Highlights of the 2019 MedPhys Match

The fifth edition of the MedPhys Match (MPM) was completed in March of 2019, and the purpose of this article is to review some highlights from this year's statistics in comparison to prior years. We will also review some different ranking statistics that will hopefully provide some insight into the importance of rank list length.

Summary of 2019 MedPhys Match Results

Every year, as soon as the results are sent out to programs and applicants, National Matching Services Inc provides year by year statistics on the MPM website.¹ There were 208 applicants that submitted rank lists this year, which was slightly higher than the 204 applicants that submitted rank lists in 2018. The number of acceptable applicants, defined as those applicants ranked by at least one program, decreased slightly from 176 to 174.² The number of matched applicants increased from 116 to 131, and the number of unmatched applicants decreased from 88 to 77. From a program perspective, the number of positions offered in the MPM increased from 129 in 2018 to 138 in 2019, and the number of unfilled positions decreased from 13 to 7. Overall, the status of the MPM appears to be quite healthy, with overall increasing program participation and more positions being offered though the MPM each year.

When combined with demographic data from MP-RAP,³ the residency application system, we can also get some insight into matching statistics for applicant subgroups, and this has been published in prior articles.⁴ Figure 1 shows statistics for the same subgroups, updated for this year's MPM data. For those with a CAMPEP graduate background (degree or certificate program), about 81–96% were ranked by at least one program. For those that are ranked at least once, CAMPEP PhD graduates have a match rate of 96%, followed by certificate graduates at 77% and CAMPEP MS graduates at 72%. Compared to 2018, the match rate for CAMPEP-accredited MS and PhD degree holders was a little higher, with certificate holders being about the same. Similar to 2018, applicants without a CAMPEP graduate or certificate background are at a disadvantage, with only 63% of applicants ranked by at least one program and only 73% of those matched to a residency position. The success rate, which is defined as the ratio of matched applicants relative to those that did not withdraw from the MPM, was also lowest for this subgroup.

If we look at applicants by reported gender, female applicants continue to have an advantage over male applicants with 86% of female applicants being ranked by at least one program and 86% of those being matched to a residency position. For male applicants, 81% of applicants are ranked by at least one program, and 77% of those applicants are matched to a residency position. Drilling down a little further (data not shown), 100% (11) of the female CAMPEP PhD applicants were ranked by at least one program and matched to a residency position.

Ranking Insights

One of the most frequently asked questions, for both applicants and programs, is about rank lists. More specifically, how long should a rank list be to provide a good result? The answer to this question is multi-faceted and depends on the definition of a good result and the quality of the applicant or program. For an applicant, a good result might be finding a residency position, or it might be finding the best position. The former may only require a short rank (or interview) list, while the latter will generally require a longer rank list. For all applicants and programs to get the best result, one could argue that all programs should review all applicants, but that is obviously impractical.

Table 1. Summary applicant match statistics for the first 5 years of the MedPhys Match (2015-2019). The first column is the number of ranks an applicant received. Subsequent columns show the number that were matched, the number that were not matched, the number that withdrew and the percentage that matched. The last column does not include the number that withdrew.

Table 1 shows summary applicant match statistics for the first five years of the MPM as a function of the number of ranks that an applicant received. Programs are generally only going to rank applicants that they interview, so the number of ranks received is a surrogate for the number of interviews. Getting ranked by as few as 6 programs gives a greater than 90% chance of being placed in a residency position. However, even with 14 ranks, there is a chance of being unmatched. It should be obvious that going on more interviews will increase the chances of being matched to a position, but the stark reality is that 10 or more interviews might not be enough. For the 2019 MPM, all applicants with 6 or more ranks were matched to a position, but this was the lowest number in the first five years. It is worth noting that the number of acceptable applicants was reasonably constant over the last 3 years, but the number of available positions has increased significantly. If this trend continues, one might expect fewer applicants with large numbers of ranks to be unmatched.

Table 2. Summary program match statistics for the first 5 years of the MedPhys Match (2015-2019). The first column is the number of ranks a program had on their list. Subsequent columns show the number that were matched, the number that were not matched, the number that withdrew and the percentage that matched. The last column does not include the number that withdrew.

Table 2 shows similar summary program match statistics for the first five years of the MPM as a function of the number of ranks in each program's rank list and data for programs with a single opening is shown separately. There are a few circumstances where a program will decide to enter a zero-length rank list and those are excluded from the table. For programs with a single residency position, rank lists with 10 applicants or greater was not enough to fill the position on 4 occasions (2015, 2016, and for 2 programs in 2018). For programs with two or more positions in a given year, you might expect that more ranks are required and the data shows that at least 5 ranked applicants are required to fill all positions. With a rank list of 10 or greater applicants, there were 4 occasions where not all positions were filled (2017, 2 programs in 2018, and 2019).

Table 3. 2019 MedPhys Match status versus number of first ranks for applicants that were ranked by at least one program. The first column is the number of times an applicant was ranked first (or ranked to match) by a program that they also ranked. Any number greater than zero guarantees a match unless the applicant withdraws from the MedPhys Match. The number in parentheses is the number of positions those first ranks represent.

Table 3 drills down a little further into the ranking statistics for applicants. For all applicants that were ranked by at least one program (174), only 66 applicants (38%) were ranked first (or ranked to match) by at least one program that they also ranked. For a program offering more than one position, a first rank is counted for the top n applicants, where n is the number of offered positions. Because of the way the matching algorithm works, those 66 applicants were guaranteed to match at least as highly as their top ranked to match position. Of the remaining 99 applicants (not counting those that withdrew), 65 (66%) were matched to positions, and 34 were not matched. There was a total of 118 positions represented by the first ranks for programs, but there were 137 positions offered. This implies that 19 applicants (14%) did not rank a program that ranked them to match!

We can go in even further (not shown) and anyone with at least 3 first ranks was ranked by 5–17 programs. However, there were also 8 applicants that were ranked by at least 10 programs that received no first ranks; therefore, getting a lot of interviews does not guarantee getting ranked to match. Of the 99 applicants that were not ranked to match, 26 of them were ranked second by a program and 24 of those were matched to a position. There was a total of 39 matched applicants that had neither a first or second rank by any program.

Table 4. 2019 MedPhys Match status versus number of first ranks for programs that were ranked by at least one acceptable applicant. The first column is the number of times a program was ranked first (or ranked to match) by an applicant that they also ranked. Any number greater than zero guarantees a match unless the program withdraws from the MedPhys Match. The number in parentheses is the number of applicants those first ranks represent.

Programs can also be ranked to match, which means that at least one of their ranked applicants ranked them first in their list, and Table 4 shows a summary of first ranks for programs. In total, 69 of 100 programs that offered positions in the MPM had at least one applicant in their rank list that ranked them first. Of the 31 programs that were not ranked to match, 26 (84%) were still able to fill all of their positions. The total number of applicants represented by the first ranks in Table 4 is 137, but there were 174 applicants that were ranked by at least one program. That means there were 37 ranked applicants (21%) whose first-ranked program did not rank them!

Assuming that all programs and applicants follow the MPM rules, applicants and programs are not aware of whether they are ranked first by anyone. However, I found it quite surprising that a significant number of applicants (14%) and programs (21%) who were ranked to match did not rank the party that ranked them to match. This points out that while an applicant (or program) might be impressed enough by a program (or applicant) to rank them first, the feeling may not be mutual. Applicants and programs should generally not make assumptions regarding how highly they are regarded and proceed accordingly.

The opinions expressed in this article belong to the author and should not be attributed to his employer or to AAPM. The author welcomes any and all constructive criticism regarding any aspect of the MPM program.

John A. Antolak, PhD
Chair, Subcommittee on the Oversight of MedPhys Match (SCOMM)⁵