Disability Measurement in Patients with Low Back Pain Using Roland-Morris Questionnaire as a Model and Studying Possible Modifications

Background: Self-reported questionnaires have become popular measures in assessing disability in patients with low back pain (LBP). Roland-Morris Questionnaire (RMQ) is one of the internationally accepted, self-reporting questionnaire which demonstrates good psychometric properties. Objectives: The present study intended to evaluate a face to face interview applying Arabic version of the RMQ, and comparing it with a modified version of RMQ (RMQV), and whether the purely subjective RMQ scores correlate with symptoms and signs which have predictive or prognostic values. Design: case-series study Methods and materials: Case-series study conducted on seventy-two patients with chronic LBP at Ibn-Sina Teaching Hospital in Mosul city. Physical examination at rheumatology outpatient clinic performed for every patient, and the patients are allowed to answer the questions of RMQ through a direct investigator-patient Arabic conversation. The disability measured by the RMQ subdivided into: mild (0-8), moderate (9-16) and severe (17-24). An individualized literature review performed for clinical features which have predictive or prognostic values in LBP, and including these features in the clinical evaluation of patients in the current study. Results: The results showed that the Arabic conversational RMQ have acceptable reliability and RMQV have excellent reliability (Cronbach’s alpha values=0.72 and 0.94 respectively). There was a significant direct correlation between these two questionnaires (r=0.861; p-value<0.001). However, we found a significant difference between them (p-value<0.01). The scores of the RMQ and RMQV correlate moderately with a score of the predictive features (r=0.503; p-value < 0.01 and 0.530; p-value<0.01 respectively). Conclusion: The study found that the modified version (RMQV) has higher reliability than the original one. Also, the RMQV showed a better correlation with the narrow-angle straight leg raising test, and its mild and moderate subgroups have significant differences regarding the duration and pain intensity of the current episode of LBP. Other measured properties look similar between the two questionnaires.


INTRODUCTION
ain at lower back is a common problem in everyday practice that reported by the majority of adults during their lifetime and is the most frequent cause of disability for individuals younger than 45 years.It is the third leading cause of disability for those older than 45 years. .The RMQ which is a self-reported instrument; its questions focus consistently on disabilities related to the back and the answers are dichotomous: yes/no 13 .The scores, therefore, range from zero (no disability) to 24 (maximum disability).The questionnaire can complete in a maximum of 5 min, and an un-weighted score can calculate in less than 1 min.No sub-scoring reported, and the administration is straightforward.The questions deal with body functions (pain, sleeping, and appetite) as well as activities (selfcare, walking, sitting, standing, lifting, dressing, stairs climbing, housework and resting)
The presence of scoliosis was assessed by using plumb-bob with the tip of the string placed at the spinous process of the first thoracic vertebra and the weight allowed to hang down to reach the buttock of the patient, if any deviation of the weight from the mid-gluteal cleft observed then the patient reported as having scoliosis 30 .Flexion of the spine was evaluated by Schober's test 31 .Any increase in the distance of less than 5 cm considers abnormal.The lateral flexion was assessed using a finger to fibula test, in which the patient asked to bend to one side as much as he can (without any bending forward).The test is considering abnormal if the distance between the tip of the third finger and the head of the fibula was more than 5 cm, then the test repeated on the other side 32 .The extension test is done by asking the patient to actively extend his back as far as tolerated with the examiner providing support for balance, the position held for 30 seconds and the test considered positive if the patient reports pain in the low back area or the lower limb 24 .While in the quadrant test the patient actively moves into lateral bending, extension, and rotation to the same side as far as tolerated, the examiner guides the patient in this motion and provides overpressure if no pain is reported with this movement.
The test considered positive if the patient reports pain or numbness in the area of the back or lower extremities 25 .
The angle of the SLR test was measured for each patient by the plastic goniometer and subdivided into 3 subgroups: patients with narrowangle SLR (0-45 degree), patients with wide angle SLR (46-70 degree) and patients with negative SLR (more than 70 degree).
Vibration sense as part of the neurological examination for both lower limbs was assessed using a 128 Hz tuning fork on the medial malleolus and the big toe.
After filling the RMQ in its original form, another investigator-patient Arabic conversation started to answer the questions of RMQ in a modified method (RMQV).In this method the answer for each item is obtained on a global visual analogue scales (VAS) of zero to 100 subdivisions.The patient was asked to determine how much his/her various activities (daily living, duties, recreation, and vitality feeling) were affected by each particular question due to the LBP, with zero means no any effect on these activities while 100 indicates the worst outcome that the patient can think of.The disability measured by the RMQV also subdivided into three subgroups: mild (0-30), moderate (31-60) and severe (61-100).

RESULTS
Seventy-two patients with the history of chronic LBP for more than three months duration participated in this study.Male 3 shows the range and mean of the pain intensity for the initial episode of LBP measured on VAS retrospectively and current episodes of LBP measured on VAS.Also range and mean for both RMQ and RMQV are shown.
Table 4 shows the different subgroups of LBP.In RMQ, we can see that the highest percentage of patients (48.6%) lie within the moderate subgroup followed by the severe subgroup (26.4%) and the lowest percentage for the mild subgroup (25%).In contrast, the distribution of patients within the severity subgroups of RMQV is clearly different, most of the patients lie within the mild subgroup (76.4%), a small number of patients within the moderate subgroup (19.4%) and only few number of patients lie within the severe subgroup (4.2%).
Some of the important clinical features of the patients are shown in the Table 5, including the number and percentage of patients having each particular feature.These clinical features were chosen because they have predictive values.The total number of positive clinical features with predictive values (16 features) for each patient was estimated and, the resulting total score of these predictive features (which ranged from 0-16) was compared with that of the RMQ and RMQV separately.The range and mean of this predictive features-score were (0-15) and 6.43±2.83respectively.Table 6 shows the correlation between the predictive features score and each of RMQ and RMQV.
Table 7 compares between the means for RMQ and RMQV questionnaires according to Pearson correlation coefficient and independent samples ttest.Both of them were significant.
Table 8 shows the proportions of patients for each particular clinical feature within the severity subgroups of the RMQV.The comparison between the percentages of patients having each particular clinical feature within the mild and moderate subgroups of the RMQV has been studied using the two-proportion test.If the number of patients for a specific clinical feature was zero in the mild and/or the moderate subgroups, then the percentage of patients for this feature is not compared.Statistical analysis of the severe subset of the RMQV was omitted because of the small number of patients in this subgroup (3 patients only).Table 9 shows the proportions of patients for each particular clinical feature within the severity subgroups of the RMQ.The comparisons between the percentages of patients having each particular clinical feature within the mild and moderate subgroups of the RMQ have been studied using the two-proportions z-test.If the number of patients for a particular clinical feature was found to be zero in the mild and/or the moderate subgroups, then the percentage of patients for that feature has not been compared.A comparison with the severe subgroup was not performed as it has not been performed in that of the RMQV one (the preceding table ).
Table 10 shows the different duration categories of LBP for each severity subgroup of the RMQV and RMQ.Statistical analysis between the severe subgroups of the RMQ and RMQV was omitted because of the small number of patients in the severe subgroup of RMQV (3 patients only).Table 11 shows the correlations between the severity subgroups of both (RMQ and RMQV) and the pain intensity of the current episode of LBP (Pearson's correlation), and between the severity subgroups of both RMQ and RMQV, and the SLR test subgroups (Spearman's rank correlation).S** = significant difference at p-value <0.01; NS = not significant (using independent samples t-test (>0.05).

DISCUSSION
The self-reported questionnaires of disability are of great relevance in assessing the severity of low back pain (LBP).The Roland-Morris Questionnaire (RMQ) is considered one of the most commonly studied self-reported disability measures 8, 33 . This questionnaire is considered thoroughly validated, have acceptable reliability and is recommended and referred to as a tool of choice in the assessment of the severity of disability caused by LBP 34 .Also, it is quickly completed, easy to be understood and scored, and broadly consistent with the WHO ICF definition of activity limitation.
But among the weakening points of this questionnaire are: being purely subjective and lack any physical signs, dichotomous responses only with no ability to rate the degree of limitation for a Ann Coll Med Mosul June 2019 Vol.41 No. 1 given functional activity, providing equal marks to mild and severe symptoms, patients rate limitation just in last 24 hours, depends mainly on daily living activities rather than duties and work activities (such as lifting, carrying, pushing or pulling objects) 35,36 .Moreover, the high levels of illiteracy in many countries make the completion of such self-reported questionnaire by the patient alone very difficult.Therefore in this study, an Arabic conversational version of the RMQ was chosen, which has been adapted in the rheumatology unit of Ibn-Sina teaching hospital.
The mean age of seventy-two participated patients was 37.23 years, and this is within the range of age considered to have the highest frequency of LBP 37 .Their mean BMI was also high (29.3kg/m 2 ) and the high BMI is considered as a risk factor for chronicization of acute and subacute LBP 20 .The mean duration of LBP since its initial onset was 50.31 months and the mean duration of the initial episode of LBP was 29.16 days; whereas the mean duration of the current episode of LBP was 22.26 days.All the patients were having a history of more than one previous episode of LBP with the mean number equal to 11.52 episodes.
The mean intensity of the pain for initial episodes measured on global VAS scale (0-100 scores) was (71.52/out of 100 scores) and, there were 4 patients (5.6%) having mild pain, 20 patients (27.8%) having moderate pain and, 48 patients (66.7%) having severe pain.The mean intensity of the current episode pain which also measured on global VAS scale was 59.58 and, there was only one patient (1.4%) having mild pain, 53 patients (73.6%) having moderate pain and 18 patients (25%) having severe pain.The range and mean for RMQ were 2-22 and 12.38 respectively; and there were 18 patients (25%) lie within its mild subgroup, 35 patient (48.6%) within its moderate subgroup, and nineteen patients (26.4%) within the severe subgroup.While for RMQV the range and mean were 0 -66.67 and 21.71 respectively; and there were 55 patients (76.4%) lie within its mild subgroup, 14 patients (19.4%) within its moderate subgroup, and only 3 patients (4.2%) within the severe subgroups.The mean scores for the RMQ reflect a moderate level of disability while that for the RMQV reflect a mild degree of disability.
It appears that the differences in the distribution of patients within the severity subgroups of RMQ and RMQV are remarkable.Most patients lie within the mild subgroup of the RMQV and a minimal number within the severe one, while most of the patients lie in the moderate subgroup of RMQ followed by the severe, then the mild ones.These may indicate that a high percentage of the patients who considered to have moderate or severe disabilities according to the RMQ was really only mildly disabled according to the RMQV.Therefore these two questionnaires are different from each other, and this difference was confirmed by comparing the means for both questionnaires using independent samples t-test which showed a significant difference between them (p-value< 0.01).Yet they were found to have a strong direct correlation with each other (r=0.861;p-value< 0.001).The reliability for the Arabic conversational version of the RMQ was estimated using Cronbach's alpha, and it was 0.73, and for the RMQV the reliability was higher (Cronbach's alpha = 0.94).
Collectively, these predictive features have an impact on the management of patients with LBP.A set of clinical features (16 features) with predictive and prognostic values was used for comparison with both RMQ & RMQV separately.The total numbers of positive predictive features were estimated for each patient, and the final score was used to compare it with that of RMQ and RMQV for the same patient.
The correlation between these predictive features score and each of the RMQ and RMQV were found to be moderate and significant (r=0.503,p-value < 0.01 and r=0.530, p-value < 0.01 respectively).The number and percentages of patients having each particular clinical feature in each severity subgroup of the RMQ and RMQV were estimated.The two-proportion test was used to compare the percentages of patients having each of these clinical features in the mild subgroup and those having the same clinical features in the moderate subgroup of the RMQV.The same comparisons also performed for the patients within the mild and moderate subgroups of the RMQ.The results revealed that there were significant differences between the mild and moderate subgroups of RMQV in the frequency of stress pain and painful lateral bending (p-value<0.001 and < 0.05 respectively).Similar results occurred between mild and moderate subgroups of RMQ (pvalue = 0.03 and 0.003 respectively).Furthermore, the moderately severe current back pain episode was significantly more common in the mild RMQV subgroup (p-value=0.002);while the severe current episode was significantly more common in the moderate RMQV subgroup (p-value =0.001).In the RMQ, on the other hand, the severity of current back pain episode was not different between mild and moderate subgroups.There were no significant differences between the percentages of patients with the remaining clinical features in both the RMQ and RMQV.
The mean duration of LBP since its onset was found to be longer in the mild subgroup (49.34 months) than the moderate subgroup (44.35 months) of the RMQV, while it was longer in the moderate (51.34 months) than the mild subgroups (44.94 months) of the RMQ.The mean duration of the initial episodes was found to be longer in the moderate subgroup (34.28 days) than in the mild one (24.58days) in the RMQV, while it was longer in the mild (30.94 days) than the moderate subgroups (18.94 days) of the RMQ.In case of the current episode, its mean duration was found to be longer in the moderate subgroups (39.78 days) than in the mild one (18.56days).Also, its mean duration was longer in the mild (17.88 days) than the moderate subgroups (15.97 days) of the RMQ.These differences in the duration categories were found to be significant only for the mean duration of the current episode of LBP between the severity subgroups of RMQV with p-value < 0.01 (using independent samples t-test).
Furthermore, the intensity of the LBP was found to have strong correlations with RMQ and RMQV (r=0.625 at p-value < 0.01 and r=0.634 at p-value < 0.01 respectively).The pain intensity of the LBP considered having predictive values such as being a predictor of short-term outcome after lumbar discectomy 39 , also it is considered to be correlated with greater disability in patients with lumbar spinal stenosis 40 .A narrow angle SLR test also has predictive values such as being a predictor of poor outcome after lumbar disc herniation surgery 41 .Spearman's rank correlation was used to compare the severity subgroups of both the RMQ and RMQV with the SLR test subgroups, a significant moderate correlation was found between the SLR test subgroups and the RMQV subgroups (rho= 0.262 at p-value<0.05),whereas the relationship between the SLR test subgroups and the RMQ subgroups was weak and not significant (rho = 0.154).

CONCLUSIONS
Arabic translation and face to face interview instead of the original patient report can maintain the excellent internal consistency of the questionnaires.
The differences between the original RMQ and the VAS-graded modification (RMQV) are prominent in that the RMQV have better reliability, better correlation with the narrow-angle straight leg raising test, and its mild and moderate subgroups have significant differences regarding the duration and pain intensity of the current episode of low back pain.
The scores of the two versions RMQ correlated somewhat moderately with the predictive and prognostic clinical feature.

Table 1 ,
to female percentage was 41.7: 58.3.As shown in the the mean age and BMI of the studied group were 37.23 ±9.14 and 29.3± 4.46 respectively.

Table 1 :
Demographic features of the patients.

Table 2 :
Different duration categories of LBP and number of the prior episodes.
SD = Standard deviation

Table 3 :
Total pain intensity of initial and current episodes and disability scores.

Table 5 :
Some of the essential clinical features of the patients.

Table 6 :
Correlation (Pearson's coefficient r) between predictive features score and the disability scores (RMQV and RMQ).

Table 7 :
The Cronbach's alpha, Pearson's correlation coefficient (r) and comparison between the means for both RMQ and RMQV.

Table 8 :
Numbers and percentages of patients having some particular clinical features within RMQV subgroups.
NC = No comparison has been done; NS = Not significant (p >0.05) according to two-proportion z-test.

Table 9 :
Numbers and percentages of patients having some particular clinical features within RMQ subgroups.
NC = No comparison has been done; NS = Not significant test (p >0.05).According to two-proportions z-test.

Table 10 :
Different duration categories for each severity subgroups of RMQ & RMQV.

Table 11 :
Correlations between subgroups of pain intensity and SLR test with that of RMQV and RMQ.