These authors have contributed equally to this work
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
A new lung allocation system was introduced in France in September 2020. It aimed to reduce geographic disparities in lung allocation while maintaining proximity. In the previous two-tiered priority-based system, grafts not allocated through national high-urgency status were offered to transplant centres according to geographic criteria. Between 2013 and 2018, significant geographic disparities in transplant allocation were observed across transplant centres with a mean number of grafts offered per candidate ranging from 1.4 to 5.2. The new system redistricted the local allocation units according to supply/demand ratio, removed regional sharing and increased national sharing. The supply/demand ratio was defined as the ratio of lungs recovered within the local allocation unit to transplants performed in the centre. A driving time between the procurement and transplant centres of less than 2 h was retained for proximity. Using a brute-force algorithm, we designed new local allocation units that gave a supply/demand ratio of 0.5 for all the transplant centres. Under the new system, standard-deviation of graft offers per candidate decreased from 0.9 to 0.5 (
Geographic model in lung allocation policies varies from one country to another. In Germany, Spain, United Kingdom, United States or France, the lung allocation models combine geography with urgency. Usually, national allocation is applied for the most urgent tier and local or regional allocation for non-urgent tiers (
In France there are 183 organ procurement centres (OPCs) and nine adult lung transplant centres (LTCs). In 2018, 419 patients were newly registered on the national waiting list and 373 transplants (5.5 per million population) were performed, which corresponds to 1.1 new candidates for one graft (
France’s allocation rules are developed by the Agence de la biomédecine, in collaboration with the transplant community. The previous two-tiered lung transplant allocation system was based on national allocation for patients with high-urgency (HU) status and local, regional, and then national allocation for elective patients.
HU status based on immediate risk of death is requested by the transplant centres and assigned by a transplant physician from another region. HU status could not be granted to candidates with chronic obstructive pulmonary disease. Lungs from donors under the age of 55 not allocated to a HU candidate are first offered to elective pediatric candidates. For elective candidates, donor lungs are allocated to LTC, with selection of the candidate at the LTC discretion.
Donor lungs are allocated to identical blood type candidates but candidates with poor access to transplant as highly sensitized patients may be assigned an exception with compatible blood type donor lung offers. For lung-kidney or lung-liver candidates, the kidney and the liver are automatically allocated with the lung.
Concerning the geographical model, local allocation was based on local allocation units. One local allocation unit comprises a network of OPCs’ that serve one LTC (
Previous
In 2018, 15% of grafts were allocated to HU candidates, 1% to pediatric candidates without HU status, 2% to lung-kidney and lung-liver transplant candidates, and 82% to elective adult candidates. Local, regional and national allocations accounted for 18%, 22%, and 43% of the transplants, respectively.
The assessment of the previous lung allocation model by the French Agence de la biomedecine identified that the sharing of brain-dead donor lungs was not fair across LTCs. Indeed, over the 2013 to 2018 period, the mean number of lungs offers per candidate, including offers to HU candidates, ranged from 1.4 to 5.2 among LTC, with a mean of 3 and a standard deviation of 1.2 (
Disparity in lungs sharing across transplant centres over the 2013–2018 period.
Transplant centre | Procurement hospitals in the local allocation unit (n) | Lungs recovered in the local allocation unit (n) | Candidates (n) | Lung offer per candidate (n) |
---|---|---|---|---|
Bordeaux | 10 | 39 | 176 | 3.5 |
Lyon | 23 | 17 | 207 | 4.4 |
Marseille | 28 | 96 | 287 | 2.9 |
Marie Lannelongue | 1 | 6 | 301 | 2.8 |
Nantes | 32 | 54 | 153 | 5.2 |
Foch | 1 | 21 | 400 | 1.4 |
Bichat | 2 | 10 | 308 | 1.9 |
Strasbourg | 26 | 72 | 296 | 2.4 |
Toulouse | 11 | 19 | 134 | 3 |
This situation resulted from several geographic disparities. First, the number of lungs recovered from brain-dead donors differed from one region to another. Secondly, the number and medical condition of candidates differed from one LTC to another. Third, the number of OPCs within local allocation units differed among LTCs from 1 to 32, with a mean of 14.9 and a standard deviation of 12.5 (
These geographic disparities, together with differences in donor selection across LTCs, might account for differences in lung offers and access to transplantation across LTCs. Indeed, over the 2013–2018 period, the 1-year cumulative incidence of transplantation estimated with competing risk analysis for newly registered candidates varied from 62% (52–77) to 97% (95–99) across LTCs (
In addition, regional allocation had several disadvantages. Indeed, it contributed to geographic disparities in graft allocation given the number of LTCs varied from 0 to 4 across the regions and regional allocation was affecting the graft offer process to the disadvantage of national allocation.
Lastly, maintaining a prominent place for national allocation was considered as the most effective way to address geographic disparities (
All data used in this study were extracted from the Cristal national database (
To develop the new allocation model, all lungs from brain-dead donors transplanted between 1 January 2013 and 31 December 2018, were included. Calculations were performed using ArcGIS Network-Analyst 10.6, numpy (
The effect of the changes in the allocation system was assessed by comparing the 8 September 2020 to 8 September 2021, post-implementation cohort of candidates (
The three-month cumulative incidence was calculated using Fine & Gray method considering transplantation and death or delisting from the waiting list for worsening reason as competing event. Cumulative incidence of transplantation and waitlist mortality or delisting for clinical worsening were assessed with the competing risk analysis (
We designed optimized local allocation units to achieve geographic equity (
For each LTC, all possible combinations of OPCs were calculated with a brute-force algorithm to determine all possible optimized local allocation units. For each combination, we calculated the new supply/demand ratio, defined as the number of donor lungs within the optimized local allocation unit divided by the total number of transplantations performed in the LTC (
The judgement criteria for choice between all combinations were: 1) a similar supply/demand ratio between all LTC and 2) a lower standard deviation of the supply/demand ratio than the previous local allocation units for the nine active LTCs (
(Continued).
Finally, the number of OPCs belonging to any local allocation unit decreased from 120 to 92 (
Mean and standard deviation for the proportion of donor lungs from the local allocation unit transplanted in the assigned transplant centre according to the geographic allocation model.
Model | Mean ( |
Standard deviation ( |
Min | Max |
---|---|---|---|---|
Previous model (model 0) | 0.87 (ref) | 0.88 (ref) | 0.11 | 3.13 |
Supply/demand ratio: 0.5 (model 1) | 0.48 (0.23) | 0.006 (<0.001) | 0.29 | 0.6 |
Final model (model 2) | 0.73 (0.7) | 0.19 (0.04) | 0.47 | 1.85 |
In addition, the allocation sequence has been simplified by removing regional allocation (
The developed algorithm is available online under a Creative Common license:
The number of candidates and transplant recipients declined respectively from 358 to 285 (−20%) and 257 to 197 (−23%) between the pre- and post-implementation periods.
Type of geographic allocation, shipping distance between OPCs and LTCs, cold ischemia time and graft offers per candidate before and after introduction of the new system are displayed in
Lung allocation metrics for grafts from brain-dead donors before and after the change in the geographic model.
Period | |||
---|---|---|---|
Pre-implementation | Post-implementation |
|
|
Percentage of transplants by type of geographic allocation for each transplant centre | |||
mean (standard deviation) | |||
Local | 25.7% (19.1) | 20.3% (9.8) | 0.48 (0.08) |
Regional and national | 74.3% (19.1) | 79.7% (9.8) | 0.48 (0.08) |
Shipping distance in km for geographic allocation (Km) | |||
Mean | 395 | 406 | 0.7 |
Standard deviation | 296 | 304 | 0.64 |
Cold ischemia time | |||
Mean | 6h11 | 5h58 | 0.08 |
Standard deviation | 1h20 | 1h13 | 0.51 |
Lung offers per candidate by transplant centre | |||
Bordeaux | 3.4 | 2.8 | — |
Lyon | 3.7 | 1.9 | — |
Marseille | 2.6 | 1.7 | — |
Marie Lannelongue | 2.3 | 2.3 | — |
Nantes | 4.1 | 2.5 | — |
Foch | 1.2 | 1.7 | — |
Bichat | 1.6 | 1 | — |
Strasbourg | 2.2 | 2.2 | — |
Toulouse | 3.2 | 2.7 | — |
Mean | 2.7 | 2.1 | 0.04 |
Standard deviation | 0.9 | 0.5 | 0.08 |
Under the new allocation system, mean and standard deviation of offers per candidate decreased at the edge of significance from 2.7 to 2.1 (
The 3-month cumulative incidence of death and delisting for worsening medical condition (0.16% vs. 0.17% before,
Three months survival before and after the new lung allocation system. Panel
In September 2020, the French Agence de la biomedicine in collaboration with the transplant community implemented a new geographic lung allocation system for non-urgent candidates. The new system aimed to address geographic disparities in lung sharing through redistricting local allocation units, removing regional allocation, and improving national distribution.
While the use of LAS was discussed with support from some French transplant centres, the system has remained an urgency tier-based system, with graft allocation first to HU candidates in an immediate life-threatening situation. The LAS usefully takes account of waiting list and post-transplant mortality within 1 year (
The primary intention of the new system was to reduce geographic disparities in lung supply to non-urgent candidates. Among all the reasons for such disparities, the local allocation units’ make-up was the most easy to change determinant. A brute-force algorithm was used to design new optimized local allocation units. All combinations ensuring a similar supply/demand ratio among LTCs were explored. Finally, new local allocation units were designed to provide each LTC with approximately 50% of their lung grafts demand.
We used the standard deviation of the mean number of lungs offers per candidate across the nine LTCs as a metric for equity. Potential changes in geographic disparity with the new system were not simulated before its implementation, since the geographic allocation is centre- and not patient-based. Indeed, amount of patients per transplant centre was too low and selection criteria varying a given day.
The main 1 year consequences of the implementation of the new geographic allocation system are that allocating lungs to LTCs according to supply/demand ratio reduces disparities in graft offers per candidate and disparities in percentage of transplants performed by local allocation across LTCs without increasing the distance traveled by lungs. Mean dropped significantly while standard deviation results are at the edge of significance due to the low sample number. A lower ratio of 0.05 points would make the differences significant. Not to mention that the post-implementation period was affected by the COVID-19 epidemic, which reduced transplant and lung procurement activity. Thus, the result of the modification in the allocation system aligns with the goals of the new policy. No unexpected changes in type of geographic allocation, cold ischemia time, and pre- and post-transplant outcomes were observed. Even if they are not significant, these indicators seem to improve with the new system.
The new system has several advantages over the previous French geographic allocation system and over geographic allocation systems based on fixed distance. First and most importantly the supply/demand ratio-based system can reduce geographic disparities in the number of grafts offered per candidate across LTCs. Secondly, short transport distances are maintained for a significant pool of transplants. Thirdly, the system is easy to adjust in case of local lung procurement and transplant activities change. Likewise, the local allocation units can be modified if a transplant centre opens or closes. Finally, regional allocation has now been cancelled, speeding up the lung allocation process.
The new system has also several limitations. We designed local allocation units using the number of transplants rather than the number of candidates as an index of demand. The reason was to take account of differences in graft selection among LTCs. Indeed, assessment of the previous system indicated that the rate of lung discard ranged between 34% and 85% among the nine transplant centres (data not shown). Another limitation is the modification of local allocation units by the LTCs resulting in a mean supply/demand ratio of 0.73 instead of 0.5. Indeed, the modification of the allocation system required a general acceptance of all LTCs, some of which are attached to their historical local allocation unit. An additional limitation was the algorithmic method used for the construction of the geographic model. The addition of new OPC to the model increases the calculation time in an exponential manner. There is therefore a nondeterministic polynomial time concern, like the knapsack problem (
A new geographic lung allocation system based on supply/demand ratio was introduced in France in September 2020. The new system was expected to reduce geographic disparity in the number of grafts offered per candidate to non-urgent patients while maintaining proximity. The expected changes were apparent 1 year after the implementation of the new system. Long- term comprehensive monitoring of the allocation policy change is underway.
The datasets presented in this article are not readily available because Data not available due to legal restrictions. Requests to access the datasets should be directed to FB,
FB developed the theoretical formalism and performed the analytic calculations. CC, CL, and FK contributed to the final version of the manuscript. RD and CJ supervised the project.
This study was conducted in accordance with the French legislation stating that research studies based on Cristal which are part of donation and graft allocation system assessment do not require approval by an institutional review board.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We would like to thank the following clinical and surgical centres as well as the medical supervisors of the transplant units participating in the Cristal national registry.
The Supplementary Material for this article can be found online at:
ECMO, extracorporeal membrane oxygenation; HU, high-urgency; LAS, lung allocation score; LTC, lung transplant centre; OPC, organ procurement centre.