGM use (7), we did not find significant improvements in glucose control. This is consistent with a recent meta-analysis and systematic review of telemedicine use among patients with T1D, which concluded that there was insufficient evidence to support telemedicine use for glucose control with a mild reduction in HbA1c (0.18%) and found that studies with longer duration were associated with larger effects (14).

Our study has limitations. First, a sample selection bias might still exist because patients self-selected for participation in CoYoT1. However, the major demographic characteristics of the study groups were balanced. Second, our study may be underpowered because of missing data. We used the multiple imputation method to address the problem of missing data, and its results (Supplementary Table 10) were consistent with our main findings.

Based on this single-center trial, the CoYoT1 care model may help transition-age young adults with T1D maintain a higher QoL with no increase in costs, with an accompanying shift to more routine diabetes care while decreasing acute care visits (e.g., urgent care, emergency department, and hospitalizations). Additional trials with larger patient numbers, longer-term follow-up, and more structured training for shared telemedicine visits are needed.

Funding. This study was supported by grants from Helmsley Charitable Trust (2015 PG-T1D059) and the National Institute of Diabetes and Digestive and Kidney Diseases (P30-DK-092949 and K24-DK-105340 [to E.S.H.]).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. W.W., A.G.N., M.R.S., P.Z., M.W.R., J.K.R., and E.S.H. contributed to study concept and design; were responsible for acquisition, analysis, or interpretation of data and critical revision of the manuscript for important intellectual content; and provided administrative, technical, or material support and study supervision. W.W. and M.R.S. were responsible for statistical analysis. W.W. and E.S.H. were responsible for drafting of the manuscript. J.K.R. and E.S.H. obtained funding. W.W., M.R.S., M.W.R., J.K.R., and E.S.H. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in poster form at the 78th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018.

1.
Peters
A
,
Laffel
L
;
American Diabetes Association Transitions Working Group
.
Diabetes care for emerging adults: recommendations for transition from pediatric to adult diabetes care systems: a position statement of the American Diabetes Association, with representation by the American College of Osteopathic Family Physicians, the American Academy of Pediatrics, the American Association of Clinical Endocrinologists, the American Osteopathic Association, the Centers for Disease Control and Prevention, Children with Diabetes, The Endocrine Society, the International Society for Pediatric and Adolescent Diabetes, Juvenile Diabetes Research Foundation International, the National Diabetes Education Program, and the Pediatric Endocrine Society (formerly Lawson Wilkins Pediatric Endocrine Society)
.
Diabetes Care
2011
;
34
:
2477
2485
[PubMed]
2.
Weigensberg
MJ
,
Vigen
C
,
Sequeira
P
, et al
.
Diabetes Empowerment Council: integrative pilot intervention for transitioning young adults with type 1 diabetes
.
2018
;
7
:2164956118761808
[PubMed]
3.
Raymond
JK
,
Berget
CL
,
Driscoll
KA
,
Ketchum
K
,
Cain
C
,
Fred Thomas
JF
.
CoYoT1 clinic: innovative telemedicine care model for young adults with type 1 diabetes
.
Diabetes Technol Ther
2016
;
18
:
385
390
[PubMed]
4.
Viiginia
AL
.
The challenges of diabetes management for emerging young adults
.
Diabetes Spectr
2011
;
24
:
4
5
5.
Los
E
,
Ulrich
J
,
Guttmann-Bauman
I
.
Technology use in transition-age patients with type 1 diabetes: reality and promises
.
J Diabetes Sci Technol
2016
;
10
:
662
668
[PubMed]
6.
Miller
KM
,
Foster
NC
,
Beck
RW
, et al.;
T1D Exchange Clinic Network
.
Current state of type 1 diabetes treatment in the U.S.: updated data from the T1D Exchange clinic registry
.
Diabetes Care
2015
;
38
:
971
978
[PubMed]
7.
Reid
MW
,
Krishnan
S
,
Berget
C
, et al
.
CoYoT1 clinic: home telemedicine increases young adult engagement in diabetes care
.
Diabetes Technol Ther
2018
;
20
:
370
379
[PubMed]
8.
Sanders
GD
,
Neumann
PJ
,
Basu
A
, et al
.
Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: second panel on cost-effectiveness in health and medicine
[published correction appears in JAMA 2016;316:1924].
JAMA
2016
;
316
:
1093
1103
[PubMed]
9.
Husereau
D
,
Drummond
M
,
Petrou
S
, et al.;
.
Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement
.
Value Health
2013
;
16
:
e1
e5
[PubMed]
10.
Manca
A
,
Hawkins
N
,
Sculpher
MJ
.
Estimating mean QALYs in trial-based cost-effectiveness analysis: the importance of controlling for baseline utility
.
Health Econ
2005
;
14
:
487
496
[PubMed]
11.
van Asselt
,
van Mastrigt
GA
,
Dirksen
CD
,
Arntz
A
,
Severens
JL
,
Kessels
AG
.
How to deal with cost differences at baseline
.
Pharmacoeconomics
2009
;
27
:
519
528
[PubMed]
12.
Henriksson
M
,
Jindal
R
,
Sternhufvud
C
,
Bergenheim
K
,
E
,
Willis
M
.
A systematic review of cost-effectiveness models in type 1 diabetes mellitus
.
Pharmacoeconomics
2016
;
34
:
569
585
[PubMed]
13.
Chew
BH
,
Mohd-Sidik
S
,
Shariff-Ghazali
S
.
Negative effects of diabetes-related distress on health-related quality of life: an evaluation among the adult patients with type 2 diabetes mellitus in three primary healthcare clinics in Malaysia
.
Health Qual Life Outcomes
2015
;
13
:
187
[PubMed]
14.
Lee
SWH
,
Ooi
L
,
Lai
YK
.
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
:
330
[PubMed]