Use of Telehealth Expedites Evaluation and Listing of Patients Referred for Liver Transplantation

Abstract

Background & Aims:

Liver transplantation is the only treatment that increases survival times of patients with decompensated cirrhosis. Patients who live further away from a transplant center are disadvantaged. Health care delivery via telehealth is an effective way to remotely manage patients with decompensated cirrhosis. We investigated the effects of telehealth on the liver transplant evaluation process.

Methods:

We performed a retrospective study of 465 patients who underwent evaluation for liver transplantation at the Richmond Veterans Affairs Medical Center from 2005 through 2017. Of these, 232 patients were evaluated via telehealth, and 233 via in-person evaluation. Using regression models, we evaluated the differential effects of telehealth vs. usual care on placement on the liver transplant waitlist. We also investigated the effects of telehealth on time from referral to initial evaluation by a transplant hepatologist, liver transplantation, and mortality.

Results:

Patients in the telehealth group were evaluated significantly faster than patients evaluated in person, without or with adjustment for potential confounders (21.7 days vs. 79.5 days; P<.01). Telehealth was also associated with a significantly shorter time placement on the liver transplant waitlist (138.8 vs. 249 days; P<.01). After propensity-matched analysis, telehealth was associated with a reduction in the time from referral to evaluation (HR 0.15, 95% CI 0.09, 0.21, p<0.01) and listing (HR 0.26, 95% CI 0.12, 0.40, p<0.01), but not to transplantation. In the intent to treat analysis of all referred patients, we found no significant difference in pre-transplant mortality between patients evaluated via telehealth vs in person. There was statistically significant interaction between model for end-stage liver disease (MELD)-Na scores and time to evaluation (P=.009) and placement on the transplant waitlist (P=.002) with Telehealth offering greater benefits to patients with low MELD-Na scores.

Conclusions:

Use of telehealth is associated with a substantial reduction in time from referral to initial evaluation by a hepatologist and placement on the liver transplant waitlist—especially for patients with low MELD scores, with no changes in time to transplantation or pre-transplant mortality. More studies are needed, particularly outside of the Veterans Administration Health System, to confirm that telehealth is a safe and effective way to expand access for patients undergoing evaluation for liver transplantation.

Introduction:

Liver transplantation is the only curative therapy for patients with decompensated cirrhosis and offers the best chance of long-term survival (1,2). The decision to list and transplant is made on the basis of Model of End Stage Liver Disease (MELD) score, which correlates well with mortality in patients with decompensated cirrhosis (1). While a MELD threshold of 12-15 is often used to begin liver transplant evaluation, patients are often not transplanted until the MELD scores are in the mid-20s to 30s based on organ availability (3). While liver transplantation has shown a mortality benefit in patients with higher MELD scores, distance from a transplant center is a negative prognostic indicator and associated with higher mortality (4,5). This represents a major challenge in clinically managing patients who live remotely, and better tools are necessary to standardize care.

Health care delivery via telehealth is increasingly being incorporated into clinical practice. Telemedicine is defined as the remote diagnosis and treatment of patients using telecommunication technology. Telehealth is a broader term that includes telemedicine but also telementoring (provider education and support using telehealth) and remote patient monitoring. Our group and others have previously shown that provider-to-provider interaction via Specialty Care Access Network- Extension for Community Health care Outcomes (SCAN-ECHO) has been shown as an effective way of triaging patients for liver transplantation and remotely managing patients with decompensated cirrhosis (6,7,8). Incorporation of telehealth via telemedicine at transplant centers has the potential to improve access and expedite liver transplant evaluation, especially in patients who do not reside near a transplant center. Theoretically, early involvement of a transplant team early may bring additional expertise in managing these sick patients and offer transplantation if they develop acute on chronic liver failure. Thus, while the idea of remote liver transplant evaluation via telehealth is an attractive option, it is unclear if telehealth offers benefits in those with a lower MELD scores (15-18). The aim of our study was to compare the impact of using telehealth on the time from referral to initial transplant hepatologist evaluation, on listing, transplantation and mortality in patients referred for liver transplantation.

Methods

This is a retrospective study evaluating the impact of the method of liver transplant evaluation on time from referral to evaluation, listing, transplantation and pre-transplant mortality. The study was reviewed and approved by Institutional Review Board (IRB). The manuscript was reviewed and approved by all authors prior to submission.

Study Population

In the VA system, liver transplantation is offered at one of six VATCs across the United States, including Richmond, VA (9). The process is initiated at the patient’s local VA and is commonly done by a gastroenterologist or primary care physician, in the absence of a transplant hepatologist. In patients with cirrhosis, referral is often initiated when the MELD-Na score≥15 or with a condition eligible for MELD exemption (1). However, there is no national VA standard and there are center-to center variations in terms of time of initiation of a transplant referral. A standardized checklist including labs, imaging, non-invasive cardio-pulmonary testing and psychosocial evaluations are completed by referring provider in preparation for liver transplant evaluation. The results from these tests are assembled and forwarded electronically to a VATC based on patient/provider choice. The work up is remotely reviewed by a transplant hepatologist and are either provisionally accepted or rejected. Those who are provisionally accepted then undergo an initial evaluation with the transplant hepatologist. This evaluation was done either by a traditional in-person evaluation or by telehealth.

All patients referred for liver transplant evaluation to the Veteran Health Administration transplant center (VATC) in Richmond, VA are prospectively enrolled in the natural history of cirrhosis study. The present study is a sub-study of the larger study evaluating all patients referred for liver transplant evaluation between January 1^st, 2005 and December 31^st, 2017.

Initial Evaluation In-person (Usual Care) or by Telehealth

Prior to 2/1/2011, the initial transplant hepatologist evaluation required a traditional in-person visit, which is considered usual care (UC). After 2/1/2011 the Richmond VA introduced telehealth, and all patients who met criteria for telehealth were completed in this fashion. Patients who did not meet criteria for telehealth evaluation included those requiring expedited inpatient evaluation (acute liver failure or MELD>35) or local patients from Richmond or neighboring VAs (<100 miles). The patients were brought into the local VA for the telehealth visit, where they were met by a telehealth licensed practical nurse or technician for documentation of vital signs and to help with the technology. The telehealth visit with the transplant hepatologist included a detailed history of the patient’s liver disease, co-morbidities, social history particularly substance abuse and social support, counseling about the transplant listing process, waiting list, MELD scores, transplant procedure, immunosuppression, complications and posttransplant outcomes. After the initial evaluation, both groups of patients undergo an in-person listing visit, and their candidacy discussed at the multidisciplinary selection committee. This time from evaluation to listing includes the time patients waited for specialized testing (such as Cardiac catheterization), additional treatments (such as down staging of a patient outside Milan with locoregional therapy and confirmation prior to listing), presentations at multidisciplinary tumor board and transplant selection committees. In some cases, listing was deferred if MELD scores improved after referral till scores went up again. If approved, patients are listed and allowed to return home, where they are co-managed by the referring physician and the transplant center. Those Veterans not deemed to be acceptable candidates for liver transplant waitlist are sent back to the referring VA hospital.

The primary endpoint of the study was the time from referral to initial evaluation by the transplant hepatologist. The secondary endpoints evaluated in the present study included the time from referral to listing (or listing decision), time from referral to transplantation and mortality from the time of referral till transplantation.

Statistical Analysis:

Data were summarized using mean (standard deviation) when normally distributed otherwise median (interquartile range) for continuous variables and number and percentage for categorical variables. Differences between the groups were examined using Student’s t test for normally distributed continuous variables, Kruskal-Wallis equality-of-populations rank test for skewed continuous variables, and chi-square test for binary and categorical variables. To examine the relationship between telehealth group status and time to different milestones in evaluation, we used generalized linear regression models with logarithmic link functions. Generalized linear regression models were used instead of the ordinary linear regression because they handled the log transformation of the data better and accurately estimated the standard errors (10). The outcome variables were time from referral to evaluation, listing, and transplantation. In addition to the unadjusted analysis, a second model adjusted for patient level covariates (race, age, hepatitis C virus [HCV] infection, MELD Na score and distance from the transplant center (in miles)). In addition, MELD exemption status was included in the model for time to transplantation alone. A propensity-weighted analysis was performed. Propensity score was estimating using a logit model, predicting probability of being in Telehealth group by etiology, Distance in miles, HCV, blood group, race, insurance status, marital status, gender and education. We specifically did not include the year of referral because of the high correlation between telehealth and the year of referral (as all patients referred after 2011 and met the inclusion criteria were evaluated by telehealth). A nominal P-value < 0.05 was considered as statistically significant.

Cox proportional hazard models were used to test for the effect of telehealth on pre-transplant mortality based on an intention to treat analysis. The proportional hazard assumptions were visually tested using the log(−log) plots and by testing for the nonzero slope in a GLM of the scaled Schoenfeld residuals on time (11). All analyses were performed using Stata 14.0 (StataCorp, College Station, Texas).

RESULTS

Study Cohort

During the study period, 715 patients were referred for liver transplant evaluation to the Richmond VATC from 2005 to 2017, of whom, 465 met the inclusion criteria. {Fig 1}. Of these, 232 were evaluated by telehealth and 233 by usual care. These patients were referred from 60 unique VA Medical centers across 29 states. Briefly, the population was predominantly male (98%), non-Hispanic Caucasian, with a median age of 59 years (Table 1). The telehealth and usual care groups were similar with regards to distribution of Gender and ethnicities, however, patients evaluated by telehealth were older (61 vs. 57 years, P<0.01), with lower incidence of chronic HCV infection (60 vs. 69%, p=0.04) and a lower median distance from the transplant center (372 vs. 459 miles, p=0.047). MELD-Na scores at referral were slightly lower in the telehealth (15 vs. 17, p<0.001), with no differences in patients qualifying for MELD exemptions.

Figure 1:

Flow of patients referred for liver transplantation (n=715)

Table 1:

Baseline characteristics of patients undergoing transplant evaluation by telehealth and in-person evaluation.

Variables	Telehealth (N=232)	In-person evaluation (N=233)	P-value
Age (median, IQR)	61(7)	57(7)	<.0001
Gender(N, %)
Female	7(3.02)	10(4.29)	<.0001
Male	225(96.98)	223(95.70)
Race (N, %)
White	139(59. 91)	146(62.66)	0.34
Black	59(25.43)	42(18.03)
Hispanic	24(10.34)	32(13.73)
Other	2(0.86)	5(2.15)
Unanswered	8(3.44)	8(3.43)
Education (N, %)
Beyond High School	93(40.09)	71(30.47)	0.03
High School	115(49.57)	143(61.37)
Unknown	24(10.34)	19(8.15)
Insurance (N, %)
No Insurance	135(58.19)	145(62.23)	0.55
Insurance	97(41.81)	88(37.77)
HCV Infection (N, %)
No-HCV	93(40.09)	73(31.33)	0.04
HCV	139(59.91)	160(68.67)
Approval for Listing (N, %)
Not Approved	112(48.28)	81(34.76)	0.003
Approved	120(51.72)	152(65.24)
MELD (median, IQR)	14(8)	16(9)	<.0001
MELD-Na (median, IQR)	15(9)	16.5(15)	<.0001
MELD by Exemption (N, %)
No	181(78.02)	194(83.3)	0.15
Yes	51(21.98)	39(16.7)
Distance in miles (median, IQR)	371.5(514.5)	459(601)	0.047
Days from referral to initial evaluation (median, IQR)	22 (9)	54 (77)	<.0001
Days from referral to listing (median, IQR)	95 (106)	149 (217)	<.0001
Days from referral to transplant (median, IQR)	218 (247)	244 (343)	0.08
Days from Referral to Death (median, IQR)	381.5 (641)	703.5 (1161)	<.0001

Of the 465 patients enrolled in the study, 272 were approved and listed, while 193 were denied. Patients in either groups were denied because of standard clinical or psychosocial contraindications that were discovered during the evaluation process.

Of the approved patients, 186 were transplanted during the study period, 51 died on the waiting list and 35 remained on the list awaiting transplantation. Of the 193 that were deemed not transplant candidates, 118 died after evaluation while 75 remained alive at the end of the study period (Figure 1).

Telehealth Reduces Time to Initial Evaluation

Patients in the telehealth group were evaluated significantly faster than patients in usual care group (22 vs 54 days; P<0.001). After adjusting for age, race, presence of HCV infection, MELD-Na, and the distance from the transplant center, telehealth evaluation remained highly significantly associated with a decrease in time from referral to evaluation (21.7 days vs. 79.5 days; p<0.01) (Table 2). After propensity-matched analysis, telehealth was associated with an 85% reduction in the time from referral to evaluation (HR 0.15, 95% CI 0.09, 0.21, p<0.01) (Table 3).

Table 2:

Adjusted time from referral to evaluation, listing and transplantation in patients evaluated by telehealth and in-person evaluation respectively.

	Telehealth (days)	In-person evaluation (UC) (days)	P Value
Referral to evaluation	21.7	79.5	0.000
Referral to listing	138.8	249.0	<0.01
Referral to transplant	324.7	409.0	0.165

Table 3:

Propensity-score Weight Based Time to Initial Evaluation, Listing, Transplant

	Time to Evaluation		Time to Listing		Time to Transplant
	eβ	se	β	se	β	se
Main
UC	Reference		Reference		Reference
Telehealth	0.149*	(0.0317)	0.256*	(0.071)	0.407	(0.196)
MELD/MELD-Na	0.954*	(0.007)	0.917*	(0.008)	0.896*	(0.030)
Telehealth#	1.041*	(0.013)	1.058*	(0.018)	1.051	(0.030)
MELDMELD-Na
Age	1.003	(0.005)	0.994	(0.007)	0.985	(0.010)
White	0.980	(0.034)	0.948	(0.047)	0.871	(0.094)
No HCV	0.964	(0.077)	0.970	(0.102)	0.971	(0.136)
No MELD					0.690*	(0.111)
Exemption
Distance (miles)	1.000	(0.000)	1.000	(0.000)	1.000	(0.000)
N	465		465		186
adj. R2
BIC	8746.8		11402.8		4991

e^β: Exponentiated coefficients; se: standard errors in parentheses.

^*:p < .05.

In addition, there was a significant interaction between method of the transplant evaluation (telehealth or usual care) and MELD-Na. Specifically, this meant that time from referral to evaluation was different between the telehealth and usual care groups, but the difference varied based on the MELD scores at the time of referral. Time from referral to evaluation was consistently low in the telehealth group across all MELD scores, while that for the usual care was significantly shorter in patients with high MELD scores (Figure 2A and Supplementary Figure 3A). This interaction was statistically significant (coefficients of interaction 0.03, P value=0.009) and was adjusted for in the multivariable model. We also looked at time to evaluation by the year of evaluation as shown in supplementary figure 2A. This figure indicates that while there were wide fluctuations in the time to evaluation based on the year of referral in the usual care group, this was stable and low in the telehealth group and indicates that the differences between the two groups were not due to temporal trends.

Figure 2:

Time from referral to evaluation (panel A), listing (panel B) and transplantation (panel C) in patients evaluated by Telehealth versus in-person, by MELD-Na

Telehealth Reduces Time from Referral to Listing

Patients who underwent the initial evaluation by telehealth were listed significantly earlier than the usual care group (95 vs 149 days; p<0.001). After adjusting for confounders, telehealth evaluation remained highly significantly associated with a decrease in time to listing (139 vs. 249 days, p<0.01) (Table 2). After propensity-matched analysis, telehealth was associated with a 74% reduction in time to listing (HR 0.26, 95% CI 0.12, 0.40, p<0.01). In addition, there was a significant interaction between method of the transplant evaluation and MELD-Na. Time from referral to listing was fairly stable in the telehealth group across all MELD scores, while that for the usual care was significantly higher in patients with low MELD scores (Figure 2B and Supplementary Figure 3B). For example, the time from referral to listing in the telehealth and usual care groups were 150 and 311 days respectively in patients with MELD<15, but were 59.8 and 52.5 days respectively for MELD>25. This interaction was statistically significant (coefficients of interaction 0.05, P value=0.002). We also looked at the time from referral to listing by the year of evaluation as shown in supplementary figure 2B. This figure indicates that while there were wide fluctuations in the time to evaluation based on the year of referral in the usual care group, this was stable and low in the telehealth group and indicates that the differences between the two groups were not due to temporal trends that preceded telehealth implementation.

Telehealth does not affect time from Referral to Transplantation

The median time to transplant was not significantly different between the two groups on unadjusted (218 vs 244 days; P=0.084) or adjusted analysis (325 vs. 409 days; P=0.08). The difference in the time to transplantation between the two groups did not have a statistically significant coefficient of interaction with MELD at referral (coefficient of interaction 0.04, P=0.18).

Impact of Telehealth on Pre-transplant Mortality

Overall, 169 patients out of the 465 (51 on waiting list and 118 who were not listed) who were referred, died without receiving a liver transplant. There was no difference in pretransplant mortality between patients evaluated by telehealth or usual care in unadjusted analysis (Figure 3). Factors associated with pre transplant mortality included MELD-Na (HR=1.18, P<0.001), age and white race. On multivariable cox proportion hazard model (using Breslow method), the method of transplant evaluation was not associated with pre-transplant mortality (HR for telehealth=1.27, P=0.17) (Supplementary Figure 1). We also conducted a subgroup analysis among the 272 patients who were listed for transplantation. On multivariable cox regression analysis, age at the time of evaluation (HR 1.06, P=0.02), MELD-Na (HR 1.18, P=0.0001) and presence of HCV infection (HR 2.58, P=0.03) were associated with mortality on the transplant list, while the mode of evaluation by telehealth (HR 1.05, P=0.89), MELD exemption and the distance from the transplant center had no impact. It is to be noted that our sample size was probably underpowered to detect a small effect size on pretransplant mortality. For example, we need a sample of 1971 patients to be able to detect a change in mortality rate from 20% to 15% with 80% power and 95% confidence.

Figure 3:

Kaplan-Meier survival curve between those evaluated by Telehealth versus in-person evaluation (usual care)

Discussion:

Access to specialty care has been associated with improved survival in patients with liver disease but local access to liver transplantation is not always feasible (8). The increasing popularity of telehealth offers an accessible (12,13), convenient (14), and cost-effective (15) way to access transplantation (16).

In an elegant meta-analysis (6), Serper and Volk identified 20 published studies on the use of telemedicine in chronic liver disease prior to May 2017. While there have been several studies using telemedicine in HCV treatment (17,18), to aid in procedural or surgical management (19,20) and in the management of HCC (21), no studies where telehealth was used in the pre-transplant evaluation were identified. Most recently, researches from Los Angeles and Cincinnati have described the successful use of telehealth in the management of post liver transplant patients with good outcomes (22, 23). We have reported that the use of telementoring using telehealth based technology designed to transfer subspecialty hepatology knowledge to non-transplant providers was highly effective in identifying non-candidates for liver transplantation (7).

Our work from the current study is an extension of the above work describing the use of telehealth in the evaluation of patients for liver transplantation. Our results demonstrate that the use of telehealth was associated with a markedly shorter time to transplant evaluation as well as the time to listing. Not surprisingly, this did not translate into a decrease in time to transplantation, likely because the latter is a complex metric that is driven primarily by organ availability. Our data also showed that the median distance from transplant center was shorter in the telehealth compared to usual care group. This is likely a reflection of the distribution of the referral centers by the addition of two new VA transplant centers at Houston, TX and Madison, WI in the latter half of the study period.

We also find that pretransplant mortality were similar in the two groups either on an intention to treat analysis or on a subgroup analysis of patients who were listed. This is similar to a landmark study by Su et al, who found that mortality of patients evaluated using a telementoring program, SCAN-ECHO, was lower than patients who were not evaluated by a specialist, and equivalent to those undergoing in-person specialist evaluation (8).

While this study was done within the VA system where transplantation is centralized, telehealth can improve access even in the non-Veteran population, since access to transplant centers are not uniformly available throughout the United States. In fact, 13 out of the 50 states do not have liver transplant centers (24). The benefits of telehealth include a reduction in travel, a decreased need for both patients and their caregivers to take time off, and improved overall access. According to the American Time Use survey data, the average medical in-person appointment that involves travel, waiting, payments and completion of paperwork takes an average of 121 minutes with only 20 minutes of direct time with the provider (25). Our study shows that telehealth is not only a convenient method associated with improvement in access to care but leads to an expedition of both transplant evaluation and listing. This is particularly helpful in an era where transplant centers are doing outreach, traveling to cities hundreds of miles away (26). Telehealth is an attractive method to provide transplant services for these patients, and demonstration of expedited evaluations and listing while maintaining clinical outcomes, should help to open up insurance coverage for transplant evaluation via telehealth.

Although we have provided robust data, the study has limitations. First, it is an observational study and patients were not randomized to either intervention. However, all patients prior to 2011 were evaluated in-person and all referred after 2011 who met study inclusion criteria were evaluated by telehealth and the analysis was propensity weighted. Second, the study was done within the VA system where transplantation is centralized, and patients are referred across greater distances compared to the non-VA setting (5). However, access to transplantation is not uniform across the United States and many patients have to travel to access transplant centers even in the non-VA setting. While the time from referral to evaluation and listing are available in the VA setting, such data is not available from non-VA transplant centers. Therefore, we do not know if our findings can be generalizable to the non-VA setting. Lack of reimbursement and licensing issues across state lines were not an issue in the VA system, but could pose potential barriers in the private sector. Finally, while this is the largest study looking at the role of telehealth in liver transplantation, it is possible that our sample size was underpowered to estimate a difference.

We believe that while this data is important to show benefits of telehealth in transplant evaluation, more studies are needed, particularly from non-VA centers. Some non-VA academic centers are at the forefront of telehealth implementation, where they use grants or income from clinical transplant services, to support a telehealth program (16,22,23).

In summary, our results support a transplant hepatologist evaluation using telehealth was associated with significantly reduced time to evaluation and listing without adversely affecting pre transplant mortality compared to the current standard of care of in-person evaluation at a transplant center. The ability to successfully evaluate and list patients via telehealth and obtain the same outcomes in terms of time to transplant and pre-transplant mortality is significant because of the numerous advantages that telehealth offers to improve overall access to transplantation. However, more studies are needed, particularly outside of the Veterans Administration Health System, to confirm that telehealth is a safe and effective way to expand access for patients undergoing evaluation for liver transplantation.

Supplementary Material

1

Supplementary Figure 1: Pre-transplant mortality in patients evaluated by telehealth and in-person evaluation on an intention to treat analysis.

Supplementary Figure 2A and B: Time to evaluation (Fig 2A) and listing (Fig 2B) by year of Referral in patients in evaluated by Usual Care (2005-2011)) and Telehealth (2011-2017).

Supplementary Figure 3A and B: Time to evaluation (Fig 3A) and listing (Fig 3B) categorized by MELD scores (<15, 15-25, >25) in patients evaluated by Usual Care and Telehealth.

What You Need to Know.

Background: Patients with cirrhosis who live far from transplant centers are less likely to receive liver transplants. Health care delivery via telehealth is an effective way to remotely manage patients with decompensated cirrhosis.

Findings: In a retrospective analysis of patients who underwent evaluation for liver transplantation at a Veterans Affairs Medical Center, the use of telehealth was associated with substantial reductions in time from referral to initial evaluation by a hepatologist and placement on the liver transplant waitlist, with no changes in time to transplantation or pre-transplant mortality.

Implications for patient care: Telehealth appears to be a safe and effective way to expand access for patients undergoing evaluation for liver transplantation, although more studies are needed