This application presents the SRTR evaluation of the new liver and intestine allocation policy, referred to as ''acuity circles'' (AC). The new liver allocation system was implemented on Feburary 04, 2020. The evaluation used data between February 04, 2019 and February 03, 2021.
The evaluation investigated the effect of AC on three distinct components of the liver transplant system:
Each section shows descriptives, separate adjusted analyses, and relevant analytical details.
Note. Under normal circumstances, the liver allocation system would likely take several months to reach an equilibrium. The emergence of COVID-19 may confound many of the analyses included in the evaluation. The individual sections discuss the potential effect of COVID-19 on the evaluation.
This section describes the state of the waiting list during the months before and after AC implementation, including the number of removals due to death or becoming too sick to undergo transplant. Percentage of wailist time is the number of days spent at the level of interest during the month divided by the total number of days on waiting list; eg, the percentage of all waitlist time for candidates with any HCC exception. The information can be presented across several stratification variables, including allocation MELD.
Figure notes
This section describes the unadjusted deceased donor transplant rate, that is, deceased donor transplants for every 100 person-years. This section includes inactive candidates; for example, see the Current PELD/MELD stratification variable.
Figure notes
This section presents the effect of AC on adjusted deceased donor transplant rates, that is, the effect of the stratification variables on the rate of deceased donor transplants after statistically adjusting for the other stratification variables. The first and second figures present the differences in deceased donor transplant rates before and after AC implementation. The third figure presents the difference-in-differences analysis, which is the primary analysis for understanding the effect of AC on deceased donor transplant rates.
Interpretation of before implementation figure. A transplant rate ratio above 1 indicates that, before implementation of AC, the transplant rate was higher for the level of interest than the reference level.
Interpretation of after implementation figure. A transplant rate ratio above 1 indicates that, after implementation of AC, the transplant rate was higher for the level of interest than the reference level.
Interpretation of change before and after implementation figure. A transplant rate ratio above 1 indicates that the difference in the transplant rate between the level of interest and the reference level was larger after implementation than before.
Effect of COVID-19. These analyses have a relatively low risk of confounding from COVID-19, because they compare the relative rates of deceased donor transplant during the same periods of time. However, if the emergence of COVID-19 had different effects across the levels of the stratification variable, then the post-AC comparisons would be confounded, ie, include the effect of both AC and COVID-19.
Cohort definitions: The unadjusted analyses presented the deceased donor transplant rates for the months before and after AC implementation. These months included candidates on the waiting list at some point during the period. The figures for the unadjusted analyses stated the specific dates for a given month. The adjusted analyses used candidates on the waiting list between February 4, 2019, and February 03, 2021, which was a slightly different cohort from the unadjusted analyses. The adjusted analyses set the before AC period as between February 4, 2019, and February 3, 2020, and the after AC period as between February 4, 2020, and February 03, 2021.
Adjusted analyses: A difference-in-differences analysis investigated the relative change in deceased donor transplant rates for each stratification variable. The underlying model was a piecewise exponential model with the timescale set to calendar time. The baseline hazard included different components for each month before and after AC implementation. Importantly, the model censors for removal from the waiting list for reasons other than deceased donor transplant, and therefore estimated the cause-specific hazard of deceased donor transplant rate. The model adjusted for the stratification variables and included an indicator for post-implementation of AC. Interactions between the stratification variables and the indicator for post-implementation estimated the difference-in-differences before and after AC for each stratification variable.
Handling of inactive status: The unadjusted deceased donor transplant rates and the adjusted analyses did not include inactive time.
This section describes the offer rates for each month; that is, the average number of deceased donor offers within the top 1, 5, or 10 spots on a match run a candidate receives per person-year. For example, the offer rate for 1st offers is the average number of offers at the top of the match run a candidate would receive over 1 year. Similarly, the offer rate for the top 10 offers is the average number of offers in the top 10 spots of a match run a candidate would receive over 1 year.
Figure notes
This section presents the effect of AC on adjusted offer rates for deceased donors. Offer rates are the average number of deceased donor offers a candidate receives per person-year. The offer number of interest determines whether the user wants the offer rates for the top 1, 5, or 10 offers of a match run. The first and second figures present the differences in offer rates before and after implementation of AC. The third figure presents the difference-in-differences analysis, which is the primary analysis for understanding the effect of AC on adjusted offer rates.
Interpretation of before implementation figure. An offer rate ratio above 1 indicates that, before implementation of AC, the offer rate was higher for the level of interest than the reference level.
Interpretation of after implementation figure. An offer rate ratio above 1 indicates that, after implementation of AC, the offer rate was higher for the level of interest than the reference level.
Interpretation of change before and after implementation figure. An offer rate ratio above 1 indicates that the difference in the offer rate between the level of interest and the reference level was larger after implementation than before.
Effect of COVID-19. These analyses have a relatively low risk of confounding from COVID-19, because they compare the relative offer rates during the same periods of time. The time period effects would identify any systematic changes in the offer rates after COVID-19; eg, systematically fewer offers due to fewer donors. However, if the emergence of COVID-19 had different effects across the levels of the stratification variable, then the post-AC comparisons would be confounded, ie, include the effect of both AC and COVID-19.
Cohort definitions: The unadjusted analyses presented the offer rates for months before and after AC implementation. The figures for the unadjusted analyses stated the specific dates for a given month. The adjusted analyses used match runs with submit dates between February 4, 2019, and February 03, 2021, which were slightly different from the unadjusted analysis. Importantly, the adjusted analyses defined the before AC period as between February 4, 2019, and February 3, 2020, and the after AC period as between February 4, 2020, and February 03, 2021.
The cohort only included match runs with at least one acceptance, and offers after the last acceptance on a match run were removed.
Adjusted analyses: A difference-in-differences analysis investigated the change in offer rates for each stratification variable. A Poisson model estimated the rate of offers for a candidates during a period of time. The outcome was the number of offers a candidate received during a status update. A status update was the period of time a candidate spent in a particular allocation PELD/MELD or status 1A/1B. The model used an offset equal to the natural log of days in a status, and an overdisperson term allowed for a more flexible mean-variance relationship than a typical Poisson model. The model adjusted for monthly effects before and after AC implementation, an indicator for post-implementation, and every stratification variable. Interactions between the stratification variables and the indicator of post-implementation estimated the difference-in-differences for each stratification variable.
Aligning match runs and status histories: The match submit date determined the timing of the offers and was compared with the start and stop times of candidate status histories. Importantly, only calendar dates tracked the status history of a candidate, not the date and time of the status update. Thus, inactive candidates could receive offers before the specific time of inactivation on a given day. This will appear in the analysis as inactive candidates receiving an offer.
This section describes the population of donors during the months before and after AC implementation. The information can be presented by several stratification variables.
Figure notes
This section presents the effect of AC on donor liver yield, which is the average number of transplanted livers from a donor with any organ recovered for the purposes of transplantation. The first figure presents the odds ratios of the linear trend (per 1-week difference) before and after AC, and includes donors recovered between February 4, 2019, and February 03, 2021. The rightmost part of the figure is the difference in the odds ratios before and after implementation and is the primary effect of AC on liver yield. An odds ratio above 1 would indicate that the trend of liver yield increased after implementation compared to before. The second figure presents the non-linear trend of liver yield over the study period. The reference level is the date of AC implementation.
Effect of COVID-19. These analyses have a relatively high risk of confounding from COVID-19, because they compare the rate of change in yield before and after AC. If the emergence of COVID-19 affects liver yield, then the slope after implementation of AC will include the effect of AC and COVID-19.
Cohort definitions: The descriptive analyses used donors recovered during monthly intervals before and after AC implementation. The descriptive figures stated the specific dates for a given month. The adjusted analyses (Liver yield tab) used donors recovered between February 4, 2019, and February 03, 2021. The adjusted analyses set the before AC period as between February 4, 2019, and February 3, 2020, and the after AC period as between February 4, 2020, and February 03, 2021.
Donor yield: A logistic regression estimated the effect of AC on the rate of change in liver yield. Specifically, a linear effect estimated the effect of calendar time on liver yield with an interaction with the implementation date for acuity circles. These effects were translated into odds ratios per week. So the primary effect of AC was the difference in odds of liver yield per week before and after implementation. A separate logistic regression estimated the non-linear time trend of liver yield (The change in liver yield over time) with smoothing splines. Both regressions adjusted for the stratification variables in the descriptives section.
This section describes the population of deceased donor transplants during the months before and after AC implementation, including cold ischemia time and the distance between donor and transplant hospitals, which was the straight line distance between the latitude and longitude of the two hospitals. The information can be presented by several stratification variables.
Figure notes
This page presents the number of deceased donor transplants in each DSA before and after AC. The before AC cohort included between February 03, 2019, and February 03, 2020. The after AC cohort included between February 04, 2020, and February 03, 2021. It is the same cohort used to determine the MMAT.
This page presents the number of heptocellular carcinoma (HCC) deceased donor transplants in each DSA before and after AC. The before AC cohort included between February 03, 2019, and February 03, 2020. The after AC cohort included between February 04, 2020, and February 03, 2021. It is the same cohort used to determine the MMAT.
The median allocation PELD/MELD at transplant (MMAT) was calculated for transplant programs and donation service areas (DSAs). Because the precision of the median depends on sample size, the cohort before AC was restricted to the same number of days as the post AC period. Specifically, the before AC cohort included transplants between February 03, 2019, and February 03, 2020. The after AC cohort included transplants between February 04, 2020, and February 03, 2021.
Effect of COVID-19. These analyses have a relatively moderate risk of confounding from COVID-19. The emergence of COVID-19 may reduce the prevalence of lower PELD/MELD transplants in affected areas. If this occurred differentially across transplant programs, then the distribution of PELD/MELD at transplant could differ before and after AC due partly to COVID-19.
The evaluation of AC will not include posttransplant outcomes at this time. The emergence of COVID-19 likely confounds such evaluations, because even recipients who underwent transplant before COVID-19 were at risk of negative COVID-19-related outcomes due to known risk factors, eg, immunosuppression.
This page provides a list of definitions of terms and abbreviations used throughout the evaluation:
The cohort for the unadjusted analyses are updated every month. This tab only describes changes beyond the updated cohorts.