Educational disclaimer: This article is for informational and educational purposes only. It is not financial, legal, tax, or professional advising, and it should not substitute for individualized guidance from your school, GME office, specialty advisors, or other qualified professionals when evaluating residency programs.
Attrition matters more than reputation. That is the blunt truth.
Applicants routinely treat established programs as safe and new programs as risky. The data shows that this shortcut is weak. Program age, by itself, does not tell you enough. What matters is whether residents are actually staying, why they are leaving, and whether the departures reflect a structural problem or simple statistical noise.
Let us define the metric clearly. Attrition means a resident leaves a program before completion. That may be voluntary, involuntary, or a transfer depending on the case, but from a workforce standpoint the effect is immediate: one fewer resident covering call, one smaller peer group, one more disruption to training continuity. In a small cohort, that impact is not subtle. It is huge. A single departure in a six-resident class is 16.7%. In a class of twenty, it is 5%. Same event. Very different headline.
That is why I do not trust anecdotes. I have heard applicants dismiss a new program because “two people left.” Fine. Out of how many? Over how many years? For what reasons? I have also seen famous, long-established programs quietly bleed residents while preserving a polished reputation on interview day. Prestige can hide bad systems for a long time.
You should care because attrition can signal something real: unstable leadership, poor onboarding, weak mentorship, malignant service pressure, bad scheduling, or inadequate resident support. But it can also reflect nothing more than tiny denominators. The job is to separate signal from noise. That takes numbers, not gossip.
How Attrition Is Measured: What the Numbers Actually Capture
Start with the best available sources. The data usually comes from three places:
- Program-reported outcomes on websites or during interviews
- ACGME accreditation and public program information where available
- Institutional GME reports, annual quality documents, or internal recruitment materials
No source is perfect. Program self-reporting can be selective. Institutional summaries may lag. Public accreditation data can be broad rather than applicant-friendly. Still, triangulating these sources gives you a far better read than Reddit threads and hallway rumor.
You also need clean definitions, because applicants constantly mix up unlike events.
- Attrition: resident leaves before completing the program
- Transfer: resident moves to another program; this may still count as attrition for the original program
- Dismissal: resident is removed, usually for performance or professionalism reasons
- Remediation: resident stays but receives formal support or corrective oversight; not attrition
- Fellowship departure: completion of residency followed by fellowship; not attrition at all
This confusion matters. I have seen applicants panic over “three residents did not stay,” only to learn one transferred for family reasons, one switched specialties, and one graduated on time into fellowship. Sloppy interpretation creates fake red flags.
The denominator problem is even more important. In small classes, percentages swing violently.
One departure:
- 6-resident cohort = 16.7%
- 10-resident cohort = 10%
- 20-resident cohort = 5%
The data shows why single-year attrition percentages can mislead you, especially in newer programs that often begin with smaller classes. A one-year spike may mean a real problem. Or one resident got sick, had a spouse relocate, or realized the specialty was wrong. That is why multi-year averages are the only serious way to compare programs.
My rule is simple: if you only have one year of attrition data, your confidence should be low. If you have three to five years, class size, and at least some explanation for departures, now you are analyzing something real.
What the Data Usually Shows: New Programs vs Established Programs
Here is the most defensible conclusion: new programs often show more volatility, but they do not automatically show persistently higher attrition once you account for startup phase and small class size.
That distinction matters. Volatility is not the same thing as dysfunction.
A new program starts with obvious disadvantages. Systems are still being built. Faculty are still learning how to supervise together. Call structures may look reasonable on paper and feel terrible in month three. Clinic workflows are often clunky. There may be genuine uncertainty about case distribution, didactic consistency, and backup coverage. I have seen this repeatedly in first- and second-year programs: nobody is lying, but nobody has enough historical experience yet. That creates friction, and friction can drive early exits.
Typical drivers of early attrition in new programs include:
- Weak onboarding into EMR, hospital logistics, and workflow
- Uneven faculty development and inconsistent teaching expectations
- Call schedules that are technically compliant but operationally chaotic
- Incomplete wellness, mentorship, or remediation infrastructure
- Institutional overpromising during recruitment
Established programs usually have an advantage here. Their culture is more legible. The call schedule has been stress-tested. Senior residents teach juniors. Faculty know where the bottlenecks are. Even if the program is not glamorous, predictability itself is protective.
But established does not mean healthy. That assumption is one of the dumbest habits in residency advising.
I have seen mature programs with excellent names and terrible resident retention. Why? Because reputation can outlive reality. A legacy program may have:
- overloaded inpatient services,
- chronic understaffing,
- high faculty turnover,
- poor psychological safety,
- leadership that normalizes burnout,
- or a habit of quietly pressuring struggling residents to leave.
Those programs may post lower visible attrition in some years simply because residents feel trapped, not supported. Or because departures happen through delayed pathways that do not show up cleanly in a casual applicant conversation.
The better question is not “new or established?” It is “what trend does the program show over time, normalized for size and specialty?”
An illustrative comparison helps. Suppose a new program has attrition of 8%, 6%, and 5% across its first three years, while an established program posts 4%, 4%, and 3%.
The data shows two things at once:
- The new program starts higher.
- The new program improves.
That second point is the one applicants often miss. Improvement is a powerful signal. If a new program learns fast, stabilizes leadership, and tightens support systems, its early attrition may be a startup tax rather than a chronic defect.
By contrast, an established program sitting at 4%, 4%, and 3% may look safer. Maybe it is. But if those numbers occur in a much larger program with multiple anonymous departures, hidden service strain, and flat resident satisfaction, the surface-level comfort can be misleading. Low drama is not the same as high quality.
Specialty also changes interpretation. Attrition norms differ across surgical and non-surgical fields, across highly competitive tracks versus broad-service specialties, and across urban referral centers versus community-based sites. Comparing a tiny new surgical subspecialty pipeline to a large mature internal medicine program without context is statistically lazy.
So what does the data really show? Program age is a weak standalone predictor. Stability, transparency, and trend direction matter more.
Interpreting Red Flags Without Overreacting to Noise
Not every departure is a red flag. Repeated patterns are red flags.
The patterns that deserve your attention are:
- departures across multiple consecutive classes,
- midyear exits rather than end-of-year transfers,
- spikes after a program director or chair change,
- clustering within one rotation site or service,
- and evasive answers when applicants ask about resident retention.
A single resident leaving a tiny program? That can be noise. Two or three similar departures over several years? That is signal.
The data becomes much more meaningful when paired with adjacent indicators:
- Board pass rates: low or falling rates may suggest weak educational structure
- Faculty turnover: high turnover often predicts resident instability
- Resident satisfaction: even informal trends from open houses and away rotators matter
- Duty-hour violations: repeated problems usually reflect system strain, not bad luck
- Accreditation actions or citations: these deserve direct follow-up
This is where applicants often make a basic analytical mistake. They isolate attrition as if it is a complete story. It is not. Attrition is one output metric. You need the full dashboard.
Ask sharper questions during interviews. Not hostile questions. Specific ones.
Try:
- “How many residents have left in the last three to five years, and were those transfers, specialty changes, or dismissals?”
- “What changes did the program make after those departures?”
- “Has leadership changed recently?”
- “How is backup coverage handled if a resident is out unexpectedly?”
- “What is the faculty turnover rate?”
- “Can residents describe how mentorship works when someone is struggling?”
Listen for direct answers. Good programs answer cleanly. Weak programs blur categories, hide behind vague optimism, or act offended. That defensiveness is data too.
Applicant Decision Framework: How to Weigh New vs Established Programs
You should not choose a residency based on attrition alone. You also should not ignore it. The right approach is weighted comparison.
I recommend a simple scoring lens across five domains:
- Attrition and retention trend
- Mentorship and faculty access
- Clinical volume and case mix
- Geographic and personal fit
- Career alignment for fellowship, academics, or practice goals
Give each category a score, then force yourself to justify it with evidence. Not vibe. Not prestige. Evidence.
For attrition specifically, ask for three numbers:
- Cohort size by year
- Number of departures over the last 3–5 years
- Reasons for departure, if the program is willing to disclose them
Then calculate your own normalized view. Example:
- Program A: 2 departures over 5 years, average class size 6
- Program B: 4 departures over 5 years, average class size 20
Raw counts make Program B look worse. They are not. Program A had 30 total resident slots over that period; Program B had 100. That means the crude departure burden is roughly 6.7% for Program A versus 4% for Program B. The data shows why denominator blindness leads applicants into bad conclusions.
But numbers are not the whole story. A new program with transparent leadership, stable hospital backing, honest communication, and clear evidence of improvement can be a better bet than an established program that feels tired, cynical, and unsupported. I have watched applicants choose the famous name and regret it by October. The brochure looked great. The service reality did not.
What predicts safety? Not age alone. Response quality.
Programs that handle early instability well usually do four things:
- acknowledge problems directly,
- change systems quickly,
- invest in resident support,
- and maintain faculty consistency.
Programs that handle instability badly tend to do the opposite:
- deny concerns,
- blame individual residents,
- rotate through leaders,
- and sell “resilience” as a substitute for staffing.
That is not resilience. That is mismanagement with a wellness slogan attached.
So weigh new versus established the way an analyst would. Put the prestige halo aside. Compare trend lines, denominators, and institutional behavior under stress. The strongest predictor is not whether a program is young. It is whether the program learns, adapts, and retains trust when something goes wrong.
Summary: The Evidence-Based Bottom Line
New residency programs are not automatically high-attrition programs. The data does not support that lazy conclusion.
What the numbers do show is more nuanced. New programs often have more year-to-year volatility, largely because their cohorts are small and their systems are still maturing. Established programs usually look steadier, but reputation can hide real operational problems. That is why single-year percentages are weak evidence and anecdotes are worse.
Use the right lens:
- look at 3- to 5-year trends,
- adjust for cohort size,
- separate transfers from dismissals and other departures,
- and compare attrition with board pass rates, faculty turnover, and resident experience.
That is the bottom line. Trust trends, not rumors. Trust normalized data, not prestige mythology. If you do that, you will make a better residency decision than the applicant who falls for a famous name or panics over one bad year.
