The effects of cement fixation on survival in elderly patients with hip hemiarthroplasty: a nationwide cohort study

This retrospective cohort study used claims data during the period of 1 January 2007–31 December 2015 from the National Health Insurance Database (NHID) and the National Register of Deaths Database (NRDD) for analysis. Based on the NHID, patients who were aged 80 years or older and diagnosed with femoral neck fracture (International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes: 820.xx) between January 1, 2008, and December 31, 2014, and were treated with HHA (ICD-9-CM procedure codes: 81.52) were eligible for inclusion (n = 26,247). The patients who underwent THA (ICD-9-CM procedure codes: 81.51), HHA, or revision of hip replacement (ICD-9-CM procedure codes: 81.53) during hospitalization (n = 24) or one year prior to hospitalization (n = 361) were excluded from this study. Finally, 25,862 cases were included in the analysis.

The National Health Insurance programme in Taiwan has included over 99.7% of the Taiwanese population since 2004 and provides universal, compulsory coverage with low co-payments to minimize the economic barrier to care for patients needing inpatient, outpatient, prescription, and other services. The NHID contains each patient’s demographic characteristics, medical treatment location and detailed records of outpatient visits, hospital admissions, and emergency department (ED) visits, including diagnoses, procedures, medications, providers, and expenses [16]. The NRDD monitors the completeness and accuracy of death registration data retrieved from the Ministry of Health and Welfare, and the data contained age, sex, date of death, and causes of death. Both databases are encrypted, patient data are de-identified, and the database is monitored for completeness and accuracy by Taiwan’s Ministry of Health and Welfare. The authors analysed data at the Health and Welfare Data Science Centre, which is an independent workplace managed by Taiwan’s Ministry of Health and Welfare.

The dependent variable of this study was all-cause mortality within 1 year after the operation. The duration from the first day of hospitalization to the day of death was defined as the overall survival time. Subjects who were still alive after 1 year of follow-up were censored.

The fixation method (cemented or uncemented) was the main independent variable in this study and was identified by payment codes from the Taiwan NHI Medical Service Benefits and Payment Criteria [17]. Other covariates included characteristics of patients and healthcare providers. The patients’ characteristics included sex, age group (80–84 years, 85–89 years, ≥90 years), insurance status (≥840 USD, < 840 USD) [18], Charlson comorbidity index (CCI) status (CCI = 0, CCI ≥ 1) during the last year before the operation [19‐21] and type of head used in the operation (unipolar hemiarthroplasty (UHA) or bipolar hemiarthroplasty (BHA)) [22‐25]. The insurance premiums were determined by the enrolee’s income and therefore were a proxy for income status [26]. Insurance status was classified into two groups according to the median of the insurance premiums.

The providers’ characteristics included the type of hospital (private (including corporate) or public hospital) [27, 28], hospital accreditation level (medical centre, non-medical centre) [28], the annual volumes of hip replacement for the hospital [29‐33] and surgeon [29, 30, 34]. Annual volumes of hip replacement were classified into two groups according to the median operation case number among all the providers/surgeons.

Pearson’s χ2 test was used to compare characteristics between the study participants who received cement and those who did not. The log-rank test was used to compare the mortality rates of different time intervals (7, 30, 180 days and 1 year from the operation) between participants receiving cement and non-cement HHA. To assess the one-year mortality risk factors of different fixation methods, we fitted a Cox proportional hazard regression with covariates including sex, age group, insurance status, CCI score, hospital type, hospital level, hospital volume, and surgeon volume. We also investigated multicollinearity by the variance inflation factor (VIF) using regression analysis. Because the VIF of each coefficient was less than 5, we presumed that the effect of correlation among the independent variables was not enough to distort the estimation. For data management and statistical analysis, we used the statistical software SAS 9.4 (SAS Institute, Inc., Cary, North Carolina, USA).

This study was approved by the Institutional Review Board of National Yang-Ming University in Taiwan (approval number YM105043E-3). The authors declare no any financial and non-financial competing interests.

In total, 47.8% (12,364) of patients received cemented HHA. The percentages of female patients, patients with age ≥ 90 years, patients with lower insurance premiums, patients receiving UHA, patients treated at higher-volume hospitals, and patients operated on by higher-volume surgeons were significantly higher in the cemented HHA group than in the uncemented HHA group. The CCI score was not significantly different between the two groups (Table 1).

As shown in Table 2, the overall postoperative mortality rates within 7, 30, 180 days and 1 year among senior patients who received HHA were 0.6, 3.1, 11.2, and 17.3%, respectively. The mortality rates at 7 days (0.7% vs 0.4%, p < 0.01), 30 days (3.5% vs 2.7%, p < 0.01), 180 days (11.8% vs 10.7%, p < 0.01) and 1 year (18.0% vs 16.7%, p < 0.01) were significantly higher in the cement group than in the non-cement group.

Tables 3 and 4 present the crude and adjusted hazard ratios of risk factors of mortality within 7, 30, 180 days and 1 year. The patients who received bone cement had a significantly higher mortality risk than those who did not receive bone cement within 7, 30, 180 days and 1 year (aHR = 1.8, 95% CI = 1.23–2.51, p < 0.01; aHR = 1.4, 95% CI = 1.17–1.58, p < 0.01; aHR = 1.2, 95% CI = 1.08–1.26, p < 0.01; aHR = 1.1, 95% CI = 1.04–1.18, p < 0.01), respectively. The type of head and insurance status were not risk factors for mortality (p > 0.05). Male gender and age > 85 years were risk factors for mortality within 7, 30, 180 days and 1 year (p < 0.05). CCI scores were identified as a significant risk factor for 30 days, 180 days and 1-year mortality (p < 0.05). After adjustment, hospital type and hospital level were not risk factors for mortality (p > 0.05) (Table 4). Patients who received HHA at lower-volume hospitals were 1.16 and 1.15 times more likely to die within 180 days and 1 year than those who received HHA at high-volume hospitals (95% CI = 1.05–1.29, p < 0.01; 95% CI = 1.06–1.24, p < 0.01), respectively (Table 4). Compared with patients who were operated on by high-volume surgeons, those who were operated on by lower-volume surgeons were more likely to die within 30 days (aHR = 1.22, 95% CI = 1.05–1.42, p < 0.05), 180 days (aHR = 1.16, 95% CI = 1.07–1.25, p < 0.01) and 1 year (aHR = 1.19, 95% CI = 1.12–1.27, p < 0.01). The effect of bone cement on postoperative mortality was significantly stronger within 7 days than within 30, 180 days and 1 year (Table 5).