Hip fractures are currently a public health problem, evidenced by a high and growing incidence. In Spain they increased from 26,834 in 2000 to 35,997 in 2012, with the respective impact in terms of mortality that this entails, as one in six patients dies during the first year.1,2 Further, those who survive experience a significant reduction in quality of life and autonomy measured by EuroQol-5D-3L (EQ-5D) and Barthel index, respectively.2 As the incidence is expected to increase in the near future due to population ageing, such a scenario will be associated with an even greater burden in terms of both, loss of health and loss of social and healthcare costs.1,2 In Europe, with over 610,000 cases, hip fractures accounted for more disability-adjusted life years lost than many common cancers.3 At the same time, their management entails high costs to address surgical care, medical care, and rehabilitation treatment.4 The social impact is also noteworthy, as denoted by the high percentage of community-dwelling patients which cannot be discharged to home.5

Displaced femoral neck fractures usually require replacement, most commonly with partial hip replacement, also called hip hemiarthroplasty (HA), given that it is less aggressive, achieves similar results in terms of health-related quality of life and is less expensive than total hip arthroplasty (THA).4,6 THA tends to be indicated for younger patients who have a better quality of life and longer life expectancy, as this type of replacement is more durable and associated with greater mobility.7 Nonetheless, patients of advanced age with good quality of life also receive THA.4 Questions arise regarding the indications for this type of replacement given that THA renders better health-related quality of life but with greater surgical morbidity and higher cost.6,7 Carroll et al.8 carried out a systematic review concluding that THA seems more cost-effective than HA. Nonetheless, they suggest the need for a cost-effectiveness analysis stratified as a function of various different clinical profiles.

As delivery of care should differ based on patients’ characteristics, the hypothesis was that the evaluation of resources used for hip replacement and their correspondence with patients’ needs would contribute to a better decision-making in the Basque Health Service. Moreover, the economic evaluation of hip replacement based on real-world data would allow to cover the gap of studies seeking to evaluate the efficiency of orthopaedic treatments using recommended methodologies.9 The objective of the study was to carry out a stratified cost-utility analysis as a function of the clinical subtypes of patients who underwent THA versus HA for the treatment of femoral neck fractures.

The clinical trial continues to be considered the research design providing the highest level of evidence. However, there are few economic studies of THA using individual patient data from randomised clinical trials and observational studies.10,11 Instead, the development of electronic health records has facilitated the creation of databases of clinical data linked to administrative data, enabling the recording of all patients’ interactions with the health system and the use of resources by individual patients (at different levels: primary care, emergency department, inpatient care, home care and outpatient specialist care).11 Such databases allow for the analysis of large samples of cases from clinical practice as a function of surgical procedure.9,11 But the analysis of samples of cases from clinical practice requires the use of procedures that improve the validity of the results, particularly regarding selection bias.12

The study identified 5867 patients diagnosed with femoral neck fractures, of whom 1307 and 4560 were treated with THA and HA, respectively. As shown in Table 1, the mean age of patients treated was 75.41 years for THA and 84.79 years for HA. Notably, 49% of patients below 80 years of age underwent THA and 51% HA. In contrast, only 12% of those ≥80 years old underwent THA. The univariate analysis did not show significant differences by sex or socioeconomic status between the two groups. Instead, significant differences were detected in ASA class, Charlson index, and age, the HA group patients having a higher Charlson index (mean 2.59 vs 1.97), as well as higher ASA class and age than those in the THA group. There were also significant differences in mortality, the rate being higher in the HA group (66.7% vs 35.8% in the THA group), despite the longer follow-up of patients who underwent THA (4.12 years vs 3.02 years in the HA group).

During follow-up, the percentage of patients with any type of complication during the first year after surgery was lower in the THA group than in the HA group, while the percentage of patients with some long-life complication was higher. Table 2 indicates that the costs were significantly higher for patients who underwent THA than for those who underwent HA.

Table 3 shows the cost-utility analysis carried out before and after genetic matching. In both cases, THA is more costly and more effective than HA, with incremental differences after balancing the groups of €1825 and 0.55 QALYs. In the analysis by ASA class and age subgroups, the same results for ASA class I and II patients were found. In contrast, HA was dominant for patients aged ≥80 years old and classified as ASA class III or IV.

Table 4 summarises the results of the multivariate SUR model for the incremental cost and effectiveness without any type of interaction. These results are in line with those of the univariate analysis, indicating that THA was more costly and more effective than HA, with an incremental cost of €2465 and an incremental effectiveness of 0.42 QALYs. The analysis of the subgroups derived from the interaction between ASA class and age in the SUR model (Table 5 and Fig. 1) revealed that THA was not cost effective in ASA class III-IV patients aged ≥ 80 years old. It should be noted that Tables 4 and 5 show the results of the SUR models in a summarised way. Nevertheless, the complete SUR models results are available in Tables II-IV of the online Appendix.

Figure 1

(0.2MB).

The main contribution of this economic assessment is to provide evidence that it is more efficient to perform HA in most patients with displaced femoral neck fracture and reserve THA for the youngest patients and those classified as low risk.22 This analysis has been possible thanks to the disaggregation of cost-effectiveness by subgroups as a function of age and ASA class. To restrict the economic assessment to the entire sample would have led to a misleading result, since it would not have addressed the clinical heterogeneity of patients. Further, the importance of adjusting the results is even more evident if the high cost of the surgical procedures studied is considered and that the already elevated incidence of femoral fractures is expected to grow in the future due to population ageing.23 It is also noteworthy that this is the first cost-utility analysis using individual clinical information extracted from corporate databases (real-world data) in Spain.11

The unstratified results indicated that THA was more expensive and associated with more QALYs than HA and that the ICUR was lower than the €22,000 per QALY willingness-to-pay threshold or cut-off value.24 Although the incremental cost and effectiveness were lower once the groups were balanced, the ICUR remained practically stable, only changing from €2912 per QALY to €3310 per QALY. When the SUR model was used to adjust the incremental cost-effectiveness as a function of the different variables, the results were similar, THA being more expensive and more effective than HA, with an ICUR of €5869 per QALY. On the other hand, when the interaction between ASA class and age was included, this conclusion only held for the subgroups of patients under 80 years of age. In these patients, THA was cost-effective, with both the ICUR and the confidence intervals below the cut-off value and not crossing the axes of the cost-effectiveness plane. In contrast, the results in the subgroups of patients ≥ 80 years old are harder to interpret. THA was not cost-effective in the low risk (ASA class I-II) group, the confidence intervals being above the cut-off, and was dominated by HA in the high risk (ASA class III-IV) group. As can be observed in the cost-effectiveness plane, in this subgroup, the two surgical procedures have the same effectiveness but THA has an incremental cost of €2576. In economic assessments, an intervention is considered dominated when it has a lower utility for the patient at a higher cost.17

The results of the economic evaluation fit well with the distribution of replacement type in the subsample of patients above 80 years old, only 12% of these patients undergoing THA. On the other hand, they may seem inconsistent with the distribution in under-80-year-olds, THA being the procedure of choice in 49% of cases. It is plausible that surgeon experience at the local level to carry out a more complex procedure such as THA helps to explain these differences.8 It has been reported that HA procedures are performed by a wide range of surgeons, while THA procedures in fractures tend to be carried out by fewer surgeons, with considerable experience.8 The results are in line with this. Specifically, when analysing the differences in the distribution of the replacement type as a function of hospital size, it was found that the procedures corresponded to THA in 17% of cases in large hospitals and 32% of cases in small hospitals. It is likely that one of the reasons for this difference is that patients with more comorbidities are referred to large hospitals.4

The results of studies reported in the literature are similar to ours, although they have not estimated cost-effectiveness as a function of clinical subgroup.8,25 Based on Markov models, Slover et al.25 calculated that the ICUR of THA vs HA was $1960 per QALY in the USA. They based themselves on a previous study by Keating et al.26 that indicated a better quality of life after 24 months in patients who underwent THA. The same data were used in another Markov model in the context of a systematic review to estimate an ICUR of £7952 per QALY in the United Kingdom.8 On the other hand, the improvement described by Keating et al. has not been confirmed in other studies;6 hence, in our study, the same utilities were assumed in both options.15 In the aforementioned studies, the great difference in utility scores is a key factor in the ICUR for THA being below the €22,000 per QALY cut-off, and hence, this intervention being considered cost-effective.26,27 Notably, in another study that assessed the long-term functional capacity of patients who underwent one of these two surgical interventions,28 there were no differences in the results at 12 years after surgery. Nonetheless, this study has been criticized for the methodology used to assess the functional status of patients, the lack of assessment before surgery and the fact that less than a quarter of the patients in both groups were alive at the end of the follow-up period.29

The observational design used in the study does not have the internal validity of clinical trials.11 Patient recruitment from databases that are representative of clinical practice can be seen as a limitation, given that the decision regarding type of surgery has been based on the judgement of different orthopaedic surgeons. On the other hand, seeking to support surgical decision making, observational studies that analyse real-world data are increasingly common, given their high external validity.18,30 The current availability of information from electronic medical records presents an ideal opportunity for measuring the real effectiveness and cost of different treatments. Nonetheless, given the non-random design of the study, the clinical characteristics of the two groups differ due to the selection bias. To overcome this, the recommendations of expert groups were followed regarding the use of real-world data in economic assessments.11 This approach includes propensity score analysis to ensure that the observed baseline distribution of the covariates is similar in both groups. In our case, the potential bias was addressed using a genetic matching algorithm.19 Now that the electronic health records and administrative registries facilitate access to real data from health systems, it is essential to develop and apply procedures that overcome their weaknesses and improve the validity of results obtained from their analysis.12

The main limitation of the study is the lack of information regarding utilities at the patient level. Given that the Basque Health Service does not systematically record utility scores for patients with hip prostheses, the data concerning utilities used for the calculation of QALYs was taken from a study that measured the quality of life of patients with hip prostheses.15 The same values were used in both groups of surgical patients, given that an international randomised controlled trial did not identify any significant difference in clinical improvement between THA and HA.6 In the future, the systematic recording of patient-reported outcome measures in clinical databases will allow us to standardise the use of economic assessments with data from clinical practice and enable decisions to be informed by specific data from each healthcare system.

The conclusion of this cost-utility study is that subgroup analysis supports current daily clinical practice in displaced femoral neck fractures, namely, using partial hip replacement in most patients and reserving total hip replacement for younger patients. Further, this study opens the use of real-world data for the economic evaluation of surgical procedures.

David Cantarero.

The corresponding author on behalf of the other authors guarantee the accuracy, transparency and honesty of the data and information contained in the study, that no relevant information has been omitted and that all discrepancies between authors have been adequately resolved and described.

I. Larrañaga and I. Etxebarria-Foronda conceived and designed the research. I. Etxebarria-Foronda, O. Ibarrondo and J.M. Martínez-Llorente obtained the data, and interpreted the data. I. Larrañaga, A. Gorostiza and C. Ojeda-Thies designed the methods, performed the analyses and interpreted the data. I. Larrañaga drafted the manuscript and approved the final manuscript. I. Etxebarria-Foronda critically revised the manuscript. I. Etxebarria-Foronda, O. Ibarrondo, A. Gorostiza, C. Ojeda-Thies and J.M. Martínez-Llorente revised the manuscript for important intellectual content and approved the final manuscript.

We would like to acknowledge the help of Ideas Need Communicating Language Services in improving the use of English in the manuscript.

The study was funded by the Basque Government Department of Health (grant number 2020111021). The funding source had no involvement in the study design, in the collection, analysis and interpretation of data, in the writing of the report and in the decision to submit the article.

None.