Surgical interventions for treating extracapsular hip fractures in older adults: a network meta‐analysis

Objectives

To assess the relative effects (benefits and harms) of all surgical treatments used in the management of extracapsular hip fractures in older adults, using a network meta‐analysis of randomised trials, and to generate a hierarchy of interventions according to their outcomes.

Search methods

We searched CENTRAL, MEDLINE, Embase, Web of Science and five other databases in July 2020.

Selection criteria

We included randomised controlled trials (RCTs) and quasi‐RCTs comparing different treatments for fragility extracapsular hip fractures in older adults. We included internal and external fixation, arthroplasties and non‐operative treatment. We excluded studies of hip fractures with specific pathologies other than osteoporosis or resulting from high‐energy trauma.

Data collection and analysis

Two review authors independently assessed studies for inclusion. One review author completed data extraction which was checked by a second review author. We collected data for three outcomes at different time points: mortality and health‐related quality of life (HRQoL) ‐ both reported within 4 months, at 12 months and after 24 months of surgery, and unplanned return to theatre (at end of study follow‐up).

We performed a network meta‐analysis (NMA) with Stata software, using frequentist methods, and calculated the differences between treatments using risk ratios (RRs) and standardised mean differences (SMDs) and their corresponding 95% confidence intervals (CIs). We also performed direct comparisons using the same codes.

Main results

We included 184 studies (160 RCTs and 24 quasi‐RCTs) with 26,073 participants with 26,086 extracapsular hip fractures in the review. The mean age in most studies ranged from 60 to 93 years, and 69% were women.

After discussion with clinical experts, we selected nine nodes that represented the best balance between clinical plausibility and efficiency of the networks: fixed angle plate (dynamic and static), cephalomedullary nail (short and long), condylocephalic nail, external fixation, hemiarthroplasty, total hip arthroplasty (THA) and non‐operative treatment. Seventy‐three studies (with 11,126 participants) with data for at least two of these treatments contributed to the NMA.

We selected the dynamic fixed angle plate as a reference treatment against which other treatments were compared. This was a common treatment in the networks, providing a clinically appropriate comparison.

We downgraded the certainty of the evidence for serious and very serious risks of bias, and because some of the estimates included the possibility of transitivity owing to the proportion of stable and unstable fractures between treatment comparisons. We also downgraded if we noted evidence of inconsistency in direct or indirect estimates from which the network estimate was derived. Most estimates included the possibility of benefits and harms, and we downgraded the evidence for these treatments for imprecision.

Overall, 20.2% of participants who received the reference treatment had died by 12 months after surgery. We noted no evidence of any differences in mortality at this time point between the treatments compared. Effect estimates of all treatments included plausible benefits as well as harms. Short cephalomedullary nails had the narrowest confidence interval (CI), with 7 fewer deaths (26 fewer to 15 more) per 1000 participants, compared to the reference treatment (risk ratio (RR) 0.97, 95% CI 0.87 to 1.07). THA had the widest CI, with 62 fewer deaths (177 fewer to 610 more) per 1000 participants, compared to the reference treatment (RR 0.69, 95% CI 0.12 to 4.03). The certainty of the evidence for all treatments was low to very low. Although we ranked the treatments, this ranking should be interpreted cautiously because of the imprecision in all the network estimates for these treatments.

Overall, 4.3% of participants who received the reference treatment had unplanned return to theatre. Compared to this treatment, we found very low‐certainty evidence that 58 more participants (14 to 137 more) per 1000 participants returned to theatre if they were treated with a static fixed angle plate (RR 2.48, 95% CI 1.36 to 4.50), and 91 more participants (37 to 182 more) per 1000 participants returned to theatre if treated with a condylocephalic nail (RR 3.33, 95% CI 1.95 to 5.68). We also found that these treatments were ranked as having the highest probability of unplanned return to theatre. In the remaining treatments, we noted no evidence of any differences in unplanned return to theatre, with effect estimates including benefits as well as harms. The certainty of the evidence for these other treatments ranged from low to very low.

We did not use GRADE to assess the certainty of the evidence for early mortality, but our findings were similar to those for 12‐month mortality, with no evidence of any differences in treatments when compared to dynamic fixed angle plate. Very few studies reported HRQoL and we were unable to build networks from these studies and perform network meta‐analysis.