Optimizing thre effectiveness of Organ Allocation 121 HLA mismatches. Success is affected by the sensitivity of the person, measured by the persons PRA, which we model by. adding a linear term to the success rate. Moreover, we reduce the success rate by 5 percentage points if the orga is not procured from the same center as the patient; 5% is the average effect on the success rate of increasing the delay by 10-20 hrs, according to optn data. In Phase IV, we regress the coefficients a and b of the previous section, and use this formula to calculate the probability of death. To adjust the parameters for this model, we use the OPtN national data for the national active wait list for cadaver kidneys from 1995 to 2004 and feedinto our model the number of donations for each year. Results of the basic model To quantify the quality of a network, we use: a set of objective functions, which represent various ideas about the desirability of policy outcomes. For these functions, let a be the number of"healthy"patients to receive a successful transplant each y be the corresponding number of"sick"patients(those with some terminal illness or serious medical condition) to receive a successful transplant, be the average age of the transplant recipients, and m be the maximum wait time in the queue. We examine the following objective functions a+y: This is simply the number of successful transplants per year. (100-a) x(+y): This considers the premise that transplants are more valu- able when given to young recipient a+0.5 y: This is a stylized adoption of the idea that transplants given to ter- minally ill recipients are less valuable (=+y)/max(9, m): This incorporates queue wait time. We also include a proposed tradeoff between big and small centers In a big center, the doctors are more experienced. We simulate this by de- creasing the success rate of operations at centers that do not perform a thresh- old number of operations per year. with small centers, kidneys are allocated on a more local basis, which mini- mizes deterioration of organs in transportation. We simulate this by apply ing a penalty when kidneys are moved to larger regional centers, and also when kidneys are moved between centersOpthnizing the Effectiveness of Organ Allocation HLA mismatches. Success is affected by the sensitivity of theperson, measured by the person's PRA, which we model by adding a linear term to the success rate. Moreover, we reduce the success-rate'by. percentage points if the organ is not procured from the same center as'the patient; 5%,is the average effect on the success rate of increasing the delay by 10-20 hrs,,according to OPTN data. In Phase IV, we-regress,the coefficients a and b of the previous section, and use this formula to calculate'the probability of death. To adjust the parameters for this model, we use the OPTN national data for the national active wait listlfor cadaver kidneys from 1995 to 20041and feed into our model the number of donations for each year. Results of the Basic Model To quantify the quality of a network, We use aý set of objective functions, which represent various ideas about the desirability of policy outcomes. For these functions, let x be the number of "healthy" patients to receive 'a successful transplant each year, y be the corresponding number of'"sick" patients (those with some terminal illness or serious medical condition) to receive a successful transplant, a be the average age, of the transplant.recipients, and m be themaximum wait time in the queue.. We examine the-following objective functions: x + y : This is simply the number of successful transplants per year. (100 - a) x (x +-y) : This considers the premise thattransplants are more valu￾able when given to young recipients. x + 0.5 : This is a stylized adoption of the idea that transplants given to ter￾minally ill recipients are less valuable. (x + y)/ max(9, m) : This incorporates queue walt time. We also include a proposed tradeoff between big and small centers: . In a big center, the doctors are more experienced. We simulate this by de￾creasing the success rate of operations at centers that do not perform a thresh￾old number of operations per year. * With small centers, kidneys are allocated on a more local basis,, which mini￾nmizes deterioration of organs in transportation. We simulate this by apply￾ing a penalty when kidneys are moved to larger regional centers, and also when kidneys are moved between centers. 121