Abstract
Objective
To examine whether long-term opioid use after total knee arthroplasty (TKA) is associated with subsequent risks of major adverse cardiovascular events (MACE) and all-cause mortality in patients with knee osteoarthritis (OA).
Methods
This retrospective, population-based cohort study used Taiwan’s Longitudinal Health Insurance Database (LHID) 2005 to identify patients with knee OA undergoing TKA (2002–2018). Opioid exposure was assessed during the 1-year post discharge window (long-term > 3 months vs. non-long-term < 3 months), and the index date was set at 360 days after discharge (landmark design). The primary outcome was first MACE after the index date. After 1:2 propensity score matching (PSM), adjusted hazard ratios (aHRs) were estimated using Cox models; Fine–Gray models were used for nonfatal components.
Results
In the matched cohort (2,351 long-term vs. 4,702 non-long-term users), long-term opioid use was associated with higher risks of MACE (aHR = 1.16; 95% CI, 1.06–1.26) and all-cause mortality (aHR = 1.16; 95% CI, 1.05–1.29), as well as AMI (aHR = 1.30; 95% CI, 1.03–1.64) and CHF (aHR = 1.15; 95% CI, 1.00–1.32), but not ischemic or hemorrhagic stroke. Effect modification by Charlson Comorbidity Index (CCI) was observed for MACE (P-interaction = 0.0278) and mortality (P-interaction = 0.0006), with the strongest associations in CCI ≥ 5. Time-stratified analyses for MACE showed the excess risk was greatest early after the index date and remained significant through 1–5 years, with attenuation thereafter.
Conclusions
Long-term opioid use after TKA was associated with increased risks of MACE and mortality, particularly among patients with high comorbidity burden, underscoring the importance of opioid-sparing strategies and closer cardiovascular risk assessment in patients requiring prolonged opioid therapy.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12871-026-03775-2.
Keywords: Total knee arthroplasty, Opioids, Mortality, Major adverse cardiovascular events, Osteoarthritis
Introduction
Osteoarthritis (OA) is a common disease that featured loss of articular cartilage, subchondral sclerosis, and bone spurs [1, 2].The knee OA is a common form of all types of OA which most likely to influence the patellofemoral and tibiofemoral joints [3]. The lifetime probability of developing knee OA is estimated to be 45% and prominent pain and disability could arise in those with severe knee OA [4]. Currently, the total knee arthroplasty (TKA) surgery is a well-established surgical method for the management of OA which associates with fair functional outcome and increased quality of life [4, 5].
Despite the success of TKA surgery in the treatment of most individuals with knee OA, certain complications could develop after the complement of TKA surgery. The disalignment of the knee joint is a not-uncommon complication of TKA surgery which multiple methods were proposed to reduce the incidence [6, 7]. Persistent pain is a common complication after total knee arthroplasty (TKA) and can significantly impair patients’ ability to perform daily activities and reduce their overall quality of life [8, 9]. Due to the substantial impact of postoperative pain following TKA, various analgesic medications are frequently prescribed for pain management. In some cases, the duration of analgesic use, including opioid medications, may extend beyond one year after surgery [10].
Opioid is an analgesic with dominant pain-relieving effect that is applied to control acute pain and persistent postoperative pain [11, 12]. Due to the strong analgesic effect, the opioid is commonly applied in patients received TKA surgery with a fair result [13]. However, long-term usage of opioid may result in dependence, vomiting, dizziness, and even mortality [14, 15]. Because opioids can influence the general health status of people [14, 16], the long-term application of opioids can contribute to a worse prognosis after TKA surgery in terms of mortality rate and major morbidity, such as major adverse cardiovascular events (MACE). Still, this concept needs to be elucidated.
Although previous studies have evaluated opioid use in surgical populations, evidence regarding the long-term cardiovascular and mortality risks associated with sustained opioid use after TKA remains limited. In particular, the relationship between prolonged postoperative opioid use and MACEs and all-cause mortality in this population has not been well established.
Therefore, using data from the Longitudinal Health Insurance Research Database (LHIRD), we conducted a population-based cohort study to investigate whether sustained opioid use for more than three months after TKA is associated with an increased risk of MACE and all-cause mortality. In addition, we evaluated the risks of individual cardiovascular outcomes to identify which events primarily contributed to the observed associations.
Methods
Data resource and ethics
This study used the 2005 Longitudinal Health Insurance Database (LHID 2005), a representative subset of Taiwan’s National Health Insurance Research Database (NHIRD). The LHID 2005 contains comprehensive, de-identified claims data for 2 million beneficiaries randomly sampled from the 2005 registry and provides longitudinal records from 1998 to 2021. The database includes demographic characteristics (age, sex, urbanization level, and type of insurance), diagnoses coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and Tenth Revision, Clinical Modification (ICD-10-CM), procedure codes, and prescription dispensing records classified according to the Anatomical Therapeutic Chemical (ATC) system. All-cause mortality was ascertained by linkage to the National Death Registry through the Health and Welfare Data Science Center.
The study protocol adhered to the Declaration of Helsinki and was approved by the Institutional Review Board of Chung Shan Medical University Hospital (Approval No.: CS2-23031; Date of Approval: March 13, 2023). The requirement for informed consent was waived because the analyses were based on de-identified secondary data. This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for observational studies [17].
Study design and population selection
This retrospective cohort study was conducted using the LHID2005 database. We initially identified 27,485 patients diagnosed with knee osteoarthritis (OA) between 2002 and 2018, among whom 16,517 underwent total knee arthroplasty (TKA). After applying predefined eligibility criteria, 14,163 patients were included in the cohort. Details of the patient selection process and exclusion criteria are presented in the flowchart (Fig. 1).
Fig. 1.
Flowchart of cohort identification and propensity score matching. Using LHID2005 claims data (1998–2021), we identified patients diagnosed with knee osteoarthritis between 2002 and 2018 who subsequently underwent TKA. After applying prespecified exclusion criteria, eligible patients were classified as long-term or non–long-term opioid users. A 2:1 PSM was performed based on demographics, TKA surgical characteristics, baseline comorbidities, physical therapy/rehabilitation, and baseline medication use to create matched cohorts for outcome analyses. Abbreviations: LHID2005, Longitudinal Health Insurance Database 2005; OA, osteoarthritis; TKA, total knee arthroplasty; PSM, propensity score matching
We applied a landmark design, defining the index date as 360 days after discharge from the index TKA hospitalization. This design provides a fixed 1-year postdischarge window to characterize sustained opioid exposure and reduces immortal time and reverse-causation concerns by restricting analyses to 1-year survivors. During the 1-year exposure ascertainment window, long-term opioid use was defined as receiving opioid prescriptions for > 3 months, whereas non-long-term opioid use was defined as < 3 months of exposure [18]. Among eligible individuals, 11,085 were classified as non-long-term opioid users and 3,078 as long-term opioid users.
To reduce potential confounding due to baseline differences between groups, propensity score matching (PSM) was performed. The propensity score was estimated using a logistic regression model, with long-term opioid use as the dependent variable and baseline covariates including demographic characteristics, comorbidities, and relevant clinical variables as independent variables. After matching, the final analytical cohort consisted of 2,351 patients in the long-term opioid group and 4,702 patients in the non-long-term opioid group.
Study outcomes and follow-up
The primary outcome was the first occurrence of MACE after the index date. MACE was defined as a composite endpoint comprising all-cause mortality, AMI, ischemic stroke, hemorrhagic stroke, and congestive heart failure (CHF). All-cause mortality was ascertained through linkage to the National Death Registry. The cardiovascular components were identified using diagnosis codes as follows: AMI (ICD-9-CM: 410, 411; ICD-10-CM: I21, I22, I24), ischemic stroke (ICD-9-CM: 433–435; ICD-10-CM: I63–I66), hemorrhagic stroke (ICD-9-CM: 430–432; ICD-10-CM: I60–I62), and CHF (ICD-9-CM: 428; ICD-10-CM: I50). Each component was also evaluated as a secondary outcome.
Follow-up began on the index date and continued until the earliest of the following: first occurrence of the composite outcome (for the primary analysis) or the specific event of interest, disenrollment from the National Health Insurance (NHI) program, or end of the study period (December 31, 2021).
Potential confounders
We considered a prespecified set of covariates selected on the basis of clinical relevance and prior literature [19], including demographic characteristics, TKA-related factors, comorbidities, and baseline medication use.
Demographic variables included age at the index TKA, sex, urbanization level, and insurance category. TKA-related factors included the interval from knee OA diagnosis to TKA, laterality, hospital accreditation level, LOS, and anesthesia type.
Baseline health status was summarized using the Charlson Comorbidity Index (CCI) [20]. In addition, specific comorbidities identified within 1 year before the index date included hypertension, diabetes mellitus, ischemic heart disease, hyperlipidemia, chronic pulmonary disease, chronic kidney disease (CKD), liver disease (including alcoholic liver disease), malignancy, osteoporosis, rheumatoid arthritis, and peripheral arterial disease (PAD). We also included neuropsychiatric and sleep-related conditions (anxiety, major depressive disorder, dementia, sleep disorders, and obstructive sleep apnea), as well as prior myocardial infarction (MI), stroke, and urinary tract infection (UTI).
Baseline medication use was defined as prescriptions during the index hospitalization and within 3 months after discharge. We adjusted for analgesic and adjuvant therapies (non-steroidal anti-inflammatory drugs [NSAIDs], acetaminophen, glucosamine, gabapentin/pregabalin, and antidepressants including serotonin–norepinephrine reuptake inhibitors [SNRIs] and selective serotonin reuptake inhibitors [SSRIs]), sedative-hypnotics (benzodiazepines [BZDs] and Z-drugs), and cardiometabolic medications (statins, proton-pump inhibitors [PPIs], angiotensin-converting enzyme inhibitors [ACEIs], angiotensin receptor blockers [ARBs], beta-blockers, antiplatelet agents [aspirin, clopidogrel, or ticagrelor], sodium–glucose cotransporter 2 inhibitors [SGLT-2 inhibitors], and anticoagulants [vitamin K antagonists, direct thrombin inhibitors, and direct factor Xa inhibitors]).
Statistical analysis
Descriptive statistics were used to summarize baseline characteristics. Between-group covariate balance was assessed using the absolute standardized difference (ASD), with ASD > 0.10 indicating potentially meaningful imbalance. To reduce confounding, 1:2 propensity score matching (PSM) was performed using a greedy nearest-neighbor algorithm based on demographics, TKA-related factors, comorbidities, physical therapy utilization, and baseline medications.
Time-to-event analyses were conducted using Cox proportional hazards regression to estimate adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) for the association between long-term opioid use and the risk of MACE. Covariates were prespecified a priori based on clinical relevance. Model fit was compared using the Akaike information criterion (AIC) and Schwarz Bayesian criterion (SBC). To account for within-hospital correlation, we incorporated a hospital-level random effect (frailty term). The proportional hazards assumption was evaluated using Schoenfeld residuals; when violations were detected, time-varying effects were reported.
Predefined sensitivity and subgroup analyses were conducted to evaluate robustness. First, for nonfatal cardiovascular components (AMI, ischemic stroke, hemorrhagic stroke, and CHF), we used the Fine–Gray subdistribution hazards model to estimate subdistribution hazard ratios (sHRs) treating all-cause mortality as a competing event. Second, stratified analyses were performed across prespecified subgroups, including demographics (sex, age at TKA, urbanization level, and index year), TKA-related characteristics (time from OA diagnosis to TKA, hospital accreditation level, LOS, anesthesia type, and CCI), and baseline interventions/clinical history (physical therapy, co-medications, and prior MACE history), with interaction terms evaluated where appropriate. Follow-up time was calculated from the index date to the earliest of outcome occurrence, disenrollment from the NHI program, or end of follow-up (December 31, 2021). For the composite MACE analysis, all-cause mortality was treated as an event; for analyses of individual nonfatal cardiovascular outcomes, death was treated as a competing event in the Fine–Gray models. All analyses were performed using SAS, version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results
The distribution of characteristics among study groups
Baseline characteristics before and after PSM are summarized in Table 1. In the pre-matched cohort, the mean age at knee OA diagnosis was 64.5 ± 8.8 years among non–long-term opioid users and 63.6 ± 9.0 years among long-term opioid users (ASD = 0.10). Long-term opioid users more frequently had a longer interval (≥ 10 years) from OA diagnosis to TKA (32.07% vs. 23.92%; ASD = 0.18) and had higher prevalences of several comorbidities, including hypertension, diabetes mellitus, kidney disease, malignancy, osteoporosis, and sleep disorders (all ASD > 0.10). They also more commonly received physical therapy within 1 year before surgery (30.31% vs. 24.77%; ASD = 0.12) and had higher use of several co-medications (e.g., PPIs, acetaminophen, gabapentin/pregabalin, antidepressants, BZDs, and Z-drugs; all ASD > 0.10).
Table 1.
Baseline characteristics of the study population before and after propensity score matching, by long-term opioid use status
| Before PSM | After PSM | |||||
|---|---|---|---|---|---|---|
| Non–long-term opioid users (n = 11,085) |
Long-term opioid users (n = 3,078) |
ASD | Non–long-term opioid users (n = 4,702) |
Long-term opioid users (n = 2,351) |
ASD | |
| Demographics | ||||||
| Age at knee OA diagnosis (years) | 64.5 ± 8.8 | 63.6 ± 9.0 | 0.1011 | 63.9 ± 8.9 | 63.9 ± 8.9 | 0.0000 |
| Sex | ||||||
| Male | 2,964 (26.74%) | 828 (26.90%) | 0.0037 | 1,255 (26.69%) | 614 (26.12%) | 0.0130 |
| Female | 8,121 (73.26%) | 2,250 (73.10%) | 0.0037 | 3,447 (73.31%) | 1,737 (73.88%) | 0.0130 |
| Urbanization levels | ||||||
| 1 (Highest) | 2,550 (23.00%) | 600 (19.49%) | 0.0859 | 959 (20.40%) | 485 (20.63%) | 0.0058 |
| 2 | 3,104 (28.00%) | 906 (29.43%) | 0.0317 | 1,371 (29.16%) | 702 (29.86%) | 0.0154 |
| 3 | 1,638 (14.78%) | 486 (15.79%) | 0.0281 | 721 (15.33%) | 346 (14.72%) | 0.0173 |
| 4 | 2,099 (18.94%) | 528 (17.15%) | 0.0463 | 850 (18.08%) | 411 (17.48%) | 0.0156 |
| 5 | 396 (3.57%) | 110 (3.57%) | 0.0001 | 173 (3.68%) | 90 (3.83%) | 0.0078 |
| 6 | 787 (7.10%) | 274 (8.90%) | 0.0665 | 373 (7.93%) | 190 (8.08%) | 0.0058 |
| 7 (Lowest) | 511 (4.61%) | 174 (5.65%) | 0.0473 | 255 (5.42%) | 127 (5.40%) | 0.0154 |
| Types of insurance | ||||||
| Government employee | 657 (5.93%) | 179 (5.82%) | 0.0047 | 267 (5.68%) | 134 (5.70%) | 0.0009 |
| Labor | 5,207 (46.97%) | 1,412 (45.87%) | 0.0220 | 2,178 (46.32%) | 1,099 (46.75%) | 0.0085 |
| Farmers/fishermen | 3,770 (34.01%) | 1,025 (33.30%) | 0.0150 | 1,595 (33.92%) | 784 (33.35%) | 0.0122 |
| Low income | 28 (0.25%) | 8 (0.26%) | 0.0014 | 11 (0.23%) | 5 (0.21%) | 0.0045 |
| Non-labor force/ unemployed | 1,329 (11.99%) | 426 (13.84%) | 0.0552 | 600 (12.76%) | 305 (12.97%) | 0.0064 |
| Other | 94 (0.85%) | 28 (0.91%) | 0.0066 | 51 (1.08%) | 24 (1.02%) | 0.0063 |
| TKA surgery | ||||||
| Age at first TKA surgery (years) | 70.4 ± 8.0 | 70.8 ± 8.4 | 0.0487 | 70.7 ± 8.1 | 70.7 ± 8.1 | 0.0000 |
| 40–50 | 72 (0.65%) | 35 (1.14%) | 0.0518 | 38 (0.81%) | 18 (0.77%) | 0.0048 |
| 50–60 | 962 (8.68%) | 253 (8.22%) | 0.0165 | 386 (8.21%) | 193 (8.21%) | 0.0000 |
| 60–70 | 3,791 (34.20%) | 991 (32.20%) | 0.0425 | 1,559 (33.16%) | 779 (33.13%) | 0.0005 |
| 70–80 | 4,858 (43.82%) | 1,344 (43.66%) | 0.0032 | 2,062 (43.85%) | 1,037 (44.11%) | 0.0051 |
| ≥ 80 | 1,402 (12.65%) | 455 (14.78%) | 0.0621 | 657 (13.97%) | 324 (13.78%) | 0.0055 |
| Index year | ||||||
| 2003–2008 | 2,001 (18.1%) | 197 (6.4%) | 0.3614 | 360 (7.66%) | 190 (8.08%) | 0.0158 |
| 2009–2014 | 3,649 (32.9%) | 1,002 (32.6%) | 0.0078 | 1,645 (34.99%) | 803 (34.16%) | 0.0174 |
| 2015–2020 | 5,435 (49.0%) | 1,879 (61.1%) | 0.2433 | 2,697 (57.36%) | 1,358 (57.76%) | 0.0082 |
| Interval from knee OA diagnosis to TKA (years) | ||||||
| 0–4 | 5,439 (49.07%) | 1,161 (37.72%) | 0.2305 | 1,898 (40.37%) | 934 (39.73%) | 0.0130 |
| 5–9 | 2,995 (27.02%) | 930 (30.21%) | 0.0708 | 1,444 (30.71%) | 715 (30.41%) | 0.0065 |
| ≥ 10 | 2,651 (23.92%) | 987 (32.07%) | 0.1823 | 1,360 (28.92%) | 702 (29.86%) | 0.0205 |
| Bilateral knee surgery | 499 (4.50%) | 88 (2.86%) | 0.0873 | 145 (3.08%) | 69 (2.93%) | 0.0087 |
| Hospital level | ||||||
| Medical center | 4,185 (37.75%) | 1,072 (34.83%) | 0.0609 | 1,642 (34.92%) | 815 (34.67%) | 0.0054 |
| Regional hospital | 4,148 (37.42%) | 1,302 (42.30%) | 0.0998 | 1,932 (41.09%) | 974 (41.43%) | 0.0069 |
| District hospital | 2,752 (24.83%) | 704 (22.87%) | 0.0459 | 1,128 (23.99%) | 562 (23.90%) | 0.0020 |
| Length of hospital stay (days) | 7.2 ± 2.7 | 7.3 ± 2.6 | 0.0384 | 7.2 ± 2.6 | 7.2 ± 2.5 | 0.0000 |
| 1–6 | 4,696 (42.36%) | 1,303 (42.33%) | 0.0006 | 2,008 (42.71%) | 1,007 (42.83%) | 0.0026 |
| 7–13 | 5,984 (53.98%) | 1,659 (53.90%) | 0.0017 | 2,536 (53.93%) | 1,266 (53.85%) | 0.0017 |
| 14–21 | 405 (3.65%) | 116 (3.77%) | 0.0061 | 158 (3.36%) | 78 (3.32%) | 0.0024 |
| Types of anesthesia | ||||||
| General anesthesia | 4,177 (37.68%) | 1,286 (41.78%) | 0.0838 | 1,884 (40.07%) | 925 (39.34%) | 0.0148 |
| CCI score (within 1 year before index) | ||||||
| 0 | 3,549 (32.02%) | 580 (18.84%) | 0.3060 | 1,152 (24.50%) | 550 (23.39%) | 0.0259 |
| 1 | 3,278 (29.57%) | 754 (24.50%) | 0.1145 | 1,337 (28.43%) | 665 (28.29%) | 0.0033 |
| 2 | 2,129 (19.21%) | 689 (22.38%) | 0.0784 | 1,105 (23.50%) | 543 (23.10%) | 0.0096 |
| 3 | 1,108 (10.00%) | 448 (14.55%) | 0.1393 | 589 (12.53%) | 311 (13.23%) | 0.0210 |
| 4 | 584 (5.27%) | 287 (9.32%) | 0.1564 | 303 (6.44%) | 164 (6.98%) | 0.0213 |
| ≥ 5 | 437 (3.94%) | 320 (10.40%) | 0.2522 | 216 (4.59%) | 118 (5.02%) | 0.0199 |
| Co-morbidity (within 1 year before index) | ||||||
| Asthma | 920 (8.30%) | 303 (9.84%) | 0.0538 | 427 (9.08%) | 219 (9.32%) | 0.0081 |
| Anxiety-related disorders | 1,656 (14.94%) | 602 (19.56%) | 0.1225 | 832 (17.69%) | 427 (18.16%) | 0.0122 |
| Major depressive disorder | 265 (2.39%) | 161 (5.23%) | 0.1488 | 137 (2.91%) | 72 (3.06%) | 0.0087 |
| Hypertension | 7,788 (70.26%) | 2,388 (77.58%) | 0.1674 | 3,547 (75.44%) | 1,778 (75.63%) | 0.0045 |
| Diabetes mellitus | 3,356 (30.28%) | 1,154 (37.49%) | 0.1529 | 1,596 (33.94%) | 816 (34.71%) | 0.0161 |
| Ischemic heart diseases | 2,013 (18.16%) | 678 (22.03%) | 0.0966 | 897 (19.08%) | 461 (19.61%) | 0.0135 |
| Hyperlipidemia | 3,643 (32.86%) | 1,108 (36.00%) | 0.0660 | 1,715 (36.47%) | 852 (36.24%) | 0.0049 |
| Dementia | 344 (3.10%) | 151 (4.91%) | 0.0920 | 167 (3.55%) | 86 (3.66%) | 0.0057 |
| Chronic pulmonary diseases | 1,603 (14.46%) | 554 (18.00%) | 0.0961 | 756 (16.08%) | 392 (16.67%) | 0.0161 |
| Kidney disease | 1,118 (10.09%) | 553 (17.97%) | 0.2284 | 595 (12.65%) | 320 (13.61%) | 0.0283 |
| Liver disease | 1,213 (10.94%) | 438 (14.23%) | 0.0992 | 547 (11.63%) | 295 (12.55%) | 0.0281 |
| Malignancy | 572 (5.16%) | 327 (10.62%) | 0.2037 | 307 (6.53%) | 154 (6.55%) | 0.0009 |
| Alcoholic liver disease | 27 (0.24%) | 16 (0.52%) | 0.0448 | 15 (0.32%) | 8 (0.34%) | 0.0037 |
| UTI | 1,951 (17.60%) | 726 (23.59%) | 0.1484 | 948 (20.16%) | 485 (20.63%) | 0.0116 |
| Osteoporosis | 1,440 (12.99%) | 610 (19.82%) | 0.1852 | 709 (15.08%) | 376 (15.99%) | 0.0252 |
| RA | 278 (2.51%) | 151 (4.91%) | 0.1272 | 149 (3.17%) | 77 (3.28%) | 0.0060 |
| Sleep disorder | 2,854 (25.75%) | 1,013 (32.91%) | 0.1579 | 1,395 (29.67%) | 709 (30.16%) | 0.0107 |
| MI | 154 (1.39%) | 78 (2.53%) | 0.0826 | 79 (1.68%) | 35 (1.49%) | 0.0153 |
| Stroke | 1,010 (9.11%) | 355 (11.53%) | 0.0797 | 476 (10.12%) | 240 (10.21%) | 0.0028 |
| PAD | 380 (3.43%) | 120 (3.90%) | 0.0251 | 182 (3.87%) | 85 (3.62%) | 0.0134 |
| OSA | 62 (0.56%) | 23 (0.75%) | 0.0233 | 30 (0.64%) | 12 (0.51%) | 0.0169 |
| Physical therapy and rehabilitation | ||||||
| Within 1 year before surgery | 2,746 (24.77%) | 933 (30.31%) | 0.1242 | 1,280 (27.22%) | 664 (28.24%) | 0.0228 |
| During TKA hospitalization | 7,037 (63.48%) | 2,098 (68.16%) | 0.0988 | 3,110 (66.14%) | 1,570 (66.78%) | 0.0135 |
| Within 1 year after surgery | 3,267 (29.47%) | 971 (31.55%) | 0.0451 | 1,426 (30.33%) | 735 (31.26%) | 0.0203 |
| Medication usage (within 3 months after TKA) | ||||||
| NSAIDs | 10,464 (94.40%) | 2,889 (93.86%) | 0.0229 | 4,448 (94.60%) | 2,213 (94.13%) | 0.0203 |
| PPIs | 533 (4.81%) | 301 (9.78%) | 0.1920 | 298 (6.34%) | 143 (6.08%) | 0.0106 |
| Acetaminophen | 5,431 (48.99%) | 1,694 (55.04%) | 0.1212 | 2,457 (52.25%) | 1,252 (53.25%) | 0.0200 |
| Glucosamine | 670 (6.04%) | 192 (6.24%) | 0.0081 | 297 (6.32%) | 146 (6.21%) | 0.0044 |
| Gabapentin or Pregabalin | 89 (0.80%) | 97 (3.15%) | 0.1693 | 35 (0.74%) | 23 (0.98%) | 0.0253 |
| SNRIs | 290 (2.62%) | 136 (4.42%) | 0.0980 | 133 (2.83%) | 77 (3.28%) | 0.0260 |
| SSRIs | 209 (1.89%) | 98 (3.18%) | 0.0827 | 121 (2.57%) | 58 (2.47%) | 0.0068 |
| Other antidepressants | 320 (2.89%) | 169 (5.49%) | 0.1303 | 158 (3.36%) | 79 (3.36%) | 0.0000 |
| BZDs | 3,677 (33.17%) | 1,349 (43.83%) | 0.2203 | 1,792 (38.11%) | 925 (39.34%) | 0.0253 |
| Z-drugs | 1,282 (11.57%) | 474 (15.40%) | 0.1124 | 599 (12.74%) | 321 (13.65%) | 0.0270 |
| Statins | 1,916 (17.28%) | 664 (21.57%) | 0.1085 | 953 (20.27%) | 481 (20.46%) | 0.0048 |
| ACEI | 823 (7.42%) | 237 (7.70%) | 0.0104 | 359 (7.64%) | 171 (7.27%) | 0.0138 |
| ARBs | 3,279 (29.58%) | 1,153 (37.46%) | 0.1675 | 1,655 (35.20%) | 826 (35.13%) | 0.0013 |
| Beta-blockers | 2,567 (23.16%) | 815 (26.48%) | 0.0769 | 1,171 (24.90%) | 583 (24.80%) | 0.0025 |
| Aspirin | 2,524 (22.77%) | 738 (23.98%) | 0.0285 | 1,109 (23.59%) | 562 (23.90%) | 0.0075 |
| Clopidogrel/Ticagrelor | 238 (2.15%) | 105 (3.41%) | 0.0770 | 138 (2.93%) | 65 (2.76%) | 0.0102 |
| Vitamin K antagonists | 148 (1.34%) | 48 (1.56%) | 0.0188 | 59 (1.25%) | 38 (1.62%) | 0.0304 |
| Direct thrombin inhibitors | 33 (0.30%) | 13 (0.42%) | 0.0208 | 17 (0.36%) | 8 (0.34%) | 0.0036 |
| Direct factor Xa inhibitors | 171 (1.54%) | 62 (2.01%) | 0.0357 | 79 (1.68%) | 39 (1.66%) | 0.0017 |
| SGLT-2 inhibitors | 35 (0.32%) | 21 (0.68%) | 0.0520 | 25 (0.53%) | 11 (0.47%) | 0.0090 |
Continuous variables are presented as mean ± standard deviation, and categorical variables as number (percentage)
ASDs were calculated to assess covariate balance between groups. An ASD > 0.10 was considered indicative of potentially meaningful imbalance
Abbreviations ACEI angiotensin-converting enzyme inhibitor, ARBs angiotensin II receptor blockers, ASD absolute standardized difference, BZDs benzodiazepines, CCI Charlson comorbidity index, MACE major adverse cardiovascular events, MI myocardial infarction, NSAIDs non-steroidal anti-inflammatory drugs, OA osteoarthritis, OSA obstructive sleep apnea, PAD peripheral artery disease, PPIs proton pump inhibitors, PSM propensity score matching, RA rheumatoid arthritis, SGLT-2 sodium–glucose cotransporter 2 (inhibitors), SNRIs serotonin–norepinephrine reuptake inhibitors, SSRIs selective serotonin reuptake inhibitors, TKA total knee arthroplasty, UTI urinary tract infection
After 1:2 PSM, 4,702 non–long-term opioid users and 2,351 long-term opioid users were included in the final analytical cohort. Covariate balance improved substantially, with all ASDs below 0.10 across demographics, TKA-related factors, comorbidities, physical therapy variables, and baseline medications.
The risk of mortality and MACEs in patients with long-term opioid use
In the PSM cohort (Table 2), long-term opioid use remained independently associated with an increased risk of MACE (aHR = 1.16; 95% CI, 1.06–1.26; P = 0.0007). The random-effects model produced a similar direction of effect (aHR = 1.28; 95% CI, 1.13–1.45; P = 0.0002). For individual outcomes, long-term opioid use was associated with higher risks of all-cause mortality (aHR = 1.16; 95% CI, 1.05–1.29; P = 0.0055), AMI (aHR = 1.30; 95% CI, 1.03–1.64; P = 0.0263), and CHF (aHR = 1.15; 95% CI, 1.00–1.32; P = 0.0463). In contrast, no significant differences were observed for ischemic stroke or hemorrhagic stroke, and competing-risk models for nonfatal outcomes were consistent with these findings. Kaplan–Meier curves for the matched cohort showed a higher cumulative probability of MACE and all-cause mortality among long-term opioid users (Fig. 2).
Table 2.
Mortality and cardiovascular outcomes after total knee arthroplasty in propensity score–matched cohorts
| Non–long-term opioid users (n = 4,702) |
Long-term opioid users (n = 2,351) |
P value | AIC | BIC | |
|---|---|---|---|---|---|
| MACE a | |||||
| Person-months | 320,742 | 149,832 | |||
| Event | 1,594 | 837 | |||
| Incidence rate (95% CI) b | 4.97 (4.73–5.22) | 5.59 (5.22–5.98) | 0.0062 | ||
| Crude HR (95% CI) | Reference | 1.13 (1.04–1.23) | 0.0033 | 39,307 | 39,313 |
| Adjusted HR (95% CI) c | Reference | 1.16 (1.06–1.26) | 0.0007 | 38,238 | 38,928 |
| Random-effects aHR (95% CI) d | Reference | 1.28 (1.13–1.45) | 0.0002 | ||
| All-cause mortality | |||||
| Person-months | 370,108 | 174,221 | |||
| Event | 1,026 | 547 | |||
| Incidence rate (95% CI) b | 2.77 (2.61–2.95) | 3.14 (2.89–3.41) | 0.0187 | ||
| Crude HR (95% CI) | Reference | 1.17 (1.05–1.29) | 0.0040 | 24,943 | 24,948 |
| Adjusted HR (95% CI) c | Reference | 1.16 (1.05–1.29) | 0.0055 | 24,053 | 24,691 |
| Random-effects aHR (95% CI) d | Reference | 1.25 (1.05–1.50) | 0.0133 | ||
| Acute myocardial infarction | |||||
| Person-months | 360,965 | 168,877 | |||
| Event | 205 | 119 | |||
| Incidence rate (95% CI) b | 0.57 (0.50–0.65) | 0.70 (0.59–0.84) | 0.0612 | ||
| Crude HR (95% CI) | Reference | 1.25 (0.99–1.56) | 0.0548 | 5328 | 5332 |
| Adjusted HR (95% CI) c | Reference | 1.30 (1.03–1.64) | 0.0263 | 5241 | 5691 |
| Competing-risk sHR (95% CI) e | Reference | 1.25 (0.98–1.58) | 0.0694 | 5348 | 5798 |
| Random-effects aHR (95% CI) d | Reference | 2.38 (1.39–4.08) | 0.0016 | ||
| Ischemic stroke | |||||
| Person-months | 346,641 | 164,144 | |||
| Event | 519 | 241 | |||
| Incidence rate (95% CI) b | 1.50 (1.37–1.63) | 1.47 (1.29–1.67) | 0.8019 | ||
| Crude HR (95% CI) | Reference | 0.98 (0.84–1.14) | 0.8129 | 12,480 | 12,484 |
| Adjusted HR (95% CI) c | Reference | 0.98 (0.84–1.14) | 0.7803 | 12,340 | 12,892 |
| Competing-risk sHR (95% CI) e | Reference | 0.94 (0.80–1.10) | 0.4271 | 12,585 | 13,136 |
| Random-effects aHR (95% CI) d | Reference | 1.09 (0.85–1.40) | 0.4840 | ||
| Hemorrhagic stroke | |||||
| Person-months | 363,501 | 170,895 | |||
| Event | 210 | 96 | |||
| Incidence rate (95% CI) b | 0.58 (0.50–0.66) | 0.56 (0.46–0.69) | 0.8200 | ||
| Crude HR (95% CI) | Reference | 0.98 (0.77–1.25) | 0.8917 | 4983 | 4987 |
| Adjusted HR (95% CI) c | Reference | 0.99 (0.77–1.26) | 0.9250 | 5056 | 5499 |
| Competing-risk sHR (95% CI) e | Reference | 0.92 (0.72–1.19) | 0.5327 | 5177 | 5620 |
| Random-effects aHR (95% CI) d | Reference | 1.79 (0.95–3.39) | 0.0717 | ||
| Congestive heart failure | |||||
| Person-months | 345,879 | 161,206 | |||
| Event | 612 | 312 | |||
| Incidence rate (95% CI) b | 1.77 (1.63–1.92) | 1.94 (1.73–2.16) | 0.1974 | ||
| Crude HR (95% CI) | Reference | 1.10 (0.96–1.27) | 0.1541 | 15,038 | 15,043 |
| Adjusted HR (95% CI) c | Reference | 1.15 (1.00–1.32) | 0.0463 | 14,520 | 15,095 |
| Competing-risk sHR (95% CI) e | Reference | 1.10 (0.95–1.27) | 0.1959 | 14,828 | 15,403 |
| Random-effects aHR (95% CI) d | Reference | 1.18 (0.94–1.50) | 0.1624 | ||
Abbreviations aHR adjusted hazard ratio, AIC Akaike information criterion, BIC Bayesian information criterion, BZDs benzodiazepines, CI confidence interval, HR hazard ratio, MACE major adverse cardiovascular event, NSAIDs non-steroidal anti-inflammatory drugs, PPIs proton pump inhibitors, sHR subdistribution hazard ratio, SNRIs serotonin–norepinephrine reuptake inhibitors, SSRIs selective serotonin reuptake inhibitors
a The composite MACE outcome included all-cause mortality, acute myocardial infarction, ischemic stroke, hemorrhagic stroke, and congestive heart failure
b Incidence rates were calculated as events per 1,000 person-months
c Adjusted HRs were estimated using multivariable Cox proportional hazards models adjusting for demographics, surgical characteristics, comorbidities within 1 year before the index date, physical therapy/rehabilitation, and medication use (NSAIDs, PPIs, acetaminophen, glucosamine, gabapentin/pregabalin, SNRIs, SSRIs, other antidepressants, BZDs, and Z-drugs) during hospitalization and up to 90 days post-discharge
d Random-effects (shared frailty) Cox models with institution-level clustering were used to account for within-center correlation and between-institution heterogeneity
e For non-fatal cardiovascular outcomes, competing-risk analyses were performed using the Fine–Gray subdistribution hazards model, treating all-cause mortality as the competing event (reporting sHRs)
Fig. 2.
Cumulative event probability of study outcomes in the propensity score–matched cohort. Kaplan–Meier curves show the 14-year cumulative event probability after propensity score matching, including (A) the composite MACE endpoint, (B) all-cause mortality, (C) acute myocardial infarction, (D) ischemic stroke, (E) hemorrhagic stroke, and (F) congestive heart failure. Shaded areas indicate 95% confidence intervals. Numbers at risk are shown below each panel, and P values are from the log-rank test. Abbreviations: CI, confidence interval; MACE, major adverse cardiovascular event
Subgroup analysis and sensitivity analysis
In subgroup analyses of primary outcomes (Fig. 3), long-term opioid use was generally associated with higher risks of MACE and all-cause mortality across most prespecified strata. Evidence of effect modification was observed for CCI. For MACE, the interaction with CCI was significant (P for interaction = 0.0278), with the strongest association among individuals with CCI ≥ 5 (aHR = 1.60; 95% CI, 1.26–2.02), whereas estimates in lower CCI strata were more modest. For all-cause mortality, the interaction with CCI was also significant (P for interaction = 0.0006), and the association was most pronounced in CCI ≥ 5 (aHR = 2.08; 95% CI, 1.58–2.74). No statistically significant interactions were identified for several other subgroup factors (e.g., sex, age at TKA, or hospital level). Stratified analyses for individual cardiovascular components are shown in Supplementary Figure S1.
Fig. 3.
Subgroup analyses of the primary outcomes. Forest plots show aHRs with 95% CIs for (A) the composite MACE endpoint and (B) all-cause mortality across prespecified demographic and clinical subgroups. P values for interaction are shown to assess effect modification across subgroups. Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; aHR, adjusted hazard ratio; BZD, benzodiazepine; CCI, Charlson comorbidity index; CI, confidence interval; MACE, major adverse cardiovascular event; NSAIDs, non-steroidal anti-inflammatory drugs; OA, osteoarthritis; TKA, total knee arthroplasty
Time-stratified analyses (Table 3) demonstrated that the excess risk associated with long-term opioid use was greatest early after the index date for MACE (Index–6 months: HR = 1.48; 95% CI, 1.20–1.82; 6 months–1 year: HR = 1.43; 95% CI, 1.15–1.79) and remained significant through 1–5 years (HR = 1.21; 95% CI, 1.09–1.34), with attenuation thereafter (5–10 years: HR = 1.03; 95% CI, 0.90–1.20). A similar temporal pattern was observed for all-cause mortality and AMI, with the largest point estimates occurring within the first year. By contrast, ischemic stroke, hemorrhagic stroke, and CHF did not show statistically significant time-specific associations across prespecified intervals.
Table 3.
Time-varying hazard ratios for mortality and cardiovascular outcomes across follow-up intervals
| Time-varying HRs (95% CI) | ||||
|---|---|---|---|---|
| Outcome | Index–6 months | 6 months–1 year | 1–5 years | 5–10 years |
| MACE | 1.48 (1.20–1.82) | 1.43 (1.15–1.79) | 1.21 (1.09–1.34) | 1.03 (0.90–1.20) |
| All-cause mortality | 2.97 (1.99–4.45) | 1.43 (0.96–2.14) | 1.31 (1.14–1.51) | 1.10 (0.95–1.27) |
| Acute myocardial infarction | 2.15 (1.17–3.95) | 1.84 (1.01–3.36) | 1.08 (0.82–1.43) | 0.92 (0.61–1.39) |
| Ischemic stroke | 1.09 (0.73–1.64) | 1.14 (0.75–1.73) | 1.14 (0.95–1.37) | 0.88 (0.67–1.14) |
| Hemorrhagic stroke | 1.38 (0.68–2.79) | 1.74 (0.77–3.93) | 1.03 (0.74–1.42) | 1.20 (0.84–1.70) |
| Congestive heart failure | 1.10 (0.76–1.58) | 1.28 (0.90–1.82) | 1.13 (0.95–1.34) | 1.22 (0.97–1.52) |
Abbreviations CI confidence interval, HR hazard ratio, MACE major adverse cardiovascular event
Hazard ratios were estimated across prespecified follow-up intervals using time-stratified Cox proportional hazards models and were adjusted for prespecified covariates
Discussion
In this nationwide claims-based cohort study of patients undergoing TKA for knee OA, long-term opioid use during the 1-year post discharge ascertainment window was associated with an increased subsequent risk of MACE and all-cause mortality. In the PSM cohort, long-term opioid users also showed higher risks of AMI and CHF, whereas risks of ischemic and hemorrhagic stroke were not significantly different.
The presence of OA could accompany with several co-morbidities and mortality in previous publications [21]. The OA can associate with the reduction of bone mineral density and T-score in the postmenopausal women population that imply the risk of bone fracture should be considered [22]. In previous study, the dyslipidemia and type 2 diabetes mellitus associates with a higher possibility of OA development [21]. In addition, the incidence of cerebrovascular accident, hypertension, and arterial intima media thickening are significantly higher in the population with OA [23]. Because of the reduction of general function result from the development of OA, the mortality rate is also higher in the OA population than the general population [24]. In previous literature, the OA population presented with 1.18 folds of mortality rate compared to the non-OA population which is statistically significant [25]. Despite the TKA surgery can reduce the mortality and morbidity of OA, the mortality and MACEs can still occurred in OA patients after TKA surgery [26]. On the other side, the application of opioid can contribute to a prominent number of disorders including respiratory depression and sedation [27]. The usage of opioid can alter the central nervous system via the opioid receptor and toxic dose of opioid can cause multifocal myoclonus and seizures [28]. Also, the overdose and opioid withdrawal could alter the function of cardiovascular system and the cardiovascular disorders like the AMI, heart failure, and arrhythmias develops may occur in the individuals with opioid misuse and overdose [29]. In addition, the mortality rate of individuals with opioid usage elevated for about 50% than the non-opioid usage population [30]. The OA patients is a vulnerable population to MACE and mortality and the risk factors of OA including metabolic syndrome and atherosclerosis [21, 26], and opioid exposure has been associated with adverse neurologic and cardiopulmonary effects and increased mortality risk [16, 27, 29]. Accordingly, we hypothesized that the persistent opioid application in OA population with the analgesia requirement like those receive TKA surgery would elevate the risk of mortality and major morbidities. These findings support our a priori hypothesis that sustained opioid exposure after TKA is associated with increased long-term risks of mortality and adverse cardiovascular outcomes.
The risk of mortality was significantly higher in the long-term opioid users than the non-long term opioid users after the TKA surgery for OA management. In the previous research, the long-term application of opioid is associated with higher chance of neurological disorder such as bladder dysfunction and respiratory depression [14]. Also, the persistent usage of opioid for chronic non-cancer pain causes more than 10,000 deaths per year [31]. Nevertheless, the negative effect of opioid, if existed, on the postoperative outcome had seldom been reported. Evidence regarding long-term postoperative outcomes among patients who continue opioids after TKA remains limited. Using a landmark design, we restricted analyses to 1-year survivors to ensure that exposure ascertainment preceded outcome assessment and to reduce immortal time and reverse-causation concerns. Furthermore, we adjusted several confounders that may elevated the risk of death after TKA surgery including hypertension, diabetes mellitus, ischemic heart diseases, respiratory disease and malignancy in the Cox proportional hazard regression [32]. Accordingly, long-term opioid use after TKA was associated with a higher risk of all-cause mortality. Compared to the short-term usage of opioid, the application of opioid for months to years cause significant adverse effect in previous researches [15, 31]. This association may reflect a combination of opioid-related physiologic effects and underlying patient vulnerability in the post-TKA OA population [25]. As a consequence, the mortality rate of long-term opioid user after TKA surgery was significantly elevated compared to the non-long-term opioid users. In the matched cohort, long-term opioid users showed a higher cumulative probability of all-cause mortality than non–long-term users, and time-stratified analyses suggested that the excess risk was most pronounced early after the index date and attenuated over longer follow-up. These findings further highlight the importance of limiting prolonged opioid use after TKA.
Prior analyses have reported that MACE is a major contributor to adverse outcomes among patients with OA undergoing TKA [33]. Consistent with evidence linking opioid exposure to cardiovascular risk [29], our findings indicate that long-term opioid use after TKA was associated with an increased risk of MACE, particularly driven by AMI and CHF, whereas stroke components were not significantly different. These associations were directionally consistent in sensitivity analyses that accounted for competing risks. Although TKA may improve functional status and has been associated with reductions in some adverse outcomes in OA populations [34], our results suggest that prolonged opioid exposure remains associated with excess long-term cardiovascular risk. Prior literature has reported associations between opioid misuse/overdose and cardiovascular events such as AMI, heart failure, and arrhythmias [29].
About the epidemiological issue, the OA is a prevalent disease in which the annual incidence of OA ranges from 6.3 to 68.4% throughout the world [2, 35]. In the American population, there were more than 20 million person complaining about arthritis-attributable motion limitations [3]. Among the OA population, the TKA surgery is a frequent management in which 150–200 per 100,000 OA inhabitants received TKA surgery in the industrialized countries [7], and the individuals living with previous TKA surgery achieves 10% in the United States according to previous study [36]. On the other side, the opioid is a widely-applied analgesia which was applied in patients received orthopedic surgeries [37]. The pain control is a major issue after TKA surgery and the opioid prescription account for a prominent number of analgesia [10]. Although the application of opioid did not illustrate superior pain-control effect than other analgesia, the opioid dependence and its complications may reduce the health of patients [38]. Because the persistent usage of opioid could contribute to major health issue [16], any complications of opioid in the patients frequently applied it like those received TKA surgery could be illustrated.
Our subgroup analyses demonstrated clinically meaningful heterogeneity by comorbidity burden. The association of long-term opioid use with both MACE and all-cause mortality was strongest among patients with high CCI, suggesting that medically complex individuals may be particularly vulnerable to adverse outcomes associated with sustained opioid exposure. Previous research indicated that younger opioid users were at greater risk of overdose before entering opioid maintenance treatment [39] and that a high-dose increasing trajectory of opioids was positively associated with mortality [40]. Additionally, the proportion of deaths related to opioid use was significantly higher in individuals under 65 compared to those aged 65 and above [41]. Although prior studies have reported that opioid-related mortality patterns may vary by age and exposure trajectories, such mechanisms cannot be directly evaluated in our claims data and should be interpreted as contextual evidence rather than subgroup-specific causal inference. The time-stratified analyses suggested that the excess risk associated with long-term opioid use was most pronounced early after the index date and attenuated over longer follow-up. This temporal pattern may reflect a period of heightened vulnerability in the early post-landmark phase, when persistent pain, functional limitation, and comorbidity-related instability may coincide with sustained opioid exposure. It may also reflect time-varying residual confounding (e.g., pain severity, frailty, functional status) and changes in exposure status over time—factors not fully captured in administrative data.
Strengths and limitations
This study has several strengths. The large, representative longitudinal database enabled extended follow-up and evaluation of both composite and component outcomes. The landmark design provided a fixed exposure ascertainment window and helped reduce immortal time and reverse-causation concerns by restricting analyses to 1-year survivors. In addition, extensive covariate adjustment, PSM, and sensitivity analyses—including competing-risk models for nonfatal cardiovascular outcomes—were used to evaluate robustness.
Several limitations should be considered. First, as a claims-based observational study, residual confounding is possible, particularly confounding by indication, because pain severity, functional status, frailty, lifestyle factors, and detailed clinical measurements are not available in the database. Second, prescription dispensing does not guarantee medication consumption, and we lacked detailed information on opioid dose, route, and cumulative exposure (e.g., morphine milligram equivalents), which may influence risk. Third, some opioid exposure may be missed because patient-controlled analgesia can be self-paid in Taiwan and may not be fully captured, potentially leading to exposure misclassification. Fourth, outcomes were identified using diagnosis codes, and misclassification is possible despite the use of established coding definitions. Finally, because the index date in this landmark design was defined as 360 days after discharge from the index TKA hospitalization, our findings apply to patients who survived the first postoperative year; however, the consistency of results across the PSM cohort and sensitivity analyses, including time-stratified estimates, supports the robustness of the observed associations.
Conclusion
In this nationwide claims-based cohort of patients undergoing TKA for knee OA, long-term opioid use during the first postoperative year was associated with increased subsequent risks of MACE and all-cause mortality. In the PSM cohort, long-term opioid users also had higher risks of AMI and CHF, whereas ischemic and hemorrhagic stroke risks were not significantly different. The excess risk was most pronounced early after the index date and attenuated over longer follow-up, and the association with both MACE and mortality was strongest among patients with higher comorbidity burden (CCI ≥ 5). These findings underscore the importance of postoperative opioid stewardship, including multimodal analgesia and opioid-sparing strategies, and support closer clinical follow-up, particularly cardiovascular risk assessment and monitoring, for patients requiring long-term opioids after TKA, especially those with substantial comorbidities.
Supplementary Information
Acknowledgements
This study was partly based on data from the NHIRD provided by the NHI Administration, Ministry of Health and Welfare, and managed by the National Health Research Institutes. The interpretation and conclusions contained herein do not represent those of the NHI Administration, Ministry of Health and Welfare, or National Health Research Institutes.
Authors’ contributions
Han-Fang Tseng: This author designed the study, searched the literature, responsible for data interpretation, prepared the manuscript draft, and was approved for the final version of the manuscript. Yat-Yin Law: This author designed the study, searched the literature, responsible for data interpretation, prepared the manuscript draft, and was approved for the final version of the manuscript. Pei-Ni Chen: This author designed the study, prepared the manuscript draft, and was approved for the final version of the manuscript. Chia-Yi Lee: This author designed the study, prepared the manuscript draft, and was approved for the final version of the manuscript. Jing-Yang Huang: This author was responsible for data collection, data analysis, performed the statistical analyses, and approved the final version of the manuscript. Shun-Fa Yang: This author designed the study, was responsible for data collection, and data analysis, and approved the final version of the manuscript. Chao-Bin Yeh: This author designed the study, was responsible for data analysis, and data interpretation, refined the manuscript, and approved the final version of the manuscript.
Funding
None.
Data availability
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The study adhered to the Declaration of Helsinki of 1964 and its amendments. The current study was agreed upon by the National Health Insurance Administration of Taiwan and the Institutional Review Board of Chung Shan Medical University (approval number: CS2-23031; Data of approval: Mar 13, 2023).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Shun-Fa Yang, Email: ysf@csmu.edu.tw.
Chao-Bin Yeh, Email: sky5ff@gmail.com.
References
- 1.Leopold SS. Minimally invasive total knee arthroplasty for osteoarthritis. N Engl J Med. 2009;360:1749–58. [DOI] [PubMed] [Google Scholar]
- 2.Pereira D, Ramos E, Branco J. Osteoarthr Acta Med Port. 2015;28:99–106. [DOI] [PubMed] [Google Scholar]
- 3.Mandl LA. Osteoarthritis year in review 2018: clinical. Osteoarthritis Cartilage. 2019;27:359–64. [DOI] [PubMed] [Google Scholar]
- 4.Becker R. Total knee arthroplasty 2010. Knee Surg Sports Traumatol Arthrosc. 2010;18:851–2. [DOI] [PubMed] [Google Scholar]
- 5.Yeh HW, Chan CH, Yang SF, Chen YC, Yeh YT, Yeh YT, et al. Total knee replacement in osteoarthritis patients on reducing the risk of major adverse cardiac events: a 18-year retrospective cohort study. Osteoarthritis Cartilage. 2022;30:416–25. [DOI] [PubMed] [Google Scholar]
- 6.Rivière C, Iranpour F, Auvinet E, Howell S, Vendittoli PA, Cobb J, et al. Alignment options for total knee arthroplasty: A systematic review. Orthop Traumatol Surg Res. 2017;103:1047–56. [DOI] [PubMed] [Google Scholar]
- 7.Alrawashdeh W, Eschweiler J, Migliorini F, El Mansy Y, Tingart M, Rath B. Effectiveness of total knee arthroplasty rehabilitation programmes: A systematic review and meta-analysis. J Rehabil Med. 2021;53:jrm00200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Momoli A, Giarretta S, Modena M, Micheloni GM. The painful knee after total knee arthroplasty: evaluation and management. Acta Biomed. 2017;88:60–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lewis GN, Rice DA, McNair PJ, Kluger M. Predictors of persistent pain after total knee arthroplasty: a systematic review and meta-analysis. Br J Anaesth. 2015;114:551–61. [DOI] [PubMed] [Google Scholar]
- 10.Lavand’homme PM, Kehlet H, Rawal N, Joshi GP. Pain management after total knee arthroplasty: PROcedure SPEcific Postoperative Pain ManagemenT recommendations. Eur J Anaesthesiol. 2022;39:743–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Che T, Roth BL. Molecular basis of opioid receptor signaling. Cell. 2023;186:5203–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yeh PH, Yeh HW, Yang SF, Wang YH, Chou MC, Tsai PK, et al. No association of postoperative opioid usage with long-term surgery outcomes in patients with liver cancer: a population-based retrospective cohort study. Pain. 2023;164:848–54. [DOI] [PubMed] [Google Scholar]
- 13.Derogatis MJ, Sodhi N, Anis HK, Ehiorobo JO, Bhave A, Mont MA. Pain Management Strategies To Reduce Opioid Use Following Total Knee Arthroplasty. Surg Technol Int. 2019;35:301–10. [PubMed] [Google Scholar]
- 14.Benyamin R, Trescot AM, Datta S, Buenaventura R, Adlaka R, Sehgal N, et al. Opioid complications and side effects. Pain Physician. 2008;11:S105–20. [PubMed] [Google Scholar]
- 15.Paul LA, Li Y, Leece P, Gomes T, Bayoumi AM, Herring J, et al. Identifying the changing age distribution of opioid-related mortality with high-frequency data. PLoS ONE. 2022;17:e0265509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Salazar CI, Huang Y. The burden of opioid-related mortality in Texas, 1999 to 2019. Ann Epidemiol. 2022;65:72–7. [DOI] [PubMed] [Google Scholar]
- 17.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370:1453–7. [DOI] [PubMed] [Google Scholar]
- 18.Rosenberg JM, Bilka BM, Wilson SM, Spevak C. Opioid Therapy for Chronic Pain: Overview of the 2017 US Department of Veterans Affairs and US Department of Defense Clinical Practice Guideline. Pain Med. 2018;19:928–41. [DOI] [PubMed] [Google Scholar]
- 19.Yeh PH, Yang SF, Huang JY, Yeh CB. Prediction Model for Factors Associated With Long-Term Opioid Use Following Total Knee Arthroplasty: A Retrospective Population-Based Study. Eur J Pain. 2026;30:e70195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–83. [DOI] [PubMed] [Google Scholar]
- 21.Abramoff B, Caldera FE. Osteoarthritis: Pathology, Diagnosis, and Treatment Options. Med Clin North Am. 2020;104:293–311. [DOI] [PubMed] [Google Scholar]
- 22.Stamenkovic BN, Rancic NK, Bojanovic MR, Stojanovic SK, Zivkovic VG, Djordjevic DB, et al. Is Osteoarthritis Always Associated with Low Bone Mineral Density in Elderly Patients? Medicina (Kaunas). 2022;58. [DOI] [PMC free article] [PubMed]
- 23.Al-Khazraji BK, Badrov MB, Kadem M, Lingum NR, Birmingham TB, Shoemaker JK. Exploring Cerebrovascular Function in Osteoarthritis: Heads-up. Physiol Rep. 2019;7:e14212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hochberg MC. Mortality in osteoarthritis. Clin Exp Rheumatol. 2008;26:S120–4. [PubMed] [Google Scholar]
- 25.Veronese N, Cereda E, Maggi S, Luchini C, Solmi M, Smith T, et al. Osteoarthritis and mortality: A prospective cohort study and systematic review with meta-analysis. Semin Arthritis Rheum. 2016;46:160–7. [DOI] [PubMed] [Google Scholar]
- 26.Okpara S, Lee T, Pathare N, Ghali A, Momtaz D, Ihekweazu U. Cardiovascular Disease in Total Knee Arthroplasty: An Analysis of Hospital Outcomes, Complications, and Mortality. Clin Orthop Surg. 2024;16:265–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Stein C. New concepts in opioid analgesia. Expert Opin Investig Drugs. 2018;27:765–75. [DOI] [PubMed] [Google Scholar]
- 28.Trescot AM, Datta S, Lee M, Hansen H. Opioid pharmacology. Pain Physician. 2008;11:S133–53. [PubMed] [Google Scholar]
- 29.Krantz MJ, Palmer RB, Haigney MCP. Cardiovascular Complications of Opioid Use: JACC State-of-the-Art Review. J Am Coll Cardiol. 2021;77:205–23. [DOI] [PubMed] [Google Scholar]
- 30.Santo T Jr., Clark B, Hickman M, Grebely J, Campbell G, Sordo L, et al. Association of Opioid Agonist Treatment With All-Cause Mortality and Specific Causes of Death Among People With Opioid Dependence: A Systematic Review and Meta-analysis. JAMA Psychiatry. 2021;78:979–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Von Korff M, Kolodny A, Deyo RA, Chou R. Long-term opioid therapy reconsidered. Ann Intern Med. 2011;155:325–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Bemenderfer TB, Rozario NL, Moore CG, Karunakar MA. Morbidity and Mortality in Elective Total Knee Arthroplasty Following Surgical Care Improvement Project Guidelines. J Arthroplasty. 2016;31:202–6. [DOI] [PubMed] [Google Scholar]
- 33.Berstock JR, Beswick AD, López-López JA, Whitehouse MR, Blom AW. Mortality After Total Knee Arthroplasty: A Systematic Review of Incidence, Temporal Trends, and Risk Factors. J Bone Joint Surg Am. 2018;100:1064–70. [DOI] [PubMed] [Google Scholar]
- 34.Ravi B, Croxford R, Austin PC, Lipscombe L, Bierman AS, Harvey PJ, et al. The relation between total joint arthroplasty and risk for serious cardiovascular events in patients with moderate-severe osteoarthritis: propensity score matched landmark analysis. BMJ. 2013;347:f6187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Glyn-Jones S, Palmer AJ, Agricola R, Price AJ, Vincent TL, Weinans H, et al. Osteoarthr Lancet. 2015;386:376–87. [DOI] [PubMed] [Google Scholar]
- 36.Aujla RS, Esler CN. Total Knee Arthroplasty for Osteoarthritis in Patients Less Than Fifty-Five Years of Age: A Systematic Review. J Arthroplasty. 2017;32:2598–603.e1. [DOI] [PubMed]
- 37.Colvin LA, Bull F, Hales TG. Perioperative opioid analgesia-when is enough too much? A review of opioid-induced tolerance and hyperalgesia. Lancet. 2019;393:1558–68. [DOI] [PubMed] [Google Scholar]
- 38.Chaturvedi R, Tram J, Chakravarthy K. Reducing opioid usage in total knee arthroplasty postoperative pain management: a literature review and future directions. Pain Manag. 2022;12:105–16. [DOI] [PubMed] [Google Scholar]
- 39.Clausen T, Waal H, Thoresen M, Gossop M. Mortality among opiate users: opioid maintenance therapy, age and causes of death. Addiction. 2009;104:1356–62. [DOI] [PubMed] [Google Scholar]
- 40.Binswanger IA, Shetterly SM, Xu S, Narwaney KJ, McClure DL, Rinehart DJ, et al. Opioid Dose Trajectories and Associations With Mortality, Opioid Use Disorder, Continued Opioid Therapy, and Health Plan Disenrollment. JAMA Netw Open. 2022;5:e2234671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Gomes T, Tadrous M, Mamdani MM, Paterson JM, Juurlink DN. The Burden of Opioid-Related Mortality in the United States. JAMA Netw Open. 2018;1:e180217. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.