Experience Is Quantifiable

The Core Principle

Surgeons and hospitals that perform an operation more often tend to record better outcomes with it. This is not a marketing line. It is one of the most repeatedly tested findings in health services research, first quantified in the New England Journal of Medicine in 1979 and re-examined in joint replacement ever since. Every figure below comes from a named published study and uses the comparison groups that study itself defined. All of it is observational evidence, and the limits are set out further down this page.

Hip replacement

Katz and colleagues analysed Medicare claims covering 58,521 elective primary total hip replacements and 12,956 revisions. Patients whose surgeon performed more than 50 primary hip replacements a year in Medicare beneficiaries had a 90 day dislocation rate of 1.5%, against 4.2% for surgeons performing five or fewer (adjusted odds ratio 0.49, 95% confidence interval 0.34 to 0.69). Hospitals performing more than 100 a year had a mortality rate of 0.7%, against 1.3% at hospitals performing ten or fewer (adjusted odds ratio 0.58, 95% confidence interval 0.38 to 0.89). The authors note their analysis lacks operative detail and preoperative functional status. Journal of Bone and Joint Surgery, 2001.

Knee replacement

The same group analysed Medicare claims for elective primary total knee replacement. Patients of surgeons performing more than 50 knee replacements a year in Medicare recipients had a 28% lower risk of pneumonia in the first 90 days (odds ratio 0.72, 99% confidence interval 0.54 to 0.95) and a 19% lower risk of any of the adverse outcomes studied, which were death, pneumonia, pulmonary embolus, acute myocardial infarction and deep infection (odds ratio 0.81, 99% confidence interval 0.68 to 0.98), compared with patients of surgeons performing twelve or fewer. Journal of Bone and Joint Surgery, 2004.

Hip and knee together

Siddiqi and colleagues linked American Joint Replacement Registry records from 2012 to 2017 with Medicare claims and the National Death Index. Measured against the highest volume surgeon and hospital pairing, the lowest volume pairing carried 1.63 times the adjusted odds of all cause revision after hip replacement and 1.72 times after knee replacement, 2.50 times the odds of early revision for periprosthetic joint infection after hip replacement, 2.47 times the odds of early hip instability or dislocation, and 1.72 times the odds of 90 day mortality after hip replacement and 1.47 times after knee replacement. Turned around, that is roughly 39% lower odds of revision after hip replacement for the highest volume pairing. Journal of the American Academy of Orthopaedic Surgeons, 2022.

Shoulder replacement

Jain and colleagues examined shoulder arthroplasty in the Nationwide Inpatient Sample from 1988 to 2000. Mortality after total shoulder arthroplasty was 0.20% for surgeons performing four or more a year, against 0.36% for surgeons performing fewer than two. Risk adjusted complications after hemiarthroplasty were 0.97% for surgeons performing five or more a year, against 1.68% for surgeons performing fewer than two. The volume categories that separated better from worse shoulder outcomes are an order of magnitude smaller than the hip and knee thresholds above, because far fewer shoulder replacements are performed. Borrowing a knee threshold and applying it to shoulder surgeons would be a mistake. Journal of Bone and Joint Surgery, 2004.

What happens when the knee studies are pooled

Lau and colleagues systematically reviewed the published studies on surgeon volume in primary total knee arthroplasty. Lower surgeon volume was associated with a higher infection rate (0.26% to 2.8% higher, as the review states it), longer operating time (165 minutes against 135), longer hospital stays (0.4 to 2.13 days longer), a higher transfusion rate (13% against 4%) and worse patient reported outcomes. The same review found no association between surgeon volume and mortality, implant survivorship or thromboembolic events, and its authors conclude that the findings “suggest a trend towards better outcomes for higher volume surgeons, but results must be interpreted with caution.” The reviewed studies also disagreed sharply on what counts as low volume, ranging from fewer than 3 to fewer than 52 knee replacements a year. BMC Musculoskeletal Disorders, 2012.

Where the idea came from

Luft, Bunker and Enthoven examined 12 operations across 1,498 hospitals in the paper that opened this literature. Hospitals performing 50 to 100 total hip replacements a year “attained a mortality rate for this procedure almost as low as that of hospitals doing 200 or more,” while some operations, cholecystectomy among them, showed no volume relationship at all. The authors also named the problem that has shadowed the field ever since: the results “may reflect the effect of volume or experience on mortality, or referrals to institutions with better outcomes, as well as a number of other factors, such as patient selection.” New England Journal of Medicine, 1979.

Volume Benchmarks to Watch

There is no single number that divides a high volume surgeon from a low volume one. Each study set its own threshold, and the thresholds differ by joint. These are the comparisons the published research actually made.

The published standards

The Leapfrog Group, whose hospital standards are used by large employers and health plans, sets a minimum annual volume of 50 total hip replacements and 50 total knee replacements per hospital, and 25 of each per surgeon for privileging. Its stated basis is that “Three decades of research have consistently demonstrated that patients that have their high-risk surgery at a hospital and by a surgeon that have more experience with the procedure have better outcomes, including lower mortality rates, lower complication rates, and a shorter length of stay than for patients who have their surgery done at a hospital or by a surgeon with less experience.” Note that Leapfrog’s surgeon minimum is 25 a year, half the threshold the Medicare studies above used. Serious bodies read the same evidence and land on different numbers. Leapfrog Group, 2026 hospital survey standards.

How to Use the Data

1. Read the count as Medicare volume, because that is what it is. The numbers on this site are Medicare fee-for-service claims, and our state rankings sort by them. That is close to how the Katz studies defined surgeon volume, which was procedures performed in Medicare patients, so their thresholds are the right ones to hold these counts against. One caveat: those studies ran in 2000 and 2001, when almost every Medicare patient was in fee-for-service. Slightly more than half are now in Medicare Advantage, whose claims never enter our file, so a modern fee-for-service count is a narrower slice of a surgeon’s Medicare work than theirs was.
2. Use the threshold for the right joint. More than 50 a year is the hip and knee figure. For shoulder replacement the published comparisons sit at four or more a year, and applying a knee threshold to shoulder surgeons would exclude most of the field for no evidential reason.
3. Layer other signals. Board certification and subspecialty training, hospital quality ratings and infection metrics, and whether the surgeon participates in your insurance.
4. Schedule consultations. Meet two or three surgeons. Volume is a filter for building a shortlist, not a substitute for the conversation that decides who operates on you.

Important Context

• Volume is a correlation, not a mechanism. Every study on this page is observational. Higher volume surgeons may do better because repetition builds skill, or because surgeons with good results attract more referrals, or because busy centres differ in staffing, anaesthesia, protocols and which patients they accept. Luft and colleagues flagged that ambiguity in 1979 and no study since has fully resolved it. Nothing here demonstrates that moving an individual patient to a busier surgeon causes a better outcome for that patient.
• The ratios look large; the absolute differences are small. Where these studies report absolute rates, the gaps are narrow. Shoulder replacement mortality was 0.20% against 0.36%. Hip replacement mortality was 0.7% against 1.3%. The widest absolute gap on this page is hip dislocation, 1.5% against 4.2%. Joint replacement is a safe operation at almost any volume, and volume shifts the odds at the margin rather than deciding the result.
• Medicare coverage. Our rankings count Medicare fee-for-service claims only. That stream is an estimated 25% to 33% of national hip replacement volume, 23% to 28% of knee replacement volume and 29% to 37% of shoulder replacement volume. A surgeon’s true all-payer caseload is therefore higher than the count shown here, by an amount that depends on their payer mix. The full derivation, with its sources and its uncertainty, is on our methodology page.
• Complex cases. Revision surgery and higher risk patients are where the volume differences in this research are widest, so travelling to a regional centre can be worth the effort for those cases specifically.
• Individual fit. Volume is a starting point, not the only decision factor. Partner with your primary care physician and weigh your own comfort with each surgeon.

Dig Deeper

• Browse the medical evidence library
• Review the data audit process
• Read the all-payer derivation behind the Medicare shares above
• Use our surgeon search to filter by procedure and location

Volume is the one quality signal that is published, countable and comparable across every surgeon in the country, which is why this site is built on it. It is not the only signal and it is not proof. Use it to build a shortlist, then decide the rest with your primary care physician and the surgeons themselves.

Corrections

This page was reviewed and rewritten on 1 August 2026, and four previously published claims were withdrawn. A claim that high volume surgeons recorded “63% fewer revision surgeries” across “4+ million procedures” inverted the direction of an odds ratio: the registry study reported that the lowest volume pairing carried 1.63 times the odds of revision, which is about 39% lower odds for the highest volume pairing, and that study’s patient count is not stated in its published abstract. A claim of “60% lower infection rates” attributed to the Journal of Bone and Joint Surgery does not appear in the paper it was credited to and has been removed. A claim that Medicare fee-for-service represents “approximately 19.5% of total U.S. joint replacements” had no source and has been replaced with the per procedure shares derived on our methodology page. A banded table of “0 to 25”, “25 to 49” and “50+” procedures a year, including a “47% lower 90-day mortality” figure that also inverted an odds ratio, has been replaced with the thresholds the cited studies actually used.