Abstract
Objective
To determine the impact of telemental health (TMH) use on total healthcare costs and mental health (MH)-related costs paid by a third party among adults with mental health conditions (MHC).
Method
This study employed a pre-post design with a non-equivalent control group. The cohort comprised adults with MHCs identified using diagnosis codes from de-identified claims data of the Optum Clinformatics DataMart (01/01/2010 - 06/30/2017). We identified mental health (MH) service users and TMH users (N = 348) based on procedure codes. Non-users (N = 238,595) were defined as those who only used in-person MH services. A Difference-in-Differences (DID) analysis was performed within a multivariable two-part model (TPM) framework to examine the impact of TMH use on adjusted standardized costs (2018 US $) of all healthcare services and MH services. Patient-level and state-level factors were adjusted in TPM.
Results
TMH use was associated with significantly higher MH-related costs [Marginal effect = $461.3, 95% confidence interval: $142.4 – $780.2] and an excess of $370 increase in MH-related costs at follow-up as compared to baseline. However, TMH use was not associated with an increase in total third-party healthcare costs nor with changes in total costs from baseline to follow-up.
Conclusions
Despite having a higher likelihood of MH services use and MH-related costs, TMH users did not have higher total costs as compared to adults using only in-person MH services. Our findings suggest that TMH can increase access to MH care without increasing total healthcare costs among adults with MHC. Future studies exploring whether TMH use can lead to cost-savings over a longer period are warranted.
Keywords: telemental health, mental health service, healthcare cost, difference-in-differences
Introduction
Mental health conditions (MHCs) are the most expensive conditions among ten conditions with fast spending growth (i.e., heart conditions, trauma, cancer) in the United States, with an estimated excess cost of $38 billion in direct healthcare expenses in 2013 [1]. The high MHC prevalence coupled with continued undertreatment of MHC has contributed significantly to this high economic burden [2-4]. Telemental health (TMH) has emerged as an alternative care delivery approach to meet the growing demand of MH care and shortages of MH providers [5]. TMH is the use of remote technologies such as video-conferencing systems to deliver MH care [6]. Clinical trials have shown that TMH is well accepted [7-9] and has comparable efficacy in the evaluation and treatment of MHC patients across a broad range of demographic groups [10-14].
The comparative efficacy of TMH in delivering MH care relative to in-person care is well-documented in clinical trials [10-14] and systematic reviews [14-16]. A growing body of literature performing economic evaluation of TMH programs also indicated that TMH could be cost-effective in various health settings across different countries [17-19]. However, studies on outcomes of TMH use among adults with MHCs in real-world settings in the US are sparse, limited primarily to those using data from the Veterans Health Administration. For instance, a study examining outcomes of 98,609 veterans receiving TMH services reported a 25% reduction in psychiatric admissions and hospital stays after initiating TMH over four years [20]. However, the study did not compare the changes in healthcare utilization and costs between individuals with TMH use and those with only in-person MH services use [20], limiting interpretation of findings.
To establish the true value of TMH services, studies comparing healthcare utilization and costs among MHC patients receiving TMH and in-person MH services are warranted. In response to the rising healthcare costs, the Institute of Health Improvement developed the “Triple Aim” to define high-value care as care that can improve population health and individual care experience with reduced per-capita costs [21]. In alignment with the “Triple Aim,” the National Quality Forum (NQF) and the American Telemedicine Association use healthcare costs as one of the key measurements in their newly developed telehealth value framework [22]. However, previous studies have primarily focused on indirect medical costs such as time and expenses related to travel [23-25]. No study has compared the direct healthcare costs of individuals with MHC who used TMH services with that of individuals with MHC who used only in-person MH services. Understanding the relative utilization and direct care costs will help inform and guide the decisions of healthcare payers regarding the expansion of TMH coverage [26].
Therefore, the primary objective of this study was to determine the impact of TMH use on total healthcare costs and mental health (MH)-related costs among adults with mental health conditions (MHC) using data from a large commercial insurer.
Methods
Data Source
Data (January 2010 through June 2017) were derived from Optum Clinformatics ® Data Mart (Eden Prairie, MN, USA), a large commercial health plan in the US Data included de-identified medical claims of physician, hospital, and prescription drug services as well as enrollment information for covered individuals from all 50 states and the District of Columbia [27]. We linked claims data with that from the 2016-2017 Area Health Resources File (AHRF) [28] and the 2010-2016 National Mental Health Services Survey (N-MHSS) [29] using state identifiers. The linkage was conducted to obtain state-level information on poverty, rurality, and MH provider shortage from the AHRF, as well as the TMH capacity from N-MHSS files.
Study Design
We used a pre-post study design with a non-equivalent control group. The pre- (baseline) and post- (follow up) periods each consisted of six months. For TMH users, the first TMH service date after 12 months of MHC diagnosis was used as the index date. For those who used only in-person MH services (i.e., non-users), a pseudo-index date was randomly selected from all MH service dates between 2010 and 2016 to allow a six-month follow-up. Standardized costs were assessed for both six months before (i.e., baseline) and six months after (i.e., follow-up) the index/pseudo-index dates.
Study Cohort
The study cohort comprised adults (> 18 years old at baseline) who were diagnosed with any MHC between 2010 and 2016, who had at least one MH service encounter 12months after the MHC diagnosis, and who were continuously enrolled in any health plan during the baseline and follow-up periods. We identified adults with MHC based on the primary and secondary diagnosis codes [the International Classification of Diseases, Ninth Revision (ICD-9) codes: 291-292; 295-316; ICD-10: F10-F99] in inpatient and outpatient medical claims [30]. Adults with MCH were defined as those with an MCH diagnosis from at least two outpatient visits or one inpatient admission [30]. MH service encounters were identified with procedure codes (Supplemental Table 1) or primary/secondary diagnostic codes of any MHC [30]. Data on adults with MCH for each year (2010-16) were combined to form the study cohort. The cohort identification process is illustrated in Figure 1. The final analytic cohort included 238,943 unique adults with MHC.
Figure 1:
Cohort identification process.
Abbreviations: MHC: mental health condition; MH: mental health; TMH: Telemental health; TMH user: adults with MHC who had at least 1 TMH service encounter during the observation period (2010/06/01-2016/12/31); Non-users: adults with MHC who had never used any TMH services during the observation period.
Measures
Outcomes
Standardized total healthcare costs.
Total insurer-allowed payments for inpatient stays, facility outpatient visits, professional services, and prescription drugs during the 6-month period were assessed. To account for differences in allowed payments across health plans and provider contracts, standardized prices were used to represent allowed payments. Standardized prices for medical services and prescription drugs were estimated using algorithms developed by the Optum researchers, accounting for the type and quantity of services as well as the relative resource costs involved in providing the service. For example, standardized costs of a hospitalization were calculated by multiplying length of stay by per diem costs, which was estimated with an pricing model, accounting for aggregated diagnostic/service category, presence of may surgery (yes/no), and length of stay. All the costs were normalized to 2018 US dollars using the standard prices and cost factors provided by Optum Clinformatics ® DataMart.
MH-related costs.
All allowed payments from the insurer for MH services using standard prices and cost factors were estimated. MH services were identified by procedure or primary/secondary diagnostic codes as described above (study cohort part).
Key Independent Variable: TMH Use versus (vs.) No TMH Use
We defined comparison groups based on TMH use (yes/no), identified from MH service claims using the relevant Healthcare Common Procedure Coding System (HCPCS)/Current Procedural Terminology (CPT) codes (“0188T”, “0189T”, “G0406 - 0408”, “G0425 - 0427”, “G0459”) and/or modifiers (“GT”, “GQ”, “95”) [30,31]. These codes were listed on the policy document of the insurer and have been used in a previous study of TMH among Medicare beneficiaries [30,31].
The TMH user group included adults with MHC who had any record of TMH use; non-user group included those who used only in-person MH services.
Other Independent Variables
We selected other independent variables based on the adapted Andersen’s Behavioral Model of Health Services Use [32]. The model posits that individual healthcare utilization is a function of multiple factors, including 1) predisposing factors (age, sex); 2) enabling factors [TMH use (yes/no), type of insurance (Medicare Advantage/others), type of health plan (Health Maintenance Organization/others); 3) need factors [severe MHC diagnosis (yes/no), mental health specialist visits (yes/no); the number of physical chronic conditions (none/one or two/three or more); or the presence of polypharmacy (use of > 5 drug classes in a 90-day period [33]), obesity, and/or any substance (i.e., alcohol, tobacco, drug) use disorders (SUD)]; and 4) environmental factors (the percentage of: counties under poverty, rural counties, counties with MH provider shortage, and MH facilities with TMH adoption by state). We also adjusted for the year of TMH use to account for the effect of time. Time from MHC diagnoses to index/pseudo-index dates (136 ± 104 days) was also included in the multivariable analyses.
All need factors were identified from inpatient and outpatient claims using ICD9/ICD10 diagnostic codes. An algorithm developed by researchers of the Healthcare Cost and Utilization Project was used to define the severity of MHCs using ICD9/10 diagnostic codes [34]. The severity classes derived from this algorithm were strong predictors of hospitalizations after an Emergency department visit [34]. The number of chronic physical conditions [range 0-13, based on a list developed by the Multiple Chronic Conditions (MCC) working group [35]] were also ascertained using ICD9/ICD10 codes from claims data. All patient-level factors were measured at the baseline and follow-up period. Environmental factors, measured at the state level, were ascertained using data from the AHRF.
Statistical Analyses
Chi-square tests and independent t-tests were conducted to compare the characteristics of TMH users and non-users at baseline and follow-up, respectively. Standard linear regression is usually not appropriate for modeling costs due to certain common characteristics of cost data, including zero mass, non-negative values, and positive skewness [36]. Therefore, we used a mixed distributions approach to model costs. Specifically, we used two-part models (TPM), in which the probability of non-zero costs (yes/no) was modeled with logistic regressions, and the magnitude of costs was modeled within Generalized Linear models (GLMs) using gamma distributions and log-link functions [36]. The marginal effect (ME) and associated 95% confidence interval (95%CI) were used to represent the parameter estimate. As need factors were derived from claims of service encounters, we adjusted for need factors in GLM but not in the logistic regression analyses. We adjusted for all other independent variables in both parts of TPM; these variables include the year of TMH use, time from MHC diagnosis to index/pseudo-index dates, and predisposing-, enabling-, and environmental factors. A Difference in Difference (DID) estimate was included in the TPMs to estimate the excess change in both total third-party healthcare costs and MH-related costs due to TMH use. The DID estimate is an interaction term of TMH use (yes vs. no) and time (follow-up vs. baseline), representing the effect of using TMH on cost changes from baseline to follow-up [37]. For all analyses in this study, a p-value less than 0.05 was considered statistically significant. All data management and analyses were performed with SAS 9.4 (SAS Institute Inc., Cary, NC) and Stata 14 (StataCorp LLC, College Station, TX).
Results
Sample Description
A total of 238,943 adults with MHC were identified. The most prevalent MHC was anxiety disorder (42.4%). About one-third of the study sample had serious MHCs such as schizophrenia (1.4%), bipolar disorder (7.6%), and major depressive disorder (24.7%). The sample was predominantly female (61.5%), with a mean age of 48.5 years (SD = 18.3).
Overall, only 15 in 10,000 adults with MHCs used TMH (Table 1). At both baseline and followup, TMH users differed significantly from non-users in age and in all enabling-, need-, and environmental factors and non-users (Table 1). For example, at baseline, a higher percentage of TMH users had received a serious MHC diagnosis (75.6% vs. 32.2%, p < 0.001) and had visited a mental health specialist (51.2% vs. 19.2%, p < 0.001) as compared to non-users. Furthermore, a greater percentage of TMH users than non-users had a record of polypharmacy (54.3% vs. 27.7%, p < 0.001) and had been diagnosed with at least two comorbid chronic conditions (61.2% vs. 32.7%, p < 0.001). The distribution of all patient-level factors within each group was similar at baseline and follow-up (Table 1).
Table 1:
Characteristics of adults with mental health conditions (MHCs) who used telemental health (TMH) and those used in-person mental health (MH) services at baseline and follow-up, Optum Clinformatics ® DataMart, 2010-2017.
| Baseline | Follow-up | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| TMH users | Non-users | TMH users | Non-users | ||||||||
| N | Col. % | N | Col. % | Sig. | N | Col. % | N | Col. % | Sig. | ||
| Total | 348 | 100.00 | 238,595 | 100.00 | 348 | 100.00 | 238,595 | 100.00 | |||
| Predisposing factors a | |||||||||||
| Age [mean (SD)], years | 58.18 (15.51) | 48.42 (18.33) | *** | 58.23 (9.78) | 48.46 (18.34) | *** | |||||
| Sex | |||||||||||
| Female | 207 | 59.48 | 146,692 | 61.48 | 207 | 59.48 | 146,692 | 61.48 | |||
| Male | 141 | 40.52 | 91,903 | 38.52 | 141 | 40.52 | 91,903 | 38.52 | |||
| Enabling factors a | |||||||||||
| Medicare Advantage | *** | *** | |||||||||
| Yes | 296 | 85.06 | 67,545 | 28.31 | 296 | 85.06 | 67,537 | 28.31 | |||
| No | 52 | 14.94 | 171,050 | 71.69 | 52 | 14.94 | 171,058 | 71.69 | |||
| Plan type | *** | *** | |||||||||
| HMO | 153 | 43.97 | 52,808 | 22.13 | 153 | 43.97 | 52,759 | 22.11 | |||
| Non-HMO | 195 | 56.03 | 185,787 | 77.87 | 195 | 56.03 | 185,836 | 77.89 | |||
| Need factors a | |||||||||||
| Any severe mental health diagnosis | *** | *** | |||||||||
| Yes | 263 | 75.57 | 76,835 | 32.20 | 253 | 72.70 | 55,149 | 23.11 | |||
| No | 85 | 24.43 | 161,760 | 67.80 | 95 | 27.30 | 183,446 | 76.89 | |||
| Any mental health specialist visit | *** | *** | |||||||||
| Yes | 178 | 51.15 | 45,772 | 19.18 | 207 | 59.48 | 42,480 | 17.80 | |||
| No | 170 | 48.85 | 192,823 | 80.82 | 141 | 40.52 | 196,115 | 82.20 | |||
| Number of physical chronic conditions b | *** | *** | |||||||||
| Zero | 70 | 20.11 | 112,805 | 47.28 | 78 | 22.41 | 124,098 | 52.01 | |||
| One | 65 | 18.68 | 47,726 | 20.00 | 53 | 15.23 | 42,051 | 17.62 | |||
| Two or more | 213 | 61.21 | 78,064 | 32.72 | 217 | 62.36 | 72,446 | 30.36 | |||
| Polypharmacy c | *** | *** | |||||||||
| Yes | 189 | 54.31 | 66,130 | 27.72 | 197 | 56.61 | 62,898 | 26.36 | |||
| No | 159 | 45.69 | 172,465 | 72.28 | 151 | 43.39 | 175,697 | 73.64 | |||
| Any tobacco/alcohol use disorder | ** | ** | |||||||||
| Yes | 77 | 22.13 | 40,161 | 16.83 | 73 | 20.98 | 24,202 | 10.14 | |||
| No | 271 | 77.87 | 198,434 | 83.17 | 275 | 79.02 | 214,393 | 89.86 | |||
| Any drug use disorder | *** | *** | |||||||||
| Yes | 47 | 13.51 | 13,677 | 5.73 | 42 | 12.07 | 9,608 | 4.03 | |||
| No | 301 | 86.49 | 224,918 | 94.27 | 306 | 87.93 | 228,987 | 95.97 | |||
| Obesity | *** | *** | |||||||||
| Yes | 57 | 16.38 | 19,235 | 8.06 | 47 | 13.51 | 16,571 | 6.95 | |||
| No | 291 | 83.62 | 219,360 | 91.94 | 301 | 86.49 | 222,024 | 93.05 | |||
| Environmental factors | |||||||||||
| Poverty d | ** | ** | |||||||||
| Low (1st & 2nd Quartile) | 165 | 47.41 | 92,997 | 38.93 | 165 | 47.41 | 92,997 | 38.93 | |||
| High (3rd & 4th Quartile) | 183 | 52.59 | 145,869 | 61.07 | 183 | 52.59 | 145,869 | 61.07 | |||
| Rurality d | *** | *** | |||||||||
| Low (1st & 2nd Quartile) | 174 | 50.00 | 149,879 | 62.75 | 174 | 50.00 | 149,879 | 62.75 | |||
| High (3rd & 4th Quartile) | 174 | 50.00 | 88,987 | 37.25 | 174 | 50.00 | 88,987 | 37.25 | |||
| Mental health provider shortage d | *** | *** | |||||||||
| Low (1st & 2nd Quartile) | 101 | 29.02 | 130,580 | 54.67 | 101 | 29.02 | 130,580 | 54.67 | |||
| High (3rd & 4th Quartile) | 247 | 70.98 | 108,286 | 45.33 | 247 | 70.98 | 108,286 | 45.33 | |||
| TMH capacity e | *** | *** | |||||||||
| 1st Quartile | 22 | 6.32 | 62,555 | 26.19 | 22 | 6.32 | 62,555 | 26.19 | |||
| 2nd Quartile | 86 | 24.71 | 72,528 | 30.36 | 86 | 24.71 | 72,528 | 30.36 | |||
| 3rd Quartile | 145 | 41.67 | 48,624 | 20.36 | 145 | 41.67 | 48,624 | 20.36 | |||
| 4th Quartile | 95 | 27.30 | 55,159 | 23.09 | 95 | 27.30 | 55,159 | 23.09 | |||
| Lagging time [mean (SD)], days f | 179.0 (105.0) | 136.2 (104.4) | *** | 179.0 (105.0) | 136.2 (104.4) | *** | |||||
Abbreviations: Col.%: Column%; SD: standard deviation; Sig.: significance level; HMO: health maintenance organization
Notes: TMH users were defined as adults with MHC used any TMH services; non-users were defined as adults with MHC who only used in-person MH services.
all patient-level factors were measured in the six-month period before and after index date based on de-identified claims of Optum Clinformatics ® Data Mart. The first TMH use date was used as index date for users and a pseudo-index date was randomly selected from all mental health service dates among non-users.
chronic conditions examined included arthritis, asthma, coronary artery disease, cardiac arrhythmias, congestive heart failure, chronic kidney disease, chronic obstructive pulmonary disease, dementia and related disorders, diabetes, hypertension, hyperlipidaemia, osteoporosis, stroke.
polypharmacy was defined as having five or more medications from different therapeutic classes in the 90 days before index date.
poverty, rurality, and mental health provider shortage were measured as average percentages of counties under qualified for each condition from 2010-2016 based on data from the 2016-2017 Area Health Resources File.
TMH capacity was measured as average percentages of mental health facilities with telemedicine from 2010-2016 based on data from the 2010-2016 National Mental Health Services Survey.
lagging time reflects the number of days between first observed MHC diagnosis in claims to index date/pseudo-index date.
p < 0.01
p < 0.001.
Unadjusted and Adjusted MH-related Costs
As illustrated in Table 2, “within” group differences in MH-related costs varied between TMH users and non-users. While a similar percentage of TMH users received MH services at baseline and follow-up (29.9% vs. 33.3%, p = 0.548), the percentage of non-users receiving MH services declined significantly (from 13.9% to 8.3%, p < 0.001) from the baseline to the follow-up period. The median MH-related costs for TMH users changed from $3,198.9 (Interquartile range: $524.1-$11845.9) at baseline to $2,290.1 ($234.8-$14,632.1) at follow-up. For the nonusers, the median MH-related costs were $2,059.0 ($369.3-$8,548.9) at baseline and $2,408.0 ($385.2-$10,439.3) at follow-up.
Table 2:
Unadjusted median and interquartile range (IQR) of baseline and follow-up costs (2018 US $) of adults with mental health conditions (MHC) who used Telemental health (TMH) services and those used in-person mental health (MH) services, Optum Clinformatics® DataMart, 2010-2017.
| All adults with MHC | ||||||||
|---|---|---|---|---|---|---|---|---|
| TMH users | Non-users | |||||||
| Baseline | Follow-up | Baseline | Follow-up | |||||
| N | Median (IQR) |
N | Median (IQR) |
N | Median (IQR) |
N | Median (IQR) |
|
| Total a | 348 | $4,232.4 ($1,040.6- $12,466.1) | 348 | $4,051.1 ($1,339.6- $13,764.0) | 238,595 | $1,232.5 ($243.1- $4,909.3) | 238,595 | $1,105.3 ($218.8- $4,260.5) |
| MH b | 348 | $0 ($0- $158.2) | 348 | $0 ($0- $234.8) | 238,595 | $0 ($0-$0) | 238,595 | $0 ($0-$0) |
| Adults with MHC with specific service use | ||||||||
| TMH users | Non-users | |||||||
| Baseline | Follow-up | Baseline | Follow-up | |||||
| N (%) |
Median (IQR) |
N (%) | Median (IQR) |
N (%) | Median (IQR) |
N (%) | Median (IQR) |
|
| Total a | 327 (94.0%) | $4,949.7 ($1,281.0- $12,974.9) | 328 (94.2%) | $4,660.3 ($1,513.8- $15,220.4) | 221,638 (92.9%) | $1,482.1 ($372.5- $5,493.6) | 219,822 (92.1%) | $1,349.1 ($346.3- $4,807.8) |
| MH b | 105 (30.2%) | $3,199.0 ($524.1- $11,845.9) | 116 (33.3%) | $2,290.1 ($234.8- $14,632.1) | 33,304 (13.9%) | $2,059.0 ($369.3- $8,548.9) | 19,742 (8.3%) | $2,408.0 ($385.2- $10,439.3) |
Notes: All outcomes were measured based on de-identified claims of Optum Clinformatics ® Data Mart. All the costs were standardized to 2018 US dollar. Baseline and follow-up period were defined as the six-month period before and after index/pseudo-index date, respectively. The first TMH use date was used as index date for users and a pseudo-index date was randomly selected from all mental health service dates among non-users.
TMH users were defined as adults with MHC used any TMH services; non-users were defined as adults with MHC who only used in-person MH services.
total healthcare costs included the standard costs for all healthcare services paid by payers, including those for outpatient visits, inpatient stays, and prescription medications.
MH-related utilizations and costs were identified based on primary/secondary diagnoses of any MHC.
%MH costs were calculated as the average percentage of MH-related costs among the total healthcare costs for individuals with positive MH-related costs.
ns, p > 0.05
p < 0.001.
Between group comparisons using DID analysis indicated that TMH users were over twice as likely than non-users to utilize MH services after adjustment for other factors (AOR: 2.06, 95%CI: 1.62 – 2.63). Based on the marginal effect estimate, TMH use was also associated with $461.3 higher MH-related costs. Furthermore, after adjusting for other factors, TMH user had increased MH-related costs at follow-up as compared to baseline, and these increased costs were $369.6 higher than those among non-users, after adjusting for other factors. (Table 3).
Table 3:
Parameter estimates of telemental health use (TMH) from difference-indifferences (DID) analyses for total healthcare costs and total costs for mental health services (MHS) at follow-up among adults with mental health conditions (MHC), Optum Clinformatics® DataMart, 2010-2017.
| Total healthcare costs a | |||
|---|---|---|---|
| Logistic | GLM | Marginal effect | |
| AOR (95% CI) | Adjusted Beta (SE) | Mean (95%CI), $ | |
| TMH use | |||
| Yes vs. No | 1.06 (0.68, 1.66) | −0.043 (0.104) | −253.9 (−1,601.5, 1,093.7) |
| Time | |||
| Follow-up vs. baseline | 0.89 (0.88, 0.91) *** | 0.023 (0.009) ** | −197.1 (−307.0, −87.2)*** |
| DID c | 1.18 (0.70, 2.00) | 0.074 (0.121) | 554.3 (−1,005.2, 2,113.8) |
| Total MH-related costs b | |||
| Logistic | GLM | Marginal effect | |
| AOR (95% CI) | Adjusted Beta (SE) | Mean (95%CI), $ | |
| TMH use | |||
| Yes vs. No | 1.84 (1.44, 2.34) | 0.130 (0.203) | 461.3 (142.4, 780.2) *** |
| Time | |||
| Follow-up vs. baseline | 0.56 (0.55, 0.57)*** | 0.108 (0.016) *** | −269.9 (−294.3, −245.5) *** |
| DID c | 2.06 (1.62, 2.63) *** | −0.097 (0.232) | 369.6 (13.9, 725.4)* |
Abbreviations: AOR(95%CI): adjusted odds ratio and associated 95% confidence interval were estimated from logistic regressions on the probability of non-zero costs; SE: standard error was estimated from two-part models, where the first part estimated the probabilities of having non-zero costs in a logistic regression and the second part estimated positive costs in a generalized linear model (GLM) with log-link functions and gamma distributions.
Notes: All outcomes were measured based on de-identified claims of Optum Clinformatics ® Data Mart. All the costs were standardized to 2018 US dollar. Baseline and follow-up period were defined as the six-month period before and after index/pseudo-index date, respectively. The first TMH use date was used as index date for users and a pseudo-index date was randomly selected from all mental health service dates among non-users.
Total healthcare costs included the standard costs for all healthcare services paid by payers, including those for outpatient visits, inpatient stays, and prescription medications.
Total costs for mental health services were identified based on primary/secondary diagnoses of any MHC from outpatient and inpatient claims.
DID is the interaction term of TMH use and Time, representing the impact of TMH use on the changes of outcomes from baseline to follow-up.
p < 0.01
p < 0.001.
Other factors associated with higher MH-related costs included MA enrollment, all need factors such as having polypharmacy, any SUD, and multiple chronic physical conditions, and living in states with high levels of poverty (Supplemental Table 2).
Unadjusted and Adjusted Total Healthcare Costs
Table 2 presents the median unadjusted total costs and MH-related costs among TMH users and non-users at baseline and follow-up. When comparing follow-up utilization and costs to those at baseline, TMH users had similar percentages of individuals with zero costs (5.7% vs. 6.0%) at baseline and follow-up. However, among non-users, a significantly higher percentage of individuals had zero costs at follow-up than at baseline (7.9% vs. 7.1%, p < 0.001).The median total healthcare costs at baseline were $4,949.7 ($1,081.0-$12,974.9) among TMH users and $1,482.1 ($372.5-$5,493.6) among non-users. The median total healthcare costs appeared to be lower at follow-up for both TMH users ($4,660.3 [$1,513.8-$15,220.4]) and non-users ($1,349.1 [$346.3-$4,807.8]).
Group difference in differences from baseline to follow-up were examined with multivariable DID analysis within a TPM framework. Results from DID analysis indicated that TMH use (ME = −$253.9, 95%CI: −$1,601.5 - $1,093.7) was not associated with total healthcare costs. Furthermore, TMH use (ME of DID estimate = $554.3, 95% CI: −$1,005.2 – $2,113.8) had no significant impact on the change in healthcare costs from baseline to follow-up (Table 3). Other factors contributing to higher total third-party costs were similar to those for higher MH-related costs (e.g., all need factors).
Discussion
Our study is the first to compare direct total third-party healthcare costs and MH-related costs among adults with MHCs who used TMH services and those used only in-person MH services (i.e., non-users). It has been reported that payers may be less likely to adopt TMH due to the fear of higher utilization and costs associated with TMH [38]. Our findings suggest that TMH use did not have an impact on total healthcare costs. However, TMH use was associated with a greater likelihood of using MH services, suggesting that TMH may have increased access to MH services for adults with MHC. Although our study did not have information on wait time, published studies suggest that TMH may contribute to increased utilization by reducing the wait time for MH services [39].
It is worth noting that, in our study, TMH was associated with increased MH-related costs but not total healthcare costs, suggesting that reduced use of non-MH services may offset the incremental costs from using MH services. Although we were not able to analyze the source of this cost-offset (e.g., reduced inpatient use, reduced emergency room visits) due to small sample sizes, it is plausible that TMH may have positive spill-over effects on physical health, resulting in decreased use of non-MH services. Such “cost-offset effect” following MH treatment has been reported in other clinical populations, including adults with co-existing depression and diabetes [40]. As a key priority of the patient-centered medical home model (PCMH) is to integrate mental and physical health, TMH may offer unique opportunities to achieve such integration [41,42]. Future study is needed to explore how to incorporate TMH in the PCMH model.
The increased costs of MH services might also be offset by TMH-facilitated medication management. Polypharmacy was prevalent among adults with MHC in our study, characterizing 27.8% of adults with MHC overall and 56.6% of those with TMH use (Table 1). We found that the presence of polypharmacy was significantly associated with both higher healthcare costs and MH-related costs (Supplemental Table 2). Without proper management, polypharmacy has been shown to increase healthcare costs due to elevated risks of adverse drug reactions and drug-drug interactions [43]. Medication management is one of the most commonly used TMH services [44]. Studies have demonstrated the efficacy of TMH in improving medication adherence and associated clinical outcomes [45,46]. Furthermore, TMH allows for frequent short visits if necessary, such as those for reviewing medications, a feature that is especially valuable for patients for whom travel is difficult due to distance, weather, poor health, or other challenges (i.e., obesity).
Several limitations of the study should also be noted. First, our study estimated only direct healthcare costs and our study population was restricted to adults enrolled in a single private insurance plan. Indirect healthcare costs such as those related to travel of patients, their family members, and providers were not considered. Given that multiple studies have demonstrated that TMH can reduce indirect healthcare costs due to lost productivity (e.g., time off from work) and transportation expenses [14,47,48], indirect MH costs are likely to be comparable or lower for MHC patients who used TMH as compared to those used in-person MH services. Second, we only estimated costs for MH services. Costs of prescription medications used to treat MHCs were not included because some of these medications (e.g., antidepressants) are often used to treat physical conditions such as pain and insomnia [49]. Third, due to the limited sample size, we did not conduct subgroup analyses to compare the costs of specific MH services delivered via TMH vs. in-person. Furthermore, as this study compared short-term (i.e., six months) healthcare costs between TMH users and non-users. The long-term impact of TMH use on healthcare costs among adults with MHCs thus warrants further examination. It is also important to note that our study was not designed to perform an economic evaluation of TMH due to the limitations discussed above. To encourage the usage of TMH by payers and patients, future studies following published good practice criteria (e.g., Consolidated Health Economic Evaluation Reporting Standards) are warranted to evaluate the economic value of TMH from payers’ and patients’ perspective. Finally, although our study cohort was drawn from enrollees in one of the largest commercial plans in the country, our results may not be generalizable to MHC patients insured by other commercial plans.
Conclusion
This is the first study to examine the association between TMH use and direct healthcare costs over time from a private payer's perspective. TMH use was associated with an increased utilization of MH services and a corresponding excess in MH-related costs due to. However, TMH use did significantly affect total third-party healthcare costs. Future studies should examine the sources of this apparent “cost-offset” and explore whether TMH use can lead to cost-savings over a longer period.
Supplementary Material
Acknowledgements
No assistance in the preparation of this article is to be declared.
Declaration of funding
Research reported in this manuscript was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number U54GM104942, WVCTSI. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Declaration of financial/other relationships
The authors report no conflicts of interest. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.
Data Availability Statement
This retrospective database study used commercial claims data from the Optum Clinformatics® Data Mart (Eden Prairie, MN, USA) spanning January 1, 2010, through June 30, 2017. The claims data that support the findings of this study are from a proprietary administrative claims database and are not publicly available. However, summary data tables are available from the authors upon reasonable request.
References
- 1.Roehrig C Mental disorders top the list of the most costly conditions in the United States: $201 billion. Health Affairs. 2016;35(6):1130–1135. [DOI] [PubMed] [Google Scholar]
- 2.Kisely S, Scott A, Denney J, et al. Duration of untreated symptoms in common mental disorders: association with outcomes: international study. The British Journal of Psychiatry. 2006;189(1):79–80. [DOI] [PubMed] [Google Scholar]
- 3.Gleason Lauren. Community Leaders Identify Mental Health System As One Of Most Overburdened Community Resources; Consequences Include Stressed Emergency Medical And Education Systems. National Alliance on Mental Illness; 2003. [Google Scholar]
- 4.Park-Lee E, Lipari RN, Hedden SL, et al. Receipt of services for substance use and mental health issues among adults: results from the 2016 National Survey on Drug Use and Health. 2018. [PubMed] [Google Scholar]
- 5.Flaum M Telemental health as a solution to the widening gap between supply and demand for mental health services. New York, Elsevier; 2013. [Google Scholar]
- 6.Clarke G, Yarborough BJ. Evaluating the promise of health IT to enhance/expand the reach of mental health services. General hospital psychiatry. 2013;35(4):339–344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Zarate CA Jr, Weinstock L, Cukor P, et al. Applicability of telemedicine for assessing patients with schizophrenia: Acceptance and reliability. The Journal of clinical psychiatry. 1997. [DOI] [PubMed] [Google Scholar]
- 8.Sheeran T, Rabinowitz T, Lotterman J, et al. Feasibility and impact of telemonitor-based depression care management for geriatric homecare patients. Telemedicine and e-Health. 2011;17(8):620–626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Yuen EK, Herbert JD, Forman EM, et al. Acceptance based behavior therapy for social anxiety disorder through videoconferencing. Journal of Anxiety Disorders. 2013;27(4):389–397. [DOI] [PubMed] [Google Scholar]
- 10.Norman S The use of telemedicine in psychiatry. Journal of psychiatric and mental health nursing. 2006;13(6):771–777. [DOI] [PubMed] [Google Scholar]
- 11.Langarizadeh M, Tabatabaei MS, Tavakol K, et al. Telemental Health Care, an Effective Alternative to Conventional Mental Care: a Systematic Review. Acta informatica medica : AIM : journal of the Society for Medical Informatics of Bosnia & Herzegovina : casopis Drustva za medicinsku informatiku BiH. 2017;25(4):240–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hilty DM, Ferrer DC, Parish MB, et al. The effectiveness of telemental health: a 2013 review. Telemedicine journal and e-health : the official journal of the American Telemedicine Association. 2013;19(6):444–454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hernandez-Tejada MA, Zoller JS, Ruggiero KJ, et al. Early Treatment Withdrawal from Evidence-Based Psychotherapy for PTSD: Telemedicine and in-Person Parameters. The International Journal of Psychiatry in Medicine. 2014;48(1):33–55. [DOI] [PubMed] [Google Scholar]
- 14.Bashshur RL, Shannon GW, Bashshur N, et al. The empirical evidence for telemedicine interventions in mental disorders. Telemedicine and e-Health. 2016;22(2):87–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Hubley S, Lynch SB, Schneck C, et al. Review of key telepsychiatry outcomes. World J Psychiatry. 2016;6(2):269–282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Mistry H, Garnvwa H, Oppong R. Critical appraisal of published systematic reviews assessing the cost-effectiveness of telemedicine studies. Telemedicine and e-Health. 2014;20(7):609–618. [DOI] [PubMed] [Google Scholar]
- 17.Davalos ME, French MT, Burdick AE, et al. Economic evaluation of telemedicine: review of the literature and research guidelines for benefit-cost analysis. Telemedicine journal and e-health : the official journal of the American Telemedicine Association. 2009. December;15(10):933–48. [DOI] [PubMed] [Google Scholar]
- 18.Lokkerbol J, Adema D, Cuijpers P, et al. Improving the cost-effectiveness of a healthcare system for depressive disorders by implementing telemedicine: a health economic modeling study. Am J Geriatr Psychiatry. 2014;22(3):253–262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Moirangthem S, Rao S, Kumar CN, et al. Telepsychiatry as an economically better model for reaching the unreached: A retrospective report from South India. Indian journal of psychological medicine. 2017;39(3):271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Deen TL, Godleski L, Fortney JC. A description of telemental health services provided by the Veterans Health Administration in 2006–2010. Psychiatric Services. 2012;63(11): 1131–1133. [DOI] [PubMed] [Google Scholar]
- 21.Donald M Berwick TWN, John Whittington,. The Triple Aim: Care, Health, And Cost. Health Affairs. 2008;27(3):759–769. [DOI] [PubMed] [Google Scholar]
- 22.Frandsen BR, Joynt KE, Rebitzer JB, et al. Care fragmentation, quality, and costs among chronically ill patients. The American journal of managed care. 2015;21(5):355–362. [PubMed] [Google Scholar]
- 23.RJ E, MR R, Louise D VA Telemedicine: An Analysis of Cost and Time Savings. Telemedicine and e-Health. 2016;22(3):209–215. [DOI] [PubMed] [Google Scholar]
- 24.Cota A, Tarchala M, Parent-Harvey C, et al. Review of 5.5 Years' Experience Using E-mail-Based Telemedicine to Deliver Orthopedic Care to Remote Communities. Telemedicine journal and e-health : the official journal of the American Telemedicine Association. 2017. January;23(1):37–40. [DOI] [PubMed] [Google Scholar]
- 25.Hatcher-Martin J, Anderson E, Factor S. Patient Acceptance and Potential Cost-Savings of Teleneurology in an Academic Outpatient Movement Disorders Practice (P1.022). Neurology. 2016;86(16 Supplement). [Google Scholar]
- 26.Congressional Budget Office. Answers to questions fo the record following a hearing by the house committee on the budget on the work of the Congressional Budget Office. Congressional Budget Office website 2015. [Google Scholar]
- 27.Wallace PJ, Shah ND, Dennen T, et al. Optum Labs: building a novel node in the learning health care system. Health Affairs. 2014;33(7):1187–1194. [DOI] [PubMed] [Google Scholar]
- 28.Health Resources & Services Administration (HRSA). Area Heatlh Resources Files [cited 2019 Jan 27]. Available from: https://data.hrsa.gov/topics/health-workforce/ahrf [Google Scholar]
- 29.Substance Abuse and Mental Health Services Administration. National Mental Health Services Survey (N-MHSS): 2016 Data on Mental Health Treatment Facilities. BHSIS Series S-98: Substance Abuse and Mental Health Services Administration; 2017. [Google Scholar]
- 30.Mehrotra A, Huskamp HA, Souza J, et al. Rapid growth in mental health telemedicine use among rural Medicare beneficiaries, wide variation across states. Health Affairs. 2017;36(5):909–917. [DOI] [PubMed] [Google Scholar]
- 31.UnitedHealthcare. Telehealth and Telemedicine Policy, Professional. 2019. [Google Scholar]
- 32.Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? Journal of health and social behavior. 1995:1–10. [PubMed] [Google Scholar]
- 33.Masnoon N, Shakib S, Kalisch-Ellett L, et al. What is polypharmacy? A systematic review of definitions. BMC geriatrics. 2017;17(1):230–230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Coffey RM, Kassed CA, Mhsa CS, et al. A severity-of-illness classification for mental and substance-use disorders for use with hospital administrative data. 2011. [Google Scholar]
- 35.Goodman RA, Posner SF, Huang ES, et al. Peer Reviewed: Defining and measuring chronic conditions: Imperatives for research, policy, program, and practice. Preventing chronic disease. 2013;10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Baser O Modeling Healthcare Costs. STATinMED Research; 2012. [Google Scholar]
- 37.Lechner M The estimation of causal effects by difference-in-difference methods. Foundations and Trends® in Econometrics. 2011;4(3):165–224. [Google Scholar]
- 38.Lustig TA. The role of telehealth in an evolving health care environment: workshop summary. National Academies Press; 2012. [PubMed] [Google Scholar]
- 39.Neufeld J, Case R. Walk-in telemental health clinics improve access and efficiency: a 2-year follow-up analysis. Telemedicine journal and e-health : the official journal of the American Telemedicine Association. 2013;19(12):938–941. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Katon W, Unutzer J, Fan MY, et al. Cost-effectiveness and net benefit of enhanced treatment of depression for older adults with diabetes and depression. Diabetes care. 2006. February;29(2):265–70. [DOI] [PubMed] [Google Scholar]
- 41.McWilliams JK. Integrating Telemental Healthcare with the Patient-Centered Medical Home Model. Journal of child and adolescent psychopharmacology. 2016. April;26(3):278–82. [DOI] [PubMed] [Google Scholar]
- 42.Hilty DM, Sunderji N, Suo S, et al. Telepsychiatry and other technologies for integrated care: evidence base, best practice models and competencies. International review of psychiatry (Abingdon, England). 2018. December;30(6):292–309. [DOI] [PubMed] [Google Scholar]
- 43.Kojima G, Bell C, Tamura B, et al. Reducing cost by reducing polypharmacy: the polypharmacy outcomes project. Journal of the American Medical Directors Association. 2012;13(9):818.e11–818.e8.18E15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Deen Tisha L., Ph.D. ,, Godleski Linda, M.D., and, Fortney John C., Ph.D. A Description of Telemental Health Services Provided by the Veterans Health Administration in 2006–2010. Psychiatric Services. 2012;63(11):1131–1133. [DOI] [PubMed] [Google Scholar]
- 45.Hommel KA, Hente E, Herzer M, et al. Telehealth behavioral treatment for medication nonadherence: a pilot and feasibility study. European journal of gastroenterology & hepatology. 2013;25(4):469–473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Frangou S, Sachpazidis I, Stassinakis A, et al. Telemonitoring of medication adherence in patients with schizophrenia. Telemedicine Journal & E-Health. 2005;11(6):675–683. [DOI] [PubMed] [Google Scholar]
- 47.Rabinowitz T, Murphy KM, Amour JL, et al. Benefits of a Telepsychiatry Consultation Service for Rural Nursing Home Residents. Telemedicine and e-Health. 2010;16(1):34–40. [DOI] [PubMed] [Google Scholar]
- 48.Spaulding R, Belz N, DeLurgio S, et al. Cost savings of telemedicine utilization for child psychiatry in a rural Kansas community. Telemedicine and e-Health. 2010;16(8):867–871 [DOI] [PubMed] [Google Scholar]
- 49.Wong J, Motulsky A, Abrahamowicz M, et al. Off-label indications for antidepressants in primary care: descriptive study of prescriptions from an indication based electronic prescribing system. BMJ. 2017;356:j603. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.