Introduction

Adult scoliosis occurs in a heterogenous population, with diversity of age, diagnosis and magnitude [1]. One’s own interpretation of their functional demands, pain and body image further differentiates clinical presentations. Adult spinal deformity (ASD) can be evaluated morphologically by performing full-spine x-rays, in the standing position, to measure specific radiographic parameters. HRQL (Health-related quality of life) questionnaires including Oswestry disability index (ODI), Scoliosis Research Society 22 Questionnaire (SRS-22) and Short Form 36 (SF36) are commonly used to assess adult spinal patients for pain, function and quality of life, with varying degrees of diversity, sensitivity and specificity.

HRQLs overall have been shown to be improved with surgery, at high volume centres, whereas conservative treatments (including physical therapy, chiropractic treatment, bracing and injections) have not shown the same benefit [2,3,4]. However, surgical treatment of ASD will likely cause more complications than non-operative treatments. Thus, the decision to surgically correct adult scoliosis should weigh the potential benefits, including the improvement in HRQL, against the potential complications of the proposed surgical treatment [5]. These decisions are therefore complex and patients need to understand both the chances of improvement and the burden associated with operative and non-operative approaches in order to offer informed consent [6, 7].

While global improvement in HRQL is useful for patients’ knowledge, they are more interested in the specific impact it will have on them. Thus, it is important for patients to understand what HRQL subclasses are more likely to be affected and from what aspects of deformity correction. This cohort evaluated adult scoliosis patients to identify what significant HRQL sub-class changes are likely after surgery and whether they relate to the degree of sagittal or coronal plane correction.

Patients and Methods

Patient Cohort

This study was an analysis of a multicentre prospective database of consecutive ASD patients, who had been evaluated and had undergone surgical treatment at five European spine centres, from June 2007 to June 2016. Each enrolled site obtained institutional review board approval according to the common protocol. The inclusion criteria were patients older than 18 years with whole spine radiographs confirming a coronal Cobb angle ≥ 20°, sagittal vertical axis (SVA) ≥ 5 cm; thoracic kyphosis (TK) ≥ 60°; or pelvic tilt (PT) ≥ 25° who had surgical deformity correction. Exclusion criteria were all non-scoliotic patients (Schwab coronal N), single level surgery or non-deformity correcting surgeries. Demographic data of the patients, including age, gender and body mass index (BMI) were collected.

Radiographic Data

All patients had obtained standing postero-anterior and lateral full spine radiographs at baseline and at 1- and 2-year follow up. Coronal plane measurements included coronal Cobb angle for the major curve. Sagittal plane measurements included relative spinopelvic alignment (RSA). The relative spinopelvic alignment (RSA) parameter is a PI based global parameter that evaluates the amount of malalignment based on a patient’s ideal global tilt (GT) (RSA = GT- ideal GT with ideal GT = 0.48xPI-15) [8, 9]. Relative Sagittal Alignment (RSA) was used in this analysis as it is sufficient as a single sagittal parameter to take into account spinal malalignment and pelvic compensation [10].

HRQL Scores

All patients were asked to complete the Oswestry disability index (ODI) questionnaire, 36-item short-form health survey (SF-36), and Scoliosis Research Society-22 score (SRS-22) at enrolment. The Short Form (SF)-36, Oswestry disability index (ODI) and Scoliosis Research Society (SRS-22) score are universally used for evaluating ASD. The SRS-22 is the only disease specific instrument for ASD, despite being originally developed for adolescent idiopathic scoliosis. The SF-36 health survey comprises 36 items that measure eight sub-classes: physical functioning, role-physical, body pain, general health, vitality (VT), social functioning, role-emotional, and mental health. For each scale, a score ranging from 0 (worst measured health) to 100 (best measured health) was calculated. The ODI contains 10 sections: pain intensity, personal care, lifting, walking, sitting, standing, sleeping, social life, sex life, and traveling. For each subclass, a score ranging from 0 (best measured health) to 5 (worst measured health) was calculated; to calculate the level of disability, the points for each section were added and used in the following formula: point total/50 × 100 = % disability. SRS-22 scores have been shown to be reliable and with good-to-excellent internal consistency and strong test–retest reliability [11]; it exhibits concurrent validity with the corresponding SF-36, SF-12, and ODI domains. Although the SF-36 is a general health instrument and ODI is an assessment tool that is specific to back pain, four of the five subclasses from SRS- 22 were used: pain, function/activity, self-image, and mental health domains, which reflect the diverse symptoms in this population while satisfaction with management was excluded.

Statistical Analysis

SPSS version 17.0 (SPSS Inc., Chicago, IL) was used for the statistical analyses.

A multivariate analysis was designed to provide the impact of each independent variable on the dependent variable, to allow the statistical analysis of many variables at once.

Multivariate analysis included:

  • One dependent variable: dHRQL = HRQL postop – HRQL preop (d, delta)

  • Multiple independent variables: dRSA = RSA postop – RSA preop, dCobb = Cobb postop – Cobb preop, age, gender, BMI.

Multivariate analysis for each HRQL item: dHRQL = (a × dRSA) + (b × dCobb) + (c × Age) + (d × Gender) + (e × BMI).

Means and standard deviations (SDs) were used to describe continuous variables. Changes from the baseline to the outcomes at 1- or 2-years were evaluated using a paired t test analysis, and group comparisons were conducted using an unpaired t-test analysis. The significance level was set at 0.05. A standardized beta coefficient (Std. beta) was used to compare the strength of the effect of each individual independent variable to the dependent variable, so that the higher the absolute value of the beta coefficient, the stronger the effect. A negative beta coefficient would indicate that for every 1-unit increase in the predictor variable, the outcome variable would decrease by the beta coefficient value.

As the two most common cohorts were Idiopathic and Degenerative ASD, these were analysed in terms of mean values, standard distribution, distribution (Kolmogorov–Smirnov test of normality, KS) and for significant difference between the mean baseline-2 year changes (independent t-test). Correlations (Pearson’s co-efficient) were calculated between increasing age and mean baseline-2 year HRQL subclass changes, to identify if the surgery had an increasing effect on subjects as they got older.

Results

A total of 353 patients were included in this study, comprising of 289 females and 64 males. The average age was 49.1 (SD 19.5) and BMI was 24.3 (SD 4.4). Table 1 demonstrates the changes in radiographic and HRQL scores over the 2-year period. All HRQL total scores significantly improved postoperatively, including ODI, SRS-22 and SF36.

Table 1 Univariate analysis
Full size table

Tables 2, 3 and 4 demonstrate the beta coefficient results for the ODI, SRS22 and SF36, respectively. Table 5 summarizes the significant findings. HRQL subclasses which displayed persistent (1- and 2-year) improvements which correlated to dRSA included sex life from ODI, self-image from SRS22 and fatigue, vitality and social functioning from SF36. The only HRQL subclass improvement that correlated with dCobb was self-image from SRS22. No other change in HRQL subclass correlated with change in dCobb.

Table 2 Oswestry Disability Index subclass pre- to post-operative changes (dODI) due to coronal (dCobb) and sagittal (dRSA) corrections (standardized beta coefficient, Std. beta). The only subclass to consistently demonstrate significance of effect was sagittal correction on sex life. No coronal correction affected any HRQL subclass
Full size table
Table 3 Scoliosis Research Society (SRS22) subclass changes relative to coronal (dCobb) and sagittal (dRSA) corrections (standardized beta coefficient, Std. beta). HRQL subclass variability was persistently associated with dRSA and dCobb for self-image
Full size table
Table 4 SF36 subclass changes relative to coronal (dCobb) and sagittal (dRSA) corrections (standardized beta coefficient, Std. beta). HRQL subclass variability was persistently associated with dRSA for reductions in fatigue, vitality and social functioning. No coronal correction affected any HRQL subclass
Full size table
Table 5 Summary of HRQL subclass changes relative to coronal (dCobb) and sagittal (dRSA) corrections (standardized beta coefficient, Std. beta)
Full size table

Age profiles of Idiopathic and Degenerative cohorts were means (standard deviation) of 42.4 18.1) and 66.5 (10.3) years, respectively. KS test statistics were 0.12 and 0.07, p-values were 0.11 and 0.72 indicating normal distribution. Mean values for dODI, dSRS, dSF36PCS, dSF36MCS were −10.1 and −14.1, 0.6 and 0.7, 5.7 and 6.8 and 4.0 and 6.0 without significant differences evident for any group (p > 0.05). Correlation scores, r assessing increasing age and dODI sex life, dSRS self-image, dSF36 Fatigue, dSF36 vitality and dSF36 social functioning were 0.46, −0.14, −0.17, −0.15 and −0.19, respectively.

Discussion

As adult spinal deformities are increasing in prevalence in western societies, particularly in ageing population groups, HRQLs have gained in popularity as an effective method for evaluating the burden of this disease and benefits of treatment. ASD has been shown to display significantly lower HRQL scores than the other chronic conditions including arthritis, chronic lung disease, diabetes and congestive heart failure [12]. Consistent with previous data from within this and other databases, HRQLs and their sub-classes improve post scoliotic correction in adult spinal deformity cases [2, 13]. Relative sagittal alignment (RSA) was used in this analysis as it is sufficient as a single sagittal parameter to take into account spinal malalignment and pelvic compensation [10]. With overall improvement in HRQL, a limited number of sub-class improvements correlated with improvements in degrees of sagittal correction (dRSA) but only self-image improved relative to coronal correction (dCobb) (Table 5).

The amount of RSA correction persistently correlated with improvements in HQRL for sex life, fatigue and vitality, without corresponding correlations in coronal Cobb angle correction. Social functioning from SF36 also correlated with sagittal correction but was relatively unchanged from 1 to 2 years, although it was also evident from the ODI after 2 years. Walking distances also correlated with sagittal correction from the SF36, albeit expressed as walking “several blocks” at 1-year and “more than a mile” at 2 years. This was not reflected in the ODI and did not correlate with climbing several flights of stairs, perhaps because of other limitations. Pre-existent patient factors, complications, revision surgery and neurologic pain, etc. are also key factors that if experienced effect the HRQLs even with sufficient correction [3, 14,15,16]. While self-image correlated with both RSA and Cobb changes, it was not as lasting as other significant correlative parameters as it had a lower co-efficient at 2 years, indicating that the satisfaction achieved with post-operative appearance is unlikely to improve with further healing or rehabilitation, as expected. Self-image, if reported as a predominant symptom preoperatively, would indicate that measures to achieve a greater Cobb correction are a priority surgical objective. Otherwise, achieving coronal correction in ASD remains important overall, but the degree to which it is achieved will not correlate with gains in sub-classes of HRQL.

The relative influence of clinical and radiological factors on the decision-making process has recently been reported. Some studies have shown that coronal deformity is an essential factor for decision-making in ASD [5, 17]. Proven characteristics that aid the decision-making process in ASD correction in patients under 40 years include self-image score in the SRS-22 score, coronal Cobb angle, PI-LL mismatch, and RSA [5]. In those older than 40 years, pain and self-image domains in the SRS-22 score, the coronal Cobb angle and RSA were reliably the most predictive scores for the selection of surgical management [18]. In both cohorts, coronal correction was linked to sagittal correction. It must be considered that this study focussed on subjects with a preoperative Cobb angle of > 30° residual curve. A residual coronal curve may be an acceptable outcome (Figs. 1 and 2), in the knowledge that a greater coronal angular correction does not yield greater improvements in HRQL.

Fig. 1
figure 1

Clinical photograph of female patient with coronal and sagittal deformity; preoperative

Full size image
Fig. 2
figure 2

Post-operative photograph. Residual coronal curve is less important than achieving head-over-pelvis coronal-plane correction

Full size image

Identification of a threshold beyond which coronal correction correlates with HRQL sub-class improvement is the subject of ongoing work. Differential item functioning analysis from the same database reveals that coronal balance is not associated with HRQL outcomes but a cobb correction greater than 33° is associated with a poorer SRS-22 score [15]. Cohorts of Idiopathic and Degenerative scoliosis demonstrated different age patterns, as expected. They did not yield significant differences in baseline-2 years changes in HRQLs nor were there strong correlations between increasing age and the baseline-2 years changes for sex life, self-image, fatigue, vitality or social functioning.

This study has some limitations warranting consideration. Firstly, the mean age of patients from this report is 49 years and standard distribution is 19.1. It is likely that the findings of sex life and self-image may vary according to age which was not assessed. Coronal deformity has more recently been classified according to coronal translation, stiffness, mobility and degeneration of the lumbosacral junction, however these parameters were not analysed in the current study [19, 20]. Previous HRQL subclass analysis on 170 patients from an earlier version of the same database demonstrated the restrictive effects of instrumentation extending to the pelvis [3]. That study demonstrated that personal care and lifting from the ODI were not improved after 1 year. These disadvantages were correlated to sagittal modifiers of SRS-Schwab classification similar to other HRQL. The degree of personal care disadvantage was mainly dependant on the lower instrumented vertebra (LIV) location and preoperative pathology. These parameters were not included in our study. While single level and selective fusions were excluded it cannot be affirmed that all patients were complaining of malalignment. Coronal translation was not included in this study which remains an important aspect of coronal correction, particularly where there is coronal decompensation.

Conclusion

Adult scoliosis correctional surgery improves overall HRQL with a limited number of sub-classes demonstrating improvements relative to the amount of sagittal correction. Achieving coronal correction remains important overall, but the degree to which it is achieved, apart from self-image, will not correlate with gains in sub-classes of HRQL. Aggressive corrections of coronal deformity may not be required in ASD surgery.