Predicting Infections After Total Joint Arthroplasty Using a Prescription Based Comorbidity Measure

Integrating Genome-Wide Polygenic Risk Scores With Nongenetic Models to Predict Surgical Site Infection After Total Knee Arthroplasty Using United Kingdom Biobank Data

BACKGROUND

Prediction of the risk of developing surgical site infection (SSI) in patients following total knee arthroplasty (TKA) is of clinical importance. Genetic susceptibility is involved in developing TKA-related SSI. Previously reported models for predicting SSI were constructed using nongenetic risk factors without incorporating genetic risk factors. To address this issue, we performed a genome-wide association study (GWAS) using the UK Biobank database.

METHODS

Adult patients who underwent primary TKA (n = 19,767) were analyzed and divided into SSI (n = 269) and non-SSI (n = 19,498) cohorts. Nongenetic covariates, including demographic data and preoperative comorbidities, were recorded. Genetic variants associated with SSI were identified by GWAS and included to obtain standardized polygenic risk scores (zPRS, an estimate of genetic risk). Prediction models were established through analyses of multivariable logistic regression and the receiver operating characteristic curve.

RESULTS

There were 4 variants (rs117896641, rs111686424, rs8101598, and rs74648298) achieving genome-wide significance that were identified. The logistic regression analysis revealed 7 significant risk factors: increasing zPRS, decreasing age, men, chronic obstructive pulmonary disease, diabetes mellitus, rheumatoid arthritis, and peripheral vascular disease. The areas under the receiver operating characteristic curve were 0.628 and 0.708 when zPRS (model 1) and nongenetic covariates (model 2) were used as predictors, respectively. The areas under the receiver operating characteristic curve increased to 0.76 when both zPRS and nongenetic covariates (model 3) were used as predictors. A risk-prediction nomogram was constructed based on model 3 to visualize the relative effect of statistically significant covariates on the risk of SSI and predict the probability of developing SSI. Age and zPRS were the top 2 covariates that contributed to the risk, with younger age and higher zPRS associated with higher risks.

CONCLUSIONS

Our GWAS identified 4 novel variants that were significantly associated with susceptibility to SSI following TKA. Integrating genome-wide zPRS with nongenetic risk factors improved the performance of the model in predicting SSI.

The Incidence and Risk Factors of Early Periprosthetic Joint Infections

Objective

Periprosthetic joint infections (PJI) represent major complications in arthroplasty, contributing to increased patient morbidity and imposing substantial financial burdens. Meticulous surveillance of PJI occurrences and identification of associated risk factors is imperative for accurately gauging the incidence rates and implementing proactive infection control measures. This study aimed to ascertain the early incidence of PJI and elucidate the key risk factors involved in its occurrence.

Materials and Methods

This monocentric, prospective descriptive study conducted between June 2018 and June 2019, including all patients aged 18 years and above who underwent hip and knee arthroplasty. The research documented and evaluated patient demographic characteristics, clinical findings, laboratory results, treatment practices, and potential risk factors associated with the surgical process. After the 90-day postoperative period, patients were categorized into PJI and non-PJI groups, allowing for a comprehensive comparison of identified risk factors.

Results

This study identified a cohort of 590 patients, of whom 185 underwent hip arthroplasties (31.4%) and 405 underwent knee arthroplasties (68.6%). The average age of the patients was 65.2 years, with females constituting 80.2% of the population. The overall incidence of early PJI was found to be 2.88% (n=17). Following hip arthroplasties, the PJI incidence was 4.86%, while knee arthroplasties exhibited a lower incidence of 1.9%. Several potential risk factors associated with PJI were identified, including comorbid diseases (adjusted odds ratio [aOR]=3.35, 95% confidence interval [CI]=1.18-9.47), preoperative length of stay (aOR=0.89, 95% CI=0.79-1.01), postoperative erythrocyte suspension replacement (aOR=1.96, 95% CI=0.71-5.44), and a National Nosocomial Infections Surveillance System (NNIS) score of 1 or higher (aOR=3.10, 95% CI=1.10-8.71). These factors were identified as potential contributors to the risk of PJI in patients undergoing hip and knee arthroplasties.

Conclusion

Compared to other reported outcomes in the literature, this study observed a higher incidence of early-stage PJI. The higher incidence may be due to PJI surveillance deficiencies such as difficulty in post-discharge surgical site infection (SSI) follow-up, reporting, and bacterial sampling. This discrepancy underscores the importance of actively monitoring patients with risk factors for PJI development, including medical comorbidities and a high NNIS score. Implementing prospective active surveillance in such cases is deemed crucial for the timely identification and management of PJI.

The Use of Artificial Intelligence for the Prediction of Periprosthetic Joint Infection Following Aseptic Revision Total Knee Arthroplasty

Periprosthetic joint infection (PJI) following revision total knee arthroplasty (TKA) for aseptic failure is associated with poor outcomes, patient morbidity, and high health care expenditures. The aim of this study was to develop novel machine learning algorithms for the prediction of PJI following revision TKA for patients with aseptic indications for revision surgery. A single-institution database consisting of 1,432 consecutive revision TKA patients with aseptic etiologies was retrospectively identified. The patient cohort included 208 patients (14.5%) who underwent re-revision surgery for PJI. Three machine learning algorithms (artificial neural networks, support vector machines, k-nearest neighbors) were developed to predict this outcome and these models were assessed by discrimination, calibration, and decision curve analysis. This is a retrospective study. Among the three machine learning models, the neural network model achieved the best performance across discrimination (area under the receiver operating characteristic curve = 0.78), calibration, and decision curve analysis. The strongest predictors for PJI following revision TKA for aseptic reasons were prior open procedure prior to revision surgery, drug abuse, obesity, and diabetes. This study utilized machine learning as a tool for the prediction of PJI following revision TKA for aseptic failure with excellent performance. The validated machine learning models can aid surgeons in patient-specific risk stratifying to assist in preoperative counseling and clinical decision making for patients undergoing aseptic revision TKA.

External Validation of Prediction Models for Surgical Complications in People Considering Total Hip or Knee Arthroplasty Was Successful for Delirium but Not for Surgical Site Infection, Postoperative Bleeding, and Nerve Damage: A Retrospective Cohort Study

Although several models for the prediction of surgical complications after primary total hip or total knee replacement (THA and TKA, respectively) are available, only a few models have been externally validated. The aim of this study was to externally validate four previously developed models for the prediction of surgical complications in people considering primary THA or TKA. We included 2614 patients who underwent primary THA or TKA in secondary care between 2017 and 2020. Individual predicted probabilities of the risk for surgical complication per outcome (i.e., surgical site infection, postoperative bleeding, delirium, and nerve damage) were calculated for each model. The discriminative performance of patients with and without the outcome was assessed with the area under the receiver operating characteristic curve (AUC), and predictive performance was assessed with calibration plots. The predicted risk for all models varied between <0.01 and 33.5%. Good discriminative performance was found for the model for delirium with an AUC of 84% (95% CI of 0.82-0.87). For all other outcomes, poor discriminative performance was found; 55% (95% CI of 0.52-0.58) for the model for surgical site infection, 61% (95% CI of 0.59-0.64) for the model for postoperative bleeding, and 57% (95% CI of 0.53-0.61) for the model for nerve damage. Calibration of the model for delirium was moderate, resulting in an underestimation of the actual probability between 2 and 6%, and exceeding 8%. Calibration of all other models was poor. Our external validation of four internally validated prediction models for surgical complications after THA and TKA demonstrated a lack of predictive accuracy when applied in another Dutch hospital population, with the exception of the model for delirium. This model included age, the presence of a heart disease, and the presence of a disease of the central nervous system as predictor variables. We recommend that clinicians use this simple and straightforward delirium model during preoperative counselling, shared decision-making, and early delirium precautionary interventions.

Prediction of Early Periprosthetic Joint Infection After Total Hip Arthroplasty

Purpose

To develop a parsimonious risk prediction model for periprosthetic joint infection (PJI) within 90 days after total hip arthroplasty (THA).

Patients and Methods

We used logistic LASSO regression with bootstrap ranking to develop a risk prediction model for PJI within 90 days based on a Swedish cohort of 88,830 patients with elective THA 2008–2015. The model was externally validated on a Danish cohort with 18,854 patients.

Results

Incidence of PJI was 2.45% in Sweden and 2.17% in Denmark. A model with the underlying diagnosis for THA, body mass index (BMI), American Society for Anesthesiologists (ASA) class, sex, age, and the presence of five defined comorbidities had an area under the curve (AUC) of 0.68 (95% CI: 0.66 to 0.69) in Sweden and 0.66 (95% CI: 0.64 to 0.69) in Denmark. This was superior to traditional models based on ASA class, Charlson, Elixhauser, or the Rx Risk V comorbidity indices. Internal calibration was good for predicted probabilities up to 10%.

Conclusion

A new PJI prediction model based on easily accessible data available before THA was developed and externally validated. The model had superior discriminatory ability compared to ASA class alone or more complex comorbidity indices and had good calibration. We provide a web-based calculator (https://erikbulow.shinyapps.io/thamortpred/) to facilitate shared decision making by patients and surgeons.

Point your SmartPhone at the code above. If you have a QR code reader the video abstract will appear. Or use:

https://youtu.be/T0qfHTvBEs4

The Use of Artificial Neural Networks for the Prediction of Surgical Site Infection Following TKA

This is a retrospective study. Surgical site infection (SSI) is associated with adverse postoperative outcomes following total knee arthroplasty (TKA). However, accurately predicting SSI remains a clinical challenge due to the multitude of patient and surgical factors associated with SSI. This study aimed to develop and validate machine learning models for the prediction of SSI following primary TKA. This is a retrospective study for patients who underwent primary TKA. Chart review was performed to identify patients with superficial or deep SSIs, defined in concordance with the criteria of the Musculoskeletal Infection Society. All patients had a minimum follow-up of 2 years (range: 2.1-4.7 years). Five machine learning algorithms were developed to predict this outcome, and model assessment was performed by discrimination, calibration, and decision curve analysis. A total of 10,021 consecutive primary TKA patients was included in this study. At an average follow-up of 2.8 ± 1.1 years, SSIs were reported in 404 (4.0%) TKA patients, including 223 superficial SSIs and 181 deep SSIs. The neural network model achieved the best performance across discrimination (area under the receiver operating characteristic curve = 0.84), calibration, and decision curve analysis. The strongest predictors of the occurrence of SSI following primary TKA, in order, were Charlson comorbidity index, obesity (BMI >30 kg/m²), and smoking. The neural network model presented in this study represents an accurate method to predict patient-specific superficial and deep SSIs following primary TKA, which may be employed to assist in clinical decision-making to optimize outcomes in at-risk patients.

Does a Prescription-based Comorbidity Index Correlate with the American Society of Anesthesiologists Physical Status Score and Mortality After Joint Arthroplasty? A Registry Study

BACKGROUND

When analyzing the outcomes of joint arthroplasty, an important factor to consider is patient comorbidities. The presence of multiple comorbidities has been associated with longer hospital stays, more postoperative complications, and increased mortality. The American Society of Anesthesiologists (ASA) physical status classification system score is a measure of a patient's overall health and has been shown to be associated with complications and mortality after joint arthroplasty. The Rx-Risk score is another measure for determining the number of different health conditions for which an individual is treated, with a possible score ranging from 0 to 47.

QUESTIONS/PURPOSES

For patients undergoing THA or TKA, we asked: (1) Which metric, the Rx-Risk score or the ASA score, correlates more closely with 30- and 90-day mortality after TKA or THA? (2) Is the Rx-Risk score correlated with the ASA score?

METHODS

This was a retrospective analysis of the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR) database linked to two other national databases, the National Death Index (NDI) database and the Pharmaceutical Benefits Scheme (PBS), a dispensing database. Linkage to the NDI provided outcome information on patient death, including the fact of and date of death. Linkage to the PBS was performed to obtain records of all medicines dispensed to patients undergoing a joint replacement procedure. Patients were included if they had undergone either a THA (119,076 patients, 131,336 procedures) or TKA (182,445 patients, 215,712 procedures) with a primary diagnosis of osteoarthritis, performed between 2013 and 2017. We excluded patients with missing ASA information (THA: 3% [3055 of 119,076]; TKA: 2% [4095 of 182,445]). This left 127,761 primary THA procedures performed in 116,021 patients (53% [68,037 of 127,761] were women, mean age 68 ± 11 years) and 210,501 TKA procedures performed in 178,350 patients (56% [117,337 of 210,501] were women, mean age 68 ± 9 years) included in this study. Logistic regression models were used to determine the concordance of the ASA and Rx-Risk scores and 30-day and 90-day postoperative mortality. The Spearman correlation coefficient (r) was used to estimate the correlation between the ASA score and Rx-Risk score. All analyses were performed separately for THAs and TKAs.

RESULTS

We found both the ASA and Rx-Risk scores had high concordance with 30-day mortality after THA (ASA: c-statistic 0.83 [95% CI 0.79 to 0.86]; Rx-Risk: c-statistic 0.82 [95% CI 0.79 to 0.86]) and TKA (ASA: c-statistic 0.73 [95% CI 0.69 to 0.78]; Rx-Risk: c-statistic 0.74 [95% CI 0.70 to 0.79]). Although both scores were strongly associated with death, their correlation was moderate for patients undergoing THA (r = 0.45) and weak for TKA (r = 0.38). However, the median Rx-Risk score did increase with increasing ASA score. For example, for THAs, the median Rx-Risk score was 1, 3, 5, and 7 for ASA scores 1, 2, 3, and 4, respectively. For TKAs, the median Rx-Risk score was 2, 4, 5, and 7 for ASA scores 1, 2, 3, and 4, respectively.

CONCLUSION

The ASA physical status and RxRisk were associated with 30-day and 90-day mortality; however, the scores were only weakly to moderately correlated with each other. This suggests that although both scores capture a similar level of patient illness, each score may be capturing different aspects of health. The Rx-Risk may be used as a complementary measure to the ASA score.

LEVEL OF EVIDENCE

Level III, therapeutic study.

Comorbidity indices in orthopaedic surgery: a narrative review focused on hip and knee arthroplasty

Comorbidity indices currently used to estimate negative postoperative outcomes in orthopaedic surgery were originally developed among non-orthopaedic patient populations.

While current indices were initially intended to predict short-term mortality, they have since been used for other purposes as well.

As the rate of hip and knee arthroplasty steadily rises, understanding the magnitude of the effect of comorbid disease on postoperative outcomes has become increasingly more important.

Currently, the ASA classification is the most commonly used comorbidity measure and is systematically recorded by the majority of national arthroplasty registries.

Consideration should be given to developing an updated, standardized approach for comorbidity assessment and reporting in orthopaedic surgery, especially within the setting of elective hip and knee arthroplasty.

Cite this article: EFORT Open Rev 2021;6:629-640. DOI: 10.1302/2058-5241.6.200124

A narrative review of using prescription drug databases for comorbidity adjustment: A less effective remedy or a prescription for improved model fit?

BACKGROUND

The use of claims data for identifying comorbid conditions in patients for research purposes has been widely explored. Traditional measures of comorbid adjustment included diagnostic data (e.g., ICD-9-CM or ICD-10-CM codes), with the Charlson and Elixhauser methodology being the two most common approaches. Prescription data has also been explored for use in comorbidity adjustment, however early methodologies were disappointing when compared to diagnostic measures.

OBJECTIVE

The objective of this methodological review is to compare results from newer studies using prescription-based data with more traditional diagnostic measures.

METHODS

A review of studies found on PubMed, Medline, Embase or CINAHL published between January 1990 and December 2020 using prescription data for comorbidity adjustment. A total of 50 studies using prescription drug measures for comorbidity adjustment were found.

CONCLUSIONS

Newer prescription-based measures show promise fitting models, as measured by predictive ability, for research, especially when the primary outcomes are utilization or drug expenditure rather than diagnostic measures. More traditional diagnostic-based measures still appear most appropriate if the primary outcome is mortality or inpatient readmissions.

The Impact of Patient and Surgical Factors on the Rate of Postoperative Infection After Total Hip Arthroplasty-A New Zealand Joint Registry Study

BACKGROUND

Periprosthetic joint infection (PJI) is a devastating complication after total hip arthroplasty (THA). The potential to define and modify risk factors for infection represents an important opportunity to reduce the incidence of PJI. This study uses New Zealand Joint Registry data to identify independent risk factors associated with PJI after primary THA.

METHODS

Data on 91,585 THAs performed between 2000 and 2014 were analyzed. Factors associated with revision for PJI within 12 months were identified using univariate and multivariate analyses.

RESULTS

Revision rates for PJI were 0.15% and 0.21% at 6 and 12 months, respectively. Multivariate analysis showed significant associations with the American Society of Anesthesiologists grade (odds ratio [OR] 6.13, 95% confidence interval [CI] 1.28-29.39), severe or morbid obesity (OR 2.15, CI 1.01-4.60 and OR 3.73, CI 1.49-9.39), laminar flow ventilation (OR 1.98, CI 1.38-2.85), consultant-supervised trainee operations (OR 1.94, CI 1.22-3.08), male gender (OR 1.68, CI 1.23-2.30) and anterolateral approach (OR 1.62, CI 1.11-2.37). Procedures performed in the private sector were protective for revision for infection (OR 0.68, CI 0.48-0.96).

CONCLUSIONS

The PJI risk profile for patients undergoing THA is constituted of a complex of patient and surgical factors. Several patient factors had strong independent associations with revision rates for PJI. Although surgical factors were less important, these may be more readily modifiable in practice.

Systematic review of risk prediction scores for surgical site infection or periprosthetic joint infection following joint arthroplasty

Accurate identification of individuals at high risk of surgical site infections (SSIs) or periprosthetic joint infections (PJIs) influences clinical decisions and development of preventive strategies. We aimed to determine progress in the development and validation of risk prediction models for SSI or PJI using a systematic review. We searched for studies that have developed or validated a risk prediction tool for SSI or PJI following joint replacement in MEDLINE, EMBASE, Web of Science and Cochrane databases; trial registers and reference lists of studies up to September 2016. Nine studies describing 16 risk scores for SSI or PJI were identified. The number of component variables in a risk score ranged from 4 to 45. The C-index ranged from 0·56 to 0·74, with only three risk scores reporting a discriminative ability of >0·70. Five risk scores were validated internally. The National Healthcare Safety Network SSIs risk models for hip and knee arthroplasties (HPRO and KPRO) were the only scores to be externally validated. Except for HPRO which shows some promise for use in a clinical setting (based on predictive performance and external validation), none of the identified risk scores can be considered ready for use. Further research is urgently warranted within the field.

Moderating effects of immunosuppressive medications and risk factors for post-operative joint infection following total joint arthroplasty in patients with rheumatoid arthritis or osteoarthritis

OBJECTIVE

Inconclusive findings about infection risks, importantly the use of immunosuppressive medications in patients who have undergone large-joint total joint arthroplasty, challenge efforts to provide evidence-based perioperative total joint arthroplasty recommendations to improve surgical outcomes. Thus, the aim of this study was to describe risk factors for developing a post-operative infection in patients undergoing TJA of a large joint (total hip arthroplasty, total knee arthroplasty, or total shoulder arthroplasty) by identifying clinical and demographic factors, including the use of high-risk medications (i.e., prednisone and immunosuppressive medications) and diagnoses [i.e., rheumatoid arthritis (RA), osteoarthritis (OA), gout, obesity, and diabetes mellitus] that are linked to infection status, controlling for length of follow-up.

METHODS

A retrospective, case-control study (N = 2212) using de-identified patient health claims information from a commercially insured, U.S. dataset representing 15 million patients annually (from January 1, 2007 to December 31, 2009) was conducted. Descriptive statistics, t-test, chi-square test, Fisher's exact test, and multivariate logistic regression were used.

RESULTS

Male gender (OR = 1.42, p < 0.001), diagnosis of RA (OR = 1.47, p = 0.031), diabetes mellitus (OR = 1.38, p = 0.001), obesity (OR = 1.66, p < 0.001) or gout (OR = 1.95, p = 0.001), and a prescription for prednisone (OR = 1.59, p < 0.001) predicted a post-operative infection following total joint arthroplasty. Persons with post-operative joint infections were significantly more likely to be prescribed allopurinol (p = 0.002) and colchicine (p = 0.006); no significant difference was found for the use of specific disease-modifying anti-rheumatic drugs and TNF-α inhibitors.

CONCLUSION

High-risk, post-operative joint infection groups were identified allowing for precautionary clinical measures to be taken.

Evaluation of three co-morbidity measures to predict mortality in patients undergoing total joint arthroplasty

OBJECTIVE

To evaluate the 90 days and 1 year mortality predictive ability of the RxRisk-V, Charlson, and Elixhauser co-morbidity measures in total hip arthroplasty (THA) and total knee arthroplasty (TKA) patients.

METHOD

A retrospective study of 11,848 THAs and 18,972 TKAs (2001-2002) was conducted. Death within 90 days and 1 year of the surgery were the main endpoints. Co-morbidity measures were calculated using either medication or hospitalisation history. Logistic regression models were employed and discrimination and calibration were assessed. Specifically, models with unweighted and weighted measure scores, models with the specific conditions, and a model combining conditions identified by all measures were assessed.

RESULTS

In THAs, the best performing prediction models included co-morbidities from all three measures (90 days: c = 0.84, P = 0.284, 1 year: c = 0.79, P = 0.158). Individually, the model with Charlson conditions performed best at 90 days mortality (c = 0.80, P = 0.777) and the Charlson and Elixhauser performed similarly at 1 year (both c = 0.77, P > 0.05). In TKAs, the best performing prediction model included co-morbidities from all measures (90 days: c = 0.82, P = 0.349, 1 year: c = 0.78, P = 0.873). Individually, the model with Elixhauser conditions performed best with 90 days mortality (c = 0.79, P = 0.435) and all performed similarly at 1 year (c = 0.74-0.75, all P > 0.05).

CONCLUSIONS

A combined model with co-morbidities identified by the Elixhauser, Charlson, and RxRisk-V was the best mortality prediction model. The RxRisk-V did not perform as well as the others. Because of the Elixhauser and Charlson's similar performance we suggest basing the choice of measurement use on factors such as the need of specific conditions and modelling limitations.