Surgery or nonsurgical care first: what the trials and registries show
Before signing surgical consent, patients ask three questions: What happens if I do a 12-week physical therapy program instead of immediate surgery? Why are some knee replacement patients dissatisfied? And how long do knee and hip implants last? Here are the published numbers, with their limits.
TODAY: Sourced outcome ranges from the NEJM randomized trials of total knee replacement (Skou 2015 & 2018) and total hip replacement (Frydendal 2024), the 2023 systematic review of TKA dissatisfaction (DeFrance & Scuderi), and The Lancet 25-year national registry meta-analyses (Evans 2019), checked 2026-10-26. Your surgeon decides with you.
Both groups improved; surgery improved more and carried more serious adverse events, and most of the nonsurgical group put off surgery.
In the MEDIC randomized trial (Skou et al., N Engl J Med 2015, PMID 26488691), 100 adults with moderate-to-severe knee osteoarthritis eligible for unilateral total knee replacement were randomized to total knee replacement plus 12 weeks of nonsurgical care vs. 12 weeks of supervised nonsurgical care alone:
Pain and function (0 to 100 KOOS4, higher is better): Both groups improved at 12 months: +32.5 points with total knee replacement vs. +16.0 points with 12-week nonsurgical care alone (supervised exercise, education, weight advice, insoles, pain medicine) — an adjusted difference of 15.8 points (95% CI, 10.0 to 21.5). Limit: 100 patients in one Danish region; compared with nonsurgical care alone, not with no treatment. PubMed 26488691
Serious adverse events in year 1: More occurred in the surgical group than the nonsurgical group (24 vs. 6, P = 0.005). Limit: 100 patients in one Danish region; compared with nonsurgical care alone, not with no treatment. PubMed 26488691
Year 1 surgical delay: In the nonsurgical group, 13 of 50 (26%) had TKR before 12 months; 74% (37 of 50) did not have surgery in year 1. Limit: 100 patients in one Danish region; compared with nonsurgical care alone, not with no treatment. PubMed 26488691
At 2 years, TKR plus nonsurgical care stayed ahead by 18.3 KOOS4 points (95% CI, 11.3 to 25.3), while the 12-week nonsurgical program beat written advice by 7.0 points (95% CI, 0.4 to 13.5). Among surgical candidates assigned to nonsurgical care first, 16 of 50 (32%) had TKR by 2 years — meaning two out of three (68%) delayed surgery for at least 2 years. Limit: the same 100 surgical candidates in one Danish region, followed 2 years. PubMed 29723634
For hips, one randomized trial: 109 adults aged 50 or older with severe hip osteoarthritis and an indication for surgery were assigned to total hip replacement or resistance training. At 6 months, the Oxford Hip Score (0 to 48, higher is better) rose 15.9 points with surgery vs. 4.5 points with resistance training (difference 11.4, 95% CI 8.9 to 14.0). In the training group, 12 of 56 (21%) had hip replacement within 6 months. Serious adverse events at 6 months were similar in the two groups; most were known complications of hip replacement. Limit: 109 patients in Denmark, 6-month results. PubMed 39476341
How many people are dissatisfied after a knee replacement, and what was associated with it.
Older papers quoted a 20% dissatisfaction rate after primary total knee arthroplasty. Across 21 studies from 2010 to 2022, average patient dissatisfaction was 10% overall, and 7.3% when excluding complications. Limit: an average across 21 studies of primary knee replacement, 2010 to 2022. PubMed 36252743
- Before surgery: milder arthritis on X-ray (lower Kellgren-Lawrence grades), depression or anxiety, and pain catastrophizing were associated with dissatisfaction. Limit: an average across 21 studies of primary knee replacement, 2010 to 2022. PubMed 36252743
- After surgery: dissatisfaction was most often associated with complications, unmet expectations, persistent pain, and stiffness. Limit: an average across 21 studies of primary knee replacement, 2010 to 2022. PubMed 36252743
- Who: age under 65, lower income, and non-White patients were the sociodemographic factors most often associated with dissatisfaction. Limit: an average across 21 studies of primary knee replacement, 2010 to 2022. PubMed 36252743
How long implants last in national joint registries, out to 25 years.
National joint registries record revision operations across whole countries; Evans et al. (The Lancet 2019) pooled them. The hip figure comes from the Australian and Finnish registries:
| Procedure | Registry sample | 25-year survival (no revision) | Primary source |
|---|---|---|---|
| Total knee (TKR) | 299,291 knees (14 registries) | 82.3% (95% CI 81.3–83.2%) | Evans Lancet 2019 (PMID 30782341) Limit: registry data; survival without revision, not freedom from pain. |
| Partial knee (UKR) | 7,714 knees (4 registries) | 69.8% (95% CI 67.6–72.1%) | Evans Lancet 2019 (PMID 30782341) Limit: registry data; survival without revision, not freedom from pain. |
| Total hip (THR) | 215,676 hips (Australia & Finland registries) | 57.9% (95% CI 57.1–58.7%) | Evans Lancet 2019 (PMID 30782340) Limit: two registries (Australia, Finland); survival without revision, not freedom from pain. |
Note: 25-year figures describe implants put in 25 or more years before the study, and the knee and hip numbers are survival without revision, which is not the same as living without pain.
Build your trade-off summary and four surgeon questions.
Pick your joint, what you have tried, and what matters most. Runs in this browser tab; nothing is sent.
Write down where you are, why, and what you still need to ask.
Tick what is true for you. You get one page to print and bring to the surgeon. It holds your own choice and reasons, the questions still open, and the trial result already on this page. It does not predict how your own operation will go. Nothing is sent.
“Keep” saves what you ticked and your optional question in this browser only (jcl.choice.v1); “Delete” removes it. The questions come from the AAOS list and the builder above.
Check what you know.
Five questions keyed to this page and its PubMed sources. Scored on this device; nothing is sent.
- Skou ST, et al. A Randomized, Controlled Trial of Total Knee Replacement. N Engl J Med 2015;373(17):1597–1606. PubMed 26488691
- Skou ST, et al. Total knee replacement and non-surgical treatment of knee osteoarthritis: 2-year outcome. Osteoarthritis Cartilage 2018;26(9):1170–1180. PubMed 29723634
- Frydendal T, et al. Total Hip Replacement or Resistance Training for Severe Hip Osteoarthritis. N Engl J Med 2024;391(17):1610–1620. PubMed 39476341
- DeFrance MJ, Scuderi GR. Are 20% of Patients Actually Dissatisfied Following Total Knee Arthroplasty? J Arthroplasty 2023;38(3):594–599. PubMed 36252743
- Evans JT, et al. How long does a knee replacement last? Lancet 2019;393(10172):655–663. PubMed 30782341
- Evans JT, et al. How long does a hip replacement last? Lancet 2019;393(10172):647–654. PubMed 30782340
General education, not medical advice. Your surgeon decides with you.