Key Summary Points
Why carry out this study? |
The COVID-19 pandemic prompted a swift adoption of telehealth across various fields, including diabetes management |
While clinical trials have validated the effectiveness of telehealth in diabetes care, there is a lack of published real-world evidence demonstrating its impact |
What was learned from the study? |
This study provides new evidence about the impact of telehealth on healthcare utilization among low-income individuals with type 2 diabetes, using Louisiana Medicaid claims data |
The findings show that telehealth was associated with increased overall utilization of healthcare services but was associated with decreased emergency department and inpatient utilization for some groups |
Introduction
In March 2020, the Louisiana governor encouraged the use of telehealth in response to the public health emergency of the coronavirus disease 2019 (COVID-19) [1]. In an effort to provide continued healthcare services, the Louisiana Medicaid program expanded telehealth coverage and reimbursements, and the Louisiana State Board of Medical Examiners granted temporary permits to out-of-state professionals, in addition to a telehealth registration process that did not require full state licensure [2].
Anzeige
Existing evidence has confirmed the feasibility and effectiveness of telehealth in diabetes care from randomized controlled trials (RCTs) or designed programs [3‐6]. However, real-world evidence relating to the impact of telehealth on people with type 2 diabetes, such as findings from electronic heath records and claims data, was impeded by the low uptake of telehealth before the pandemic. Less than 1% of evaluation and management services were provided through telehealth among Louisiana Medicaid beneficiaries before the pandemic [7].
However, the COVID-19 pandemic introduced a dramatic shift in this trend. Research indicates a swift adoption of telehealth by both patients and providers during this period [8‐13]. A recent report shows that 30% of all visits at outpatient practices were conducted through telehealth at the beginning of the pandemic [14]. Total weekly outpatient visits in Louisiana fell by about 23% between May and June 2020 [15]. Over the same period, telehealth visits accounted for approximately 16% of total outpatient visits in Louisiana [16].
The burden of type 2 diabetes, in terms of both health implications and economic costs, has been a persistent challenge in healthcare [17]. Ensuring timely and adequate care is paramount, as delays or interruptions can precipitate severe health events leading to higher rates of emergency department (ED) visits or hospitalizations. The COVID-19 pandemic further compounded this challenge with its curtailing of regular healthcare services, especially for the disadvantaged population [18]. Telehealth expansion during this period holds promise, especially for those with type 2 diabetes who often face barriers in accessing regular care due to socio-economic constraints. By offering remote consultations and follow-ups, telehealth can potentially reshape healthcare utilization patterns. Examining the association between telehealth use and healthcare utilization is crucial. Such insights can inform healthcare resource allocation, cost management strategies and policy decisions. Our study aims to explore the impact of telehealth use on healthcare utilization among Louisiana Medicaid beneficiaries with type 2 diabetes during the COVID-19 pandemic. Delving into this domain is imperative for future research to assess the quality, effectiveness and equity of care delivered through telehealth, insights from which can shape future health policies and strategies.
Methods
Study Design and Data Source
A propensity score weighted difference-in-difference analysis was used to estimate the effect of telehealth use on healthcare utilization among patients with type 2 diabetes using de-identified Medicaid claims data from March 2019 to August 2021. This model mitigates selection bias and controls for observed factors related to telehealth use. Zip code level data from the 2019 American Community Survey (ACS) were linked to Medicaid claims based on residential 5-digit zip codes to acquire environmental characteristics.
Anzeige
Sample Selection
We selected the sample using the de-identified Medicaid claims data. The sample was restricted to those who continuously enrolled in Louisiana Medicaid over the study period (March 2019 to August 2021) to avoid issues with compositional changes resulting from increased enrollment due to the COVID-19 pandemic. We excluded patients dual-eligible for both Medicare and Medicaid as we lack access to Medicare claims. Only beneficiaries with type 2 diabetes were included in this study. Type 1 diabetes and gestational diabetes were excluded using International Classification of Diseases (ICD) 10 codes.
We identified patients with type 2 diabetes according to a modified Surveillance, Prevention and Management of Diabetes Mellitus (SUPREME-DM) diabetes definition [19] because of the absence of laboratory results in Medicaid claims data: (1) one or more of the International Classification of Disease, Tenth Revision, Clinical Modification (ICD-10-CM) codes (E11.xx) for type 2 diabetes associated with inpatient encounters; (2) two or more ICD codes associated with outpatient encounters on different days within 2 years; (3) combination of any ICD codes and antidiabetic medications with outpatient encounters on different days within 2 years.
For comparison purposes, beneficiaries were segmented into two groups. The first group, termed as telehealth users or the treatment group, consisted of those with any telehealth-related claims since March 2020. Identification of these services utilized specific procedure codes appended with modifiers such as GT, GQ, 95 and a particular place of service code (02) from Louisiana Medicaid claims. All other Medicaid beneficiaries, with no telehealth claims during the stipulated period, were categorized as the control group or non-telehealth users. This binary classification was chosen to distinctly evaluate the broad impact of any telehealth usage against exclusive non-telehealth utilization. The dates of the first telehealth encountered were coded as the initiation dates (index dates). We then assigned index dates randomly for non-telehealth beneficiaries (control group) based on the distribution of initiation dates in the treated population. The baseline period was 12 months before the index date, and the selected cohort also required at least 6 months of follow-up after the index date to ensure a sufficient evaluation period. For example, patients who had their first telehealth service after February 2021 were not included in this study because they had < 6 months of follow-up data. We also required all beneficiaries to have had at least one outpatient visit at the baseline. Beneficiaries without available zip codes to be linked with ACS data were finally excluded. The sample selection is shown in Figure S1 (supplementary material).
Measurements
In assessing healthcare utilization, our primary focus was on the visit frequency of various healthcare services. This encompassed outpatient visits, which included in-person visits with claim types of outpatient hospitals, clinics and physician services. Additionally, we considered inpatient visits, Emergency Department (ED) visits—which were identified in accordance with the Healthcare Effectiveness Data and Information Set measures [20]—and hemoglobin A1C (HbA1C) tests. For each of these categories, the frequency was represented as an average number of visits per month, with separate calculations for the periods before and after the designated index date during the COVID-19 pandemic.
Our study also assessed secondary outcomes. Specifically, we considered ED and inpatient visits associated with Major Adverse Cardiovascular Events (MACE) and Ambulatory Care Sensitive Conditions (ACSC), given their potential preventability with improved primary care access.
To ensure a comprehensive assessment, we incorporated various control variables in our analysis. This included demographic factors like age, sex, and race/ethnicity. We also considered baseline healthcare utilization, represented by monthly utilization of ED visits, hospital stays, outpatient visits and HbA1C tests. Furthermore, we considered any existing comorbidities and evaluated technological accessibility at the zip code level and use rates of computer, internet and telephone in beneficiaries' residential area.
Statistical Analyses
We first used the propensity score weighting method to obtain a successful balance between treatment and control groups. To succeed in balancing, we used group-based trajectory modeling to categorize individuals into latent groups with similar patterns of outpatient visits over 12 months before the index dates. Once the best group-based trajectory model was chosen, measures of group membership were then incorporated as control variables in propensity score weighting protocols and regression models [21]. Detailed explanations of using group-based trajectory models can be found in prior work of other studies [13, 22‐29]. We then estimated the propensity scores for receiving telehealth during the pandemic using a probit regression model, controlling for baseline characteristics and a binary indicator of each trajectory group. For every weighting process, the standardized mean difference was checked between treatment and control groups before and after weighting to ensure successful weighting as defined as standardized mean differences within 10% for all baseline characteristics.
The outcomes of healthcare utilizations were estimated by the DID model as follows.
$$ Y_{{izt}} = \alpha + \beta *Effect_{{it}} + Post_{t} + Telehealth_{i} + X_{i} + \delta _{z} + \partial _{t} + \varepsilon _{{izt}},~w_{i} $$
Anzeige
The variable,\({Effect}_{it}\), was an interaction between the time indicator (\({Post}_{t}\): 0 for pre and 1 for post) and the indicator of telehealth users (\({Telehealth}_{i}\): 0 for non-telehealth users and 1 for telehealth users). The time indicator was a dichotomous variable, 0 for the pre-period (12 months before the index date) and 1 for the post-period (> = 6 months after the index date). \(\beta \) was the coefficient of interest and captured the change in outcomes in the pre- and post-period between Medicaid beneficiaries with and without telehealth use. The model also included zip code level fixed effect (\({\delta }_{z}\)) and time fixed effect (\({\partial }_{t}\)) of index to control unobservable zip code level differences and secular trends, respectively. Xi was a set of factors used in the propensity score weighting model, with the exception of the baseline outcome variables, used as control variables in our regressions to help control for additional variation that may remain after matching [30]. Weights \(,{ w}_{i}\) were odds of treatment calculated based on propensity scores: wi = 1 for treated units and wi = propensity score/(1- propensity score) for untreated units. The standard error was clustered at the zip code level to account for common variances in these observations.
Subgroup Analyses
We identified multiple subgroups of beneficiaries to determine how the impact of telehealth uptake during the COVID-19 pandemic differed by zip code-level environmental characteristics related to telehealth. These characteristics included computer use rate, internet rate and telephone rate. The sample was divided into two subgroups separated by the median of each characteristic. Since telehealth is used as an approach to improve the healthcare access of rural residents, we then examined whether telehealth impacted differently across the rural status of the county where beneficiaries lived. Other subgroup analyses were performed in different age groups (≥ 50 versus < 50 years) and racial groups (Black versus non-Black).
Sensitivity Analyses
With the lack of healthcare access due to the pandemic, patients with critically poor health could receive telehealth services for follow-up care after they had inpatient visits or emergency department visits. Therefore, we further performed the analysis after excluding those who had ED visits or hospitalizations within 30 days before the telehealth visit.
Telehealth was not exclusively expanded for diabetes care during the pandemic. A proportion of patients used telehealth services solely for non-diabetes care, such as mental health, which could attenuate any detectable impacts on diabetes-related outcomes in this study. We then repeated our main analysis only including patients who used telehealth services for any diabetes care in the treatment group.
Anzeige
The treatment sample may change with different treatment definitions and different outcome assessments; therefore, we repeated our matching process and regenerated propensity scores for non-telehealth beneficiaries to approximate the corresponding counterfactuals in each scenario. All data analyses were performed using SAS version 9.4 and Stata version 15.1 (StataCorp). Mean values were reported with standard deviations and regression coefficients were reported with 95% confidence intervals. Statistical significance was set at P < 0.05, and all tests were two-tailed.
Ethical Approval
The study was approved by the Tulane University Institutional Review Board (2021–1707). All data used in this study were deidentified; therefore, patient consent was not warranted. This study was conducted in accordance with the Helsinki Declaration of 1964 and its later amendments.
Results
Baseline Characteristics
This study identified 56,759 continuously enrolled Medicaid beneficiaries with diagnosed type 2 diabetes (Figure S1). We excluded 2286 patients because they were late telehealth users and had a follow-up period of < 6 months. We further excluded 3984 beneficiaries because they had no outpatient visits during the 12-month baseline. To get zip code level characteristics, a sample of 49,034 beneficiaries was linked with the 2019 zip code level ACS data. After propensity score weighting, we finally included 27,340 telehealth beneficiaries in the treatment group and 21,652 non-telehealth beneficiaries in the control group. We identified a three-group trajectory model for the analytic sample.
As shown in Table 1, all baseline characteristics were successfully balanced within 10% or 0.01 of a standardized mean difference after propensity score weighting. In the weighted sample, the average age was about 47 years old and nearly half of patients were Black. Hypertension, hyperlipidemia and depression were the three most common chronic conditions in this sample, at about 78%, 60% and 38%, respectively.
Table 1
Baseline characteristics before and after propensity score weighting
Variables | Unweighted | Weighted | ||||
|---|---|---|---|---|---|---|
Telehealth | In person | SMD | Telehealth | In person | SMD | |
N | 27,340 | 21,652 | – | 27,340 | 21,652 | – |
Age at first TH, years | 47.268 | 46.360 | – 0.078 | 47.268 | 47.461 | 0.017 |
Female, n (%) | 19,739 (72.2) | 13,922 (64.3) | – 0.171 | 19,739 (72.2) | 15,460 (71.4) | – 0.017 |
Race/ethnicity, n (%) | ||||||
White | 10,471 (38.3) | 7838 (36.2) | – 0.043 | 10,471 (38.3) | 8293 (38.3) | 0.000 |
Black | 14,025 (51.3) | 10,826 (50.0) | – 0.025 | 14,025 (51.3) | 11,086 (51.2) | – 0.002 |
Hispanic | 1094 (4.0) | 953 (4.4) | 0.024 | 1094 (4.0) | 866 (4.0) | 0.003 |
Other | 8475 (31.0) | 7730 (35.7) | 0.099 | 8475 (31.0) | 6799 (31.4) | 0.008 |
Chronic conditions, n (%) | ||||||
Asthma | 4046 (14.8) | 2035 (9.4) | – 0.165 | 4046 (14.8) | 2836 (13.1) | – 0.048 |
Stroke | 1941 (7.1) | 1083 (5.0) | – 0.088 | 1941 (7.1) | 1602 (7.4) | 0.014 |
Hypertension | 21,134 (77.3) | 14,983 (69.2) | – 0.185 | 21,134 (77.3) | 16,889 (78.0) | 0.018 |
Arthritis | 7792 (28.5) | 4179 (19.3) | – 0.215 | 7792 (28.5) | 6236 (28.8) | 0.009 |
Atrial fibrillation | 793 (2.9) | 433 (2.0) | – 0.062 | 793 (2.9) | 671 (3.1) | 0.013 |
Cancer | 848 (3.1) | 411 (1.9) | – 0.076 | 848 (3.1) | 671 (3.1) | – 0.001 |
COPD | 4484 (16.4) | 2598 (12) | – 0.126 | 4484 (16.4) | 3551 (16.4) | 0.001 |
Chronic kidney disease | 3281 (12) | 1905 (8.8) | – 0.106 | 3281 (12.0) | 2685 (12.4) | 0.011 |
Depression | 10,307 (37.7) | 4114 (19.0) | – 0.425 | 10,307 (37.7) | 8163 (37.7) | – 0.001 |
Heart failure | 2297 (8.4) | 1386 (6.4) | – 0.076 | 2297 (8.4) | 1840 (8.5) | 0.004 |
Hyperlipidemia | 15,939 (58.3) | 10,804 (49.9) | – 0.169 | 15,939 (58.3) | 12,775 (59.0) | 0.015 |
Coronary heart disease | 3636 (13.3) | 2165 (10.0) | – 0.103 | 3636 (13.3) | 2858 (13.2) | – 0.004 |
HCRU, visits per month per person | ||||||
OP | 1.339 | 0.842 | – 0.534 | 1.339 | 1.362 | 0.021 |
Inpatient | 0.017 | 0.012 | – 0.123 | 0.017 | 0.018 | 0.017 |
ACSC related | 0.003 | 0.002 | – 0.038 | 0.003 | 0.003 | 0.023 |
MACE related | 0.001 | 0.001 | – 0.032 | 0.001 | 0.001 | – 0.007 |
ED | 0.129 | 0.100 | – 0.201 | 0.129 | 0.126 | – 0.017 |
ACSC related | 0.034 | 0.027 | – 0.075 | 0.034 | 0.040 | 0.034 |
MACE related | 0.002 | 0.001 | – 0.048 | 0.002 | 0.002 | – 0.003 |
HbA1c tests | 0.072 | 0.050 | – 0.227 | 0.072 | 0.073 | 0.014 |
Zip code-level environmental factor, % | ||||||
Computer use rate | 83.549 | 82.893 | – 0.082 | 83.549 | 83.494 | – 0.007 |
Internet use rate | 72.651 | 71.607 | – 0.095 | 72.651 | 72.541 | – 0.010 |
Telephone use rate | 97.647 | 97.527 | – 0.070 | 97.647 | 97.627 | – 0.012 |
Trajectory group*, n (%) | ||||||
Group 1 | 9323 (34.1) | 13,489 (62.3) | 0.589 | 9323 (34.1) | 7318 (33.8) | – 0.006 |
Group 2 | 14,244 (52.1) | 7232 (33.4) | – 0.386 | 14,244 (52.1) | 11,324 (52.3) | 0.004 |
Group 3 | 3773 (13.8) | 931 (4.3) | – 0.336 | 3773 (13.8) | 3010 (13.9) | 0.003 |
Anzeige
Healthcare Utilization
Beneficiaries using telehealth during the COVID-19 pandemic had more healthcare utilization on average compared with the beneficiaries who only had in-person care (Table 2). The in-person outpatient visits significantly increased for the telehealth group versus the non-telehealth group over time by 195.049 visits (95% CI: 166.169 to 223.929, p < 0.001) per 1000 beneficiaries per month on average. The differences in the rate of growth between telehealth users and their comparators for inpatient and ED visits were 3.816 (95% CI: 2.539 to 5.093, p < 0.001) and 10.499 (95% CI: 7.287 to 13.712, p < 0.001) visits per 1000 beneficiaries per month, respectively. The number of HbA1c tests significantly increased for the telehealth group compared with the comparison group over time by 14.153 tests (95% CI: 11.431 to 16.875, p < 0.001) per 1000 beneficiaries per month on average.
Table 2
Impact of telehealth on health utilization (per 1000 beneficiaries per month)
DID estimates | |
|---|---|
195.049*** | |
OP visits | [166.169, 223.929] |
< 0.001 | |
3.816*** | |
IP visits | [2.539, 5.093] |
< 0.001 | |
0.497 | |
ACSC related | [– 0.156, 1.149] |
0.135 | |
– 0.113 | |
MACE related | [– 0.426, 0.200] |
0.480 | |
10.499*** | |
ED visits | [7.287, 13.712] |
< 0.001 | |
5.749 | |
ACSC related | [– 0.039, 11.537] |
0.052 | |
– 0.028 | |
MACE related | [– 0.513, 0.456] |
0.909 | |
14.153*** | |
HbA1c tests | [11.431, 16.875] |
< 0.001 |
For ED and inpatient visits, we checked the decomposed measurements including the visits related to ACSC and MACE. While the ACSC and MACE could be preventable by improved access to primary care, no significant impacts of telehealth on these outcomes were found for both ED visits and inpatient visits over time (Table 2).
Sensitivity Analyses
We identified 6461 beneficiaries who had any ED visits or inpatient visits in the 30 days before receiving telehealth. We excluded these patients and re-performed propensity score weighting and analyses. Outpatient visits significantly increased for the telehealth group versus the non-telehealth group over time by 133.171 visits (95% CI: 109.562 to 156.780, p < 0.001) per 1000 beneficiaries per month on average (Table 3). While we found no significant differences in the rate of growth between telehealth users and their comparators for overall inpatient visits on average, inpatient visits related to MACE decreased slightly more for the telehealth group over time by 0.383 visits (95% CI: – 0.639 to – 0.126, p = 0.004) per 1000 beneficiaries per month. ED visits significantly decreased for the telehealth group versus the non-telehealth group over time by 9.456 visits (95% CI: – 12.356 to – 6.557, p < 0.001) per 1000 beneficiaries per month on average. MACE-related ED visits also dropped by 0.490 visits (95% CI: – 0.862 to – 0.119, p = 0.010) per 1000 beneficiaries per month. The estimate for HbA1c tests was similar to that shown in Table 2 with a significant increase over time by 14.121 tests (95% CI: 11.554 to 16.689, p < 0.001) per 1000 beneficiaries per month on average, comparing telehealth users and non-telehealth users.
Table 3
Impact of telehealth on health utilization after excluding those who had any ED visits or hospitalizations in the 30 days before a telehealth visit (per 1000 beneficiaries per month)
DID estimates | |
|---|---|
133.171*** | |
OP visits | [109.562, 156.780] |
< 0.001 | |
– 0.860 | |
IP visits | [– 1.900, 0.180] |
0.105 | |
– 0.254 | |
ACSC related | [– 0.730, 0.223] |
0.296 | |
– 0.383** | |
MACE related | [– 0.639, – 0.126] |
0.004 | |
– 9.456*** | |
ED visits | [– 12.356, – 6.557] |
< 0.001 | |
– 1.178 | |
ACSC related | [– 4.050, 1.693] |
0.421 | |
– 0.490** | |
MACE related | [– 0.862, – 0.119] |
0.010 | |
14.121*** | |
HbA1c tests | [11.554, 16.689] |
< 0.001 |
After excluding patients who only used telehealth services for non-diabetes care, we rebalanced and reanalyzed the sample. We found similar results to the primary results. Beneficiaries using telehealth during the COVID pandemic had more healthcare utilization on average compared with the beneficiaries who only had in-person care, including in-person outpatient visits, ED visits, inpatient visits and HbA1c tests (Table S1).
Subgroup Analyses
We found the estimates of telehealth effects on outcomes were similar between groups by each environmental characteristic (Table 4). The impact of telehealth on outpatient visits, ED visits and inpatient visits was similar between rural and urban areas where beneficiaries lived and between Black and non-Black groups (Table 5). Telehealth was associated with a modest decrease of 1.715 visits (95% CI: – 3.321 to – 0.110, p = 0.036) in overall inpatient visits in the age group of at least 50 years old after excluding patients who had any ED visits or hospitalizations before a telehealth service. Other associations were similar between the two age groups.
Table 4
DID estimates of telehealth on health utilization by environmental characteristics (per 1000 beneficiaries per month)
Telephone | Computer | Internet | ||||
|---|---|---|---|---|---|---|
≥ 50% | < 50% | ≥ 50% | < 50% | ≥ 50% | < 50% | |
Full sample | ||||||
OP | 216.677*** | 171.152*** | 220.703*** | 173.951*** | 212.844*** | 176.242*** |
[173.912,259.441] | [130.882,211.422] | [173.034,268.373] | [141.300,206.603] | [171.446,254.241] | [137.006,215.478] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
IP | 3.904*** | 3.854*** | 4.091*** | 3.696*** | 3.650*** | 3.770*** |
[2.167,5.641] | [1.997,5.710] | [2.077,6.106] | [1.950,5.441] | [1.767,5.532] | [2.032,5.508] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
ED | 10.396*** | 10.685*** | 10.094*** | 11.588*** | 11.368*** | 9.529*** |
[5.681,15.110] | [6.261,15.110] | [5.493,14.696] | [7.049,16.126] | [6.864,15.873] | [5.045,14.012] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
13.100*** | 15.157*** | 14.768*** | 13.822*** | 14.608*** | 13.306*** | |
HbA1c tests | [9.350,16.849] | [11.345,18.970] | [11.426,18.110] | [9.186,18.458] | [11.185,18.031] | [8.990,17.622] |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
Excluding those who had any ED or hospitalization in 30 days before a telehealth visit | ||||||
OP | 148.382*** | 116.213*** | 144.477*** | 119.942*** | 144.922*** | 123.125*** |
[113.611,183.154] | [83.325,149.100] | [107.324,181.630] | [92.447,147.438] | [110.679,179.165] | [90.772,155.478] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
IP | – 0.817 | – 0.800 | – 0.587 | – 1.168 | – 0.754 | – 1.151 |
[– 2.218,0.584] | [– 2.357,0.758] | [– 2.185,1.012] | [– 2.576,0.240] | [– 2.280,0.771] | [– 2.603,0.300] | |
0.253 | 0.313 | 0.471 | 0.104 | 0.332 | 0.12 | |
ED | – 10.082*** | – 8.556*** | – 9.431*** | – 8.879*** | – 8.262*** | – 10.479*** |
[– 14.330,– 5.833] | [– 12.508,– 4.604] | [– 13.689,– 5.173] | [– 12.878,– 4.881] | [– 12.410,– 4.114] | [– 14.562,-6.395] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
12.983*** | 15.028*** | 15.150*** | 13.178*** | 14.772*** | 12.990*** | |
HbA1c tests | [9.447,16.520] | [11.416,18.639] | [12.135,18.165] | [8.858,17.498] | [11.588,17.956] | [8.884,17.096] |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
Table 5
DID estimates of telehealth on health utilization by rurality, race and age (per 1000 beneficiaries per month)
Urban | Rural | Non-Black | Black | Age < 50 | Age ≥ 50 | |
|---|---|---|---|---|---|---|
Full sample | ||||||
OP | 194.887*** | 196.671*** | 207.848*** | 181.998*** | 185.003*** | 200.044*** |
[157.080,232.694] | [150.842,242.501] | [164.450,251.245] | [140.886,223.109] | [143.040,226.966] | [159.053,241.036] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
IP | 3.968*** | 3.271*** | 4.468*** | 3.371*** | 3.620*** | 3.641*** |
[2.329,5.608] | [1.355,5.188] | [2.457,6.478] | [1.417,5.325] | [1.910,5.329] | [1.756,5.526] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
ED | 10.655*** | 10.790*** | 14.013*** | 7.594*** | 9.544*** | 11.372*** |
[6.928,14.381] | [4.903,16.676] | [8.973,19.053] | [3.265,11.922] | [4.638,14.450] | [7.217,15.527] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
18.663*** | 6.038** | 13.711*** | 14.845*** | 14.507*** | 13.711*** | |
HbA1c tests | [15.206,22.121] | [2.331,9.745] | [10.076,17.347] | [11.588,18.102] | [11.121,17.894] | [9.885,17.537] |
< 0.001 | 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
Excluding those who had any ED or hospitalization in 30 days before a telehealth visit | ||||||
OP | 122.943*** | 145.777*** | 148.468*** | 114.589*** | 118.246*** | 143.349*** |
[93.155,152.731] | [107.227,184.327] | [113.854,183.083] | [80.880,148.297] | [84.839,151.653] | [109.072,177.626] | |
< 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
IP | – 1.272 | – 0.009 | – 0.06 | – 1.439 | – 0.141 | – 1.715* |
[– 2.581,0.037] | [– 1.621,1.603] | [– 1.697,1.576] | [– 2.970,0.091] | [– 1.386,1.105] | [– 3.321,– 0.110] | |
0.057 | 0.992 | 0.942 | 0.065 | 0.825 | 0.036 | |
ED | – 10.056*** | – 7.518** | – 5.575** | – 12.799*** | – 11.185*** | – 7.787*** |
[– 13.402,– 6.710] | [– 13.005,– 2.031] | [– 9.616,– 1.535] | [– 16.943,– 8.656] | [– 15.293,– 7.077] | [– 11.546,– 4.028] | |
< 0.001 | 0.007 | 0.007 | < 0.001 | < 0.001 | < 0.001 | |
18.505*** | 5.771** | 14.003*** | 14.278*** | 14.264*** | 13.822*** | |
HbA1c tests | [15.304,21.707] | [2.016,9.527] | [10.625,17.381] | [11.034,17.522] | [11.096,17.431] | [10.358,17.286] |
< 0.001 | 0.003 | < 0.001 | < 0.001 | < 0.001 | < 0.001 | |
Discussion
Using statewide Medicaid claims, we found patients with type 2 diabetes with any telehealth services during the pandemic had more healthcare utilization over time compared with those who only had in-person care. After we removed those who had ED or inpatient visits in the 30 days prior to receiving telehealth, we found beneficiaries with any telehealth during the pandemic had slightly fewer hospitalization visits related to the MACE over time compared with those with in-person care only. ED visits also significantly dropped over time among those who received telehealth versus those who did not. This paper contributes to evidence on the impact of telehealth use during the pandemic versus traditional care on healthcare utilization among type 2 diabetes, including primary care, HbA1c tests, emergency care and hospitalization.
A recent study found an increase in outpatient visits after receiving telehealth [31]. They found a significant increase (5.36%) in the proportion of established patients with OP visits during the pandemic for the overall sample compared to the pre-pandemic. However, our slightly larger findings (approximately 15%) may be explained as we included a control group that accounted for the more limited number of OP visits available during the pandemic period. Another study also found telehealth users had more in-person outpatient visits than non-telehealth users; it was conducted among Medicare beneficiaries with type 2 diabetes [27]. These findings indicate that telehealth is a promising approach to maintain contact between patients and physicians, which could promote more in-person outpatient visits for needed care or examinations.
The estimates in the present study showed that both ED visits and hospitalizations increased on average over time among those who used any telehealth compared to those who used in-person care only during the pandemic. Our findings on ED visits and hospitalizations are inconsistent with prior work done on different diseases [32, 33]. A care coordination home-telehealth program found patients receiving telehealth services were less likely to be admitted for inpatient care [6]. However, temporary disruptions in routine healthcare services during the pandemic may have caused differences in these associations [31]. A study from the Centers for Disease Control and Prevention found that 40% of adults reported delaying or avoiding urgent or emergency medical care because of concerns related to COVID-19 [34]. Patients seeking telehealth may have more opportunities to gain timely emergency and inpatient care. Other reasons may include the limited quality of telehealth services or timely follow-up care after emergency care and hospitalization captured by telehealth. Outpatient care converted rapidly to virtual care over a short time, which may raise a concern about the quality of delivering care remotely [35]. However, we are unable to directly examine the quality of telehealth services with this rapid adoption and implementation of telehealth during this era. Instead, we removed sicker patients who had ED visits and hospitalizations within 30 days before a telehealth visit. Findings of decreases in ED and hospitalization further demonstrated sicker patients were more likely to use telehealth services. Telehealth has served as an alternative to provide care after emergency department visits and hospitalization during the pandemic. A very recent study has demonstrated its effectiveness in delivering follow-up care after hospitalization [36]. It is notable for diabetes care that MACE-related ED visits and hospitalization also decreased by telehealth in this analysis, but this decrease was modest. More analyses should be conducted to further demonstrate these associations in the future.
Our study further assessed the DID estimates on healthcare utilization across different subgroups. As we expected, most DID estimates are slightly larger in areas with relatively higher use of telephones, computers and broadband internet. We found significant estimates for outpatient visits, ED visits and hospitalizations across all subgroups by rurality, race and age. These findings indicate that the uptake of telehealth during the pandemic has induced timely healthcare services across different subgroups for patients with type 2 diabetes with Medicaid coverage in Louisiana. Disparities in the impact of telehealth may need to be further evaluated in future work across these comparison groups.
Our findings regarding increased healthcare utilization among telehealth users align with broader trends observed during the pandemic. Notably, recent studies underscored the potential benefits of telehealth in managing type 2 diabetes [27, 28]. They found that patients using telehealth services exhibited improved diabetes management, especially concerning biomarker management. This suggests that the convenience and immediacy of telehealth not only lead to increased utilization but may also contribute to better patient outcomes. As more individuals gain access to and familiarity with these services, the long-term implications for chronic disease management, like type 2 diabetes, and addressing disparities in telehealth use could be profound [37, 38]. This underscores the need for healthcare systems to invest in and prioritize telehealth infrastructure and training, ensuring that patients receive the most effective care possible and optimize treatment goals for long-term health outcomes [39], whether in person or remotely.
Several limitations are notable in this study. First, telehealth visits were not randomly assigned during the pandemic. While we used propensity score weighting to control observed factors related to telehealth use, unobserved heterogeneity between the treatment and control groups may introduce bias to our estimates. Therefore, we used the difference-in-difference model with propensity score weighting in this study, which is a useful technique when randomization of treatment is not possible. However, some unobserved factors may still bias the estimates in this study, such as patient and provider preference, performance of providers in delivering telehealth and available healthcare settings around patients’ addresses during the pandemic. For example, the quality of care delivered by telehealth may not be guaranteed because of the rapid adoption of telehealth by both providers and patients. Diabetes management is complex and is highly related to diabetes duration, disease severity, diet and nutrition, and other unobservable factors not captured by the current data. The convenience of telehealth may also introduce more healthcare utilization in telehealth group. Patient motivation is another important unobserved factor that could bias our findings. Patients who received telehealth could be sicker, high utilizers or care more about their health, which could result in the increases in healthcare utilization found in this study. While we used the propensity score weighting to balance the different characteristics between telehealth users and non-telehealth users, our model still cannot fully capture and control patient motivation. However, we have further demonstrated that telehealth use during the pandemic was not associated with more ED visits or hospitalizations by removing sicker patients who used telehealth services for healthcare after an ED visit and/or hospitalization. Second, we did not use a fixed follow-up time in this study. However, the volume of healthcare utilization may be related to the time of follow-up. Instead, we calculated the monthly healthcare utilization as the outcome. Third, the sample was composed of Medicaid beneficiaries with type 2 diabetes in Louisiana and may not extend to the low-income population or other diseases in other states not in the south where social and institutional factors may meaningfully differ. Fourth, we evaluated the impact of telehealth services delivered by either audio or video, not for a specific type of telehealth service. The results of the study on the impact of telehealth on utilization may differ based on the type of telehealth used. Our study did not have the capability to evaluate the specific impact of audio-based and video-based telehealth on utilization. Besides, this study only focused on telehealth of delivering healthcare remotely. Other innovative approaches, such as tele-assessment to monitor various health metrics remotely for type 2 diabetes [40], were not included in our study. Given that our study indicates a positive trend in healthcare utilization with telehealth, further exploration into specialized innovative tools could be the next frontier in optimizing care for patients with type 2 diabetes.
Conclusions
In this context, during the COVID-19 pandemic, we observed that telehealth was associated with more healthcare utilization in general among low-income patients living with type 2 diabetes. Further research should track the long-term impact of telehealth on healthcare utilization to gain a better understanding of its value in diabetes care. Additionally, future studies could examine ways to effectively integrate telehealth with in-person services to provide optimal healthcare for patients.
Medical Writing/Editorial Assistance
No medical writing or editorial assistance was received during the writing of this article.
Declarations
Conflict of Interest
Yixue Shao, Lizheng Shi, Elizabeth Nauman, Eboni Price-Haywood and Charles Stoecker have nothing to disclose.
Ethical Approval
The study was approved by the Tulane University Institutional Review Board (2021–1707). All data used in this study were deidentified and, therefore, patient consent was not warranted. This study was conducted in accordance with the Helsinki Declaration of 1964 and its later amendments.
Open Access This article is licensed under a Creative Commons Attribution-Non-Commercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/.