Materials and Methods
This systematic review adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines25. The review was not registered on PROSPERO, and the study protocol was not published prior to submission. A comprehensive search of electronic databases, including PubMed, Embase, Web of Science, the Cochrane Library, the World Health Organization International Clinical Trials Registry Platform (ICTRP), and the UK Clinical Trials Gateway, was performed to identify relevant studies. The search was limited to articles from database inception (i.e., the earliest available records) to March 31, 2025. The combination of search terms used to systematically retrieve pertinent studies can be found in the Appendix.
Eligibility Criteria
Articles were filtered using Population, Intervention, Comparison, Outcomes, and Study design (PICOS) criteria (Table 1). The inclusion criteria were published or unpublished randomized controlled trials or cohort studies in adults (≥18 years old) undergoing TKA or THA at any disease stage while receiving a GLP-1 RA treatment of any dosage or duration. The exclusion criteria were inaccessible full texts and crossover trials.
| Population | Adult patients undergoing hip or knee arthroplasty |
| Intervention | Weight-loss GLP-1 RA medications |
| Comparison | No weight-loss medications |
| Outcomes | Medical complication |
| Implant failure | |
| All-time implant failure | |
| All-time revision surgery | |
| Study | Randomized controlled trials or cohort studies |
Study Selection
Zotero (version 6.0.37; Corporation for Digital Scholarship [2023]) was used to remove duplicates. Two independent reviewers screened titles and abstracts, then acquired eligible full texts. The citation sections of the selected articles were examined for additional relevant literature. Disagreements between reviewers were resolved by a senior author. The first unmet criterion was recorded as the primary reason for exclusion; a detailed list of the excluded studies is provided in the Results. The selection process is summarized in the PRISMA flowchart (Fig. 1).
PRISMA flowchart of the included studies.
Data Extraction
Two independent reviewers extracted data including first author, title, publication year, study design, sample size, surgery type (THA or TKA), follow-up duration, GLP-1 RA protocol, and the rates of surgical site infection, PJI, revision surgery, and medical complications at 90 days, 1 year, and 2 years. Hospital resource utilization data were collected when available. Revision arthroplasty was defined as any operation exchanging femoral, tibial, or mobile components following TKA, or exchanging femoral, acetabular, or mobile components following THA. Data were entered electronically by 1 reviewer and verified by another.
Risk-of-Bias Assessment
Two review authors independently assessed the risk of bias for each included study with use of the Cochrane risk-of-bias tool, following the guidelines outlined in the Cochrane Handbook for Systematic Reviews of Interventions. Additionally, each study was assigned a quality rating with use of the ROBINS-I (Risk Of Bias In Non-Randomized Studies - of Interventions) risk assessment tool26. Disagreements were resolved by discussion or by deferring to a third review author. When information was missing from the published papers, the study authors were contacted. The detailed risk-of-bias assessments and judgments for each study are provided in the Appendix.
Statistical Analysis
Due to substantial heterogeneity in the study designs, populations, interventions, and reported outcomes, a qualitative synthesis approach was employed. Descriptive statistics were used to summarize study characteristics, and risk ratios (RRs), odds ratios (ORs), or hazard ratios (HRs), with 95% confidence intervals (CIs), were extracted when available. Significance was defined as p < 0.05.
Results
Upon evaluation of the full texts, 6 articles were excluded on the basis of the following criteria: 1 study was eliminated due to combined data of GLP-1 RA with other medications13, 1 study was excluded for being a transcription of an oral abstract27, 3 studies were excluded because they evaluated patients prior to hip or knee surgery rather than postoperatively28–30, and 1 study was excluded because it assessed non-orthopaedic patients31.
Study Characteristics
Eight studies examined GLP-1 RA use in patients undergoing primary THA or TKA, encompassing 22,611 GLP-1 RA users and 77,810 controls32–39 (Tables 2 and 3). Only primary THA and TKA cases were included. The mean patient age ranged from 56 to 64 years. Most studies reported a mean age of 60 to 64 years for both groups32,35–38, while 2 studies reported that the majority of the patients were in the range of 55 to 69 years of age33,34. TKA studies showed greater female representation (60% to 67% of patients) than THA studies (42% to 69% of patients)32,35,36,38. The prevalence of diabetes (37% to 100%) varied considerably on the basis of the inclusion criteria, with some studies targeting only diabetic populations33,34,37. After propensity matching, comorbidity indices were well-matched between the groups, including rates of obesity (or obesity/overweight), insulin use, and metformin use33–35,39. However, despite propensity matching, 1 study38 reported substantially higher rates of insulin use (65% versus 36%) and metformin use (68% versus 43%) in the GLP-1 RA group compared with the control group.
| Study | Year | Design | Sample Size | Type of Surgery | Follow-up | GLP-1 RA Protocol |
|---|---|---|---|---|---|---|
| Katzman et al.[38] | 2025 | Retrospective, PS matching | 865 GLP-1 RA users matched with 8,650 non-users | TKA | Mean, 2.2 years (GLP-1 RA users) vs. 2.9 years (non-users) | 6 months preop. and continued up to 3 months postop. |
| Buddhiraju et al.[35] | 2024 | Retrospective, PS matching | THA: 1,044 GLP-1 RA users matched with 1,044 non-users; TKA: 2,095 GLP-1 RA users matched with 2,095 non-users | THA and TKA | 90 days | Between 1 year and 15 days preop. |
| Magruder et al.[34] | 2023 | Retrospective, PS matching | 7,051 semaglutide users matched with 34,524 non-users | TKA | 90 days and 2 years | Active semaglutide prescription at time of TKA |
| Verhey et al.[39] | 2025 | Retrospective, PS matching | 5,345 GLP-1 RA users matched with 5,345 non-users | THA | 90 days and 2 years | GLP-1 RA therapy at time of THA |
| Heo et al.[37] | 2025 | Retrospective cohorts | 812 GLP-1 RA users, 3,248 non-users | THA | 90 days and 1 year | GLP-1 RA use (at least 3 fills within 6 months preop. or 1 fill of ≥90-day supply within 6 months preop.) |
| Magruder et al.[33] | 2024 | Retrospective, PS matching | 1,653 semaglutide users, 7,812 non-users | THA | 90 days and 2 years | Semaglutide use at time of THA |
| Kim et al.[36] | 2025 | Retrospective, 3 cohorts | Severe obesity (no GLP-1): 5,949; morbid obesity + GLP-1: 2,975; morbid obesity + no GLP-1: 2,975 | TKA | 90 days and 2 years | 3 months preop. and postop. |
| Kim et al.[32] | 2025 | Retrospective, 3 cohorts | Severe obesity (no GLP-1): 3,084; morbid obesity + GLP-1: 771; morbid obesity + no GLP-1: 3,084 | THA | 90 days and 2 years | 3 months preop. and postop. |
| Study | Surgery | No. of Patients | Mean Age(yr) | % Women | BMI(kg/m)2 | Diabetes Prevalence(%) | Key Comorbidities | Notable Outcomes |
|---|---|---|---|---|---|---|---|---|
| Katzman et al.[38] | TKA | GLP-1: 865; control: 8,650 | 64 in both groups | GLP-1: 66.0%; control: 66.3% | 1 yr preop.: GLP-1, 36.0; control, 35.7.Day of surgery: GLP-1, 35.9; control, 36.1 | GLP-1: 74.5%; control: 37.5% | • Long-term insulin use: GLP-1, 65.1%; control, 35.5%• Metformin use: GLP-1, 68.3%; control, 42.7%• Higher CCI in GLP-1 (4.4 vs. 3.2) | BMI difference between groups diminished by the day of surgery and remained similar postop. |
| Buddhiraju et al.[35] | THA | GLP-1: 1,044; control: 1,044 | GLP-1: 63.3; control: 63.5 | GLP-1: 51.1%; control: 52.0% | Not reported | GLP-1: 69.3%; control: 68.3% | • Obesity/overweight: GLP-1, 70.3%; control, 73.2%• Hypertension: GLP-1, 80.3%; control, 81.7%• HbA1c ≥7.5%: GLP-1, 41.7%; control, 39.8% | Similar HbA1c levels between groups |
| Buddhiraju et al.[35] | TKA | GLP-1: 2,095; control: 2,095 | GLP-1: 64.1; control: 64.2 | GLP-1: 60.7%; control: 60.9% | Not reported | GLP-1: 68.7%; control: 68.4% | • Obesity/overweight: GLP-1, 68.8%; control, 70.6%• Hypertension: GLP-1, 78%; control, 80.6% (p = 0.039)• HbA1c ≥7.5%: GLP-1, 37.2%; control, 33.7% (p = 0.01) | Lower infection, aspiration, and DVT rates in GLP-1 group (not significant) |
| Magruder et al.[34] | TKA | Semaglutide: 7,051; control: 34,524 | Majority 55-69 in both groups 3 | Semaglutide: 61.4%; control: 61.6% | Not reported | Complicated DM: semaglutide, 44.5%; control, 44.2% | • Obesity: semaglutide, 86.2%; control, 86.4%• Insulin use: semaglutide, 55.0%; control, 54.7%• Metformin use: semaglutide, 90.1%; control, 90.4% | Closely matched cohorts |
| Verhey et al.[39] | THA | GLP-1: 5,345; control: 5,345 | GLP-1: 57; control: 57 | After matching: GLP-1, 69%; control, 69% | Not reported | Not directly reported | After matching (comparable between groups):• Obesity: GLP-1, 68.2%; control, 68.3%• Depression: GLP-1, 33.4%; control, 33.4%• Tobacco use: GLP-1, 29.9%; control, 29.8% | Significant differences in demographics before matching; well-balanced after matching |
| Heo et al.[37] | THA | GLP-1: 812; control: 3,248 | GLP-1: 61; control: 60 | GLP-1: 41.7%; control: 42.4% | Not reported | 100% (T2DM study) | • Insulin-dependent DM: GLP-1, 20.8%; control, 17.6%• Complicated diabetes: GLP-1, 49.4%; control, 47.5%• CHF: GLP-1, 12.4%; control, 10.9% | No significant differences between groups |
| Magruder et al.[33] | THA | Semaglutide: 1,653; control: 7,812 | Majority 55-69 in both groups 5 | Semaglutide: 47.6%; control: 47.8% | Not reported | Complicated T2DM: semaglutide, 39.7%; control, 39.4% | • Obesity: semaglutide, 84.9%; control, 85.2%• Insulin use: semaglutide, 47.3%; control, 46.2%• Metformin use: semaglutide, 93.2%; control: 93.6% | Well-matched cohorts; hospital LOS not reported |
| Kim et al.[36] | TKA | Severe obesity: 5,949; morbid obesity + GLP-1: 2,975; morbid obesity + no GLP-1 (control): 2,975 | ∼62.2 in each group | ∼66.8% across all groups | Severe obesity: 35-39.9; morbid obesity: ≥40 | ∼52.8% across all groups | • CCI: ∼3.3 across all groups• Smoking: lower rate in GLP-1 group compared with the severe obesity and morbid obesity control groups (46.6% vs. 50.0% and 49.7%, respectively)• Alcohol abuse: lower in GLP-1 group (5.0% vs. 7.7% vs. 8.0%) | GLP-1 group had a shorter hospital LOS (2.7 vs. 2.7 vs. 2.9 days; p = 0.047) |
| Kim et al.[32] | THA | Severe obesity: 3,084; morbid obesity + GLP-1: 771; morbid obesity + no GLP-1 (control): 3,084 | 62.1 across all groups | GLP-1: 52.8%; control: 52.9%; severe obesity: 51.9% | Severe obesity: 35-39.9; morbid obesity: ≥40 | ∼52.3% across all groups | CCI: ∼3.3-3.4 across all groups | Significantly shorter hospital LOS in GLP-1 group (2.2 vs. 2.7 vs. 3.1 days; p = 0.001) |
Study Designs
Five studies employed propensity score matching33–35,38,39. Uniquely, Kim et al.32,36 utilized a 3-cohort design, comparing (1) morbidly obese patients (body mass index [BMI], ≥40 kg/m2) who were using GLP-1 RA, (2) morbidly obese non-users, and (3) patients with severe obesity (BMI, 35 to 39.9 kg/m2) who were not using GLP-1 RA. Heo et al.37 specifically targeted diabetic patients undergoing THA, comparing outcomes without propensity matching.
Surgical Complications () Table 4
Five of 8 studies reported significant reductions in PJI associated with GLP-1 RA use. Buddhiraju et al.35 found reduced 90-day PJI among GLP-1 RA users (RR, 0.58; 95% CI, 0.34 to 0.99; p = 0.042), and both studies by Magruder et al. demonstrated reduced 2-year PJI (30%33 and 44%34 lower odds; p < 0.001 and p = 0.005, respectively). In their 3-cohort analyses, Kim et al. found lower 90-day PJI rates among GLP-1 RA users compared with both severely obese patients and morbidly obese non-users (TKA36: 1.0% versus 1.4% versus 1.8%, respectively [p = 0.028]; THA32: 1.6% versus 2.2% versus 3.2%, respectively [p = 0.01; OR, 0.47]). However, these differences disappeared at the 2-year follow-up in both cohorts32,36. The remaining 3 studies37–39 found no significant differences in PJI rates between GLP-1 RA users and controls. Revision surgery rates were significantly lower among GLP-1 RA users compared with controls in 3 studies: both of the Magruder et al. cohorts (TKA34: OR, 0.86 [p = 0.02]; THA33: OR, 0.64 [p = 0.0257]) and Katzman et al.38 (p = 0.034).
Of the 4 studies reporting surgical site infection as an outcome33,35,37,39, none demonstrated significant differences between GLP-1 RA users and non-users. At the 2-year follow-up in the 3-cohort study on THA, the difference in component revision rates approached significance (p = 0.05), favoring GLP-1 RA users (<2.3%) over severely obese patients (3.9%) and morbidly obese non-users (3.1%)32.
| Outcome | Study | GLP-1 RA Group | Control Group | Effect Size (95% CI) | P Value | Significant 5 |
|---|---|---|---|---|---|---|
| SSI | ||||||
| 90-day SSI | Buddhiraju(THA)[35] | 1.00% | 1.00% | RR, 1.01 (0.42-2.41) | 0.988 | No |
| 90-day SSI | Verhey[39] | 0.60% | 0.40% | OR, 1.549 (0.905-2.652) | 0.141 | No |
| 90-day SSI | Heo[37] | 3.90% | 5.20% | OR, 1.39 (0.94-2.06) | 0.1 | No |
| 90-day SSI | Magruder(THA)[33] | 1.00% | 1.40% | OR, 0.74 (0.43-1.21) | 0.255 | No |
| 2-year SSI | Verhey[39] | 0.6% | 0.8% | OR, 1.471 (0.923-2.343) | 0.129 | No |
| PJI | ||||||
| 90-day PJI | Buddhiraju(THA)[35] | 2.10% | 3.60% | RR, 0.58 (0.34-0.99) | 0.042 | Yes (↓) |
| 90-day PJI | Verhey[39] | 1.20% | 1.10% | OR, 1.086 (0.761-1.550) | 0.717 | No |
| 90-day PJI | Heo[37] | 1.30% | 1.20% | OR, 0.78 (0.39-1.55) | 0.47 | No |
| 1-year PJI | Heo[37] | 1.50% | 1.40% | OR, 0.88 (0.46-1.68) | 0.69 | No |
| 2-year PJI | Katzman[38] | 1.20% | 1.30% | Not specified | 0.669 | No |
| 2-year PJI | Magruder(TKA)[34] | 2.10% | 3.00% | OR, 0.70 (0.58-0.83) | <0.001 | Yes (↓) |
| 2-year PJI | Magruder(THA)[33] | 1.60% | 2.90% | OR, 0.56 (0.37-0.82) | 0.005 | Yes (↓) |
| 2-year PJI | Verhey[39] | 2.7% | 2% | OR, 1.168 (0.881-1.550) | 0.314 | No |
| Any-time PJI | Katzman[38] | 1.20% | 1.50% | Not specified | 0.392 | No |
| Revision surgery | ||||||
| 90-day revision | Buddhiraju(THA)[35] | 1.70% | 2.80% | RR, 0.62 (0.34-1.13) | 0.117 | No |
| 90-day revision | Buddhiraju(TKA)[35] | 0.60% | 0.80% | RR, 0.72 (0.34-1.50) | 0.375 | No |
| 90-day revision | Verhey[39] | 0.70% | 1.00% | OR, 0.783 (0.516-1.186) | 0.292 | No |
| 1-year revision | Heo[37] | 2.50% | 2.70% | OR, 1.21 (0.71-2.00) | 0.48 | No |
| 2-year revision | Verhey[39] | 1.7% | 1.7% | OR, 0.989 (0.738-1.325) | 1 | No |
| 2-year revision | Magruder(TKA)[34] | 4.00% | 4.50% | OR, 0.86 (0.75-0.98) | 0.02 | Yes (↓) |
| 2-year revision | Magruder(THA)[33] | 1.80% | 2.80% | OR, 0.64 (0.43-0.93) | 0.0257 | Yes (↓) |
| 2-year all-cause revision | Katzman[38] | 2.30% | 2.60% | Not specified | 0.362 | No |
| Any-time revision | Katzman[38] | 2.70% | 3.90% | Not specified | 0.034 | Yes (↓) |
| 90-day complications | ||||||
| ED utilization | Buddhiraju(THA)[35] | 5.90% | 6.60% | RR, 0.90 (0.55-1.47) | 0.668 | No |
| ED utilization | Buddhiraju(TKA)[35] | 7.20% | 7.70% | RR, 0.93 (0.68-1.28) | 0.66 | No |
| ED visits | Verhey[39] | 4.80% | 5.80% | OR, 0.814 (0.686-0.965) | 0.02 | Yes (↓) |
| ED visits | Katzman[38] | 5.90% | 4.00% | Not specified | 0.008 | Yes (↑) |
| Readmission | Buddhiraju(TKA)[35] | 1.10% | 2.00% | RR, 0.53 (0.31-0.90) | 0.017 | Yes (↓) |
| Readmission | Buddhiraju(THA)[35] | 1.60% | 2.00% | RR, 0.81 (0.41-1.59) | 0.532 | No |
| Readmission | Verhey[39] | 4.10% | 4.50% | OR, 0.909 (0.754-1.096) | 0.341 | No |
| Readmission | Heo[37] | 8.50% | 8.70% | OR, 1.01 (0.76-1.34) | 0.95 | No |
| Readmission | Katzman[38] | 4.30% | 3.60% | Not specified | 0.168 | No |
| Readmission | Magruder(TKA)[34] | 7.00% | 9.40% | OR, 0.71 (0.64-0.79) | <0.001 | Yes (↓) |
| Readmission | Magruder(THA)[33] | 6.20% | 8.80% | OR, 0.68 (0.54-0.84) | 0.0004 | Yes (↓) |
| Mortality | Verhey[39] | 0.03% | 0.10% | OR, 0.400 (0.178-2.061) | 0.45 | No |
| DVT | Buddhiraju(THA)[35] | 1.00% | 1.10% | RR, 0.91 (0.39-2.13) | 0.831 | No |
| DVT | Verhey[39] | 0.50% | 0.60% | OR, 0.866 (0.512-1.467) | 0.688 | No |
| DVT | Heo[37] | 1.00% | 1.30% | OR, 1.21 (0.52-2.81) | 0.65 | No |
| DVT | Magruder(TKA)[34] | 0.80% | 0.50% | OR, 1.50 (1.25-2.00) | 0.007 | Yes (↑) |
| DVT | Magruder(THA)[33] | 0% | 0.70% | OR, 0.69 (0.30-1.38) | 0.3348 | No |
| PE | Buddhiraju(THA)[35] | 1.00% | 1.00% | RR, 1.00 (0.42-2.38) | 0.991 | No |
| PE | Verhey[39] | 0.10% | 0.20% | OR, 0.889 (0.343-2.305) | 1 | No |
| PE | Magruder(TKA)[34] | 0.50% | 0.60% | OR, 0.82 (0.56-1.15) | 0.277 | No |
| PE | Magruder(THA)[33] | 0% | 0.40% | OR, 0.69 (0.24-1.62) | 0.4453 | No |
| VTE | Magruder(TKA)[34] | 1.10% | 1.00% | OR, 1.10 (0.86-1.40) | 0.41 | No |
| VTE | Magruder(THA)[33] | 0.70% | 0.90% | OR, 0.74 (0.37-1.35) | 0.3601 | No |
| CVA (stroke) | Magruder(TKA)[34] | 1.20% | 0.90% | OR, 1.37 (1.07-1.74) | 0.01 | Yes (↑) |
| CVA | Magruder(THA)[33] | 0% | 0.90% | OR, 0.67 (0.32-1.25) | 0.2426 | No |
| Acute renal failure | Buddhiraju(THA)[35] | 2.00% | 1.40% | RR, 1.40 (0.65-3.00) | 0.385 | No |
| Acute renal failure | Buddhiraju(TKA)[35] | 2.20% | 2.10% | RR, 1.05 (0.66-1.66) | 0.853 | No |
| MI | Magruder(TKA)[34] | 1.00% | 0.70% | OR, 1.49 (1.13-1.94) | 0.003 | Yes (↑) |
| MI | Magruder(THA)[33] | 0% | 0.70% | OR, 0.72 (0.33-1.39) | 0.3579 | No |
| PNA | Magruder(TKA)[34] | 2.80% | 1.70% | OR, 1.67 (1.41-1.97) | <0.001 | Yes (↑) |
| PNA | Magruder(THA)[33] | 1.90% | 1.40% | OR, 1.37 (0.91-2.02) | 0.1185 | No |
| AKI | Verhey[39] | 0.50% | 0.60% | OR, 0.866 (0.512-1.466) | 0.688 | No |
| AKI | Heo[37] | 4.30% | 4.40% | OR, 0.99 (0.67-1.47) | 0.96 | No |
| AKI | Magruder(TKA)[34] | 4.90% | 3.90% | OR, 1.28 (1.13-1.44) | <0.001 | Yes (↑) |
| AKI | Magruder(THA)[33] | 2.80% | 3.90% | OR, 0.69 (0.50-0.94) | 0.0242 | Yes (↓) |
| Sepsis | Verhey[39] | 0.30% | 0.30% | OR, 1.000 (0.488-2.048) | 1 | No |
| Sepsis | Magruder(TKA)[34] | 0.00% | 0.40% | OR, 0.23 (0.09-0.48) | <0.001 | Yes (↓) |
| Sepsis | Magruder(THA)[33] | 0% | 0.40% | OR, 0.57 (0.17-1.42) | 0.2821 | No |
| Hypoglycemic event | Heo[37] | 1.10% | 0.90% | OR, 0.91 (0.42-1.97) | 0.82 | No |
| Hypoglycemic event | Magruder(THA)[33] | 0% | 1% | OR, 0.45 (0.20-0.71) | 0.0348 | Yes (↓) |
| Hospital resource utilization | ||||||
| Extended LOS (≥3 days) | Heo[37] | 24.40% | 28.50% | OR, 1.25 (1.05-1.49) | 0.01 | Yes (↓) |
| Average LOS | Magruder(TKA)[34] | 2.7 days | 3.1 days | OR, 1.02 (0.95-1.10) | 0.52 | No |
| Average LOS | Magruder(THA)[33] | 2.7 days | 2.9 days | OR, 0.99 (0.81-1.21) | 0.9334 | No |
| Average LOS | Kim(TKA)[36] | 2.7 days | 2.9 days | Not specified | 0.047 | Yes (↓) |
| Average LOS | Kim(THA)[32] | 2.2 days | 3.1 days | Not specified | 0.001 | Yes (↓) |
| Average same-day cost | Magruder(TKA)[34] | $10,671.31 | $11,484.50 | Not specified | 0.708 | No |
| Average same-day cost | Magruder(THA)[33] | $9,174.72 | $10,046.30 | Not specified | 0.5169 | No |
| Average 90-day cost | Magruder(TKA)[34] | $15,291.66 | $16,798.46 | Not specified | 0.012 | Yes (↓) |
| Average 90-day cost | Magruder(THA)[33] | $13,219.92 | $14,681.71 | Not specified | 0.0562 | Borderline (↓) |
Medical Complications () Table 4
The effects of GLP-1 RA on medical complications showed variable patterns. Magruder et al.34 reported higher rates of stroke (OR, 1.37; p = 0.01), deep vein thrombosis (OR, 1.50; p = 0.007), myocardial infarction (OR, 1.49; p = 0.003), pneumonia (OR, 1.67; p < 0.001), and acute kidney injury (OR, 1.28; p < 0.001) following TKA in GLP-1 RA users compared with controls. However, Magruder et al.33 showed contradictory results for acute kidney injury, reporting lower rates in the GLP-1 RA group (OR, 0.69; p = 0.0242) following THA. Magruder et al.34 demonstrated that GLP-1 RA users had significantly reduced odds of sepsis following TKA (OR, 0.23; p < 0.001), while Magruder et al.33 reported significantly reduced odds of hypoglycemic events following THA (OR, 0.45; p = 0.0348). In the 3-cohort analyses, Kim et al. found that GLP-1 RA users had significantly lower rates of any medical complication compared with severely obese patients and morbidly obese non-users in both TKA36 (10.6% versus 10.9% versus 12.7%; p = 0.014) and THA32 (10.5% versus 13.5% versus 14.1%; p = 0.03). GLP-1 RA users undergoing THA also had fewer hematomas (0% versus 1% versus 1.3%; p < 0.01)32.
Hospital Resource Utilization () Table 4
Hospital resource utilization outcomes generally favored GLP-1 RA users. Heo et al.37 reported significantly lower rates of an extended length of stay (≥3 days) among GLP-1 RA users compared with non-users (24.4% versus 28.5%), with non-users having higher odds of an extended stay (OR, 1.25; 95% CI, 1.05 to 1.49; p = 0.01). Similarly, Kim et al. demonstrated shorter hospital stays in both their TKA cohort36 (2.7 versus 2.7 versus 2.9 days for GLP-1 RA users, severely obese patients, and morbidly obese non-users, respectively; p = 0.047) and THA cohort32 (2.2 versus 2.7 versus 3.1 days, respectively; p = 0.001). In the 3-cohort analyses, GLP-1 RA users undergoing TKA had significantly lower 90-day readmission rates (5.3%) than severely obese patients (7.4%) and morbidly obese non-users (8.9%) (p < 0.001)36. Similarly, GLP-1 RA users undergoing THA showed lower 90-day readmission rates (6.9%) compared with severely obese patients (8.9%) and morbidly obese non-users (9.7%) (p = 0.04)32. These differences disappeared at the 2-year follow-up36. Magruder et al. found lower 90-day costs among GLP-1 RA users compared with controls in both THA33 ($13,219.92 versus $14,681.71; p = 0.0562) and TKA34 ($15,291.66 versus $16,798.46; p = 0.012).
GLP-1 RA use was associated with reduced hospital readmission rates in some studies, with Buddhiraju et al.35 and Magruder et al.33,34 reporting significant reductions ranging from 29% to 47%. Verhey et al.39 also found a significantly lower rate of outpatient visits in the GLP-1 RA group. However, Katzman et al.38 observed a higher rate of outpatient visits among GLP-1 RA users.
Methodological Quality of the Studies
The studies demonstrated uniformly moderate overall quality of reporting, as shown in the Appendix.
Discussion
Given the high prevalence of obesity among patients undergoing arthroplasty, GLP-1 RA use as a part of preoperative optimization protocols may reduce perioperative risks. However, the specific impact of GLP-1 RAs on hip and knee arthroplasty outcomes remains preliminarily studied, with definitive conclusions yet to be established. The present systematic review assessed the impact of GLP-1 RAs on total hip and knee arthroplasty outcomes by analyzing 8 retrospective studies comprising 22,611 GLP-1 RA users and 77,810 controls.
Hospital readmission rates showed the most consistently favorable results among GLP-1 RA users, with 3 studies33–35 reporting significant reductions associated with GLP-1 RA use, particularly during the 90-day postoperative period. Hospital resource utilization similarly favored GLP-1 RA therapy, with several studies documenting shorter hospital stays32,36,37 and lower 90-day costs33,34. PJI outcomes were promising but less consistent: while 5 studies demonstrated significant reductions associated with GLP-1 RA use32–36, 3 studies reported no significant differences37–39.
Revision surgery rates showed potential improvements favoring GLP-1 RA users in 3 studies33,34,38 but demonstrated no notable differences in 5 studies32,35–37,39. Medical complications yielded the most variable results, with some studies identifying increased vascular and pulmonary events among GLP-1 RA users and other studies observing reduced rates of sepsis and hypoglycemic events associated with GLP-1 RA use33,34.
Notably, even propensity score-matched studies yielded inconsistent results, suggesting genuine heterogeneity rather than study design limitations alone33–35,38,39. The 3-cohort design studies by Kim et al.32,36 demonstrated that GLP-1 RA users had superior 90-day outcomes compared with both morbidly obese non-users and severely obese patients. However, these early advantages diminished at the 2-year follow-up, suggesting that the benefits of GLP-1 RA may be limited to the immediate perioperative period rather than translating to long-term implant survival. This temporal pattern supports the proposed mechanisms of GLP-1 RA action, including improved glycemic control40,41, reduced systemic inflammation42–44, and improved wound healing45,46, indicating that the most effective role of GLP-1 RAs may lie in perioperative optimization rather than in altering long-term arthroplasty outcomes.
The heterogeneity in outcomes observed in the present systematic review likely stems from multiple sources. Patient characteristics vary considerably, with some individuals having elevated BMI but stable metabolic profiles, while others exhibit substantial metabolic dysfunction. These differences in inflammatory status and metabolic regulation may contribute to variations in perioperative risk and complication rates. Additionally, there exists wide variation in treatment protocols across clinical settings, including differences in dosing regimens, the timing of administration, and the duration of therapy. Individual GLP-1 RA agents also have distinct pharmacokinetic and pharmacodynamic properties, resulting in variable effects on weight loss, glycemic control, and inflammation. These sources of heterogeneity must be carefully considered when interpreting the current evidence and designing future research.
Our findings contrast with those of prior publications that have suggested more definitive benefits of GLP-1 RAs in arthroplasty outcomes. A recent meta-analysis reported significant reductions in PJI rates, concluding that GLP-1 RAs demonstrated perioperative benefits23. However, these apparently positive results were statistically fragile, losing significance when influential studies were removed in sensitivity analyses, and were based on a questionable quantitative pooling of methodologically heterogeneous studies. Similarly, a previous narrative review emphasized promising preclinical mechanisms and selective clinical findings while minimizing the substantial inconsistencies that were observed across human studies24.
The present systematic review has several important limitations. All included studies were retrospective, introducing inherent selection bias despite propensity score-matching efforts. The lack of randomized controlled trials limited causal inferences regarding the impact of GLP-1 RA on arthroplasty outcomes. Considerable heterogeneity was observed in treatment protocols, including in timing, duration, and the specific agents used, and there was incomplete reporting of dosage details. Inconsistent documentation of baseline BMI and glycemic control (e.g., glycated hemoglobin) further complicates the attribution of improved outcomes to weight loss versus metabolic effects. Most studies focused on short-term (90-day) outcomes, with limited long-term follow-up. These issues highlight the need for well-designed prospective trials with standardized protocols and extended follow-up.
Future research should determine whether the observed clinical benefits result from weight-mediated effects or direct pharmacological actions that are independent of BMI changes—a distinction that has crucial implications for clinical practice and patient selection criteria. Additional priorities include elucidating the mechanisms of periarticular soft-tissue effects, examining the durability of benefits beyond 2 years, and conducting appropriately powered randomized controlled trials that stratify outcomes by weight-loss response and have longer follow-up periods. Such studies could establish definitive clinical recommendations and potentially expand the therapeutic applications of GLP-1 RAs outside of metabolic disorders.
Conclusions
Although GLP-1 RA therapy was associated with reduced hospital readmissions and decreased hospital costs within 90 days postoperatively in several studies, its benefits for PJI prevention showed mixed results in both TKA and THA, with some studies demonstrating a meaningful reduction in PJI and others showing no difference. No other clinical advantages were observed at the 2-year follow-up.
Appendix
Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJS/J156).
Footnotes
Contributor Information
Joaquin Moya-Angeler, Email: jmoyaangeler@gmail.com.
Mustafa Akkaya, Email: makkaya@outlook.com.
Roberto Civinini, Email: roberto.civinini@unifi.it.
Matteo Innocenti, Email: matteo.innocenti@unifi.it.