Aseptic loosening has been reported to be one of the most common reasons for early component revision in THA [5, 7, 16, 21]. Recent research regarding aseptic loosening of the femoral stem suggests a possible association with the surgical approach, although conclusions have differed among studies. An increased risk of early femoral component loosening was found (1) using the direct anterior approach compared with other common surgical approaches (posterior or anterolateral) ; (2) using the direct anterior and the direct lateral approaches compared with the posterior approach ; and (3) using the anterolateral approach compared with the posterior approach . In contrast, Sheth et al.  reported no difference in revisions for aseptic loosening among the anterior, anterolateral, and posterior approaches.
Several reasons have been proposed that may explain an increased risk of early femoral loosening using the direct anterior approach, including the surgeon’s learning curve with a new approach [3, 13], increased surgical time , and wound complications . In addition, it can be more difficult to achieve femoral exposure and to perform femoral preparation through the direct anterior approach , which may increase the risk of malalignment or undersizing of the femoral component and subsequent risk of aseptic loosening of the stem . Although one study  has also shown an increased risk of aseptic loosening with the anterolateral approach, the same factors may not be responsible, and other research has not supported this conclusion .
Another possible contributing factor to early aseptic loosening may be the design of the femoral stem used with different approaches, because this may affect placement and exposure. In the case of cemented THA, Lindgren et al.  found that two stems with an “anatomic” design (ie, a curved lateral profile or an obtuse angle at the proximal-lateral portion of the stem) were more likely to loosen when implanted through the anterolateral approach than those same stems were to loosen when implanted through the posterior approach. We had anecdotally observed in our clinic an increased revision rate for aseptic loosening when “shoulder” stems, ie, straight stems with a proximal shoulder (Fig. 1), were inserted through an anterolateral approach (data not shown). Based on these observations, we sought to investigate the association among the shape of the femoral stem, the surgical approach, and the early revision rate for aseptic loosening.
To do this, we used a nationwide population-based registry, the Dutch Arthroplasty Register (LROI), an interinstitutional database that contains information on joint arthroplasties in The Netherlands since 2007 , to investigate the association between surgical approach and early aseptic loosening of (1) cementless straight femoral stems with a proximal shoulder; and (2) anatomically shaped femoral stems.
Patients and Methods
Data were collected from the LROI that was founded in 2007 by the Dutch Orthopaedic Association. The LROI contains data on primary hip arthroplasty as well as revisions. Its completeness has been verified to be > 95% for primary THAs and 88% for hip revision arthroplasty .
In this registry, a primary THA was defined as the first implantation of a hip prosthesis designed to replace the hip. A revision was defined as any intervention in which one or more components were exchanged (that is, placed, replaced, or removed).
We selected all primary THAs with a cementless femoral stem inserted between 2007 and 2013 through the direct anterior, anterolateral, or posterior approach. Data on revision of the prosthesis, including time since primary surgery and reason for revision, were collected up to December 31, 2015. Consequently, all patients had a minimum of 2 years followup. Femoral stems that were used in < 10 hips were excluded. This resulted in 63,354 THAs in which 45 different stem brands were inserted. Based on their design, these 45 stem brands were classified into one of three categories: 25 “anatomic,” 12 “shoulder,” and eight “other” stems (Fig. 1; Table 1). We differentiated among these types by assessing the lateral side of the femoral stem. If the lateral side was one straight vertical line between the tip of the stem (distal) and the shoulder of the prosthesis (proximal), we classified the stem as “shoulder.” If the lateral side of the stem was curved or had an obtuse angle, we classified the stem as “anatomic.” The other category consisted of stems that did not fit one of the first two categories and included stems with an overall shape like the original Zweymüller® stem (Zweymüller® Alloclassic® SL; Zimmer, Warsaw, IN, USA), short stems (implant length inserted in the bone ≤ 120 mm ), and stems primarily meant for revision arthroplasty. Three researchers (LJ, DJ, JWM) independently put the classification into categories. Differences in the classification were discussed until consensus was reached.
For our analysis, we excluded the 15,982 THAs classified as “other,” resulting in 47,372 THAs included for the analyses: 32,591 THAs (69%) were classified as “anatomic,” whereas 14,781 THAs (31%) were classified as “shoulder.” The Corail® (DePuy Synthes, Warsaw, IN, USA; 26%), Taperloc® (Biomet, Warsaw, IN, USA; 25%), Mallory Head® (Biomet; 13%), and Accolade® (Stryker, Mahwah, NJ, USA; 10%) were the most frequently used anatomic stems. The CLS® Spotorno (Zimmer, Warsaw, IN, USA; 56%), Synergy (Smith & Nephew, Memphis, TN, USA; 22%), and Symax™ (Stryker, Montreux, Switzerland; 13%) were the most frequently used shoulder stems (Table 1).
In addition to the surgical approach and femoral stem design, we gathered patient demographics such as age, sex, diagnosis, American Society of Anesthesiologists (ASA) classification and previous surgical procedures in the ipsilateral hip, and additional prosthesis information such as coating and material of the femoral stem. Diagnosis was categorized as osteoarthritis or other (mainly consisting of osteonecrosis, dysplasia, acute fracture, or late posttraumatic). ASA classification was categorized as ASA I, ASA II, or ASA III to IV. Previous operations in the ipsilateral hip mainly involved osteosynthesis or osteotomy. Coating was categorized as hydroxyapatite, porous, or other. Material of the femoral stem was either titanium or other.
The total group thus consisted of 47,372 THAs inserted in 17,714 men (37%), 29,464 women (62%), and 194 patients (0.4%) in which sex was missing. Mean age was 66 years (SD, 10) at the time of the primary THA and mean followup was 3.5 years (SD, 1.8). The anterolateral approach was used in 5698 THAs (12%), the anterior approach in 5133 THAs (11%), and the posterior approach in 36,541 THAs (77%). Aseptic loosening of the femoral component was the second most frequently reported reason for revision (after dislocation) and was present in 340 patients (0.7%): 56 THAs (1.0%) inserted through the anterolateral approach, 44 THAs (0.9%) inserted through the anterior approach, and 240 THAs (0.7%) inserted through the posterior approach had to be revised for this reason. Furthermore, 1195 patients (2.5%) were lost to followup as a result of revision THA for other reasons than aseptic loosening: 109 THAs (1.9%) inserted through the anterolateral approach, 87 THAs (1.7%) inserted through the anterior approach, and 999 THAs (2.7%) inserted through the posterior approach. In addition, 1558 patients (3.3%) were lost to followup because they had died: 278 (4.9%) in the anterolateral group, 123 (2.4%) in the anterior group, and 1157 (3.2%) in the posterior group.
All baseline characteristics differed significantly among the different approaches (Table 2) and were later tested as possible confounders in the survival analyses, described subsequently.
Baseline characteristics (age, sex, diagnosis, ASA classification, earlier surgeries, and coating and material of the femoral stem) are presented as means (medians in case of a nonparametric distribution) or percentages and compared among the anterolateral, anterior, and posterior approaches using one-way analysis of variance (or Kruskal-Wallis test in case of nonparametric distribution) and chi square tests.
Cumulative incidences of revision resulting from aseptic loosening were calculated using competing risk analysis, where death and revisions for other reasons were considered competing risks. These were calculated for the posterior, the anterolateral, and the anterior approaches separately in patients with an anatomically shaped stem and in patients with an angular shoulder stem.
Our questions were tested using a Cox proportional hazard model. The independent variables were the surgical approach (anterolateral and anterior versus posterior), the type of femoral stem (shoulder or anatomic), and the interaction between these two variables. The dependent variable was revision resulting from aseptic loosening. Hazard ratios were calculated for comparison among approaches. Covariates entered in the regression model as a result of statistical significance after univariate testing were patient age, sex, diagnosis, ASA classification, earlier surgeries, and the coating and material of the femoral stem. An initial model, in which both approach and shape of the stem were entered as independent variables, yielded a significant interaction between these two variables (hazard ratio [HR], 1.87; 95% confidence interval [CI], 1.09-3.18; p = 0.02), disabling interpretation of the main effects. Therefore, we used two separate Cox proportional hazard models, one for each shape of stem (Table 3).
All statistical methodology was carried out using SPSS Version 22 (SPSS Inc, Chicago, IL, USA). Any p values < 0.05 were considered to be statistically significant. Missing values were not replaced.
After controlling for relevant confounding variables such as gender, ASA score, previous surgery, and coating and material of the femoral stem, we found that stems with a shoulder design had a greater likelihood of early aseptic loosening when the anterolateral approach was used compared with the posterior approach (HR, 2.28; 95% CI, 1.43–3.63; p < 0.001). Shoulder stems were also more likely to loosen early when the direct anterior approach was used compared with the posterior approach (HR, 10.47; 95% CI, 2.55-43.10; p = 0.001), although this result should be interpreted cautiously, because this was based on only three revised shoulder stems inserted through the direct anterior approach out of 100 revised shoulder stems in total. Analysis of the anatomic stems yielded no association with approach (anterolateral: HR, 1.07, 95% CI, 0.70–1.63, p = 0.77; anterior: HR, 1.31, 95% CI, 0.91-1.89, p = 0.15).
In line with these findings are the results from the competing risk analyses (Fig. 2A-B). Five years after implantation of an anatomic stem, the cumulative incidences of revision for aseptic loosening were similar for the various approaches (anterolateral: 1.02%, 95% CI, 0.67-1.53; direct anterior: 1.04%, 95% CI, 0.74-1.44; posterior: 0.86%, 95% CI, 0.73-1.01; Fig. 2A). However, for the shoulder stems, the 5-year cumulative incidence was larger for the anterolateral and the direct anterior approaches compared with the posterior approach (anterolateral: 1.34%, 95% CI, 0.92-1.95; anterior: 6.67%, 95% CI, 2.23-19.89; posterior: 0.66%, 95% CI, 0.51-0.85; Fig. 2B). Again, because only three shoulder stems that were inserted through the direct anterior approach were revised, this cumulative incidence of revision for the anterior approach should be interpreted with caution.
In cementless THA, controversy exists about the association between surgical approach used for implantation and the risk of early aseptic loosening of the stem. Stem design may also be associated with a differential risk of aseptic loosening in cementless THA depending on the approach used. In this study, we investigated the association between surgical approach and the risk of early aseptic loosening in anatomic stems and shoulder stems. For cementless THA, the risk of aseptic loosening was higher in stems with a shoulder design when an anterior or anterolateral approach was used compared with a posterior approach. For stems with an anatomic design, however, no association was found between the approach and aseptic loosening.
This registry study had a number of limitations. First, only a small number of patients had a shoulder stem that was inserted through the anterior approach. As a result, only three shoulder stems were revised that were inserted through the anterior approach. We applied the rule of thumb that Cox regression models should be used with a minimum of 10 outcome events per predictor variable . Because we have nine predictor variables, and 100 revisions among shoulder stems, our model seems justified. Needless to say, we cannot make firm conclusions about the risk of aseptic loosening of shoulder stems inserted through the anterior approach based on three patients, although the HR was large and statistically significant.
Second, the mean followup was rather short (3.5 years). Because most revisions for aseptic loosening took place within the first 2 years and the minimum followup was set to 2 years, we think that our followup was sufficient to study early aseptic loosening. Third, as is the case for registry studies in general, data were collected retrospectively and our results depend on the accuracy and completeness of the registration. The completeness of the LROI for THAs ranged from 88% in 2009 to 98% in 2012, which is comparable to other Scandinavian arthroplasty registries . Fourth, only variables that are collected in the registry could be analyzed. All possible confounding variables available in the registry data were tested and, if necessary, were corrected for in multivariate survival analyses. Body mass index, smoking habits, the surgeon’s level of experience, and hospital volume were not available, and these or other variables could theoretically have confounded our results. In addition, we did not have radiographic reports in the registry. Therefore, we were unable to check possible malpositioning of the femoral stem or subsidence as a cause for aseptic loosening. A final possible limitation may be our classification of the femoral stems into anatomic or shoulder, because this was rather subjective. To our knowledge, there is no available classification in the literature classifying different stems solely based on the (proximal) shape of the stem. Kim and Yoo  classified cementless hip implants based on fixation principles; however, this was not related to the shape of the stem. Khanuja et al.  did take the shape into account, but this resulted in six different types with nine subtypes. Because we were primarily interested in the shape of the lateral side of the femoral stem and the angle toward the neck of the stem, we used a simplified classification with only two levels that were included in the analyses. This shape may influence the ease of insertion and position in the different surgical approaches. Classifying all 45 stems into one of three groups (anatomic, shoulder, or other) resulted in three stem designs (comprising 460 stems) that initially were classified differently by the three researchers. These stems were labeled as “uncertain.” Sensitivity analyses in which the Cox regression analyses were repeated while all cases that had one of these four stems were excluded did not alter our results.
Our results show that the increased revision rate resulting from aseptic loosening that was observed for the anterior and the anterolateral approaches was only present in femoral stems with a shoulder (Fig. 1B). For anatomically shaped femoral stems (Fig. 1A), however, we did not find a difference between approaches in the risk of revision resulting from aseptic loosening. This differentiation based on the shape of the stem was not made in previous studies and may be a reason why these studies found conflicting results with respect to the relation between the approach and the risk of aseptic loosening [2, 6, 10, 12, 15, 17, 19]. Until now, only a few studies have investigated the influence of surgical approach on the revision rate [2, 6, 10, 12, 15, 17, 19]. As mentioned earlier, aseptic loosening is one of the most common reasons for revision reported [5, 6, 8, 10, 15], although the relationship with the surgical procedure used has not exhibited uniform results [2, 6, 10, 12, 15, 17, 19]. In our study, aseptic loosening was the second most frequently reported reason for revision. The primary reason for revision, dislocation, was more frequent in the posterior group (data not shown) compared with the anterior or the anterolateral group; however, this difference in frequency was not associated with the design of the stem.
One possible explanation for the increased aseptic loosening with the anterior approaches may be found in the difficulty of the approach and the exposure [1, 3, 12]. Both the anterior and anterolateral approaches are associated with more difficult femoral exposure [3, 12] as a result of obstruction of the gluteus medius muscle . This may cause more anterior preparation of the femoral canal resulting in malalignment of the femoral component [1, 3, 24] or undersizing of the stem [18, 22], which are possible contributors to an increased risk of aseptic loosening [1, 12, 22]. We theorize that the shape of the femoral stem could be an important factor with anatomically shaped stems conferring an advantage during broaching and insertion compared with angular shoulder stems. Finally, for the direct anterior approach, which has gained popularity over the past years, a learning curve may have caused a further increase in the risk of aseptic loosening found for this particular approach .
In summary, we showed that femoral stems with an angular shoulder exhibit an increased risk of aseptic loosening using the anterior or anterolateral approach compared with the posterior approach. Based on our results, we would suggest the use of an anatomically shaped stem when using an anterior or anterolateral approach, because those designs were associated with a reduced early risk of aseptic loosening. Furthermore, these results underline the importance of more in-depth analysis when studying the outcome and survival of different prostheses. Instead of solely comparing crude survival data of different implant designs, future studies need to consider a more detailed analysis including other factors—such as surgical approach—that can interact in ways that contribute to the failure or success of a specific design. Finally, we suggest that additional registry work should seek to confirm these findings; specifically, larger procedure numbers are needed to analyze shouldered prosthesis performance after implantation through the direct anterior approach.
We thank the LROI, in particular Liza van Steenbergen, for delivery of the data and statistical recommendations. In addition, we thank Sue Morrenhof for language editing and Thomas Timmers for figure editing.
1. Abe H, Sakai T, Takao M, Nishii T, Nakamura N, Sugano N. Difference in stem alignment between the direct anterior approach and the posterolateral approach in total hip arthroplasty. J Arthroplasty. 2015;30:1761–1766.
2. Arthursson AJ, Furnes O, Espehaug B, Havelin LI, Soreide JA. Prosthesis survival after total hip arthroplasty–does surgical approach matter? Analysis of 19,304 Charnley and 6,002 Exeter primary total hip arthroplasties reported to the Norwegian Arthroplasty Register. Acta Orthop. 2007;78:719–729.
3. Barton C, Kim PR. Complications of the direct anterior approach for total hip arthroplasty. Orthop Clin North Am. 2009;40:371–375.
4. de Steiger RN, Lorimer M, Solomon M. What is the learning curve for the anterior approach for total hip arthroplasty? Clin Orthop Relat Res. 2015;473:3860–3866.
5. Ehlinger M, Delaunay C, Karoubi M, Bonnomet F, Ramdane N, Hamadouche M; Société française de chirurgie orthopédique et traumatologique (SoFCOT). Revision of primary total hip arthroplasty for peri-prosthetic fracture: a prospective epidemiological study of 249 consecutive cases in France. Orthop Traumatol Surg Res. 2014;100:657–662.
6. Eto S, Hwang K, Huddleston JI, Amanatullah DF, Maloney WJ, Goodman SB. The direct anterior approach is associated with early revision total hip arthroplasty. J Arthroplasty. 2017;32:1001–1005.
7. Haynes JA, Stambough JB, Sassoon AA, Johnson SR, Clohisy JC, Nunley RM. Contemporary surgical indications and referral trends in revision total hip arthroplasty: a 10-year review. J Arthroplasty. 2016;31:622–625.
8. Jameson SS, Baker PN, Mason J, Gregg PJ, Brewster N, Deehan DJ, Reed MR. The design of the acetabular component and size of the femoral head influence the risk of revision following 34,721 single-brand cemented hip replacements: a retrospective cohort study of medium-term data from a National Joint Registry. J Bone Joint Surg Br. 2012;94:1611–1617.
9. Khanuja HS, Vakil JJ, Goddard MS, Mont MA. Cementless femoral fixation in total hip arthroplasty. J Bone Joint Surg Am. 2011;93:500–509.
10. Khatod M, Cafri G, Namba RS, Inacio MC, Paxton EW. Risk factors for total hip arthroplasty aseptic revision. J Arthroplasty. 2014;29:1412–1417.
11. Kim JT, Yoo JJ. Implant design in cementless hip arthroplasty. Hip Pelvis. 2016;28:65–75.
12. Lindgren V, Garellick G, Karrholm J, Wretenberg P. The type of surgical approach influences the risk of revision in total hip arthroplasty: a study from the Swedish Hip Arthroplasty Register of 90,662 total hipreplacements with 3 different cemented prostheses. Acta Orthop. 2012;83:559–565.
13. Masonis J, Thompson C, Odum S. Safe and accurate: learning the direct anterior total hip arthroplasty. Orthopedics. 2008;31(Suppl 2). pii: orthosupersite.com/view.asp?rID=37187.
14. McTighe T, Stulberg SD, Keppler L, Keggi J, Kennon R, Aram T, McPherson E. A classification system for short stem uncemented total hip arthroplasty. Bone Joint J. 2013;95(Suppl):260.
15. Meneghini RM, Elston AS, Chen AF, Kheir MM, Fehring TK, Springer BD. Direct anterior approach: risk factor for early femoral failure of cementless total hip arthroplasty: a multicenter study. J Bone Joint Surg Am. 2017;99:99–105.
16. Munger P, Roder C, Ackermann-Liebrich U, Busato A. Patient-related risk factors leading to aseptic stem loosening in total hip arthroplasty: a case-control study of 5,035 patients. Acta Orthop. 2006;77:567–574.
17. Palan J, Beard DJ, Murray DW, Andrew JG, Nolan J. Which approach for total hip arthroplasty: anterolateral or posterior? Clin Orthop Relat Res. 2009;467:473–477.
18. Rivera F, Leonardi F, Evangelista A, Pierannunzii L. Risk of stem undersizing with direct anterior approach for total hip arthroplasty. Hip Int. 2016;26:249–253.
19. Sheth D, Cafri G, Inacio MC, Paxton EW, Namba RS. Anterior and anterolateral approaches for THA are associated with lower dislocation risk without higher revision risk. Clin Orthop Relat Res. 2015;473:3401–3408.
20. Spaans AJ, van den Hout JA, Bolder SB. High complication rate in the early experience of minimally invasive total hip arthroplasty by the direct anterior approach. Acta Orthop. 2012;83:342–346.
21. Sundfeldt M, Carlsson LV, Johansson CB, Thomsen P, Gretzer C. Aseptic loosening, not only a question of wear: a review of different theories. Acta Orthop. 2006;77:177–197.
22. Thien TM, Karrholm J. Design-related risk factors for revision of primary cemented stems. Acta Orthop. 2010;81:407–412.
23. van Steenbergen LN, Denissen GA, Spooren A, van Rooden SM, van Oosterhout FJ, Morrenhof JW, Nelissen RG. More than 95% completeness of reported procedures in the population-based Dutch Arthroplasty Register. Acta Orthop. 2015;86:498–505.
24. Vaughan PD, Singh PJ, Teare R, Kucheria R, Singer GC. Femoral stem tip orientation and surgical approach in total hip arthroplasty. Hip Int. 2007;17:212–217.
25. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165:710–718.