The current proposed total hip arthroplasty surgical planning guidelines based on classification of spine stiffness may be flawed due to incorrect assumptions

Abstract

Introduction:

The available classifications and preoperative planning tools for total hip arthroplasty (THA) assume that: 1) there is no variation in the sagittal pelvic tilt (SPT) if the radiographs are repeated, and 2) there is no significant change in the postoperative SPT postoperatively. We hypothesized that there would be significant differences in postoperative SPT tilt as measured by the sacral slope, thus rendering the current classifications and tools flawed.

Methods:

This study was a multicenter, retrospective analysis of preoperative and postoperative (1.5–6 months) full-body imaging of 237 primary THA (standing and sitting positions). Patients were categorized as 1) Stiff spine (standing sacral slope-sitting sacral slope<10°) and 2) Normal spine (standing sacral slope-sitting sacral slope≥10°). Results were compared using the paired t-test. The post hoc power analysis showed a power of 0.99.

Results:

The difference in mean standing and sitting sacral slope between the preoperative and postoperative measurements was 1°. However, in standing position, this difference was more than 10° in 14.4% of patients. In sitting position, this difference was more than 10° in 34.2% of patients and more than 20° in 9.8% of patients. Postoperatively, 32.5% of patients switched groups based on the classification, which rendered the preoperative planning suggested by the current classifications flawed.

Discussion:

Current preoperative planning and classifications are based on a single acquisition of preoperative radiographs without the incorporation of possible postoperative changes in SPT. Validated classifications and planning tools should incorporate repeated measurements to determine the mean and variance in SPT and consider the significant postoperative changes in SPT.

Introduction

Since the introduction of the hip-spine relationship, which provided a better understanding of the role of spine pathology and spine fusion in total hip arthroplasty (THA) instability [1–9], investigators have tried to simplify the process of recognizing patients at high risk, optimize implant orientation, and try to prevent hip impingement and dislocation [10–12]. The main goal of those performing THA has been the simplification of the process so that it can be performed using single preoperative lateral lumbar spine radiographs in the standing and relaxed sitting positions. This process eliminates the need for multiple radiographs, which can cause patient inconvenience, increase the cost of assessments, increase the total time spent by patients in the radiology unit, and increase the risks related to radiation. Sacral slope, which is measured relative to the horizontal line, is easy to measure and has been used in many studies; furthermore, the sacral slope has been included in the classifications used to recognize patients at risk for THA instability and dislocation [10,11]. Based on these classifications, if the change in the sacral slope from standing to sitting is < 10°, then the spine is considered stiff; if it is ≥10°, then the lumber spine is considered flexible.

There are potential flaws with this oversimplified approach, however. The main assumption of these approaches and classifications are that the sacral slope will not change after surgery. For example, the current classifications assume that if the lumbar spine flexibility is considered non-stiff based on the single preoperative acquisition of lumbar spine radiographs in the standing and sitting positions, then it will remain flexible postoperatively. The other possible flaw of these approaches is that they assume that the hip-spine relationship and complex three-dimensional pelvis and hip motions are without any variations when these motions are repeated. This approach also relies on the change in the sacral slope when assessing the lumbar spine flexibility instead of lumbar lordosis, which has been traditionally used to measure the lumbar spine flexibility [10,11]. Additionally, the relaxed seated position does not show the maximum flexibility of the spine compared to the flexed seated position when mimicking the sit-to-stand motion [13,14]

Considering all these limitations and possible flaws in the current approaches to the hip-spine relationship, we designed this study to determine whether the sagittal pelvic tilt as measured by the sacral slope changes after primary THA, and whether the preoperative classification of patients as having stiff or non-stiff spine flexibility based on the change in the sacral slope from standing to sitting remains the same postoperatively. We hypothesized that with improvements in pain and hip range of motion after THA, there would be significant differences in the sagittal pelvic tilt as measured by the sacral slope during postoperative radiographic imaging. We also hypothesized that the current simplified classifications of the hips-pine relationship based on the change in the sacral slope may be flawed because of their assumptions and, therefore, may be clinically in accurate.

Materials and Methods

This study was a multicenter, retrospective analysis of pre- and postoperative simultaneous full-body standing and relaxed sitting biplanar imaging (EOS^®; EOS imaging SA, Paris, France) of patients who underwent primary THA.

Study population

After obtaining institutional review board approval from both institutions, 237 consecutive patients who underwent primary THA at both institutions between 2013 and 2020 were reviewed during this retrospective cohort study. We included all patients, 18–90 years of age, who underwent primary THA for osteoarthritis, inflammatory arthritis or avascular necrosis and had high-quality preoperative and postoperative biplanar imaging results. Our exclusion criteria were additional spine or lower extremity surgery between preoperative and postoperative imaging, low image quality, history of ipsilateral hip fracture, revision THA or conversion of previous hip fracture surgery to THA. As the original paper did not exclude patients with previous surgery prior to preoperative hip radiographic imaging (spine or lower extremity), low back pain, gait abnormality or the use of assistive devices, we did not consider those as exclusion criteria either. All postoperative imaging acquisitions were performed between 6 and 12 weeks after primary THA. Our final study group included 125 female and 112 male patients, with mean age of 59.9 years (19.8–89).

Radiographic imaging technique

All patients get a full-body standing and relaxed sitting biplanar imaging in a routine fashion before and after THA in our institutions. For imaging, each patient stood and sat comfortably in the biplanar imaging machine, and the position was specifically checked to avoid superimposition of anatomical structures in the lateral view. During the relaxed sitting position, patients sat on a radiolucent chair with an adjustable height; their knees were bent to 90° and their feet were flat on the floor. Two well trained investigators performed all the measurements twice to ensure the accuracy.

Study variables

The sacral slope was defined as the angle between the superior plate of S1 and a horizontal line (Figure 1). The change in the sacral slope was calculated as follows: “standing sacral slope – sitting sacral slope” (figure 2 and 3). Patients were grouped based on the change in the sacral slope from standing to sitting as the non-stiff spine group (≥10° of change in the sacral slope from standing to sitting) or the stiff spine group (<10° of change in the sacral slope from standing to sitting) according to the published classification [10]. We included the same radiographic variables the above classification paper used [10]. We did not include other variables such as gait abnormality, history of back pain or use of assistive devices as these variables were not included in the proposed classification or decision-making tool.

Figure 1:

Sacral slope is the angle between the horizontal line and sacral plate.

Figure 2:

This patient tilts the pelvis posteriorly in standing position after THA [sacral slope increases from 49° to 37° (12° difference)]. Postoperatively, the patients leans back more while relaxed sitting (45°) compared to the preoperative imaging (20°). Patient’s group for sagittal spine stiffness changes from “stiff” preoperatively to “non-stiff” postoperatively.

Figure 3:

This patient tilts the pelvis anteriorly in standing position after THA (sacral slope increases from 51° to 54°). Postoperatively, the sacral slope changes 25°. Patient’s group for sagittal spine stiffness changes from “non-stiff” preoperatively to “stiff” postoperatively.

Statistical analysis

The normal distribution of values was checked using the Shapiro–Wilk normality test for each series of measurements. Data analysis was performed using the paired t-test and each patient was compared with himself/herself to minimize the variability. The significance level was set at 5% for simple statistical analyses. Statistical analysis was performed using Stata 16.1 (StataCorp LP, College Station, Texas, USA).

Power analysis

A post hoc power analysis was conducted using G*Power version 3.1.9 to determine the power of the study considering the current sample size. Using an alpha of 0.05 and an effect size of 0.89 calculated using Stata software, our total sample size of 237 provided a study power of 0.99.

Results

The mean difference in the sacral slope between the preoperative and postoperative measurements during standing was −1° (range: −47° to 20°; p=0.022). The mean difference in the sacral slope between the preoperative and postoperative measurements during sitting was 0.9° (range: −57° to 39°; p=0.25). This shows statistically significant (not clinically significant) change in the standing pelvic tilt postoperatively. However, what is most important is not the mean difference but the quantiles of difference in means (table 1). The differences in the standing sacral slope between the preoperative and postoperative measurements were >10° in 34 patients (14.2%)(figure 4). The difference between the preoperative and postoperative sitting sacral slope was even more pronounced; where 81 (34.2%) and 23 patients (9.8%) had differences of >10° and >20°, respectively (figure 5).

Table 1:

This table shows the difference in sacral slope between pre- and postoperative measurements.

Change in the sacral slope	<5°	5°–10°	10°–15°	15°–20°	20°–25°	25°–30°	30°–35°	>35°
Postoperative standing sacral slope-preoperative standing sacral slope	54.9%	30.8%	10.1%	2.5%	0.0%	0.8%	0.4%	0.4%
Postoperative sitting sacral slope-preoperative sitting sacral slope	38.0%	27.8%	14.8%	9.7%	5.5%	1.3%	0.4%	2.5%
Preoperative standing sacral slope-preoperative sitting sacral slope	15.8%	24.4%	13.1%	13.1%	12.7%	9.2%	5.2%	6.5%
Postoperative standing sacral slope-postoperative sitting sacral slope	23.1%	22.7%	20.1%	14.9%	6.1%	4.8%	3.5%	4.8%
Difference between the preoperative and postoperative standing-sacral slope change	34.9%	26.2%	21.0%	9.2%	4.8%	1.8%	1.8%	0.4%

As shown here, the difference between the “preoperative standing sacral slope-sitting sacral slope” and “postoperative standing sacral slope-sitting sacral slope) is more than 5° in 65% of the patients.

Figure 4:

This figure shows the quantiles of difference in the sacral slope between the preoperative and postoperative measurements in standing. As seen, this difference is less than 5° in 55% of patients and many patients will have more than 5° of difference postoperatively.

Figure 5:

This figure shows the quantiles of difference in the sacral slope between the preoperative and postoperative measurements in sitting. As seen, this difference is less than 5° in 38% of patients and most patients will have more than 5° of difference postoperatively.

We then compared the preoperative and postoperative change in sagittal pelvic tilt from standing to sitting. Preoperatively, the mean difference between the sitting and standing sacral slope was 14.8° (range: −10° to 55°). Postoperatively, this difference was 12.8° (range: −13° to 55°) (p=0.02). Quantiles of difference are more important than the difference in the mean. The difference between the preoperative and postoperative values were more than 5° in 161 patients (65.2%), more than 10° in 100 patients (39%), and more than 20° in 26 patients (8.8%). Postoperatively, 77 patients (32.5%) switched groups, between stiff and non-stiff, when the change in the sacral slope from standing to sitting was analyzed. The group changed from non-stiff spine to stiff spine for 46 patients (19.5%); furthermore, the group changed from stiff spine to non-stiff spine with a more sagittal pelvic tilt for 31 patients (13%). None of these patients underwent spine surgery between THA and the postoperative imaging performed 6 weeks to 6 months after the THA. This means that the preoperative planning and decision making may not address the correct deformity post operatively for over one third of the patients.

Discussion

During this study, we showed that the postoperative sagittal pelvic tilt during standing and sitting, as measured by the sacral slope, will differ significantly from the preoperative sagittal pelvic tilt. We have also showed that if we considered the published classifications of lumbar spine mobility, that are based on the change in the sacral slope from standing to sitting, then the postoperative group would change despite the patients not undergoing any spine procedures after their THA. These findings could affect both the planning and execution of the THA procedure.

Our study has several limitations. This was a retrospective analysis of imaging studies. None of the patients experienced a postoperative dislocation. Consequently, correlation of the findings with the occurrence of dislocation was not possible. None of our patients had two sets of preoperative EOS images. Thus, it was not possible to compare the two preoperative EOS images and see if pelvic tilt would change significantly if the imaging was repeated. It is very rare to repeat the preoperative EOS images on different days for clinical reasons. The difference in positioning among radiology personnel and the effect of patients’ disability and pain on positioning is always a real concern. Although we do not have repeated radiograph to show this effect, lack of incorporation of repeated measurement in proposed classification/tool will further prove our point. We only measured sagittal pelvic tilt, while coronal and axial pelvic tilt were not measured. Previous studies using computer simulations have shown that coronal and axial pelvic tilt could also affect the risk of prosthetic impingement [15–17]. We are not proposing a new classification system and guidelines for surgical planning in this project. We also did not perform a subanalysis to predict which patient population would have more or less sacral slope changes. The purpose of this project was to show the shortcomings of the proposed classification and guidelines not to propose a new one [10]. As a result, we did not include other variables such as history of spine surgery, back pain, severity of pain, severity of hip arthritis, severity of hip contracture, gait problems, contralateral hip pathology, need for assistive devices, true or perceived limb length discrepancy before and after surgery and did not repeat the EOS imaging multiple times pre- and postoperatively to assess the effect of repeated imaging on patients’ posture. These are all important variables required for a robust classification and guidelines or a prediction model to show who would have largest change in the sacral slope postoperatively. That was out of the scope of the current paper and they were not included in the methodology of the proposed classification/guidelines [10] either.

The introduction of hip-spine relationship to orthopaedic surgery provided a better understanding of the synchronized motions of the pelvis and spine to make upright standing and other motions of daily life possible [18]. Spine pathologies and fusion can result in pain and stiffness, thereby affecting the sagittal, coronal, and axial motions of the lumbar spine and spinopelvic junction. In this scenario, when patients stand up from the sitting position, they depend on more hip flexion and further anterior pelvic tilt to move the center of gravity forward in order to stand. This can increase the risks of prosthetic or bone-on-bone impingement and posterior dislocation. In some cases, the posterior pelvic tilt in the standing position increases because of the flat back deformity, which results in dependence on more hip extension with some degree of knee flexion to be able to stand upright. This increases the risks of posterior impingement and anterior dislocation. Patients with hip pathologies such as osteoarthritis and degenerative labral tears will experience pain during hip range of motion, especially deep flexion. These patients will naturally avoid deeper flexion of the hip to prevent loading the section of the joint affected by the pathology. They rely more on the spinopelvic junction and lumbar spine flexibility to move the center of the gravity forward to stand from a sitting position. They may also tilt their pelvis in the coronal and axial planes to offload the diseased hip. The severity of the pain (central/peripheral) and limited ipsilateral hip ROM and the presence of the contralateral hip pathology will affect the movement. However, this has not been sufficiently studied. Postoperative improvements in hip pain and range of motion as well as in the change in the balance between the periarticular agonist and antagonist muscles will change the motion patterns and how the patients stand, sit, pivot, and perform other activities of daily living. The postoperative degree of the change in motion patterns, how the coronal pelvic tilt changes, and how the axial pelvic tilt changes have not been sufficiently studied; most of the published data are focused on single preoperative radiographs obtained in the standing and sitting positions, thus ignoring all the other important variables.

To recognize those patients at high risk for THA dislocation and perform preoperative planning, researchers have introduced a simple classification system and guidelines for acetabular implant orientation [10]. For example, Vigdorchik et al. prospectively reviewed 3777 patients who underwent THA; of those, 2081 patients met their criteria for inclusion. All patients had one single standing radiograph and one single relaxed sitting radiograph that were used to measure the sacral slope and lumbar lordosis. Surgeries were performed by three experienced surgeons with 7 to 16 years of experience with computer navigation and robotic surgery. Patients were categorized into four groups based on the spine alignment and change in the sacral slope from standing to sitting. If the change in the sacral slope was <10° according to the single preoperative radiograph, then these patients were considered to have a stiff spine; however, if the change was ≥10°, then they were considered to have a non-stiff spine. Based on their data, they recommended a certain range for acetabular implant anteversion and inclination angle, as well as the possible use of a larger femoral head and dual-mobility bearing. Their study and classification may have significant limitations: 1) the assumption that the postoperative mobility will remain similar to the preoperative mobility, and 2) the results of measurement in one set of imaging is sufficient and repeating the imaging will provide the same pelvic tilt angles.

As we showed during the current study, when using a single preoperative image and one single postoperative image, the postoperative sacral slope was significantly different, compared to the preoperative one, in the majority of patients. These differences in the sacral slope were as high as 47° in the standing position and 57° in the relaxed sitting position. Patient groups based on the above classification changed for 32.5% of the patients. This means that the recommendation for implant position and bearing type may be inaccurate at the postoperative assessment due to changes in pelvic tilt. Sharma et al. [14] showed that a change in the sacral slope of <10° from standing to sitting was not correlated with spine stiffness when they compared the change in the sacral slope with lumbar lordosis. During their study, using the change in the sacral slope could result in a seven-fold overprediction of stiff spine and inaccurate surgical planning and execution [14].

A clinically reliable and valid prediction model requires repeated measurements in addition to including more clinically relevant data such as previous spine surgery, gait abnormalities, level of dependance on assistive devices. Patients may not have similar SPT in repeated measurement (figure 6). The best way of describing reliability and validity is through the illustration of marksman’s target (Figure 7). Figure 7A shows a measurement with good reliability but poor validity. Figure 7B shows a measurement with poor reliability and poor validity. Figure 7C shows a measurement with good validity and good reliability. Repeated measurements will provide not only the mean but also the extremes of the pelvic tilt (Figure 8). A valid and reliable model and classification of spine stiffness are necessary to ensure the accurate prediction of the motions during preoperative planning.

Figure 6:

One example of repeated preoperative EOS imaging in a patient with severe hip osteoarthritis. When repeating the EOS imaging, there was 5°difference in sacral slope in both standing (41° versus 36°) and relaxed sitting (−4° versus 1°). There was significant coronal and axial pelvic tilt as well more pronounced in relaxed sitting position. The proper orientation of the acetabular implant will not necessarily be similar in these two occasions.

Figure 7:

Figure 7A shows a measurement with good reliability but poor validity. Figure 7B shows a measurement with poor reliability and poor validity. Figure 7C shows a measurement with good validity and good reliability.

Figure 8:

A valid and reliable classification for hip-spine relation requires multiple measurement. It will consider the mean as well as the extremes of pelvic tilt and predicted postoperative change in pelvic tilt for surgical planning.

Conclusion

The classification of the lumbar spine stiffness based on a single preoperative sacral slope measurement in two positions during preoperative planning and proposed guidelines for baring type and acetabular implant orientation is flawed and may result in inaccurate assumptions. Future work and classifications should rely on either low-radiation imaging or newer alternative technologies to perform repeated measurements of the three-dimensional pelvic tilt and hip motions preoperatively. Furthermore, it is necessary to create models that could make valid assumptions regarding the degree of improvement in the postoperative range of motion and regarding the modification of the pelvic tilt and hip motions so that valid preoperative planning can be performed using computer simulations. Guidelines should include history of spine surgery or back pain as well as dynamic variables such as gait or posture abnormalities in the models.