Attrition is not background noise. It is signal.
Educational disclaimer: This article is for educational purposes only and is not financial, legal, tax, or professional advising. Residency decisions can affect future compensation, training pathways, contracts, and career options, so applicants should verify program-specific details and consult qualified mentors, advisors, attorneys, or other professionals when appropriate.
The data shows that when residents keep leaving a program, something usually is not working. Sometimes it is workload. Sometimes it is poor mentorship. Sometimes leadership has built a culture that looks polished on interview day and feels corrosive by February. I have seen applicants get hypnotized by a famous hospital name, a glossy website, or a fellowship match list, then ignore the one metric that was quietly telling the truth: people were getting out.
Residency attrition means residents leave before completing training in that program, whether by transfer, dismissal, withdrawal, or another route. And no, it is not always about “one struggling resident.” Programs love that explanation because it individualizes a systems problem. But if the same program keeps losing people across multiple years, the numbers are speaking. You should listen.
My position is simple: do not evaluate attrition in isolation, but do not minimize it either. Compare rates across years, against specialty norms, and alongside board performance, vacancies, leave patterns, and program citations. Prestige does not erase risk. In fact, prestige often hides it better.
For a broader framework on evaluating training environments, pair this with guidance on how to choose the right residency program, questions to ask residency programs, how to build a residency rank list, and how to interpret board pass rate and program outcomes.
Why Attrition Rates Can Hide Risk in Plain Sight
A raw attrition percentage can mislead you if you do not understand the denominator. In a large internal medicine program with 120 residents, losing 3 residents in a year may look numerically small. In a small surgical subspecialty with 12 residents, losing 2 is a major event. The percentage matters, but cohort concentration matters too. If a single PGY class is hit repeatedly, that is not random variation. That is a pattern.
The data shows three common ways risk gets hidden:
- Small program smoothing: one departure every year sounds manageable until you realize that in a class of four, it means 25% instability per cohort.
- Prestige masking: competitive programs in major cities continue to attract applicants despite internal churn. Demand conceals dysfunction.
- Track-level filtering: prelim years, research tracks, or quietly unsupported residents may leave in ways that are framed as “fit issues” rather than program failures.
You also need adjacent metrics. Attrition by itself is incomplete. Attrition plus weak board pass rates is worse. Attrition plus repeated unfilled spots is worse still. Add frequent leave of absence, duty-hour complaints, or ACGME citations, and the risk profile becomes hard to dismiss.
I have reviewed programs where the website showed impressive fellowships and smiling resident headshots, while the data underneath showed a more troubling picture: two residents transferred out in three years, board pass rates trended downward, and current residents gave oddly vague answers about support. That is not an optics problem. That is a systems problem.
A program can be selective and still be unstable. Those are not opposites. That is the mistake applicants make.
What the Data Shows: Red Flags That Cluster in Risky Residencies
Bad outcomes travel in packs.
The data shows that risky residencies rarely wave a single obvious red flag. Instead, they show a cluster: above-norm attrition, lower board passage, repeated vacancies, resident classes that seem thin on interview day, leadership turnover, and strangely scripted explanations from faculty. One bad year happens. Five years of friction means the machine is built badly.
Here is the cluster I trust most as a warning sign:
- Sustained attrition above specialty norm
- Board pass rates that are low or drifting down
- Repeated need to backfill vacancies
- Multiple residents taking leave with little explanation
- Resident or chief turnover that seems unusually frequent
- ACGME citations or public signs of program instability
- Resident feedback that is cautious, coded, or inconsistent
Trendlines matter more than snapshots. A single year can be noisy because of parental leave, a spouse relocation, illness, or one poorly matched recruit. Over 3 to 5 years, noise settles and structure appears. That is where the data gets honest.
Specialty context matters too. Attrition expectations differ between, say, general surgery and dermatology. You should benchmark a program against its own field, not against another department in the same hospital. I have seen applicants compare a demanding surgical program’s attrition to pediatrics down the hall. Wrong comparison. Bad denominator. Useless conclusion.
What should worry you most is not just “high.” It is consistently high relative to peers. If a program keeps landing above its field’s baseline and cannot clearly explain why things changed and improved, you should treat that as a material risk.
How to Investigate a Program Before You Rank It
This is where applicants either do real diligence or play dress-up with spreadsheets. Be the first kind.
Start with multiple sources because no single source is complete. Program websites are marketing documents. Useful, but biased. Resident forums are noisy. Useful, but biased. ACGME and FREIDA data can anchor your review, but they do not always tell the whole story in real time. You need triangulation.
Useful sources include:
- FREIDA
- ACGME public information and accreditation updates
- Program websites and annual reports
- Institutional GME reports
- Board pass disclosures when available
- Alumni and recent graduates
- Current residents, ideally off-script and away from faculty
- Trusted faculty mentors who know regional reputations
- Resident forums and applicant networks for pattern detection
If you are comparing options across specialties or regions, it also helps to review how to compare residency programs objectively, signs of malignant residency culture, and how residents should evaluate mentorship and support.
The data shows that the most reliable approach is a 3-step process.
1. Build a 3- to 5-year view
Look for trends, not anecdotes. Track:
- attrition events
- class vacancies
- board pass performance
- leadership changes
- citation history
- obvious shifts in program structure
A program with one departure in five years is different from a program with one departure every year. Same total? Maybe. Same risk? Not even close.
2. Test the story on interview day
Ask direct but neutral questions. Then compare answers across program leadership, APDs, chiefs, and residents. If the program says attrition reflected “personal reasons,” but residents describe chronic understaffing and no backup coverage, the discrepancy is the finding.
3. Corroborate externally
Talk to alumni. Ask your school’s faculty if they know the program’s reputation. Search for changes in leadership. Read between the lines in resident comments. A single complaint is noise. Five independent descriptions of poor support is data.
Do not overreact to one metric. That is amateur hour. A program can have a slightly elevated attrition rate and still be a good fit if board outcomes are strong, residents feel supported, remediation is handled well, and departures were clearly contextual rather than structural. But the reverse is also true: a “prestigious” program with middling support and repeated churn is not a bargain. It is a risk with better branding.
I have seen students talk themselves into bad programs because they were afraid to look unsophisticated. They did not want to question the famous name. That is backwards. Sophisticated applicants investigate the famous names harder.
Questions That Reveal the Real Story During Interviews
Interview day is not just about whether they like you. It is also an audit.
Ask questions that force process-based answers, not PR slogans. Good programs can answer these cleanly. Weak programs dodge, generalize, or become defensive.
Useful questions include:
- Why have residents left this program in the last 3 to 5 years?
- How often do you have unfilled or backfilled positions?
- What systems support residents who are struggling academically or personally?
- What changed after any recent increase in attrition?
- How is duty-hour compliance monitored and enforced?
- How often are residents pulled to cover staffing gaps?
- What is the board pass trend over the last several graduating classes?
Then read the room. The data is not only numeric. Qualitative signals count.
Transparent programs usually:
- give specific timelines
- name concrete changes made
- distinguish resident-specific issues from systems issues
- let residents speak candidly without visible tension
Risky programs often:
- answer with vague language like “a couple of isolated situations”
- offer timelines that do not match between speakers
- insist everything is excellent while residents look exhausted
- overrehearse the culture narrative
I pay attention to tone. If a chief resident lowers their voice before answering a question about support, that is information. If three residents independently mention schedule instability, that is information. If leadership bristles at a basic question about departures, that is information too. Evasion is a data point.
A Simple Decision Model for Ranking Safely
You do not need a perfect model. You need a disciplined one.
Create a weighted score for each program using the factors that actually affect your odds of thriving and finishing. I recommend a five-part framework:
- Attrition trend: 25%
- Board pass performance: 20%
- Resident culture and support: 25%
- Workload and operational stability: 20%
- Mentorship and career development: 10%
Then score each category on a 1 to 5 scale.
For example:
- 1 = major concern
- 3 = mixed
- 5 = clearly strong
This turns vague impressions into comparable decisions. The data shows applicants make better ranking choices when they force themselves to compare programs across the same criteria instead of relying on “gut feel” after a charming dinner.
Set personal thresholds. Mine would be blunt:
- If attrition is consistently above specialty norms for several years, that program starts in the penalty box.
- If high attrition is paired with weak board outcomes or repeated vacancies, I would be very reluctant to rank it.
- If the program is transparent, support systems are credible, and the rest of the data is strong, I would investigate further rather than automatically discard it.
That last part matters. The goal is not perfection. The goal is avoiding preventable mistakes. You are not selecting a brand. You are selecting the environment where you will train under stress for years. Stability matters. A lot.
Closing: Use Attrition as a Filter, Not a Verdict
Attrition is not a verdict by itself, but it is absolutely a filter.
The data shows that high attrition often points to deeper instability: workload strain, weak support, poor selection, or leadership problems. Applicants get in trouble when they either ignore attrition completely or treat it like the only number that matters. Both are lazy.
The smarter approach is straightforward. Benchmark attrition against specialty norms. Look at 3- to 5-year trends. Check whether board pass rates, vacancies, leave patterns, and resident feedback point in the same direction. Ask direct questions. Watch for evasive answers. Put the famous name aside and score the program like it is any other risk asset. Because that is what it is.
If you investigate attrition intelligently, you reduce your odds of ranking a residency that looks impressive on paper and unstable in real life. That is the whole game. Not glamour. Not hype. Better odds.
