Abstract
PURPOSE:
This study aimed to provide a reliable and valid translation of the Scoliosis Research Society-22 (SRS-22r) questionnaire, compare it with the EQ-5D-5 L questionnaire, and analyse health-related quality of life (HRQoL) of patients with idiopathic scoliosis (IS) in Slovenia in order to potentially improve their rehabilitation processes.
METHODS:
A matched-case-control study was performed to assess internal consistency reliability, test-retest reliability, concurrent validity, and discriminative validity. The questionnaire was returned by 25 adolescent IS patients, 25 adult IS patients, and 25 healthy controls (87%, 71%, and 100% response rate, respectively).
RESULTS:
Internal consistency was high for all four scales in the adult IS group, but lower among the adolescent patients. Test-retest reliability of the SRS-22r was high to very high in both patient groups. Correlations between SRS-22r and EQ-5D-5 L were low or close to zero among adolescent patients and moderate or high among adult IS patients. SRS-22r domain scores were statistically significantly different between adult patients and healthy controls.
CONCLUSION:
The study proved that the Slovenian version of SRS-22r has the psychometric properties needed to measure HRQoL, whereby it appears to be more reliable for adults than adolescents. When used with IS adolescents, SRS-22r is affected by a severe ceiling effect. It could be used for longitudinal follow-up of adult patients after rehabilitation treatment. Additionally, some important issues that adolescents and adults with IS are faced with were identified.
Introduction
Idiopathic scoliosis (IS) is a three-dimensional deformity of the spine and trunk, characterized by deformation in the sagittal, frontal, and transverse planes [1]. For the vast majority of patients it is not life-threating, but certainly affects their health-related quality of life (HRQoL) and psychological well-being [2, 3]. Clinical evaluation and radiologic measurements prevail as the most important outcome measures in the literature. However, Negrini et al. found that experts in rehabilitation and orthopaedic conservative treatment give importance to a wide range of outcome criteria: aesthetics, HRQoL, disability, back pain, psychological well-being, balance, and posture. Among these, clinical and radiographic issues (apart from the Cobb angle) had the lowest importance [4]. Nevertheless, no study that focused on the priorities of individuals with IS as its main outcome measure was found.
Evaluating HRQoL requires appropriate outcome measurements that are easy to administer, reliable, valid, responsive to change, and focused on the unique health issues related to IS, whereby specifics should be addressed for children/youth (treatment impact on HRQoL) and adults (long-term results and/or consequences of pathology) [4–6]. The Scoliosis Research Society-22 (SRS-22r) questionnaire is the most frequently used [7]. Another, the EQ-5D-5 L questionnaire has been recommended together with SRS-22r as the core outcome measure for adolescents and adults with spinal deformity [5].
As there was no HRQoL measure for patients with IS available in Slovenia, this study aimed to provide a reliable and valid translation of the SRS-22r and examine its concurrent validity with the EQ-5D-5 L questionnaire. This study also aimed to analyse the HRQoL of children/youth with IS, which would help get better insight into the rehabilitation process and potentially improve it.
Materials and methods
Participants
The outpatients with IS were consecutively enrolled from the two spinal deformity referral centres (University Rehabilitation Institute in Ljubljana and Valdoltra Orthopaedic Hospital, VOH) between November 2017 and November 2018. The adolescent IS (AIS) group consisted of patients between 10 and 18 years of age, treated conservatively during referral time. The adult IS (AduIS) group consisted of patients over 18 years of age who were treated conservatively in adolescent years, but during the referral time were only under observation. All the patients were treated conservatively (observation, physiotherapy, brace) according to the International Scientific Society on Scoliosis Orthopaedic and Rehabilitation Treatment (SOSORT) [8] and were able to read and speak Slovenian fluently. The healthy control (HC) group consisted of adult VOH employees, matched by age and gender to AduIS patients. They did not have any spinal deformity or history related to any other spinal condition. All eligible participants gave their written consent to participate in the study. The exclusion criteria were previous scoliosis surgery, neurological or psychiatric disorder, or severe systemic illness.
Questionnaires
The SRS-22r consists of 22 questions, divided into five domains: Function, Pain, Mental Health, Self-Image, and Satisfaction with Management [7]. Each question is answered on a five-point scale (from 1 to 5). The sum of the first four domains gives a maximum subtotal of 100 points; with the addition of the Satisfaction domain, the maximum total is 110 points. Higher scores indicate fewer problems, i.e., better HRQoL. The questionnaire was validated in adolescent and adult population [7].
The guidelines for cross-cultural validation and adaptation of HRQoL measures were followed to prepare the Slovenian translation [9]. Two translators took part in the forward translation, a spine surgeon and a physical-and-rehabilitation-medicine specialist (both Slovenian native speakers, fluent in English). The translations were evaluated and combined into the final version by an expert committee. Backward translation was provided by a professional translator, who worked independently, did not receive any information about the previous translation process, and did not have any prior knowledge of the questionnaire. Backward translation was evaluated and compared to the original by the same expert committee.
The EuroQol EQ-5D-5 L questionnaire addresses five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [10]. Each is described by five possible levels (1 to 5), with a higher level indicating worse score: no problems, slight, moderate, severe and extreme problems [11]. The answers for all five dimensions can be combined into a 5-digit profile that describes the respondent’s health state. Using utility weights, they can be transformed from the general population into a single index ranging from 0 (worst) to 1 (best). As no value set for this purpose was available for Slovenia, the value set of a closely-resembling country was used, as recommended [12]. The questionnaire includes a vertical visual analogue scale (EQ-VAS) for patients to rate their overall current health from 0 (worst imaginable health) to 100 (best imaginable health) [12].
Protocol
The final version of the Slovenian SRS-22r was administered together with the EQ-5D-5 L to 89 subjects, 64 IS outpatients (29 AIS, 35 AduIS) and 25 HC. A total of 75 subjects returned the questionnaires: 25 AIS patients (87% response rate), 25 AduIS patients (71% response rate) and 25 HC (100% response rate). SRS-22r was administered again to both patient groups two weeks after the initial assessment. The Institutional Review Boards of both medical referral centres approved the study (#6/2017).
Statistics
Domain-specific and total scores for the SRS-22r were normalised to a 1–5 scale (i.e., divided by the number of corresponding items, so that they represent mean item score). Two types of reliability were estimated: internal consistency (using Cronbach alpha) and test-retest reliability (using intraclass correlation coefficient, ICC) [13]. To assess concurrent validity, Spearman correlations were computed between SRS-22r and EQ-5D-5 L scores. Statistical significance of those correlations is not reported, because p-values for correlation matrices are problematic and difficult to adjust for multiple tests; instead, the correlations are highlighted according to the 0.4 and 0.6 absolute thresholds (which correspond approximately to unadjusted p-threshold of 0.05 and 0.01, respectively). Discriminant validity was determined by comparing domain and subtotal mean scores between AduIS patients and age- and sex-matched HC using exact Wilcoxon signed-rank test. The HC did not receive treatment and therefore had no Satisfaction domain score. No statistical comparison between genders was feasible because of the very small number of male participants (especially in both adult samples). Statistical analyses were performed using IBM SPSS Statistics 23 (IBM Corp., Armonk, NY).
Sample size was constrained by the limited size of the accessible population of patients with AIS and AduIS in Slovenia. Nevertheless, n = 25 in each patient group is close to the minimum requirement for robust estimation of the population value if the first eigenvalue of the scale is above 3, which was the case in this study for nearly all the scales [14]. Furthermore, n = 25 is sufficient for detecting ICC values above 0.5 in a test-retest study with 90% statistical power (when alpha-level is set at 0.05), and ICC values above 0.8 (which were to be expected in this study) with nearly 100% power [15]. It is also sufficient to estimate an ICC of 0.85 in a test-retest study with precision (i.e., half the width of the 95% -confidence interval) of about 0.10 [16]. At the same time, n = 25 was also sufficient for the control group because of matched design, as the power of the Wilcoxon signed-rank test for large effect sizes (at least 1.0, which were to be expected in this study when comparing scale scores between AduIS and HC) is practically 100%.
Results
A total of 75 subjects completed the questionnaires: 25 AIS patients (87% response rate), 25 AduIS patients (71% response rate) and 25 HC (100% response rate). The large majority were women (there were two men in each of the AduIS and HC groups, and six men in the AIS group). Their mean Cobb angle varied from 28° (AIS) to 47° (AduIS). All AIS patients were performing physiotherapeutic scoliosis-specific exercises. Close to half also received a brace. None of the AduIS patients were referred to rehabilitation (Table 1).
Basic characteristics of the sample
Basic characteristics of the sample
Note: * magnitude of the main spinal curve.
Descriptive statistics for SRS-22r and EQ-5D-5 L scores are reported in Table 2. With SRS-22r, no floor effects were observed. A ceiling effect was observed among the HC on the Function and Pain scales, among the AduIS patients on the Satisfaction scale, and among AIS patients on the Function and Pain scales. With EQ-5D-5 L, the concept of floor and ceiling effect is only applicable to the index score and EQ-VAS. No remarkable floor effect was observed in any group; a ceiling effect was observed among AIS patients for index score, and among HC for index score and EQ-VAS.
Descriptive statistics for SRS-22r and EQ-5D scores
Note: NA –not applicable.
The estimated internal consistency was high for all four scales among the adult patients and HC, with Cronbach alpha ranging approximately from 0.8 to 0.9. Internal consistency was not estimated for the Satisfaction scale, which comprises only two items. The Cronbach alpha values were lower for AIS patients, ranging from 0.6 to 0.8 (Table 3).
Internal-consistency reliability estimates (Cronbach alpha) of the SRS-22r scales
Internal-consistency reliability estimates (Cronbach alpha) of the SRS-22r scales
Note: Satisfaction scale comprises only two items, so no estimate was computed.
The estimated test-retest reliability of the SRS-22r was high to very high in both patient groups (ICC ranging from 0.82 to 0.95), except for the Function and Satisfaction scales in AIS patients, where it was moderate (ICC 0.70 and 0.66, respectively; Table 4).
Repeatability estimates (intraclass correlation, ICC) of the SRS-22r scales
In the AIS group, the observed correlations between SRS-22r and EQ-5D-5 L were low or close to zero. The only exceptions were a high positive correlation of SRS-22r Pain domain with EQ-5D-5 L Pain/Discomfort dimension (negative, as expected) and overall index (positive, as expected), a moderate positive correlation of SRS-22r Function domain with EQ-VAS, and a moderate positive correlation of SRS-22r Subtotal with EQ-5D-5 L index (both positive, as expected). It should be noted that the correlations in this patient group are attenuated by very low variability of EQ-5D-5 L ratings (nearly all the answers on all the five dimensions were 1 = no problems, and only a few 2 = slight problems).
In the AduIS group, the majority of correlations were moderate or high, and all of them had the expected sign. The exceptions that did not correlate with other measures were SRS-22r Satisfaction scale and the EQ-5D-5 L Self-Care dimension (Table 5).
Correlations (Spearman rho) between SRS-22r (1st assessment) and EQ-5D-5 L scores among AIS patients
Correlations (Spearman rho) between SRS-22r (1st assessment) and EQ-5D-5 L scores among AIS patients
Notes: n = 25 for both groups for all correlations; correlations with absolute value between 0.4 and 0.6 are typeset in bold, and those with absolute value > 0.6 are typeset in bold and underlined (this correspond approximately to unadjusted p-value of < 0.05 and < 0.01, respectively); for SRS-22r scales and the EQ-5D-5 L index and EQ-VAS, a higher score indicates better health-related quality of life, whereas for the five EQ-5D-5 L dimensions, a higher score indicates more problems.
The HC had significantly higher scores in all SRS-22r domains than AduIS patients (p < 0.001 for each of the five domains; Table 2).
The present study describes cross-cultural translation and adaptation of the original SRS-22r into the Slovenian language. Barkley and Furse advise that a response rate of 50% should be aimed for inpatient satisfaction studies; response rates were substantially higher in this study [17]. This was achieved by clearly explaining the aim of the study in the invitation letter and ensuring the anonymity of the participants, without sending any reminders. The comparable Danish study had a 98% response rate in the AIS group and 93% in the control group; on the other hand, the German study had only a 35% response rate [18, 19].
Idiopathic scoliosis is a chronic condition that can lead to changes in physical appearance and functioning, development of chronic pain syndromes, and changed expectations in relation to work and leisure in adolescence and adult life [3]. The SRS-22r proved to be well-tailored to such patients by addressing their physical changes according to treatment (e.g., surgical or conservative), their psychological problems, and their quality of life expectations [20–22]. However, studies using Rasch analysis flagged some important limitations of SRS-22r, which eventually prevent it from comprehensively measuring a patient’s quality of life [23]. The present study aimed to compare the psychometric properties of SRS-22r among adults and adolescents with IS during (adolescents) or after (adults) rehabilitation treatments, and to identify possible weaknesses of the rehabilitation program that need improvement.
A ceiling effect was observed in the AIS group for Function and Pain domains, which corresponds to the results of other studies [23–25]. Because of inadequate metric properties of the questionnaire, a Rasch-consistent 7-item questionnaire (SRS-7) was prepared as a provisional solution by rearranging selected items from the original SRS-22 [23]. However, the metric properties of the SRS-7 remained unsatisfactory and thus a new questionnaire measuring HRQoL in adolescents with spinal deformities was developed [26]. Because all of the patients were treated conservatively (the majority had mild IS), a relatively high percentage of high-score answers was expected [23]. Pain is often associated with scoliosis in adults, the elderly, and in those with severe curves at skeletal maturity, so a less-pronounced ceiling effect is expected among those patients [23].
The adult patients were not referred to any physiotherapy nor comprehensive therapy program, so this should change in future to lower the prevalence of pain in these patients and potentially improve their quality of life.
In the AduIS group, ceiling effect was observed only on the SRS-22r Satisfaction scale (questions #21 and #22). Only two items are probably not enough for the scale to be sensitive to small differences between participants. The patients are usually not inclined to complain, so the ceiling effect was expected. This also means that the results are not informative enough to lead to much improvement in their rehabilitation programs. On the other hand, AIS patients reported higher results on the Satisfaction scale on average, with smaller ceiling effect. Hence, adolescents were satisfied with the results and content of ongoing rehabilitation. This differs somewhat from Piantoni et al. [27], who used the Brace Questionnaire [28]. Their AIS patients treated with bracing reported a negative impact on quality of life and treatment satisfaction in terms of psychological, motor, social, and school environment aspects, so they suggested that an interdisciplinary approach would be required for integrated psychosocial care of such patients [27].
Self-esteem is defined as a complex and multidimensional perception of the inner-self, determining a favourable relation to oneself. Durmała et al. reported on the impact of the degree of trunk deformation on the level of self-esteem in women with idiopathic scoliosis, based on Rosenberg Self-Esteem Scale. Women with smaller deformations had a higher level of self-confidence and self-appraisal [29]. Hence, lower average scores for the AduIS group were expected. Although it was not explored whether these patients had satisfactory access to psychological support, they would likely need it. This supports the aforementioned conclusion about multi-disciplinary observation and treatment [27].
Internal consistency estimates indicate that the Slovenian version of the SRS-22r is suitable for inter- and intra-individual comparisons among adult patients, whereas only group-level comparisons will be reliable with adolescents. The results for the adult population are consistent with prior assessments of SRS-22r in other languages among adults [18], whereas the results for adolescents are similar only to the one report [30]. Several reasons might explain these differences. First, the item representing financial difficulties (“Are you and/or your family experiencing financial difficulties because of your back?”) in the function domain (Table 3) is not relevant for adolescents in Slovenia because the Slovenian public health care insurance system fully covers all costs of scoliosis treatment. In the AduIS group, this item stands out less because indirect financial difficulties can be more significant: most of the adults in the sample were employed and in case of health problems they would go on the sick leave resulting in reduced income.
One should keep in mind that adolescence is a developmental phase, where relationship with parents changes and peers gain more influence. The diagnosis of chronic illness during this sensitive separation period can impede this process, whilst the formation of friendship and contacts with the opposite sex are made more difficult [3]. For that reason, social desirability response bias may have implications for the self-reported psychological distress among children with chronic health conditions. Children tend to under-report psychopathology in comparison to parent reports [31]. Parents can provide an informed estimate on how a disability may impact their child’s quality-of-life. Nevertheless, parent reports can also be biased, involving both underestimation and overestimation [32]. Therefore, a multi-informant approach in rehabilitation of children with chronic illnesses is recommended [33].
The estimated test-retest reliability of the SRS-22r was high to very high in both groups of patients and moderate for the Function and Satisfaction scales in the AIS group. This is in agreement with the original questionnaire and the results of previous studies [21, 37].
Concurrent validity of the SRS-22r with the EQ-5D-5 L has so far only been investigated in Norway (albeit using an older version of the EQ-5D questionnaire) and was found to be low [37]. The results confirm that SRS-22r does not overlap with EQ-5D-5 L in IS patients, who face very few, if any, problems with general HRQoL as conceptualised by EQ-5D-5 L. The overlap between the instruments is more considerable in adult patients, but still far from complete. The EQ-5D-5 L has been validated for adult populations with back pain, but not in the younger population with spine deformity. Similarly, low correlations were reported between SRS-22r and EQ-5D for the main dimensions of pain, mobility, function, and mental health [38]. Self-image and Satisfaction with treatment domains are not comparable between the two instruments; overall scores showed moderate correlation, which supports the utility of SRS-22r in IS patients. The advantage of EuroQol approach is that it can be used for comparison with patients with other diseases and to provide a utility index in cost-effectiveness evaluation. However, EQ-5D-3 L is not necessarily suitable for children and adolescents. Hence, the EuroQol Group adapted the original EQ-5D into EQ-5D-Y and generated new value-sets for it [39]. National spine surgery registries have been initiated in five Nordic countries, and panelists from these registries agreed that both instruments can be recommended together as the core outcome measures for adolescents and adults with spinal deformity [5].
The capacity of the SRS-22r questionnaire to discriminate between patients with scoliosis and healthy subjects in terms of HRQoL was confirmed. Significant differences in all SRS-22r domains have been observed previously between healthy adults and adults with scoliosis [18]. Admittedly, the comparison of the HRQoL scores of adults with IS to those of healthy controls is not ideal, because they are not perfectly matched in terms of other factors that might explain the mean differences.
To the authors’ knowledge, this study is the first where two IS age-groups (adults and adolescents) were assessed in parallel. Its main limitation is the number of patients, especially in the light of internal-consistency testing of the scales [40]; but in Slovenia, with a total population of two million, the population of IS patients is also small. Another potential limitation is the absence of a control group of healthy adolescents. The fact that most patients were women might also be considered a limitation, but one has to consider that AIS is more progressive in female patients and that SRS-22r does not detect sex-related difference in quality of life [26, 41].
Conclusion
The Slovenian version of SRS-22r has adequate psychometric properties to measure HRQoL of patients with IS in Slovenia, whereby it appears to be more reliable for adults than for adolescents. When used with IS adolescents, the SRS-22r is affected by a severe ceiling effect, and the support for concurrent validity of the SRS-22r among adolescents is weak. The SRS-22r could be used for longitudinal follow-up of adult patients after rehabilitation treatment, as well as for future rehabilitation research aimed at clinical outcomes after treatment of IS. Important issues (pain, low self-esteem, lack of psychological support and multidisciplinary rehabilitation programs) that adolescents and adults with IS can face were identified. In-depth research on participation limitations and family burden is warranted, so that the present rehabilitation programs for these patients could be upgraded with an interdisciplinary approach and integrated psychosocial care.
Footnotes
Acknowledgments
The authors have no acknowledgments.
Conflict of interest
The authors have no conflicts of interest to report. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.