Case Management with a Diabetes Team Using Home Telemedicine: Acceptance of Treatment Recommendations by Primary Care Providers in IDEATel

Abstract

Background: The Informatics for Diabetes Education and Telemedicine (IDEATel) project demonstrated that a telemedicine intervention can improve glycemic, lipid, and blood pressure control. The focus of the current study was to evaluate factors associated with primary care providers' (PCPs') decision on whether to follow recommendations from the remote diabetes team in Upstate New York. Materials and Methods: In the intervention group, diabetes educators videoconferenced with patients monthly to download and review glucose and blood pressure readings, diabetes-related issues, and laboratory data. These were reviewed with an endocrinologist, and recommendations to change therapy were sent to the PCPs. At annual visits, participants completed the Diabetes Symptom Checklist-Type 2 symptom severity score and Impact of Telemedicine surveys. Results: Factors that increase the acceptance rate of IDEATel recommendations included longer time in the study (p=0.0052), changing medication dose as opposed to starting or stopping a medication (p<0.0001), adjusting glucose-lowering agents compared with antihypertensive or antilipid medications (p<0.0001), higher total Diabetes Symptom Checklist-Type 2 symptom severity score (p=0.045), greater number of glucose readings submitted by participants (p=0.014), and high score on surveys measuring impact of telemedicine on patient's knowledge, adherence, and satisfaction (p=0.0023). Conclusions: Recommendations for change in glycemic control medications, delivered remotely by a diabetes team to PCPs, were better accepted over time. Results support the use of a team-based telemedicine program to help PCPs improve diabetes care.

Background

Telemedicine can improve healthcare delivery to underserved populations and can be particularly helpful in the management of chronic diseases, such as diabetes mellitus. By increasing patient access to services, socioeconomic, geographic, and weather-related barriers can be overcome.

The Informatics for Diabetes Education and Telemedicine (IDEATel) demonstration project, funded by the Centers for Medicare and Medicaid Services, evaluated the feasibility, acceptability, and effectiveness of a home telemedicine intervention to improve diabetes care in ethnically diverse, medically underserved, older adults.^1

–4 IDEATel was a randomized trial comparing telemedicine case management with usual care in Medicare beneficiaries with diabetes mellitus residing in medically underserved areas of New York State. In the intervention group, nurse and dietitian case managers, trained in diabetes management and in the use of computer-based case management tools, videoconferenced with patients every 4–6 weeks to review glucose readings, diabetes-related issues, blood pressure readings, and laboratory data. The visit information was reviewed with an endocrinologist, and therapeutic recommendations were made to the primary care providers (PCPs), who retained full responsibility and control over their patients' care. Intention-to-treat mixed models showed that the telemedicine intervention reduced hemoglobin A1c (p=0.001), low-density lipoprotein cholesterol (p<0.001), and systolic and diastolic blood pressure (p=0.024 and p<0.001, respectively) over a follow-up period of 5 years.⁴

The focus of this study is the interaction between the PCPs and the diabetes case management team in Upstate New York during the telemedicine intervention over time. Almost all of the PCPs in the Upstate New York cohort had no prior association with the diabetes team in Syracuse, NY. The goal was to better understand the factors that affected the PCPs' decision on whether to follow case management team's recommendations. This is the first report to examine this issue, which is particularly important as expansion of telemedicine services to rural areas is being considered across the United States.

Materials and Methods

Design

The IDEATel study was conducted as a randomized controlled trial with blinded assessment of the primary outcomes. As previously described,^1

–4 subjects were enrolled through primary care practices in New York City, with the enrollment and intervention hub at Columbia University Medical Center, and in Upstate New York, where the enrollment and intervention hub was at State University of New York (SUNY) Upstate Medical University at Syracuse. The current study reports data from the Upstate New York (rural) cohort. The Institutional Review Boards at all participating institutions approved the study protocol. All participants provided informed consent. An independent Data and Safety Monitoring Board monitored the study to ensure participant safety and adherence to the protocol.

Sample

The IDEATel participants (n=1,665) were recruited and randomized between December 2000 and October 2002 and followed for up to 7 years. This included 695 participants from Upstate New York. Inclusion criteria were 55 years of age or older, being a current Medicare beneficiary living in the state of New York, having diabetes as defined by a physician's diagnosis and being on treatment with diet, an oral hypoglycemic agent, or insulin, residence in a federally designated medically underserved area (either of two federal designations, Medically Underserved Area or Health Professional Shortage Area), and fluency in either English or Spanish. Exclusion criteria were moderate or severe cognitive impairment, severe visual, mobility, or motor coordination impairment, severe comorbid condition, severe expressive or receptive communication impairment, lack of free electrical outlet for home telemedicine unit, and spending more than 3 months a year at a location different from their New York State residence.

Participants were recruited through PCP practices. Randomization to telemedicine case management or to usual care was assigned within the PCP practice immediately upon completion of the baseline examination.

Intervention

The telemedicine intervention in IDEATel has been described in detail.^3,4 Participants randomized to the intervention group received a home telemedicine unit (American Telecare, Inc., Eden Prairie, MN) consisting of a Web-enabled computer with modem connection to an existing telephone line. The home telemedicine unit had the following components: (a) a Web camera that allowed videoconferencing with nurse case managers at the Berrie Diabetes Center at Columbia University or the Joslin Diabetes Center at SUNY Upstate Medical University in Syracuse; (b) a home glucose meter (OneTouch^® SureStep^®; Lifescan, Inc., Milpitas, CA) and blood pressure cuff (model UA-767-PC blood pressure monitor; A&D Medical, San Jose, CA) connected to the home telemedicine unit through an RS-232 serial port, so that home fingerstick glucose and blood pressure readings could be uploaded into a clinical database; (c) access to patients' own clinical data; and (d) access to a special educational Web page created for the project by the American Diabetes Association in English and Spanish. Nurse case managers were trained in diabetes management and in the use of computer-based case management tools to facilitate interactions through videoconferencing with patients.

In the intervention group, nurse and dietitian case managers, trained in diabetes management and in the use of computer-based case management tools, videoconferenced with patients every 4–6 weeks to review glucose readings, diabetes-related issues, blood pressure readings, and laboratory data. The visit information was reviewed with an endocrinologist, and therapeutic recommendations were made to the PCPs, who retained full responsibility and control over their patients' care.

At the following visit, the nurse educator documented if a recommendation was accepted or modified. For approximately half of the participants in the intervention the PCP signed a form that allowed us to directly modify medication doses at the time of visit (i.e., increase insulin dose by 2 units).

Usual Care

Participating PCPs cared for patients in both the intervention and usual care groups, including prescribing all medications. The diabetes team never evaluated participants face-to-face. All contact with the diabetes team was by telemedicine and telephone. The PCPs received periodic mailings with current American Diabetes Association guidelines for the care of patients with diabetes. Patients in the usual care group received clinical care from their PCPs, without other guidance or direction from study personnel.

Data

The outcome of interest for these analyses was the implementation of specific IDEATel team recommendations, as a dichotomous variable (yes/no) for each recommendation and defined by treatment changes detected by questionnaires at follow-up visits. The prespecified IDEATel clinical end points were hemoglobin A1c, low-density lipoprotein cholesterol, and blood pressure levels. However, the telemedicine intervention effects on these end points have been reported previously,⁴ and for the current study we analyzed A1c as a covariate, as described below.

Follow-up examinations were conducted at 1-year intervals after randomization, with study time beginning at the baseline examination. Personnel conducting these examinations were blinded to intervention status and were not involved in supporting the clinical or technical aspects of the intervention. Data collected annually, starting at the baseline visit and including five follow-up visits through February 28, 2007 (when the intervention ended), were used in the primary analyses. Thus, the primary analyses consisted of six waves of data. At the annual visits, in addition to obtaining blood and urine samples, participant surveys and questionnaires, including the Diabetes Symptom Severity (DSC)-Type 2 symptom severity score (a higher score indicates greater diabetes-related symptoms and perceived burden of disease) and Impact of Telemedicine survey (a higher score indicates a greater beneficial impact of the telemedicine intervention on a patient's diabetes self-care skills and behavior), were completed.⁵ Survey data were also obtained from PCPs and their practices.

Statistical Methods

Descriptive summaries of PCP characteristics (mean, standard deviation, frequencies) are presented. A likelihood-based, generalized linear mixed model was fit via SAS (SAS Institute, Cary, NOC) proc GLIMMIX to estimate the likelihood of the PCP following IDEATel recommendations. A binomial distribution was assumed for the outcome variable using the logit link and Laplace estimation method. Random effects were included to model the correlation between multiple observations from a single subject, nested within PCPs. The compound symmetry covariance structure was selected to provide the best fit as measured by the Akaike Information Criterion.

Covariates that were considered for inclusion in the generalized linear mixed model model included PCP characteristics (i.e., age, type of practice), patient characteristics (i.e., age, gender, years of diabetes), glycemic, lipid, and blood pressure control, and the type of recommendation (stop, add, or change dose of medications). For covariates distributed with high skewness, such as the total DSC-Type 2 symptom severity score⁵ and number of glucose readings, log-transformations were performed. These covariates were assessed individually in separate models and collectively. Starting from all covariates with marginal p values of<0.1, the final model was built through a step-down model selection procedure with hemoglobin A1c as the measure of model quality. For all the covariates retained in the final model, the unadjusted acceptance of diabetes team recommendations was summarized by frequency and percentage. The p values for hypothesis tests with regard to the significance of each covariate and the associated 95% confidence intervals (CIs) for odds ratios were calculated based on the empirical sandwich variance estimator.

Sensitivity analyses were implemented with respect to selection of covariate, specification of covariance structure, and variance estimation. Another step-down procedure considering p values associated each covariate was used to confirm the final list of covariates included in the model. A first-order autoregressive covariance structure was considered other than the compound symmetry structure. Values of p and 95% CIs were examined using the estimated marginal variance of the pseudo-response.

Results

PCPS and PCP Practice Characteristics

There were 178 PCP participants from Upstate New York. The number of subjects per PCP ranged from 1 to 35 (mean, 3.27; standard deviation, 4.40). Over half of the PCPs were 45–54 years old, 73.8% were male, and 84.5% were white (Table 1). Their practices consisted of a mean of 3,299 patients per practice (median, 3,000), and approximately half were in solo practice (Table 2). Most PCPs were physicians and worked a mean of 48.4 h/week (median, 50.0 h/week).

Table 1.

Characteristics of Primary Care Providers (n=103)

PCP CHARACTERISTICS	N (%)
Age (years)
25–34	4 (3.92)
35–44	24 (23.53)
45–54	59 (57.84)
55–64	13 (12.75)
65–74	2 (1.96)
Gender
Male	76 (73.79)
Female	27 (26.21)
Race
White	87 (84.47)
Black or African American	1 (0.97)
Asian	10 (9.71)
Interacially mixed	1 (0.97)
Other	1 (0.97)
Refused to specify	3 (2.91)
Occupation/academic degree
MD	86 (84.31)
PA	3 (2.94)
NP	6 (5.88)
Other	4 (3.92)
DO	3 (2.94)

NP, nurse practitioner; PA, physician assistant; PCP, primary care provider.

Table 2.

Characteristics of Primary Care Practices

PCP PRACTICE CHARACTERISTICS	N (%)
Practice size
2,000 and under	33 (37.08)
2,001–4,000	39 (43.82)
4,001–6,000	11 (12.36)
6,001 and over	6 (6.74)
Employment
Institution	39 (39.80)
Self	52 (53.06)
Other	7 (7.14)
Hours spent in clinical practice per week
Mean (SD)	48.4 (19.8)
Median (IQR)	50.0 (20.0)
Approximate number of patients in panel/practice
Mean (SD)	3,299 (3,348)
Median (IQR)	3,000 (2,000)

IQR, interquartile range; PCP, primary care provider; SD, standard deviation.

Acceptance of Diabetes Team Recommendations by PCPS

The diabetes team made 6,558 recommendations, and of these 26.1% were accepted (Table 3). A likelihood-based, generalized linear mixed model (Table 3) showed that the acceptance rate of IDEATel recommendations by PCPs increased over time (p=0.0052). Participants were followed for up to 7 years. For every year of participation in IDEATel, the odds of an IDEATel recommendation being accepted increased by a factor of 1.12 (95% CI, 1.03–1.21) (Table 3).

Table 3.

Acceptance of Recommendations from the Diabetes Team by Primary Care Providers

CHARACTERISTIC	RECOMMENDATIONS ACCEPTED OR MODIFIED (N)/TOTAL RECOMMENDATIONS (%)	ADJUSTED ODDS RATIO	95% CI	P VALUE^a
Time (years) of participation in intervention		1.12	1.03–1.21	0.0052
Year 1	440/1,705 (25.8)
Year 2	582/2,247 (25.9)
Year 3	303/1,097 (27.6)
Year 4	155/617 (25.1)
Year 5	135/457 (29.5)
Year 6	85/201 (42.3)
Year 7	11/33 (33.3)
Recommendation type				<0.0001
Stop medication	158/659 (24.0)	1
Add a new medication	430/2,354 (18.3)	0.85	0.60–1.22
Change dosing of medication	1,123/3,545 (31.7)	51	1.12–2.06
Medication type				<0.0001
Antihypertensive	298/1,491 (20.0)	0.40	0.29–0.56
Insulin	692/1,668 (41.5)	1.28	0.97–1.70
Lipid-lowering	178/1,533(11.6)	0.28	0.19–0.42
Hyperglycemic control (oral)	543/1,866 (29.1)	1
Total DSC-Type 2 symptom severity score (log-transformed)		1.15	1.00–1.33	0.045
<13	322/1,386 (23.2)
13–24	462/1,801 (25.7)
24–35	352/1,426 (24.7)
>35	549/1,653 (33.2)
Impact of telemedicine on patient		1.49	1.15–1.91	0.0023
2–2.82	227/1,065 (21.3)
2.82–3.36	243/881 (27.6)
3.36–3.55	376/1,144 (32.9)
3.55–9	286/904 (31.6)
Number of glucose readings (log-transformed)	0.15	1.16	1.03–1.31	0.014
<151	230/1,056 (21.8)
151–284	355/1,464 (24.2)
284–468	391/1,585 (24.7)
>468	726/2,225 (32.6)

Type III F test for the significance of each covariate.

CI, confidence interval; DSC, Diabetes Symptom Checklist.

Recommendations consisted of stopping a medication, adding a new medication, or changing the dose of a medication. Medication changes that were followed included changes in antihypertensive, insulin, lipid-lowering, and oral antihyperglycemic medications. Both recommendations type (p<0.0001) and medication type (p<0.0001) significantly impacted whether diabetes team recommendations were accepted by PCPs. PCPs were more likely (51 times increase in odds) to change the dose of a medication than to discontinue a medication per recommendations from the diabetes team. Insulin dose changes had the greatest acceptance, 1.28 (95% CI, 0.97–1.70) times greater compared with antihyperglycemic oral medications. Relative to antihyperglycemic oral medications, the odds of antihypertensive and lipid-lowering medication recommendations being accepted were lower (0.40 [95% CI, 0.29–0.56] and 0.28 [95% CI, 0.19–0.42], respectively).

Severity of disease was associated with whether recommendations were accepted. The DSC-Type 2 assesses six dimensions of diabetes-related symptoms: hyperglycemic, hypoglycemic, ophthalmologic, psychological, cardiovascular, and neuropathy.⁵ A higher score indicates more diabetes symptoms. The greater the total DSC-Type 2 symptom severity score, the greater the odds that an IDEATel recommendation was accepted (p=0.045).

We assessed the impact of telemedicine on the PCP and their practice through the Impact of Telemedicine survey. The survey consisted of 20 questions, with answers on a 5-point Likert scale, from 0 indicating the lowest acceptability of the telemedicine intervention to 4 indicating the highest acceptability. These questions corresponded to three major domains: (1) impact on patients, as perceived by the PCP; (2) impact on the practice; and (3) acceptability to the PCP. Higher scores on the Impact of Telemedicine survey related to IDEATel recommendations being accepted (p=0.0023). Similarly, the greater the number of transmitted glucose readings, the more likely that IDEATel recommendation being accepted (p=0.014).

Discussion

Previous publications^6
–8 have reported the satisfaction and the effectiveness of the IDEATel telemedicine intervention by patients and PCPs, including improvements in glycemic control and achievement of behavioral change goals.^9,10 In this report, which to our knowledge is the first of its kind, we identified factors that were associated with PCP acceptance of recommendations from a remote diabetes care team participating in a telemedicine case management program. Most of the PCP practices were independent and unaffiliated with the diabetes team.

Only 26.1% of the recommended changes were accepted. We did not specifically investigate why PCPs did not accept many of our recommendations. We speculate that there were multiple factors involved in such decisions. However, we hypothesize that a greater trusting relationship with the diabetes team developed the longer the participants (PCPs and patients) were involved with the intervention, which led to a greater acceptance of the case management team's recommendations over time. The management team staff (registered nurses, dietitians, and endocrinologists) was stable for most of the intervention, which contributed to the growth of mutual trust.

PCPs of patients who invested more time in the intervention as demonstrated by uploading a greater number of glucose and blood pressure readings (measures of adherence) were more apt to view the recommendations positively. Another patient factor that demonstrated the acceptance of the diabetes management team's recommendation was severity of illness. PCPs of patients who had higher symptom severity scores on the DSC-Type 2 survey were also more apt to follow the diabetes management team's suggestions. This finding, along with results from the telemedicine impact survey and our previous reports of greatest improvements in glycemic control in participants with the worst control at baseline,^11,12 suggests that acceptance of recommendations is the greatest for those with more needs. The greater the impact of telemedicine as perceived by the PCP, the more likely the PCP would approve the changes proposed.

Generally, changes in medication doses were more supported than starting or discontinuing a medication. We hypothesize that there was more hesitation to starting a new medication compared with adjusting the dose of a current medication. Because the study team did not have access to the patients' medical records, there may have been contraindications to starting specifc medications about which the study team were unaware.

The PCPs did not receive any specific incentives to participate in the study other than for access for their intervention patients to high-quality patient education (nurse educators and dietitians) and the potential benefits for their patients in obtaining better diabetes outcomes (intention-to-treat mixed models showed that the telemedicine intervention reduced hemoglobin A1c [p=0.001], low-density lipoprotein cholesterol [p<0.001], and systolic and diastolic blood pressure [p=0.024 and p<0.001, respectively] over a follow-up period of 5 years).

Using telemedicine and specialized remote diabetes teams to improve diabetes care in rural areas is feasible and effective. This is also scalable and can be extended to other rural as well as urban settings. Although telemedicine case management was not associated with a reduction in Medicare claims in this medically underserved population,¹³ diabetes performance measures were improved. The cost of implementing the telemedicine intervention was high in IDEATel due to the costs of special hardware and software needed at the time of the study. Technology costs have decreased dramatically since the study was initiated, and we expect that implementation costs would be significantly less today.

This study had several limitations. PCPs took care of both intervention and control participants. It is likely that the recommendations made for their patients in the intervention group influenced their decision-making for patients in the usual care group. The usual care group may have benefited indirectly from the intervention. We did not have the availability of sharing electronic health records. The diabetes team therefore did not have complete patient information. If this capability was available, communication between the PCP and the diabetes management team would have been faster, more efficient, and possibly would have led to more trust and better management decisions. The sharing of live electronic data and medical records would have aided both parties in a greater understanding of treatment recommendations. E-mail capability, which could have facilitated communications, was available but was not part of the PCP workflow and therefore was not frequently used by the PCPs. It is possible that regularly scheduled live video teleconferencing with the PCPs to discuss specific patients' needs would have been helpful. This could be tested in future projects. Another limitation of this study was that it involved only the Upstate New York cohort (more rural). The addition of urban participants may have added a different perspective.

Another concern is whether participants accepted the recommendations that were accepted by their PCPs. In this study we do not have any measure of patient adherence to the recommendations proposed. However, in a previously published article by Trief et al.,¹⁴ we reported that patient adherence to diabetes-related tasks improved in the intervention group in IDEATel compared with the usual care group.

Lastly, the assessment of acceptance of remotely delivered diabetes team recommendations by PCPs was not a predetermined outcome for IDEATel. Our present research investigation was an ancillary study of IDEATel. It would be important for future telemedicine programs to further examine this issue so that programs can be adapted and targeted to those who would derive greatest benefit.

Conclusions

In the IDEATel telemedicine program PCP acceptance of treatment recommendations provided remotely by a diabetes team, particularly changes in glycemic control medications including insulin dosing, were better accepted over time. Results suggest that a team-based telemedicine program is acceptable and can be helpful to PCPs in improving diabetes care.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Shea

, Starren

, Weinstock

, et al. Informatics for Diabetes Education and Telemedicine (IDEATel) project: Rationale and design. J Am Med Inform Assoc, 2002; 9:49–62.

Starren

, Hripcsak

, Sengupta

, et al. Columbia University's Informatics for Diabetes Education and Telemedicine (IDEATel) project: Technical implementation. J Am Med Inform Assoc, 2002; 9:25–36.

Shea

, Weinstock

, Starren

, et al. A randomized trial comparing telemedicine case management with usual care in older, ethnically diverse, medically underserved patients with diabetes mellitus. J Am Med Inform Assoc, 2006; 13:40–51.

Shea

, Weinstock

, Teresi

, et al. A randomized trial comparing telemedicine case management with usual care in older, ethnically diverse, medically, underserved patients with diabetes mellitus: 5 year results of the IDEATel Study. J Am Med Inform Assoc, 2009; 16:446–456.

Grootenhuis

, Snoek

, Heine

, Bouter

. Development of a type 2 diabetes symptom checklist: A measure of symptom severity. Diabet Med, 1994; 11:253–261.

Tudiver

, Thomas Wolff

, Morin

, Teresi

, Palmas

, Starren

, Shea

, Weinstock

. Primary care providers' perceptions of home diabetes telemedicine care in the IDEATel Project. J Rural Health, 2007; 23:55–61.

Palmas

, Teresi

, Weinstock

, Shea

. Acceptability to primary care providers of telemedicine in diabetes case management. J Telemed Telecare, 2008; 14:306–308.

Bakken

, Grullon-Figueroa

, Izquierdo

, Lee

, Morin

, Palmas

, Teresi

, Weinstock

, Shea

, Starren

; IDEATel Consortium. Development, validation and use of English and Spanish versions of the Telemedicine Satisfaction and Usefulness Questionnaire. J Am Med Inform Assoc, 2006; 13:660–667.

Weinstock

, Teresi

, Goland

, Izquierdo

, Palmas

, Eimicke

, Ebner

, Shea

; IDEATel Consortium. Glycemic control and health disparities in older ethnically diverse underserved adults with diabetes: Five-year results from the IDEATel study. Diabetes Care, 2011; 34:274–279.

10.

West

, Lagua

, Trief

, Izquierdo

, Weinstock

. Goal setting using telemedicine in rural underserved older adults with diabetes: Experiences from the IDEATel project. Telemed J E Health, 2010; 16:405–416.

11.

Weinstock

, Teresi

, Goland

, Izquierdo

, Palmas

, Eimicke

, Ebner

, Shea

12.

Shea

, Kothari

, Teresi

, King

, Eimicke

, Lantigua

, Palmas

, Weinstock

. Social impact analysis of the effects of a telemedicine intervention to improve diabetes outcomes in an ethnically diverse and medically underserved population: Findings from the IDEATel Study. Am J Public Health, 2013; 103:1888–1894.

13.

Palmas

, Shea

, Starren

, Teresi

, Ganz

, Burton

, Pashos

, Blustein

, Field

, Morin

, Izquierdo

, Silver

, Eimicke

, Lantigua

, Weinstock

; IDEATel Consortium. Medicare payments, healthcare service use, and telemedicine implementation costs in a randomized trial comparing telemedicine case management with usual care in medically underserved participants with diabetes mellitus (IDEATel). J Am Med Inform Assoc, 2010; 17:196–202.

14.

Trief

, Izquierdo

, Eimicke

, Teresi

, Goland

, Palmas

, Shea

, Weinstock

. Adherence to diabetes self care for white, African-American and Hispanic American telemedicine participants: 5 year results from IDEATel project. Ethn Health, 2013; 18:83–96.