Arthroplasty is currently the gold standard for treatment of femoral neck fractures in older patients. It is known to help improve the quality of life of patients. However, there is no consensus on the use of cement during arthroplasty. In this study, we synthesized evidence from 16 studies that compared cemented with uncemented arthroplasty for treatment of femoral neck fractures in older patients. We found that cemented arthroplasty was superior to uncemented arthroplasty with regards to the reoperation rate, complications related to prosthesis, and residual pain; however, the former required longer operation time. There was no significant difference between the two procedures with respect to mortality, duration of hospital stay, hip function, or general and local complications.
Compared with previous systematic reviews [9, 11, 12, 31–34], our meta-analysis improved certain aspects and provided some new insights. First, a larger sample of studies (16 studies) was included in our analysis, which helped decrease the selection bias. Second, the need for reoperation is a key concern both for surgeons and patients; previous systematic reviews including one latest review  found no differences between the two treatments with respect to reoperation rate. However, after inclusion of larger number of studies with longer follow-up period, we found that cemented arthroplasty may significantly decrease the reoperation rate; this finding is different from those reported by Li et al  and Luo et al . Third, we included only RCTs and prospective cohort studies, which yielded higher quality data. Fourth, we performed subgroup analysis and sensitivity analysis and explored potential factors that may have affected our results. Fifth, half of the included studies had a follow-up period of more than two years, which provided more robust evidence with respect to reoperation rate, mortality, residual pain, and hip function. Lastly, each outcome was assessed using the GRADE system.
Based on analysis of reoperation data from eleven studies, uncemented arthroplasty was associated with a significantly higher reoperation rate (5.8%) as compared to cemented group (3.2%). The higher reoperation rate after uncemented arthroplasty may be attributable to loosening of the prosthesis, intraoperative fracture, or dislocation. This outcome was different from previous systematic reviews [9, 11, 36, 37] where researchers found no significant differences between the cemented and uncemented groups. The difference may be due to a larger sample size and longer follow-up period in our study.
We also observed a significant difference between the cemented and uncemented groups with respect to the incidence of complications. There was significant heterogeneity and subgroup analysis was conducted (I2 = 93.6%). Complications related to prosthesis were less frequent in the cemented group, which is supported by previous studies ; this may explain the higher reoperation rate in the uncemented group. Local and general complications in the cemented group were also less frequent than those in the uncemented group, although the between-group difference was not statistically significant. These findings are consistent with those of previous studies [12, 30, 38].
Mortality is another key concern in patients with femoral neck fracture. Thirteen studies were analyzed and the rate of mortality was comparable in the two groups. Many studies support our findings [12, 27]. In the study by Talsnes et al, the hazard ratio was 0.77 with no significant difference in mortality after long-term follow-up (p = 0.233). However, a study conducted in 1994 showed that use of cement may increase the intraoperative mortality rate due to the increased risk of cardiovascular disease on insertion . In our study, one patient developed heart failure  and three patients developed cardiac arrest or myocardial infarction and died within 72 hours after cemented arthroplasty [13, 40]. The age of patients and preexisting cardiopulmonary symptoms may also affect the mortality rate .
Residual pain and hip function are key determinants of the quality of life of patients. In our study, cemented arthroplasty was significantly associated with less postoperative pain and similar hip function compared to uncemented group. This finding is consistent with the results reported by Ning et al and Bagaric et al [12, 32]. In the study by Figved et al, residual pain in the cemented group was greater than that in the uncemented group, although the difference was not statistically significant. This difference may be attributable to the different materials (hydroxyapatite-coated) used in the study populations included in the respective reviews. For analysis of HHS, there was significant heterogeneity among the included studies (55%); based on the results of sensitivity analysis, exclusion of one study  from the meta-analysis eliminated the heterogeneity (I2 = 0%). On further review of the literature, we found that in the study by Langslet et al, the HHS was significantly different between the two groups only in the fifth year. However, less than half of the original study population was followed-up. The heterogeneity may be attributable to the large number of patients who were lost to follow-up.
Our study demonstrated that cemented arthroplasty increases the operation time and this could be related to the additional procedures for cement insertion, which is consistent with many previous studies [18, 27, 29, 38]. With respect to the duration of hospital stay, our study found no significant difference between cemented and uncemented arthroplasty, which is consistent with Ning et al . There was significant heterogeneity among the studies included in the meta-analysis of operation time and duration of hospital stay. Among these studies, data from the study by Gavaskar et al  was very concentrated with a very small standard deviation. Based on the results of sensitivity analysis, exclusion of the study by Gavaskar et al  reduced the heterogeneity among the remaining studies. Thus our results were credible.
Several limitations of our study should be considered while interpreting the results. First, some unpublished studies and non-English language studies were not included in our study, which could lead to bias. Second, information pertaining to random sequence generation and allocation concealment was not available for many of the included studies, which may lead to misjudgment of the quality. Third, the included studies involved the use of many different types of prosthesis; this may have introduced an element of bias. Further analysis is required to provide stronger evidence for clinical treatment.
In our meta-analysis, we used the GRADE system to assess the evidence grading of the outcomes. The quality of evidence pertaining to the outcomes of reoperation rate, operation time, complications related to prosthesis, general complications, mortality, and residual pain was high or moderate. However, the evidence grading for hospital stay and local complications was low due to the following reasons. First, four of the included studies [3, 28–30] were prospective cohort studies and not RCTs. Second, the 95% confidence intervals around relative effects were very wide (range, 0.73–1.29) or there was considerable heterogeneity.