Abstract
Background
We sought to replicate recent findings that total knee arthroplasty (TKA) or total hip arthroplasty (THA) surgery substantially reduces the risk of serious cardiovascular events among osteoarthritis patients in a UK general population.
Methods
We conducted a time-stratified propensity score-matched cohort study for the outcome of myocardial infarction (MI). The study population included individuals aged ≥50 years who had a Readcode diagnosis of knee osteoarthritis (to evaluate TKA) or hip osteoarthritis (to evaluate THA) between January 2000 and December 2012.
Results
Among 13,849 patients who underwent TKA and 13,849 matched non-TKA controls 306 and 286 developed MI during the follow-up, respectively. During the first postoperative month, the risk of MI was substantially increased among TKA group compared with non-TKA group (hazard ratio 8.75; 95% CI, 3.11–24.62), and then gradually declined during the subsequent follow-up. The HR of the entire follow-up was 0.98 (95% CI, 0.82–1.18). The corresponding HRs for THA (n=6,063) compared with non-THA were 4.33 (95% CI, 1.24–15.21) and 0.87 (95% CI, 0.66–1.15), respectively. Using venous thromboembolism as a positive control outcome, both the first month and overall HRs of MI were substantially increased for TKA and for THA, respectively.
Conclusion
These findings provide the first general population-based evidence that TKA and THA among osteoarthritis patients are associated with a substantially increased risk of MI during the immediate postoperative period. However, its overall long-term impact was null, unlike the risk of venous thromboembolism that remained years after the procedure.
Keywords: Total Knee Arthroplasty, Total Hip Arthroplasty, Myocardial Infarction, Propensity Score
INTRODUCTION
Total knee or hip arthroplasty is a common surgical procedure for end-stage osteoarthritis; indeed, these procedures have increased dramatically over the last decade such that approximately 1.8 million procedures are performed annually worldwide.(1–3) Many studies have documented that total joint arthroplasty procedures significantly improve patients’ pain, function, and health-related quality of life in patients with osteoarthritis, (4–7) whereas their impact on cardiovascular outcomes is largely unknown.
To date, only one cohort study (with a control group) examined patients with knee or hip osteoarthritis for the potential cardiovascular impact of total joint arthroplasty (TJR) and reported a 44% reduced risk of serious cardiovascular events over a median follow-up of 7 years.(8) However, the study design led to the exclusion of short-term cardiovascular events occurring after the procedure from the analysis.(8) Thus, not only did the study fail to address the short-term risk of cardiovascular events occurring after the procedure, but also the exclusion of short-term cardiovascular events could have introduced selection bias (9, 10). As individuals who were susceptible to the cardiovascular events shortly after total joint arthroplasty were excluded from the analysis, the differential depletion of susceptibles to myocardial infarction between the total joint arthroplasty cohort and non- total joint arthroplasty cohort may have biased effect estimates of total joint arthroplasty towards the null or in a protective direction. Indeed, this possibility is supported by a nationwide retrospective cohort study (although not limited to those with osteoarthritis) that found a 25- to 31- fold increased cardiovascular risk during the first weeks after TJR.(11) No study has examined the short- and long-term cardiovascular risk among patients with osteoarthritis.
To address this key knowledge gap in the field, we conducted a propensity score-matched cohort study to evaluate the relation between total joint arthroplasty (i.e., total knee arthroplasty and total hip arthroplasty) with the short- and long-term risk of myocardial infarction among patients with knee and hip osteoarthritis in a general population context.
METHODS
Data Source
The Health Improvement Network (THIN) is a computerized medical record database from general practices in the United Kingdom (UK).(12) Data on approximately 10.2 million patients from 580 general practices are systematically recorded by general practitioners and sent anonymously to THIN. Because the National Health Service in the UK requires every individual to be registered with a general practitioner regardless of health status, THIN is a population-based cohort representative of the UK general population. The computerized information includes demographics, details from general practitioner visits, diagnoses from specialists’ referrals and hospital admissions, results of laboratory tests, and additional systematically recorded health information including height, weight, blood pressure, smoking status, and vaccinations. The Read classification is used to code specific diagnoses,(13) and a drug dictionary based on data from the Multilex classification is used to code drugs.(14) Health information is recorded on site at each practice using a computerized system with quality control procedures to maintain high data completion rates and accuracy.
Study Design and Cohort Assembling
Total Knee Arthroplasty and the Risk of Myocardial Infarction
We conducted a time-stratified propensity-matched cohort study to examine the association between total knee arthroplasty and the risk of myocardial infarction. Our study population included individuals aged ≥50 years who had a diagnosis of knee osteoarthritis between January 2000 and December 2012. The diagnosis of knee osteoarthritis was defined by Read codes indicating specific knee joints (N053611: Patellofemoral osteoarthritis; N05z611: Knee osteoarthritis NOS; N05zL00: Osteoarthritis NOS, of knee). Patients with only osteoarthritis coded as general irrespective of joint site were ineligible for analysis. A single medical visit with a diagnosis of osteoarthritis is a standard definition adopted in previous studies (15, 16) and is preferred in previous validation studies to case definitions based on medical visits, referrals, or prescription records (17, 18). Indeed, Prieto-Alhambra et al.(19) found that only 1.3% of cases defined in this way were subsequently given an alternative diagnosis. Study cohort members were further required to have ≥2 years of continuous enrollment with the general practice before entering the study cohort. Patients underwent total knee arthroplasty were identified by Read codes. Previous studies have shown the validity of this approach in electronic medical record database in United Kingdom.(20, 21) To help ensure the comparability of patients who underwent total knee arthroplasty and those who did not, we excluded subjects who might be considered potentially ineligible for total knee replacement surgery, including those with a body mass index (BMI) >40, history of joint infection, cancer (pancreatic, esophageal, gastric, or other metastatic cancers), end-stage renal disease on dialysis, use of nasal cannula oxygen, or disease-modifying anti-rheumatic drug therapy. Patients with a history of myocardial infarction were also excluded for this analysis.
As confounding by indication can be a major concern in epidemiological studies for evaluating the effect of total joint arthroplasty (i.e., the distribution of risk factors for myocardial infarction between total joint arthroplasty patients and comparison patients may systematically differ, causing a lack of comparability between the two groups), we employed propensity score-matching to minimize potential confounding by indication. Compared with traditional method of controlling for confounders (e.g. multivariable logistic regression model), propensity score analyses allows us to control for a large number of covariates at one time, and gives a more robust, less biased estimate when the number of outcome events is low relative to the number of confounders (22). Further, to closely account for potential secular trends in total joint arthroplasty in relation to various covariates at different calendar times (23), matched cohorts were constructed within 1-year blocks of calendar time (i.e., 13 blocks from January 2000 to December 2012). Within each 1-year time block we calculated propensity scores for receiving total knee arthroplasty for each eligible individual using a logistic regression model. The date of total knee arthroplasty was used as the index date for that patient, and a random date within the one-year block was assigned as the index date for the matched subject who did not receive total knee arthroplasty. The variables included in the logistic regression model consisted of osteoarthritis duration prior to the index date, socio-demographic factors, BMI (8% missing), lifestyle factors, comorbidities, medication use, and healthcare utilization. Specifically, socio-demographic information included age at the index date, sex, and the Townsend Deprivation Index score, an indicator for economic status (24, 25). Lifestyle factors included smoking status (2% missing) and alcohol use (8% missing). We grouped smoking status and alcohol status into three categories each: current smoker, past smoker or never smoker and current drinker, past drinker or never drinker, respectively. Comorbidities included hypertension, diabetes, hyperlipidemia, ischemic heart disease, heart failure, atrial fibrillation, stroke, dementia/cognitive impairment, depression, seizure disorder, peripheral vascular disease, venous thromboembolism, chronic obstructive pulmonary disease, chronic kidney disease (stage ≥3), cancers (with the exception of skin cancer), cellulitis, falls, hip fracture, anemia, and peptic ulcer disease. Medications included non-steroidal anti-inflammatory drugs, aspirin, angiotensin-converting-enzyme inhibitors, beta-blockers, calcium channel blockers, loop diuretics, hydrochlorothiazide, thiazide-like diuretics, potassium-sparing diuretics, cholesterol-lowering drugs, insulin/oral hypoglycemics, glucocorticoids, anticoagulant and antiplatelet. Healthcare utilization variables included the number of primary care visits and hospitalizations, as assessed over the two years prior to the index date Since the percentage of subjects with missing values of covariates (i.e. BMI, smoking status and alcohol consumption) that were used to generate propensity score was low, we excluded them from the analysis.
After calculation of propensity score using all of the listed covariates within each accrual time block, we identified a propensity score-matched subject who did not receive total knee arthroplasty (i.e., a subject in the comparison cohort) for each patient receiving his/her first total knee arthroplasty (i.e., a subject in the exposed cohort) using a greedy matching algorithm (23, 26). Specifically, the non-total knee arthroplasty subject whose propensity score was closest to that of a randomly selected total knee arthroplasty subject was chosen as a match to this total knee arthroplasty subject. This process was then repeated until the non- total knee arthroplasty subjects had been matched to all total knee arthroplasty subjects, or until we exhausted the list of total knee arthroplasty subjects for whom a matched non- total knee arthroplasty subject could be found (26).
Total Hip Arthroplasty and the Risk of Myocardial Infarction
We used the same approach to assemble the exposed and comparison cohorts for total hip arthroplasty. Only Read codes diagnosis indicating specific hip joints were used to define hip osteoarthritis (N053512: Hip osteoarthritis NOS; N05z511: Hip osteoarthritis NOS; N05zJ00: Osteoarthritis NOS, of hip). Read codes were used to identify patients undergoing total hip arthroplasty and have been shown to be valid by previous studies using electronic medical record database in United Kingdom (20, 21). Further, patients with a history of hip fracture prior to the index date were excluded from the total hip arthroplasty analysis.
Use of Venous Thromboembolism as a Positive Control Outcome
The association between total joint arthroplasty and venous thromboembolism is well established (27); thus, we used the same approach to additionally construct a positive control cohort that examines the effect of total joint arthroplasty on the risk of incident venous thromboembolism. Patients with a history of venous thromboembolism were excluded from this analysis. In addition, when calculating the propensity score for this analysis, myocardial infarction was added as an additional covariate.
Assessment of Outcomes
A patient was considered to have had an incident myocardial infarction at the first recording of any Read term synonymous with following diagnoses: myocardial infarction, heart attack, and any code pertaining to a specific anatomical site of the myocardium or specified pattern on an electrocardiogram(28–31). This approach has been previously validated by investigators with a confirmation rate of over 90% (30, 32). A patient was considered to have developed venous thromboembolism when he or she had a recorded diagnosis of pulmonary embolism or deep vein thrombosis and received anticoagulant therapy (e.g., heparin, warfarin, or a similar agent) (33). This venous thromboembolism definition has been found to have a confirmation rate of 94% in the UK General Practice Research Database (33), and other studies have also confirmed the validity of the diagnosis of venous thromboembolism in the same database (34).
Statistical Analysis
Total knee arthroplasty and matched comparison cohorts began accruing risk time from the index date until they experienced myocardial infarction, left the THIN database, died, or the follow-up ended (December 31, 2012), whichever came first. Incidence rates for each cohort were calculated by dividing the number of cases of myocardial infarction by the total person-years of follow-up. We plotted a cumulative incidence curve adjusting for the competing risk of death (35).
Time-stratified Cox proportional hazard models were used to estimate the hazard ratio of total knee arthroplasty for the risk of myocardial infarction and its 95% confidence intervals. In addition, we assessed whether the risk ratio varied according to the length of follow-up (i.e., ≤1 month, ≤3 months, ≤6 months, ≤1 year, ≤2 years, and ≤3 years). As a sensitivity analysis, we used symmetric trimming with the cut point at the 5th percentiles and 95th percentiles of the propensity score matching.(36) Furthermore, we conducted subgroup analyses stratified by sex, age group (<75 years vs. ≥75 years), the Townsend Deprivation Index score (i.e., ≤2 vs. > 2), and history of cardiovascular disease (i.e., ischemic heart disease, hypertension, or stroke). We used the same methods to examine the relation of total hip arthroplasty to the risk of myocardial infarction. Finally, we repeated same analysis in our positive control cohorts to examine the relation of total joint arthroplasty to the risk of venous thromboembolism. All statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC, USA).
RESULTS
Our primary analysis included 13,849 patients who received total knee arthroplasty and 13,849 propensity score matched controls, as well as 6,063 patients who received total hip arthroplasty and 6,063 matched controls. These matching rates represented 89% of 15,566 cases of total knee arthroplasty and 84% of 7,215 cases of total hip arthroplasty. The baseline characteristics were well balanced between the matched comparison cohorts (Table 1).
Table 1.
Baseline Characteristics in the Total Knee Arthroplasty (TKA) and Total Hip Arthroplasty (THA) Propensity-Score Matched Cohorts among Knee and Hip Osteoarthritis Patients
| Baseline Characteristics | TKA Propensity Score-Matched Cohort
|
THA Propensity Score-Matched Cohort
|
||
|---|---|---|---|---|
| Incident TKA (n=13,849) | Non-TKA (n=13,849) | Incident THA (n=6,063) | Non-THA (n=6,063) | |
| Demographics | ||||
| Age, years | 70.8 | 70.7 | 70.7 | 70.8 |
| Male, % | 42.3 | 42.5 | 39.1 | 39.3 |
| Socio-Economic Deprivation Index | ||||
| Score * | 2.6 | 2.6 | 2.5 | 2.5 |
| Osteoarthritis duration, years | 5.3 | 5.2 | 3.0 | 3.1 |
| Lifestyle factors | ||||
| BMI, kg/m2 | 29.2 | 29.2 | 27.6 | 27.5 |
| Smoking | ||||
| Current, % | 7.9 | 7.8 | 10.6 | 9.8 |
| Past, % | 32.5 | 32.4 | 29.9 | 29.6 |
| None, % | 59.6 | 59.9 | 59.6 | 60.8 |
| Alcohol Use | ||||
| Current, % | 79.7 | 80.5 | 81.2 | 81.1 |
| Past, % | 2.3 | 2.0 | 2.0 | 2.3 |
| None, % | 18.0 | 17.5 | 16.8 | 16.6 |
| Comorbidities | ||||
| Hypertension, % | 56.1 | 55.9 | 52.4 | 53.3 |
| Stroke, % | 2.9 | 2.6 | 2.6 | 2.8 |
| Ischemic Heart Disease, % | 11.5 | 11.0 | 9.7 | 9.4 |
| Atrial fibrillation, % | 5.7 | 5.5 | 5.6 | 5.1 |
| Diabetes, % | 12.6 | 12.4 | 9.5 | 8.7 |
| Hyperlipidemia, % | 17.9 | 18.1 | 15.0 | 14.8 |
| COPD, % | 3.7 | 3.8 | 3.8 | 3.4 |
| Chronic Kidney Disease (≥ stage 3), % | 8.9 | 9.1 | 8.3 | 8.2 |
| Cancer, % | 12.4 | 12.0 | 11.9 | 12.7 |
| Depression, % | 12.2 | 12.0 | 11.2 | 11.1 |
| Cellulitis, % | 15.2 | 14.9 | 12.7 | 12.6 |
| Venous Thromboembolism, % | 4.8 | 4.9 | 4.4 | 4.1 |
| Hip Fracture, % ** | 0.9 | 1.0 | - | - |
| Falls, % | 15.3 | 15.3 | 12.9 | 13.3 |
| Seizure, % | 0.4 | 0.4 | 0.4 | 0.5 |
| Dementia/Cognitive Impairment, % | 0.4 | 0.4 | 0.3 | 0.4 |
| No. of primary care visits | 13.6 | 13.3 | 12.8 | 13.0 |
| No. of hospitalizations | 0.9 | 0.8 | 0.9 | 0.9 |
| Medications | ||||
| NSAIDs, % | 55.1 | 56.3 | 55.1 | 58.7 |
| ACEI, % | 25.2 | 24.9 | 22.9 | 23.2 |
| Aspirin, % | 25.2 | 24.8 | 22.2 | 22.6 |
| Beta-blockers, % | 20.6 | 20.4 | 20.2 | 21.0 |
| CCBs, % | 26.9 | 27.0 | 23.5 | 23.5 |
| Loop diuretics, % | 10.2 | 9.9 | 9.6 | 9.3 |
| Thiazide, % | 25.9 | 26.3 | 24.8 | 25.2 |
| Thiazide-like diuretics, % | 3.3 | 3.6 | 2.5 | 3.1 |
| Potassium-sparing diuretics, % | 3.3 | 3.2 | 2.6 | 2.5 |
| Anti-diabetic medicines, % | 8.3 | 8.2 | 6.0 | 5.6 |
| Lipid-lowering, % | 33.9 | 33.7 | 28.7 | 28.6 |
| Glucocorticoids, % | 5.9 | 6.0 | 5.7 | 5.5 |
| Anticoagulant, % | 4.6 | 4.5 | 4.6 | 3.8 |
| Antiplatelet, % | 26.1 | 25.8 | 23.3 | 23.7 |
Socio-Economic Deprivation Index Score was measured by the Townsend Deprivation Index, which was grouped into quintiles from 1 (least deprived) to 5 (most deprived).
Patients with a history of hip fracture were excluded from the Total Hip Arthroplasty cohort
Over the follow-up period (median 4.2 years), 306 and 286 cases of myocardial infarction occurred in patients who underwent total knee arthroplasty and in those who did not, respectively. Accordingly, 128 and 138 cases of myocardial infarction occurred in patients who underwent total hip arthroplasty and in those who did not, respectively. As shown in Figure 1A, the cumulative incidence of myocardial infarction was higher among patients who underwent total knee arthroplasty during the first 12 months, after which it declined to a lower level than that of those who did not undergo the surgery during the rest of the follow up period. Compared with those who did not undergo total knee arthroplasty, the hazard ratio of myocardial infarction for patients who underwent the surgery was 8.75 (95% confidence interval: 3.11 to 24.62) during the first month after surgery and then attenuated with increasing follow-up time (Table 2). The hazard ratio of myocardial infarction for total knee arthroplasty became insignificant when follow-up was longer than 6 months, and was 0.98 (95% confidence interval: 0.81 to 1.18) over the entire follow-up period (Table 2). After we excluded subjects whose propensity score was either below the 5th or above 95th percentile, the results remained similar (First month HR 8.50, 95% CI 3.02 to 23.95, Overall HR 1.04. 95% CI 0.85 to 1.26).
Figure 1.
Time to Myocardial Infarction for the Propensity Score Matched Cohorts among (A) Total Knee Arthroplasty (TKA) patients and (B) Total Hip Arthroplasty (THA) patients adjusting for competing risk of death.
Table 2.
Hazard Ratios for Myocardial Infarction Associated with Total Knee Arthroplasty (TKA) and Total Hip Arthroplasty (THA) according to the Follow-up Period
| TKA Propensity Score-Matched Cohort
|
THA Propensity Score-Matched Cohort
|
|||||
|---|---|---|---|---|---|---|
| Incident TKA (n=13,849) | Non-TKA (n=13,849) | Hazard Ratio (95% CI) | Incident TKA (n=6,063) | Non-TKA (n=6,063) | Hazard Ratio (95% CI) | |
| Myocardial Infarction Cases (n) | ||||||
| Total follow-up | 306 | 286 | 0.98 (0.82 to 1.18) | 128 | 138 | 0.87 (0.66 to 1.15) |
| 1 month follow-up | 35 | 4 | 8.75 (3.11 to 24.62) | 13 | 3 | 4.33 (1.24 to 15.21) |
| 3 month follow-up | 40 | 9 | 4.44 (2.16 to 9.16) | 15 | 10 | 1.50 (0.74 to 3.34) |
| 6 month follow-up | 50 | 29 | 1.85 (1.16 to 2.96) | 18 | 15 | 1.20 (0.61 to 2.38) |
| 1 year follow-up | 80 | 63 | 1.30 (0.93 to 1.82) | 30 | 28 | 1.00 (0.57 to 1.76) |
| 3 year follow-up | 160 | 159 | 1.01 (0.80 to 1.28) | 69 | 71 | 0.93 (0.66 to 1.31) |
| 5 year follow-up | 232 | 222 | 1.00 (0.82 to 1.23) | 88 | 106 | 0.80 (0.59 to 1.09) |
A similar pattern of the risk of myocardial infarction was observed between patients who underwent total hip arthroplasty and their comparison cohort (Figure 1B). The cumulative incidence of myocardial infarction was higher among patients who underwent total hip arthroplasty during the first 6 months, after which it declined to a lower level than that of those who did not undergo the surgery during the rest of the follow up period. The hazard ratio of myocardial infarction was 4.33 (95% confidence interval: 1.24 to 15.21) during the first month after surgery, then attenuated with increasing follow-up time (Table 2) and became insignificant when follow-up was longer than 1 month. The hazard ratio of myocardial infarction for subjects over the entire follow-up period was 0.87 (95% confidence interval: 0.66 to 1.15) (Table 2). After we excluded subjects whose propensity score was either below the 5th or above 95th percentile, the results remained similar (First month HR 5.00, 95% CI 1.10 to 22.81, Overall HR 0.86. 95% CI 0.64 to 1.15).
The relation of total knee arthroplasty or total hip arthroplasty to the risk of myocardial infarction did not vary materially according to sex, age group (<75 and ≥75 years), osteoarthritis duration, the Townsend Deprivation Index Score, and history of cardiovascular disease (Table 3).
Table 3.
Association between Total Knee Arthroplasty (TKA) and Total Hip Arthroplasty (THA) and Risk of Myocardial Infarction according to Various Risk Factors
| TKA Hazard Ratio (95% CI) | THA Hazard Ratio (95% CI) | |
|---|---|---|
| All patients | 0.98 (0.82 to 1.18) | 0.87 (0.66 to 1.15) |
| By age, years | ||
| 50–74 | 1.00 (0.73 to 1.36) | 0.95 (0.60 to 1.50) |
| 75–90 | 1.07 (0.71 to 1.62) | 0.91 (0.50 to 1.67) |
| By sex, | ||
| Male | 0.90 (0.61 to 1.31) | 1.00 (0.54 to 1.86) |
| Female | 1.03 (0.73 to 1.46) | 0.81 (0.49 to 1.31) |
| By Socio-Economic Deprivation Index Score | ||
| ≤ 2 | 0.98 (0.68 to 1.42) | 0.93 (0.55 to 1.58) |
| > 2 | 1,02 (0.72 to 1.44) | 1.09 (0.61 to 1.95) |
| By osteoarthritis duration, years | ||
| ≤ 3.5 | 1.16 (0.81 to 1.65) | 0.77 (0.55 to 1.10) |
| > 3.5 | 0.81 (0.59 to 1.11) | 1.39 (0.68 to 2.83) |
| By history of hypertension, stroke, or ischemic heart disease, | ||
| No | 0.94 (0.58 to 1.53) | 0.74 (0.37 to 1.47) |
| Yes | 0.94 (0.70 to 1.25) | 0.90 (0.57 to 1.41) |
In our comparison analysis using venous thromboembolism as a positive control, the hazard ratios of venous thromboembolism among patients who underwent total knee arthroplasty were 63.00, 19.36, 10.31, 6.26, 2.98, 2.43, and 2.18 over 1 month, 3 months, 6 months, 1 year, 2 years, 3 years, and the entire follow up period, respectively (Table 4). Similar results were also observed when effect of total hip arthroplasty on the risk of venous thromboembolism was examined (Table 4).
Table 4.
Hazard Ratios for Venous Thromboembolism (Positive Control) associated with Total Knee Arthroplasty (TKA) and Total Hip Arthroplasty (THA) according to Follow-up Period
| TKA Propensity-Score Matched Cohort | THA Propensity-Score Matched Cohort | |||||
|---|---|---|---|---|---|---|
| Incident TKA (n=14,031) | Non-TKA (n=14,031) | Hazard Ratio (95% CI) | Incident TKA (n=6,768) | Non-TKA (n=6,768) | Hazard Ratio (95% CI) | |
| Venous Thromboembolism Cases (n) | ||||||
| Total follow-up | 598 | 296 | 2.18 (1.87 to 2.54) | 253 | 160 | 1.75 (1.41 to 2.18) |
| 1 month follow-up | 190 | 3 | 63.00 (20.14 to 197.10) | 78 | 1 | 78.00 (10.85 to 560.65) |
| 3 month follow-up | 273 | 15 | 19.36 (11.31 to 33.12) | 132 | 6 | 26.39 (10.81 to 64.48) |
| 6 month follow-up | 303 | 31 | 10.31 (7.04 to 15.09) | 149 | 15 | 11.38 (6.46 to 20.07) |
| 1 year follow-up | 337 | 56 | 6.26 (4.69 to 8.37) | 157 | 38 | 4.43 (3.07 to 6.39) |
| 3 year follow-up | 464 | 158 | 2.98 (2.47 to 3.59) | 193 | 86 | 2.44 (1.87 to 3.19) |
| 5 year follow-up | 528 | 226 | 2.43 (2.06 to 2.87) | 223 | 124 | 1.93 (1.53 to 2.45) |
DISCUSSION
In this large general practice cohort representative of the UK population, we found a substantially increased risk of myocardial infarction shortly after total joint arthroplasty. The risk ratio of myocardial infarction decreased afterwards and became insignificant 6 months after total knee joint arthroplasty and 1 month after total hip arthroplasty in our study. There was no apparent overall protective effect of either total knee arthroplasty or total hip arthroplasty on the risk of myocardial infarction over the entire follow up period. Such findings also persisted in subgroup analyses stratified by sex, age group, social deprivation index, and history of cardiovascular disease.
Contradictory to recently-published findings (8), our study indicates that total joint arthroplasty surgical procedures do not provide an overall protective effect on the risk of myocardial infarction (8). The major difference between the previous study and ours is that cardiovascular events occurring shortly after total joint arthroplasty were excluded from the previous study (8). Since the risk of myocardial infarction increased substantially shortly after total joint arthroplasty (11, 37, 38), excluding cases that occurred shortly after surgery would have led to the differential selection of less susceptible individuals to myocardial infarction over time, due to the differential depletion of susceptibles. Consequently, analyses based on individuals who were free of myocardial infarction after a certain period of time after total joint arthroplasty provide a spurious inverse association with the risk of cardiovascular events that can be entirely explained by this “built-in” selection bias (8). For example, a study conducted using the Danish national registry found that despite a 9-fold increased risk of myocardial infarction within 6 weeks after total knee arthroplasty, the risk ratio over the 6 to 52 week period after surgery dropped to 0.70 (95% confidence interval: 0.53 to 0.92) (11). Similarly, a study based on Medicare enrollees reported an adjusted hazard ratio for mortality over a period of 3 months to 5 years after total hip arthroplasty of 0.65 (95% confidence interval: 0.63 to 0.68) (39). Finally, when we repeated our analyses by starting the follow-up time at 1 month after the surgical procedure, the hazard ratios for myocardial infarction were 0.85 for total knee arthroplasty and 0.77 for total hip arthroplasty. The caveat is that these period-specific inverse associations can all be explained by the aforementioned built-in selection bias and thus, not by a biologic benefit of total joint arthroplasty.
This “built-in” selection bias is well-recognized in epidemiology research of drugs or interventions (i.e., pharmacoepidemiology). The most notable example has been that of the impact of hormonal replacement therapy (HRT) on cardiovascular risk among postmenopausal women. Many prior observational studies that investigated the effects of prevalent HRT use (or mixture of prevalent and incident uses) had shown a seemingly inverse association with the risk of cardiovascular events, which led to the widespread use of HRT in clinical practice. However, the Women’s Health Initiative (WHI) Trial has shown that HRT use increases the cardiovascular risk (40). Subsequently, it was demonstrated that when observational cohort studies were analyzed such that they emulated the design of an RCT using incident exposure (i.e., initiators of HRT), the results converged with that of the WHI trial (41). The same analytic principle and caveat should apply to this research context for the impact of total joint arthroplasty.
The potential underlying biological mechanism behind the increased risk of postoperative myocardial infarction includes cardiac or hemodynamic stressors associated with surgery (e.g., the effects of anesthesia on the cardiovascular system, blood loss, fluid shifts, arrhythmias, and hypoxia) as well as fat embolization (particularly after total hip arthroplasty). It remains unclear whether the perioperative management of antithrombotic agents (i.e., discontinuation of low-dose aspirin and initiation of inpatient anticoagulation prophylaxis) can explain the increased risk of myocardial infarction after total joint arthroplasty surgery in osteoarthritis patients. Regardless, our findings suggest that the immediate postoperative risk of myocardial infarction following total joint arthroplasty may have been previously underappreciated, and further measures to prevent this serious event may need to be considered. In contrast, improvements in patients’ pain, function, and health-related quality of life, as well as the reduced use of non-steroidal anti-inflammatory drugs after total joint arthroplasty (4–7) might lead to potential long-term benefits on cardiovascular outcomes; however, documenting such benefits are challenging, as discussed above.
Our study has several strengths and limitations. First, our study was based on a large electronic medical record database representative of the general population; thus, findings from our study are likely to be more generalizable. Further, the lack of data on important risk factors of myocardial infarction such as BMI and cardiovascular medicines have often been cited as major limitations by previous studies (11, 42); conversely, these covariates were available and adjusted for in our analyses. Uncertainty surrounding diagnostic accuracy is a potential concern in studies that identify cases from administrative databases. However, our databases are actually electronic medical records used for patients’ care and thus, the overall accuracy is expected to be higher as reflected by many validation studies of important disease outcomes (32, 43–49). The use of venous thromboembolism as a positive control (27, 50) further confirmed the validity of our study design and methodology. Moreover, considering secular trend of myocardial infarction and risk factors in UK population matching within 1-year time blocks flexibly accounted for changes in the relative importance of confounding variables at different calendar times. Nevertheless, as in other observational studies, we cannot rule out the possibility of residual or unmeasured confounders. For example, although we have adjusted for OA duration and drug use reflect OA severity to certain extent, our analyses were not adjusted for direct OA severity measures, physical activity level, and function, as these variables have not been consistently collected in THIN. On the other hand, it is perhaps reasonable to speculate that the surgeon would preferentially select osteoarthritis patients for total joint arthroplasty, i.e., individuals who are frail or at a high risk of myocardial infarction may be less likely to receive total joint arthroplasty than those who have better general health status. To that effect, the risk of myocardial infarction associated with total joint arthroplasty may have been underestimated in both our study as well as in previous studies. Excluding those with conditions that can seriously influence the decision for THA and those with missing covariates may limit the generalizability; however, our results including those with exclusion criteria did not change materially.
In conclusion, our general population-based study indicates that total knee arthroplasty and total hip arthroplasty are associated with a substantially increased risk of myocardial infarction during the immediate postoperative period among osteoarthritis patients. However, there was no overall long-term impact on the risk of myocardial infarction, unlike the increased risk of venous thromboembolism that remains even years after the procedure.
Figure 2.
Time to venous thromboembolism (Positive Control) for the Propensity Score Matched Cohorts among (A) Total Knee Arthroplasty (TKA) patients and (B) Total Hip Arthroplasty (THA) patients adjusting for competing risk of death.
Acknowledgments
Funding: This work was supported in part by grants from NIAMS (P60AR047785, R01AR056291, and R01AR065944).
Footnotes
Disclosure: The authors have no financial or academic interests to report that might be regarded as potential conflicts of interest in relation to this manuscript.
References
- 1.Cram P, Lu X, Kates SL, Singh JA, Li Y, Wolf BR. Total knee arthroplasty volume, utilization, and outcomes among Medicare beneficiaries, 1991–2010. JAMA : the journal of the American Medical Association. 2012;308(12):1227–36. doi: 10.1001/2012.jama.11153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Healthcare Cost and Utilization Project 2009. Rockvile, MD: Agency for Healthcare Research and Quality; 2009. [PubMed] [Google Scholar]
- 3.Report: Hip and Knee Replacement Market: Product Code: DMHC2264. London, England: Datamonitor Healthcare; 2006. [Google Scholar]
- 4.Harris WH, Sledge CB. Total hip and total knee replacement (1) The New England journal of medicine. 1990;323(11):725–31. doi: 10.1056/NEJM199009133231106. [DOI] [PubMed] [Google Scholar]
- 5.Harris WH, Sledge CB. Total hip and total knee replacement (2) The New England journal of medicine. 1990;323(12):801–7. doi: 10.1056/NEJM199009203231206. [DOI] [PubMed] [Google Scholar]
- 6.Ethgen O, Bruyere O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature. The Journal of bone and joint surgery American volume. 2004;86-A(5):963–74. doi: 10.2106/00004623-200405000-00012. [DOI] [PubMed] [Google Scholar]
- 7.Bruyere O, Ethgen O, Neuprez A, Zegels B, Gillet P, Huskin JP, et al. Health-related quality of life after total knee or hip replacement for osteoarthritis: a 7-year prospective study. Archives of orthopaedic and trauma surgery. 2012;132(11):1583–7. doi: 10.1007/s00402-012-1583-7. [DOI] [PubMed] [Google Scholar]
- 8.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: 10.1136/bmj.f6187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 1983;1(11):710–9. doi: 10.1200/JCO.1983.1.11.710. [DOI] [PubMed] [Google Scholar]
- 10.Dafni U. Landmark analysis at the 25-year landmark point. Circulation Cardiovascular quality and outcomes. 2011;4(3):363–71. doi: 10.1161/CIRCOUTCOMES.110.957951. [DOI] [PubMed] [Google Scholar]
- 11.Lalmohamed A, Vestergaard P, Klop C, Grove EL, de Boer A, Leufkens HG, et al. Timing of acute myocardial infarction in patients undergoing total hip or knee replacement: a nationwide cohort study. Archives of internal medicine. 2012;172(16):1229–35. doi: 10.1001/archinternmed.2012.2713. [DOI] [PubMed] [Google Scholar]
- 12.Bourke A, Dattani H, Robinson M. Feasibility study and methodology to create a quality-evaluated database of primary care data. Informatics in primary care. 2004;12(3):171–7. doi: 10.14236/jhi.v12i3.124. [DOI] [PubMed] [Google Scholar]
- 13.National_Health_Service(NHS) Read Codes - NHS Connecting for Health. [cited 2012 April 10]; Available from: http://www.connectingforhealth.nhs.uk/systemsandservices/data/uktc/readcodes.
- 14.FirstDataBank. Drug Databases - Multilex Drug Data File. [cited 2012 April 10]; Available from: http://www.firstdatabank.co.uk/8/multilex-drug-data-file.
- 15.van der Waal JM, Bot SD, Terwee CB, van der Windt DA, Schellevis FG, Bouter LM, et al. The incidences of and consultation rate for lower extremity complaints in general practice. Annals of the rheumatic diseases. 2006;65(6):809–15. doi: 10.1136/ard.2005.036988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Sacks JJ, Luo YH, Helmick CG. Prevalence of specific types of arthritis and other rheumatic conditions in the ambulatory health care system in the United States, 2001–2005. Arthritis care & research. 2010;62(4):460–4. doi: 10.1002/acr.20041. [DOI] [PubMed] [Google Scholar]
- 17.Lix L, Yogendran M, Burchill C, Metge C, McKeen N, Moore D, et al. Defining and Validating Chronic Diseases: An Administrative Data Approach. Winnipeg: Manitoba Centre for Health Policy; Jul, 2006. [Google Scholar]
- 18.Lix L, Yogendran M, Mann J. Defining and validating chronic diseases: an administrative data approach-an update with ICD-10-CA. 2008 Nov; Available at: http://umanitoba.ca/faculties/medicine/units/mchp/projects/media/ICD10_Final.pdf.
- 19.Prieto-Alhambra D, Judge A, Javaid MK, Cooper C, Diez-Perez A, Arden NK. Incidence and risk factors for clinically diagnosed knee, hip and hand osteoarthritis: influences of age, gender and osteoarthritis affecting other joints. Annals of the rheumatic diseases. 2014;73(9):1659–64. doi: 10.1136/annrheumdis-2013-203355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Prieto-Alhambra D, Javaid MK, Judge A, Maskell J, Cooper C, Arden NK. Hormone replacement therapy and mid-term implant survival following knee or hip arthroplasty for osteoarthritis: a population-based cohort study. Annals of the rheumatic diseases. 2015;74(3):557–63. doi: 10.1136/annrheumdis-2013-204043. [DOI] [PubMed] [Google Scholar]
- 21.Culliford DJ, Maskell J, Beard DJ, Murray DW, Price AJ, Arden NK. Temporal trends in hip and knee replacement in the United Kingdom: 1991 to 2006. The Journal of bone and joint surgery British volume. 2010;92(1):130–5. doi: 10.1302/0301-620X.92B1.22654. [DOI] [PubMed] [Google Scholar]
- 22.Cepeda MS, Boston R, Farrar JT, Strom BL. Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders. American journal of epidemiology. 2003;158(3):280–7. doi: 10.1093/aje/kwg115. [DOI] [PubMed] [Google Scholar]
- 23.Seeger JD, Williams PL, Walker AM. An application of propensity score matching using claims data. Pharmacoepidemiology and drug safety. 2005;14(7):465–76. doi: 10.1002/pds.1062. [DOI] [PubMed] [Google Scholar]
- 24.Morris R, Carstairs V. Which deprivation? A comparison of selected deprivation indexes. Journal of public health medicine. 1991;13(4):318–26. [PubMed] [Google Scholar]
- 25.Townsend P, Phillimore P, Beattie A. Health and deprivation : inequality and the North. London ; New York: Croom Helm; 1988. [Google Scholar]
- 26.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate behavioral research. 2011;46(3):399–424. doi: 10.1080/00273171.2011.568786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR, et al. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition) Chest. 2008;133(6 Suppl):381S–453S. doi: 10.1378/chest.08-0656. [DOI] [PubMed] [Google Scholar]
- 28.Hammad TA, McAdams MA, Feight A, Iyasu S, Dal Pan GJ. Determining the predictive value of Read/OXMIS codes to identify incident acute myocardial infarction in the General Practice Research Database. Pharmacoepidemiology and drug safety. 2008;17(12):1197–201. doi: 10.1002/pds.1672. [DOI] [PubMed] [Google Scholar]
- 29.Hernandez-Diaz S, Varas-Lorenzo C, Garcia Rodriguez LA. Non-steroidal antiinflammatory drugs and the risk of acute myocardial infarction. Basic & clinical pharmacology & toxicology. 2006;98(3):266–74. doi: 10.1111/j.1742-7843.2006.pto_302.x. [DOI] [PubMed] [Google Scholar]
- 30.Garcia Rodriguez LA, Tacconelli S, Patrignani P. Role of dose potency in the prediction of risk of myocardial infarction associated with nonsteroidal anti-inflammatory drugs in the general population. Journal of the American College of Cardiology. 2008;52(20):1628–36. doi: 10.1016/j.jacc.2008.08.041. [DOI] [PubMed] [Google Scholar]
- 31.Ruigomez A, Martin-Merino E, Rodriguez LA. Validation of ischemic cerebrovascular diagnoses in the health improvement network (THIN) Pharmacoepidemiology and drug safety. 2010;19(6):579–85. doi: 10.1002/pds.1919. [DOI] [PubMed] [Google Scholar]
- 32.Garcia Rodriguez LA, Varas-Lorenzo C, Maguire A, Gonzalez-Perez A. Nonsteroidal antiinflammatory drugs and the risk of myocardial infarction in the general population. Circulation. 2004;109(24):3000–6. doi: 10.1161/01.CIR.0000132491.96623.04. [DOI] [PubMed] [Google Scholar]
- 33.Huerta C, Johansson S, Wallander MA, Garcia Rodriguez LA. Risk factors and short-term mortality of venous thromboembolism diagnosed in the primary care setting in the United Kingdom. Archives of internal medicine. 2007;167(9):935–43. doi: 10.1001/archinte.167.9.935. [DOI] [PubMed] [Google Scholar]
- 34.Lawrenson R, Todd JC, Leydon GM, Williams TJ, Farmer RD. Validation of the diagnosis of venous thromboembolism in general practice database studies. British journal of clinical pharmacology. 2000;49(6):591–6. doi: 10.1046/j.1365-2125.2000.00199.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. Journal of the American Statistical Association. 1999;94:496–509. [Google Scholar]
- 36.Sturmer T, Rothman KJ, Avorn J, Glynn RJ. Treatment effects in the presence of unmeasured confounding: dealing with observations in the tails of the propensity score distribution--a simulation study. American journal of epidemiology. 2010;172(7):843–54. doi: 10.1093/aje/kwq198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Singh JA, Jensen MR, Harmsen WS, Gabriel SE, Lewallen DG. Cardiac and thromboembolic complications and mortality in patients undergoing total hip and total knee arthroplasty. Annals of the rheumatic diseases. 2011;70(12):2082–8. doi: 10.1136/ard.2010.148726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Parvizi J, Mui A, Purtill JJ, Sharkey PF, Hozack WJ, Rothman RH. Total joint arthroplasty: When do fatal or near-fatal complications occur? The Journal of bone and joint surgery American volume. 2007;89(1):27–32. doi: 10.2106/JBJS.E.01443. [DOI] [PubMed] [Google Scholar]
- 39.Barrett J, Losina E, Baron JA, Mahomed NN, Wright J, Katz JN. Survival following total hip replacement. The Journal of bone and joint surgery American volume. 2005;87(9):1965–71. doi: 10.2106/JBJS.D.02440. [DOI] [PubMed] [Google Scholar]
- 40.Manson JE, Hsia J, Johnson KC, Rossouw JE, Assaf AR, Lasser NL, et al. Estrogen plus progestin and the risk of coronary heart disease. The New England journal of medicine. 2003;349(6):523–34. doi: 10.1056/NEJMoa030808. [DOI] [PubMed] [Google Scholar]
- 41.Hernan MA, Alonso A, Logan R, Grodstein F, Michels KB, Willett WC, et al. Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease. Epidemiology. 2008;19(6):766–79. doi: 10.1097/EDE.0b013e3181875e61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Mantilla CB, Wass CT, Goodrich KA, Johanns CJ, Kool ML, Zhu X, et al. Risk for perioperative myocardial infarction and mortality in patients undergoing hip or knee arthroplasty: the role of anemia. Transfusion. 2011;51(1):82–91. doi: 10.1111/j.1537-2995.2010.02797.x. [DOI] [PubMed] [Google Scholar]
- 43.Seminara NM, Abuabara K, Shin DB, Langan SM, Kimmel SE, Margolis D, et al. Validity of The Health Improvement Network (THIN) for the study of psoriasis. The British journal of dermatology. 2011;164(3):602–9. doi: 10.1111/j.1365-2133.2010.10134.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Martin-Merino E, Garcia-Rodriguez LA, Masso-Gonzalez EL, Roehrborn CG. Do oral antimuscarinic drugs carry an increased risk of acute urinary retention? The Journal of urology. 2009;182(4):1442–8. doi: 10.1016/j.juro.2009.06.051. [DOI] [PubMed] [Google Scholar]
- 45.Meal A, Leonardi-Bee J, Smith C, Hubbard R, Bath-Hextall F. Validation of THIN data for non-melanoma skin cancer. Quality in primary care. 2008;16(1):49–52. [PubMed] [Google Scholar]
- 46.Martin-Merino E, Ruigomez A, Johansson S, Wallander MA, Garcia-Rodriguez LA. Study of a cohort of patients newly diagnosed with depression in general practice: prevalence, incidence, comorbidity, and treatment patterns. Primary care companion to the Journal of clinical psychiatry. 2010;12(1):PCC 08m00764. doi: 10.4088/PCC.08m00764blu. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Prieto-Alhambra D, Javaid MK, Judge A, Murray D, Carr A, Cooper C, et al. Association between bisphosphonate use and implant survival after primary total arthroplasty of the knee or hip: population based retrospective cohort study. BMJ. 2011;343:d7222. doi: 10.1136/bmj.d7222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Quint JK, Herrett E, Bhaskaran K, Timmis A, Hemingway H, Wedzicha JA, et al. Effect of beta blockers on mortality after myocardial infarction in adults with COPD: population based cohort study of UK electronic healthcare records. BMJ. 2013;347:f6650. doi: 10.1136/bmj.f6650. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Idris I, Warren G, Donnelly R. Association between thiazolidinedione treatment and risk of macular edema among patients with type 2 diabetes. Archives of internal medicine. 2012;172(13):1005–11. doi: 10.1001/archinternmed.2012.1938. [DOI] [PubMed] [Google Scholar]
- 50.Bjornara BT, Gudmundsen TE, Dahl OE. Frequency and timing of clinical venous thromboembolism after major joint surgery. The Journal of bone and joint surgery British volume. 2006;88(3):386–91. doi: 10.1302/0301-620X.88B3.17207. [DOI] [PubMed] [Google Scholar]