Are functional outcomes and reoperation rates really better than those of cementless stems in displaced femoral neck fractures? An updated systematic review and meta-analysis of randomized controlled trials in current-generation stem designs

Abstract

Background:

The aim of the study was to conduct a systematic review and meta-analysis to compare the functional outcomes and reoperation rates of cemented and cementless hip arthroplasty for treating displaced femoral neck fractures in elderly patients.

Methods:

Systematic searches were conducted of literature up to December 2018 on PubMed, Embase, Cochrane, and Web of Science for randomized controlled trials (RCTs) concerning current-generation stem designs only. Two reviewers independently determined eligibility, extracted the outcome data and assessed the risk of bias of eligible studies. The follow-up data and complication rates were pooled by using random-effects models and fixed-effects models, with mean differences and risk ratios for continuous and dichotomous variables, respectively.

Results:

Eight RCTs involving 1361 patients (1361 hips) were included in the meta-analysis. Cemented stems were associated with fewer implant-related complications (odds ratio [OR] = 0.303; 95% confidence interval [CI], 0.185%-0.496%; p < 0.001) and reoperations (OR = 0.492; 95% CI, 0.247%-0.977%; p = 0.043). There were no statistically significant differences between groups in functional outcomes, including those assessed by the EuroQol(EQ)-5D and Harris Hip Score, mortality rates, major systemic complications, minor local complications, operation times, intraoperative blood losses, and lengths of hospital stays.

Conclusion:

In treating displaced femoral neck fracture in elderly patients with hip arthroplasty with current-generation stems, cemented stems were found to have fewer implant-related complications and reoperations than those of cementless stems. Functional outcomes and mortality rates were similar between the groups.

1. INTRODUCTION

As the mean age of a population increases, the problem of hip fractures in the elderly has shifted from being strictly medical related to one involving public health issues. Femoral neck fractures account for a substantial proportion of these hip fractures and lead to increased morbidity and higher public health expenditures.^1,2

Hemiarthroplasty is now the mainstay treatment for geriatric displaced femoral neck fractures, although the use of total hip arthroplasty has increased in recent years.^2–4 The use of a cemented or cementless stem in hip arthroplasty remains a controversial issue,^5–9 and several systemic reviews have compared the two types.^10–14 The reviews that were of relatively outdated prostheses, such as the Austin Moore and Thompson stems, have found that better mobility, less residual pain, and fewer implant-related complications were noted in patients with cemented stems.^10–12 Recently, reviews that have compared current-generation stems have found that cemented stems were associated with fewer implant-related complications without an increased mortality rate.¹⁴

However, because of the limited number of available trials for analysis, there are still unanswered questions raised by these studies. In our study, we reviewed updated trials and compared clinical outcomes between cemented and cementless arthroplasty using current-generation stems in the treatment of elderly patients with displaced femoral neck fractures. The specific study aims were to answer the following questions: (1) Are functional outcomes better for cemented stems than for cementless stems? (2) Are there differences in the number of implant-related complications and reoperation rates between the two groups? (3) Does use of bone cement lead to more perioperative adverse events and higher mortality rates over different postoperative follow-up periods? and (4) Which type of stem is associated with longer operation time and greater intraoperative blood loss?

2. METHODS

The study was structured according to the Preferred Reporting Items for Systemic Reviews and Meta-Analysis (PRISMA) statement.

2.1. Eligibility Criteria

The inclusion criteria were randomized controlled trials (RCTs) that compared cemented and cementless stems in the treatment of femoral neck fractures. We excluded studies that investigated implants that were no longer unavailable in current medical practice (e.g., Austin Moore and Thompson stems) but that did investigate mortality rates, functional performance, numbers of complications and reoperations, lengths of hospital stays, intraoperative blood losses, and operation times. These outcomes were generally used to evaluate the clinical outcomes after hip surgery, and amenable quantitative parameters were used in the meta-analyses. Since total hip replacement was still considered to be a treatment option for displaced femoral neck fractures, we retained the studies that included total hip arthroplasty treatment.

2.2. Identifying Studies – Information Sources/Search

We searched the databases of PubMed (from 1975 to 2018), Embase (from 1974 to 2018), Cochrane (from 1982 to 2018), and Web of Science (from 1991 to 2018) for literature up to December 2018. The following medical terms were searched:hip fracture, femoral neck fracture, hip arthroplasty, hip replacement, cemented, cementless, and uncemented. Animal studies and publication types unlikely to contain relevant information (news, comments, letters to the editor, and editorials) were excluded.

2.3. Study Selection Processes

Two independent reviewers (H.-H.M. and C.-W.F.) evaluated the eligibility of the selected studies. When necessary, we obtained full-text articles to determine eligibility for inclusion. Disagreements were resolved by discussion.

2.4. Data Collection Processes

All of the relevant data were extracted from the selected RCTs by two independent reviewers. Any disagreement and data inconsistency were resolved by discussion.

2.5. Data Items

Methodological data included the number of centers involved in the study, method of enrollment, patient’s characteristics in each group (cemented versus cementless), type of stem, and principle of data analysis (intent-to-treat or per-protocol).

In outcome measures, mortality rates, and functional scores recorded at 1 year, 2 years, and >4 years were extracted and analyzed. The numbers of complications and reoperations, intraoperative blood losses, and lengths of hospital stays were not bound to a certain point in time. We classified the complications into four categories: implant-related complications, reoperation for any reason, major systemic complications, and minor local complications.¹⁵ Implant-related complications included intraoperative and postoperative periprosthetic fractures, prosthesis loosening, and dislocation. The major systemic complications included intraoperative cardiac arrest, acute arrhythmia, myocardial infarction, pulmonary embolus, and hypotensive crisis. The minor local complications included superficial and deep-wound infections, muscular spasms, imbalance gait, and persistent local pain.

2.6. Risk of Bias Assessment in Individual Studies

We critically accessed the risk of bias of the selected studies by using the Cochrane Risk of Bias Tool¹⁶ and assessed all publications, including protocols. All the studies were assessed according to the aspects of random-sequence generation, allocation concealment, blinding of outcome assessments, incomplete outcome data, selective reporting, and other biases. Each of the aspects was rated as having a “low risk,” “unclear risk,” or “high risk” of bias. Disagreements were resolved by discussion and consultation with the senior author. The figure of publication bias assessment was performed by using Review Manager Software (RevMan 5.2; Cochrane, Haymarket, London, UK).

2.7. Specification of Outcomes and Effect Measures/Synthesis Methods

In the outcome data analysis, we synthesized the continuous outcome data by using the mean difference and standard deviation. The standardized mean differences (SMDs) of the extracted data were indicated to be a favorable treatment option. For dichotomous outcome data, we used the odds ratio (OR) for synthesis. A random-effects model was used to pool individual SMDs and ORs. The 95% confidence intervals (CIs) were calculated for each outcome. Between-trial heterogeneity was determined by performing the I² test; values >50% were regarded as indicating considerable heterogeneity. All analyses were performed by using Comprehensive Meta-Analysis software (Version 3.3.070; Biostat, Englewood, NJ, USA).

3. Results

3.1. Study Selection

The studies search, exclusion criteria, and final selection of studies are presented in a flow diagram from the PRISMA guideline (Fig. 1). Two studies were published repeatedly with different lengths of follow-up; thus, we extracted the data from the most recent of them.^9,17 We included eight RCTs published between 2005 and 2018 for final meta-analysis. A total of 1361 patients were included. The study characteristics are presented in Table 1. Different outcome measures and lengths of follow-up times between the studies are listed in Table 2.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analysis flow diagram for the searching and identification of included studies.

Table 1.

Characteristics of included trials

	Study design	Study period	Patients (n)			Mean age (year)			Gender: n (female %)		Type of prosthesis		ITT/PP	Other
			Total	Cementless	Cemented	Total	Cementless	Cemented	Cementless	Cemented	Cementless	Cemented
Barenius et al (2018)	RCT, one center	October 2009–April 2013	141	74	67	81.3	81.3	81.2	53	46	Hydroxyapatite-coated Bimetric stem (Zimmer Biomet, Warsaw, Indiana)	Exeter stem (Stryker, Kalamazoo, Michigan)	ITT	THA %, Cemented: 42%, Cementless: 41%
Moerman et al (2017)	Multicenter parallel-RCT, three center	August 2008–June 2012	201	91	110	83.5	84	83	61	82	DB-10 (Zimmer- Biomet, 1800 West Center St. Warsaw, Indiana, USA). HA coated	Müller Straight Stem (Zimmer – Biomet, 1800 West Center St. Warsaw, Indiana, USA)	ITT
Chammout et al (2016)	RCT, one center	September 2009–March 2014	69	34	35	73	73	72	25	22	Bi-Metric stem (Biomet, Warsaw, IN)	modular CPT (Zimmer, Warsaw, IN)	ITT	THA
Langslet et al (2014)	RCT, two centers	September 2004–August 2006	220	108	112	83	83.0	83.4	74.0	78.0	Corail (Depuy)	Spectron (Smith & Nephew, London, United Kingdom)	PP
Talsnes et al (2013)	RCT, two centers	2005–2010	334	172	162	84.1	84.0	84.3	78.5	72.4	Corail (Depuy, Warsaw, Indiana)	Titan (Depuy)	PP
Taylor et al (2012)	RCT, one center	May 2006–November 2008	160	80	80	85.2	85.1	85.3	66.3	71.3	Zweymüller Alloclassic (Zimmer)	Exeter (Stryker, Kalamazoo, Michigan)	ITT
DeAngelis et al (2012)	RCT, one center	March 2005–May 2008	130	64	66	82.3	82.8	81.8	75.0	78.8	VerSys Beaded FullCoat (Zimmer)	VerSys LD/Fx (Zimmer)	ITT
Santini et al (2005)	RCT, one center	September 2000–December 2001	106	53	53	80.9	82.1	79.7	79.2	75.5	Bipolar stem (Zimmer, Warsaw, Indiana)	Bipolar stem (Zimmer)	PP

ITT = intention-to-treat; PP = per protocol; RCT = randomized controlled trial; THA = total hip arthroplasty.

Table 2.

Outcome measures and follow-up of the studies

	Study outcomes	Described complication	Follow-up
Barenius et al (2018)	HHS, EQ-5D, SMFA, mortality, complication (adverse event), radiological assessment	Reoperation due to periprosthetic fracture Revision to total hip arthroplasty Trendelenburg limp, Trochanteric pain	4 months, 1 y, 2 y, and 4 y
Moerman et al (2017)	TUG score, GARS, NMS, SF-12, mortality, complication, operation time, intraoperative blood loss, length of hospital stay, transfusion rate, radiograph assessment	Categorized in: major systemic, minor systemic, major local, minor local	1 y
Chammout et al (2016)	ADL, HHS, EQ-5D, PNRS, mortality, complications, intraoperative blood loss, operation time, and intraoperative vital signs, CRP, d-dimer, radiographic assessment	Intraoperative and postoperative periprosthetic fracture, dislocations, superficial/deep wound infection, early and late loosening, and reoperation of the hip for any reason	2 y
Langslet et al (2014)	Barthel index, HHS, EQ-5D, use of walking aids; mortality, complications, intraoperative blood loss, operation time, length of hospital stay, residual pain	Intraoperative cardiac arrest, MI, intraoperative severe decrease in blood pressure, periprothetic fracture, dislocation, pneumonia, pulmonary embolism, reoperation, thrombosis, wound infection	7 d, 30 d, 90 d, 1 y, 2 y, and 5 y
Talsnes et al (2013)	Mortality, operation time, intraoperative blood loss	N/A	Operation d, 1 y
Taylor et al (2012)	Oxford Hip Score, pain, use of walking aids, mortality, complications, intraoperative blood loss, operation time, length of hospital stay	Cardiovascular complications, respiratory/urinary infection, superficial/deep wound infection, dislocation, periprothetic fractures, reoperation	6 wk, 6 months, 1 y, 2 y
DeAngelis et al (2012)	ADL, physical activity, fatigue/level of energy, mortality, complications, operation time, intraoperative blood loss	ICU stay, pneumonia, MI, wound infection reoperation, CVA	In hospital, 30 d, 60 d, 1 y
Santini et al (2005)	VELCA, mortality, complications, operation time, length of hospital stay	Arrhythmia, MI, pulmonary embolism, respiratory/urinary infection, gastric disease, iatrogenic femur fractures, dislocation, wound infection, decubitus	1 y

ADL = activities of daily living; CRP = C-reactive protein; CVA = cerebrovascular accident; EQ-5D = EuroQol-5D; GARS = Groningen activity restriction scale; HHS = Harris hip score; ICU = intensive care unit; MI = myocardial infarction; NMS = new mobility score; PNRS = pain numerical rating scale; Qol = quality of life; SMFA = short musculoskeletal function assessment questionnaire; SF-12 = physical and mental health summary scales short form-12; TUG = timed-up and-go; VELCA = Verona Elderly Care Study.

3.2. Study Characteristics and Risk of Bias

The appraisals of the methodological quality of the studies are presented in Figs 2 and 3. Two of the studies used computer number generator sequence randomization.^6,18 Another three studies used a blinded block design, but the randomization method was not clearly described.^5,19,20 One study randomized the patients according to the day of admission, which may have led to a high risk of selection bias.²¹ For allocation methods, five studies used a sealed envelope that was opened in the operating theatre.^{6,18,19,22,23} The other three studies did not clearly describe the allocation method.^5,20,21 Two studies mentioned that the outcome assessors were aware of the patient allocation, which might have caused high detection bias.^19,23 We presented details on the risks of bias associated with the methodological quality of the studies in Fig. 2 and a summary in Fig. 3.

Fig. 2.

Assessment of risk of bias in the selected studies; ‘‘+’’: low risk of bias, ‘‘?’’: unclear risk of bias, and ‘‘–’’: high risk of bias.

Fig. 3.

Summary of the assessment of risk of bias.

3.3. Results of Individual Studies

3.3.1. Functional Outcome

Seven of the selected studies investigated the functional outcome. However, because of the various outcome parameters used for postoperative hip function assessment, we could only pool the data of three studies that adopted both the Harris Hip Score (HHS) and health-related quality of life (EuroQol(EQ)-5D). Comparison of postoperative hip function using HHS showed that there were no significant between-group differences at the 1-year (95% CI, −0.252% to 0.203%; p = 0.832), 2-year (95% CI, −0.569% to 0.225%; p = 0.397) and ≥4-year (95% CI, −0.681% to 0.945%; p = 0.751) follow-ups (Fig. 4a–c).

Fig. 4.

(A) Forest plot of the Hip Harris score at posteroperative 1 year, Random, Heterogeneity: Tau² = 0.006; Chi² = 2.313, df = 2 (p = 0.315); I² = 14%. Test for overall effect: Z = –0.213 (p = 0.832). (B) Forest plot of Hip Harris score at posteroperative 2 years, Random, Heterogeneity: Tau² = 0.033; Chi² = 1.660, df = 1 (p = 0.198); I² = 40%. Test for overall effect: Z = –0.847 (p = 0.397). (C) Forest plot of Hip Harris score at posteroperative 4 years and beyond, Random, Heterogeneity: Tau² = 0.294; Chi² = 6.790, df = 1 (p = 0.009); I² = 85%. Test for overall effect: Z = 0.318 (p = 0.751).

Additionally, functional assessments using EQ-5D showed no significant between-group differences at the 1-year (95% CI, −0.722% to 0.048%; p = 0.086), 2-year (95% CI, −0.963% to 0.8%; p = 0.855) and ≥4-year (95% CI, −0.416% to 0.515%; p = 836) follow-ups (Fig. 5a–c).

Fig. 5.

(A) Forest plot of EQ-5D at posteroperative 1 year, Random, Heterogeneity: Tau² = 0.072; Chi² = 5.370, df = 2 (p = 0.068); I² = 63%. Test for overall effect: Z = –1.715 (p = 0.086). (B) Forest plot of EQ-5D at posteroperative 2 years, Random, Heterogeneity: Tau² = 0.349; Chi² = 7.181, df = 1 (p = 0.007); I² = 86%. Test for overall effect: Z = –0.182 (p = 0.855). (C) Forest plot of EQ-5D at posteroperative 4 years and beyond, Random, Heterogeneity: Tau² = 0.061; Chi² = 2.158, df = 1 (p =0.142); I² = 54%. Test for overall effect: Z = 0.207 (p = 0.836).

3.3.2. Mortality

None of the selected studies reported significant differences in mortality between the cemented and cementless groups at any postoperative follow-up time. Because of the different set points of follow-up time, we pooled the data of mortality at postoperative 3 months, 1 year, 2 years, and ≥4years. The pooled data revealed no significant differences between the two groups for mortality at postoperative 3 months (OR = 0.45; 95% CI, 0.425% to 1.305%; p = 0.303), 1 year (OR = 0.802; 95% CI, 0.617% to 1.041%; p = 0.097), 2 years (OR = 0.921; 95% CI, 0.630% to 1.347%; p = 0.673) and ≥4 years (OR = 0.874; 95% CI, 0.565% to 1.352%; p = 0.545) (Fig. 6a–d).

Fig. 6.

(A) Forest plot of mortality in posteroperative 3 months, fixed, Heterogeneity: Tau² = 0.000; Chi² = 0.953, df = 2 (p = 0.621); I² = 0%. Test for overall effect: Z = –1.030 (p = 0.303). (B) Forest plot of mortality in posteroperative 1 years, fixed, Heterogeneity: Tau² = 0.000; Chi² = 6.259, df = 7 (p = 0.510); I² = 0%. Test for overall effect: Z = –1.660 (p = 0.097). (C) Forest plot of mortality in posteroperative 2 years, fixed, Heterogeneity: Tau² = 0.000; Chi² = 1.159, df = 3 (p = 0.763); I² = 0%. Test for overall effect: Z = –0.422 (p = 0.673). (D) Forest plot of mortality at posteroperative 4 years and beyond, fixed, Heterogeneity: Tau² = 0.000; Chi² = 0.010, df = 1 (p = 0.922); I² = 0%. Test for overall effect: Z = –0.605 (p = 0.545).

3.3.3. Implant-related Complications

Seven of the eight selected studies investigated implant-related complications. The pooled data revealed a significantly higher number of implant-related complications in the cementless groups (Fig. 7a) (OR = 0.303; 95% CI, 0.185% to 0.496%; p < 0.001). Excluding dislocation, the remaining complications related to the implants had a significantly higher rate in the cementless group (Fig. 7b) (OR = 0.229; 95% CI, 0.121% to 0.435%; p < 0.001).

Fig. 7.

(A) Forest plot of implant related complications, Random, Heterogeneity: Tau² = 0.000; Chi² = 4.091, df = 6 (p = 0.664); I² = 0%. Test for overall effect: Z = –4.738 (p < 0.001). (B) Forest plot of implant related complications (excluded data of dislocation), Random, Heterogeneity: Tau² = 0.000; Chi² = 5.024, df = 6 (p = 0.541); I² = 0%. Test for overall effect: Z = –4.500 (p < 0.001).

3.3.4. Reoperation

Six of the eight selected studies investigated the number of reoperations. None of the studies reported significant differences between the two groups. However, the pooled data revealed a significantly higher number of reoperations in the cementless group (OR = 0.492; 95% CI, 0.247% to 0.977%; p = 0.043) (Fig. 8).

Fig. 8.

Forest plot of reoperations, Random, Heterogeneity: Tau² = 0.000; Chi² = 2.813, df = 5 (p = 0.729); I² = 0%. Test for overall effect: Z = –2.027 (p = 0.043).

3.3.5. Major Systemic Complications

Six studies reported major systemic complications. Pooling did not reveal a significant difference between the two groups (Fig. 9) (OR = 1.347; 95% CI, 0.804% to 2.258%; p = 0.258).

Fig. 9.

Forest plot of major systemic complications, Random, Heterogeneity: Tau² = 0.000; Chi² = 1.211, df = 5 (p = 0.729); I² = 0%. Test for overall effect: Z = 1.132 (p = 0.258).

3.3.6. Minor Local Complications

Seven studies mentioned minor local complications. None of the studies reported significant differences between the two groups, even after pooling of the data (Fig. 10) (OR = 1.258; 95% CI, 0.711% to 2.225%; p = 0.43).

Fig. 10.

Forest plot of minor local complications, Random, Heterogeneity: Tau² = 0.000; Chi² = 2.592, df = 6 (p = 0.858); I² = 0%. Test for overall effect: Z = 0.789 (p = 0.430).

3.3.7. Operation Time

Seven studies investigated the operation time. Two of the studies revealed no significant difference between the two groups.^18,20 Four studies found a longer operation time in the cemented group.^5,6,19,21 Only one study reported a longer operation in the cementless group.²³ Pooling showed no significant between-group differences in operation time (Fig. 11) (SMD: 0.289; 95% CI, –0.017 to 0.595; p = 0.064, random model).

Fig. 11.

Forest plot of operation time, Random, Heterogeneity: Tau² = 0.142; Chi² = 40.804, df = 6 (p < 0.001); I² = 85%. Test for overall effect: Z = 1.852 (p = 0.064).

3.3.8. Intraoperative Blood Loss

Six studies reported the results of intraoperative blood loss. Three of these studies reported no significant difference in intraoperative blood loss between the two groups.^18–20 Two studies found less intraoperative blood loss in the cementless group,^5,6 and one study reported less intraoperative blood loss in the cemented group.²³ There was no significant between-group difference in the intraoperative blood loss (Fig. 12) (SMD: 0.058; 95% CI, –0.236% to 0.352%; p = 0.7, random model).

Fig. 12.

Forest plot of Intraoperative blood loss, Random. Heterogeneity: Tau² = 0.108; Chi² = 28.474, df = 5 (p < 0.001); I² = 82%. Test for overall effect: Z = 0.386 (p = 0.700).

3.3.9. Length of Hospital Stay

Four studies mentioned the length of hospital stay, and none of them revealed significant between-group differences. Pooling also showed no significant difference between the two groups (Fig. 13) (SMD: –0.02; 95% CI, –0.171% to 0.13%; p = 0.791).

Fig. 13.

Forest plot of length of hospital stay, Random. Heterogeneity: Tau² = 0.000; Chi² = 0.444, df = 3 (p = 0.931); I² = 0%. Test for overall effect: Z = –0.265 (p = 0.791).

3.3.10. Risk of Bias Across Studies

Egger’s test revealed no significant publication bias regarding the overall SMD of the functional score [HHS (p =0.826) and EQ5D (p = 0.469)] at 1 year, operation time (p = 0.948), intraoperative blood loss (p = 0.318), length of hospital stay (p = 0.704), and overall OR of mortality at 3 months (p = 0.334), 1 year (p = 0.416) and 24 years (p = 0.794); implant-related complications (p = 0.902; p = 0.444 for the excluded data of dislocation); reoperation (p = 0.701); major systemic complications (p = 0.934); and minor local complications (p = 0.953). The funnel plots for the above subgroup analyses are shown in Supplemental material 1 http://links.lww.com/JCMA/A93.

4. DISCUSSION

This is the most comprehensive meta-analysis of patients treated with arthroplasty using current-generation stems for femoral neck fracture and includes the most patients among all meta-analyses published to date. The analysis included three more of the most recent RCTs^19,22,23 for data synthesis than the number of studies in the most recent meta-analysis concerning the current-generation stems by Veldman et al, which included five RCTs.¹⁴ The review of Veldman et al in 2017 found that patients with cemented stems had a smaller number of total complications and implant-related complications but longer operating times; however, no significant differences were noted in the mortality rates, cardiovascular complications, number of reoperations, and intraoperative blood losses.¹⁴ Earlier systemic reviews investigated the trials, which included data from outdated stems (e.g., Austin Moore and Thompson stems). The Cochrane review conducted by Parker et al¹⁰ in 2010 stated that cemented hemiarthroplasty was associated with better mobility and less pain. The systemic review conducted by Luo et al in 2012 concluded that patients with cemented stems had less residual pain but found no significant differences in the mortality rates and numbers of complications and reoperations. The review of Li et al¹² in 2013 found that patients with cemented stems achieved better hip function, lower residual pain and fewer implant-related complications without an obvious increased risk in mortality rates, cardiovascular and cerebrovascular complications, general complications, local complications, and reoperation rates. Ning et al¹³ in 2014 stated that cemented hemiarthroplasty was associated with longer operation time than that of cementless hemiarthroplasty, whereas, no significant between-group differences were noted in mortality rates, hospital stay lengths, blood losses, residual pain, and complications.

Because of the variety of parameters used for postoperative functional outcome assessment, it is always difficult to pool these data. We extracted the data from three selected studies that assessed functional outcomes by the HHS and EQ-5D. According to the synthesized results, we found no significant between-group differences in short- and mid-term functional outcomes. This result slightly differed from those of a recently published RCT, which found that cemented stems had better short-term functional outcome than that of cementless stems, but the difference became insignificant by the mid-term follow-up.²² Better short-term functional outcomes in patients with cemented stems were also noted in an earlier study,¹⁸ but the difference became insignificant in further follow-up. An inferior outcome has been noted at the 5-year follow-up by Langslet et al. Although the pooled data showed no obvious between-group difference in a functional outcome comparison, the recent evidence still revealed a trend towards better short-term postoperative functional outcomes of cemented stem, but this trend became insignificant at longer follow-up times.

Mortality was documented inconsistently and ranged from 6 weeks to 5 years among the studies. The definition of perioperative mortality varied; therefore, we categorized those lethal events as major systemic complication. We found no between-group differences in the mortality rates at postoperative 3 months, 1 year, 2 years, and 4 years beyond. This result was consistent with those of previous reviews,^10–14 which indicated that the use of cement had no detrimental effect on short-term and mid-term mortalities.

In our study, we found that implant-related complications and reoperations were higher for cementless stems than for cemented stems. Since we focused on current-generation stem-related complications, such as periprosthetic fractures and prosthesis loosening, we performed both subgroup analyses with and without data of implant dislocation to eliminate the bias caused by the use of different implants (hemiarthroplasty or total hip arthroplasty). The results of both subgroup analyses showed identical tendencies that were consistent with the findings of the most recent review concerning current-generation stems.¹⁴ Furthermore, significantly higher numbers of reoperations in the cementless group emphasized the association between cementless stems and a higher rate of implant-related complications.

There has always been much concern that cement use may increase the risk of cardiovascular or cerebrovascular events, which has been described as “bone cement implantation syndrome.”²⁴ However, we found no significant between-group differences in the numbers of major systemic complications, including intraoperative cardiac arrest, acute arrhythmia, myocardial infarction, pulmonary embolus, and hypotensive crisis. These results were consistent with the results of most previous meta-analyses^10–14 no matter which kind of stem was used; nevertheless, a significantly higher rate of mortality within 24 hours after surgery associated with cemented stems was found by a study involving 60,848 patients and data from a national database.⁸ This inconsistency between the studies pointed out the uncertain incidence of bone cement implantation syndrome,^7,24 and more high-quality evidence studies are required to clarify this issue. Moreover, the incidences of minor local complications were similar between the two groups, which indicated that cement use had a lesser effect on local complications.

In our study, the operation times were similar between the two groups. This result differed from those of previous studies, which found that the use of cemented stems was associated with longer operation times, partly because of the time required for solidification of the cement.^12–14 We think that the different data synthesis results may be related to the high heterogeneity among the studies; in other words, the method for measuring operation time was affected by multiple factors. Additionally, we found no significant between-group differences in intraoperative blood losses and lengths of hospital stays, which was consistent with the results of previous reviews.^12–14

Our study may provide the most comprehensive data synthesis concerning the comparison of cemented and cementless hip arthroplasty using current-generation stems for treatment of displaced femoral neck fracture in elderly patients. The study results extend previous research by including current high-quality RCTs and by pooling comparable postoperative functional outcomes. Our findings are consistent with the findings of previous systemic reviews that the use of cemented stems was associated with fewer implant-related complications without an increased risk of mortality and life-threatening complications. The reoperation rate was found to be significantly higher in the cementless stem group than in the cemented stem group in our study, which is not mentioned in previous reviews investigating current-generation stem.

There were several limitations of the present meta-analysis. First, although more high-quality RCTs were included, there were still only eight included in our study, which may have led to unreliable results caused by publication bias assessments using only funnel plots. Additionally, because of variations in the parameters used to evaluate the clinical outcomes and in follow-up times among the trials, it was difficult to compile the preferred data for analysis from all of the trials. For example, the mortality rates in 2 years and lengths of hospital stay were identified in only four trials, and functional outcomes could be compiled only from three trials. In addition, in the measurements of functional outcomes, the pooled data included data from patients who underwent hemiarthroplasty and from those who underwent total hip arthroplasty, which means that there was a latent risk of selection bias since some of that evidence suggested that functional outcomes were better after total hip replacement than after hemiarthroplasty.^10,25 Lastly, we thought that the differences in implant design and surgical technique use led to the high heterogeneity among studies when measuring the operation times and intraoperative blood losses.

In conclusion, our comparison of cemented and cementless hip arthroplasty using current-generation stems found that cemented stems were associated with fewer implant-related complications and reoperations than those associated with cementless stems. There were no significant between-group differences in functional outcomes, mortality rates, major systemic, and minor local complications, operation times, intraoperative blood losses, and lengths of hospital stays. Some of the issues examined in our study will require more high-quality evidence before reliable conclusions can be drawn.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data related to this article can be found at http://doi.org/10.1097/JCMA.0000000000000264.