One-year mortality and Periprosthetic infection rates after Total knee Arthroplasty in Cancer patients: a population-based cohort study

Through the National Health Insurance (NHI) program, established in 1995, the Taiwan Department of Health covers the health care of 22.9 million residents of Taiwan, which is > 99% of the total population. The medical claims from 1997 to 2012 of these insurants are encrypted and released for research by the Taiwan National Health Research Institutes as the NHIRD. The Department of Health and the NHI Administration Bureau of Taiwan ensure the completeness and accuracy of the NHIRD.

The study data were obtained from the NHIRD. All patients included in the analysis were followed for outcome identification by using the International Classification of Disease, Ninth Revision, Clinical Modification, codes until the end of 2012. Because the NHIRD contains encrypted and deidentified data, this study was exempted from a full ethics review. This study was approved by the institutional Review Board of E-Da Hospital, Taiwan (EMRP-103-011; EMRP-103-012) and the Taiwan NHRI (NHIRD-103-116).

Patients diagnosed with cancer before receiving TKA between January 1, 1997, and December 31, 2011, were identified from the NHIRD and included in the TKA group. Patients with musculoskeletal cancers were not enrolled. We also excluded those whose cancer was diagnosed more than 5 years before undergoing TKA, because the cancer status might be considered as having been cured in some of these patients. Their inclusion would be inappropriate given that our analysis aimed to inform the decision making on TKA surgery in cancer patients, instead of those who had cured a cancer. Taking into account that data in the NHIRD could not ascertain whether a cancer was cured, we restricted the eligibility to emphasize our study population. The control group comprised patients without cancer who underwent TKA. One-year postoperative infection rates and mortality rates in the two groups were evaluated and compared (Fig. 1). The one-year postoperative mortality rates was calculated from the analysis of survival rate of patients (with/without cancer) who underwent TKA. Periprosthetic joint infection was defined as the occurrence of surgically treated osteomyelitis or septic arthritis in the vicinity of the joint implants. Surgical interventions included debridement, prosthesis removal, or resection arthroplasty (Appendix). Conditions existing prior to TKA were classified as comorbidities according to Charlson’s score [16]. Among the cancer patients, the number with and without metastasis and those who had received chemotherapy at the time of TKA were also calculated (Appendix). All patients were followed until death, withdrawal from the NHI program, or the end of the study period (December 31, 2012). Figure 1 illustrates the study flow chart.

Subcohorts were defined to evaluate the overall 5-year survival rate of cancer patients who underwent TKA. In addition, the 1- and 5-year relative survival rates were calculated [17, 18].

Pearson’s chi-square test and Fisher Yates continuity correction were used to compare categorical variables. Student’s t test was used for continuous ones. Traditional survival analysis only considers one event at a time (e.g., death or infection). Thus, certain events may be overlooked, and the resulting risk values may be overestimated. Therefore, these results should not be directly interpreted and applied in clinical settings. Our study considered mortality and the competing risk survival analysis using the Fine and Gray regression model [19] to calculate subdistribution hazards, and p-values were determined using Gray’s test. P < 0.05 was considered significant. All statistical tests and subdistribution hazard ratio (sHR) calculations were performed using Statistical Analysis Software, Version 9.4 (SAS Institute, Cary, NC, USA).

A total of 2294 cancer patients who underwent TKA were identified and included in the TKA group, whereas the control group comprised 131,849 patients. Score matching for gender and Charlson’s comorbidity index yielded 2196 patients in the TKA group and 8784 in the control group. Among these 10,980 patients, 3355 (30.56%) were male and 7625 (69.44%) female, and 10,236 (93.22%) patients were older than 60 years. The baseline characteristics and comorbidities of all patients are listed in Table 1. Among the cancer patients, 7 (0.32%) cases had metastatic diseases and 708 (32.34%) of them ever received chemotherapy when TKA was performed. The most common cancers were breast, colon, prostate, cervical, and rectal cancers (Fig. 2). The subcohort groups comprised 1100 and 4400 patients with and without cancer, respectively.

The number of patients with infected prosthetic joints 1 year after TKA were 38 (1.73%) and 164 (1.87%) in the cancer and control groups, respectively (Table 1). A multivariate-adjusted model revealed no association between infection and cancer (Table 2). In the unadjusted analysis, diabetes with end organ damage was associated with prosthetic joint infection (sHR, 1.77; 95% CI, 1.03–3.06); however, this association was nonsignificant in the Fine and Gray regression model (Table 2). The status of metastatic disease but not the use of chemotherapy was associated with prosthetic joint infection (Table 2 and Table 3).

The 1-year mortality rate after TKA was significantly higher (p < 0.001) in the cancer group (90 patients; 4.10%), (1-year cumulative incidence of 1.73%; 95% confidence interval [CI], 1.26–2.37%) than in the control group (146 patients; 1.66%), (1-year cumulative incidence of 1.71%; 95% CI, 1.46–2.0%) (Tables 1, 4 and Fig. 3). The 1-year relative survival rate was 97.52%. The Fine and Gray regression model did not show a significant association between metastasis and postoperative mortality (Table 4).

In the 5-year follow-up, a significantly lower overall survival rate was observed in cancer patients as compared with the controls (Fig. 4). The overall 5-year survival rate was 89.36% in the cancer cohort and was relatively 93.10% as compared with the non-cancer controls.