Neuroradiology DOI 10.1007/s00234-013-1294-y
DIAGNOSTIC NEURORADIOLOGY
Disc degeneration and chronic low back pain: an association which becomes nonsignificant when endplate changes and disc contour are taken into account Francisco M. Kovacs & Estanislao Arana & Ana Royuela & Ana Estremera & Guillermo Amengual & Beatriz Asenjo & Helena Sarasíbar & Isabel Galarraga & Ana Alonso & Carlos Casillas & Alfonso Muriel & Carmen Martínez & Víctor Abraira
Received: 24 August 2013 / Accepted: 14 October 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract Introduction The objective of this study was to assess the association between severe disc degeneration (DD) and low back pain (LBP). Methods A case–control study was conducted with 304 subjects, aged 35–50, recruited in routine clinical practice across six hospitals; 240 cases (chronic LBP patients with a median pain duration of 46 months) and 64 controls (asymptomatic subjects without any lifetime history of significant LBP). The following variables were assessed once, using previously validated methods: gender, age, body mass index (BMI), lifetime smoking exposure, degree of physical activity, severity of LBP, disability, and findings on magnetic
resonance (MRI) (disc degeneration, Modic changes (MC), disc protrusion/hernia, annular tears, spinal stenosis, and spondylolisthesis). Radiologists who interpreted MRI were blinded to the subjects' characteristics. A multivariate logistic regression model assessed the association between severe DD and chronic LBP, adjusting for gender, age, BMI, physical activity, MC, disc protrusion/hernia, and spinal stenosis. Results Severe DD at ≥1 level was found in 46.9 % of the controls and 65.8 % of the cases. Crude odds ratio (95 % CI), for suffering chronic LBP when having severe DD, was 2.06 (1.05; 4.06). After adjusting for “MC” and “disc protrusion/ hernia,” it was 1.81 (0.81; 4.05).
Electronic supplementary material The online version of this article (doi:10.1007/s00234-013-1294-y) contains supplementary material, which is available to authorized users. F. M. Kovacs Departamento Científico, Fundación Kovacs, Paseo de Mallorca 36, 07012 Palma de Mallorca, Spain
A. Estremera : G. Amengual : H. Sarasíbar : C. Martínez Hospital Son Llàtzer, Ctra. De Manacor, Km. 4, 07198 Palma de Mallorca, Spain
F. M. Kovacs : E. Arana : A. Estremera : G. Amengual : B. Asenjo : H. Sarasíbar : I. Galarraga : A. Alonso : C. Casillas : C. Martínez Spanish Back Pain Research Network, Fundación Kovacs, Paseo de Mallorca 36, 07012 Palma de Mallorca, Spain
B. Asenjo Hospital Carlos Haya, Avda Carlos Haya s/n, 29011 Málaga, Spain
E. Arana (*) Servicio de Radiología, Fundación Instituto Valenciano de Oncología, Prof Beltrán Báguena, 19, 46009 Valencia, Spain e-mail: [email protected] A. Royuela : A. Muriel : V. Abraira CIBER Epidemiología y Salud Pública (CIBERESP), Madrid, Spain A. Royuela : A. Muriel : V. Abraira Unidad de Bioestadística Clínica, IRYCIS, Hospital Ramón y Cajal, Ctra. Colmenar Km. 9.1, 28034 Madrid, Spain
I. Galarraga Hospital de Manacor, Ctra. De Manacor s/n, 07500 Manacor, Mallorca, Spain A. Alonso Fundación Jiménez Díaz, Avda. Reyes Católicos 2, 28040 Madrid, Spain C. Casillas Instituto de Traumatología Unión de Mutuas, Av. del Lledó, 67, 12004 Castellón, Spain
Neuroradiology
Conclusions The association between severe DD and LBP ceases to be significant when adjusted for MC and disc protrusion/hernia. These results do not support that DD as a major cause of chronic LBP.
by the institutional review boards of the participating hospitals.
Keywords Low back pain . Vertebral endplate changes . Disk degeneration . Magnetic resonance imaging . Lumbar spine
All subjects (cases and controls) were aged 30–50 years and had been prescribed an MRI. Cases were submitted to lumbar MRI for chronic LBP (i.e., lasting ≥90 days) [12], with or without leg pain. Controls had been prescribed a cranial MRI for headache which turned out to be normal, did not present LBP and, according to their statements and medical records, either had no history of LBP or had experienced only one episode in their life, lasting less than 7 days. Exclusion criteria were: ethnically not Spanish, history of spine surgery, current pregnancy, vertebral fractures, “red flags” for fractures or potential underlying systemic diseases, and signs suggesting cauda equina syndrome [8]. Sample size was established at 294 subjects (235 cases and 59 controls) based on the following assumptions: the prevalence of DD among chronic LBP patients would be 60 %, and 30 % among asymptomatic controls [13, 15], with an alpha error of 0.05, a statistical power of 95 %, a ratio of cases per controls of 4:1, and a percentage of subjects with missing data precluding inclusion in the regression models ≤25 %.
Abbreviations BMI Body mass index DD Disc degeneration LBP Low back pain P25 25th percentile P75 75th percentile RMQ Roland–Morris Questionnaire VAS Visual analog scale
Introduction Low back pain (LBP) has a lifetime prevalence of up to 84 % [1] and is the main cause of disability worldwide [2]. The process of disc degeneration disease (DD) is similar to aging, although at a faster pace [3, 4]. Some clinicians consider DD a potential “cause” of chronic LBP [5]. However, virtually all healthy subjects above a certain age show DD on magnetic resonance imaging (MRI) [3], and none of the currently available diagnostic procedures allow identifying the subjects in whom it may cause LBP [6]. In fact, controversy exists regarding the correlation between DD and current or future symptoms [3, 7, 8]. Inconsistencies in the definitions of DD and LBP, use of non-validated methods for assessing DD and LBP, and lack of adjustment for potential confounders, as well as other methodological shortcomings, impede drawing firm conclusions from the case–control and prospective studies which have assessed their association and led to inconsistent results [7]. Moreover, these studies were conducted with North American and Scandinavian subjects [9–12], while genetic factors influence DD and no study has assessed this association in Southern Europe. Therefore, the objective of this study was to assess the association between severe DD and chronic LBP in Spanish subjects, following standardized definitions, using previously validated methods and adjusting for potential confounders.
Methods This was a case–control study conducted in radiological departments at six hospitals across six cities in Spain, between June and December 2012. The study protocol was approved
Subjects
Procedure All subjects referred to the radiology departments of the participating hospitals for a lumbar MRI for LBP, or for a cranial MRI for headache, were screened. Those complying with the inclusion criteria were included in the study after signing the corresponding informed consent. Neither subjects nor physicians received compensation for participating in this study. Controls were invited to participate once the recruiting radiologist had established that the cranial MRI was normal. Once included, all cases and controls underwent a lumbar MRI and were assigned a number code, which substituted their name on the self-administered questionnaire and the images. Only the study coordinator (FMK), who did not interpret any images, had access to the name corresponding to each code. Radiologists read images at their own institutions and were blinded to patients' identities and clinical data. Data were introduced into a database at a central coordination office by two administrative assistants who separately double checked that the data introduced coincided with the questionnaires. At the analysis phase, the code assigned to each subject made it possible to pair each subject's clinical data with the MRI. When data analysis started, the statisticians were given the key, contained within the code, which made it possible to distinguish “cases” from “controls.” This procedure ensured that (a) radiologists were unaware of whether the images they were reading corresponded to a
Neuroradiology
case or a “control”; (b) they were blinded to subject's demographic and clinical characteristics; (c) the researchers who were aware of the code identifying each subject as a case or a control did not read the images; and (d) only the statisticians and the auxiliary personnel who handed the selfadministered questionnaire to the subjects had access to subjects' demographic and clinical data. Clinical assessment All subjects (cases and controls) underwent the same assessment. The self-administered questionnaire gathered data on subjects' age (date of birth), gender, height (centimeters), weight (kilogram), lifetime smoking exposure (pack-years smoked), duration of pain (asking separately whether pain had lasted for ≥90 days and date of appearance), degree of physical activity, current pain severity, and disability. Physical activity was assessed through a previously validated questionnaire, which gathered data on job, physical requirements during work, daily activities and leisure, and mean time per week spent on activities implying an increase in heart and breathing rates [16]. Pain intensity was measured with a 10-cm visual analog scale (VAS) and LBP-related disability with the validated Spanish version of the Roland– Morris questionnaire (RMQ) [17, 18]. Subjects completed all the self-administered questionnaires by themselves, in private. The only instructions they received were those included in the validated instruments measuring physical activity, pain intensity, and disability [16–18]. MRI evaluation All examinations were performed on six 1.5 T MR imaging systems with a six-channel phased-array spine coil and with patients in supine position, following standard imaging protocols (Supplementary Table) [12, 14]. Findings on MRI were reported using the previously validated Spanish version of the “Nordic Modic Consensus Group Classification,” which includes the standardized report of findings related to disc degeneration (Pfirrmann's grades “I,” “II,” “III,” “IV,” or “V”), vertebral endplate changes at each vertebral level (type, Modic types I, II, and III, intravertebral location, intravertebral volume, maximum height, endplate area, osteophytes, localized endplate defects, Schmorl's nodes, and irregular endplates), disc protrusion (“normal disc contour,” “bulging, ” “focal protrusion,” or extrusion”), spondylolisthesis (yes/ no), and spinal stenosis (yes/no) (Figs. 1 and 2) [19–21]. MRIs were interpreted by eight radiologists. Their postresidency experience as radiologists ranged from 13 to 19 years, and their experience interpreting spine imaging ranged between 8 and 13 years. They were trained in different institutions without fellowships. The mean (25th percentile (P25); 75th percentile (P75)) kappa values for their inter-
Fig. 1 Sagittal T1- (a) and T2- (b) weighted images from a symptomatic subject (“case”). Modic change types I and III, and disc bulging, are seen at the L1–L2 level. At the L5–S1 level, Modic change type II and a localized endplate defect are seen
observer agreement when assessing disc degeneration, disc contour abnormalities, and Modic changes using the methods implemented in this study have shown to be 0.478 (0.393;
Fig. 2 Sagittal T1- (a ) and T2- (b ) weighted images from an asymptomatic subject (control). Modic change type II, disc bulging, focal protrusion, irregular endplate, localized endplate defects, and osteophytes are seen at the L4–L5 level. At the L5–S1 level, disc bulging, Modic type II changes, and osteophytes are seen
Neuroradiology
0.553), 0.480 (0.401; 0.547), and 0.420 (0.325; 0.559), respectively [19], which is in line with the values reported in the literature irrespective of the classification system used [22, 23]. Data analysis At the analysis phase, the degree of physical activity was classified as “extremely intense” vs. “not extremely intense” [16], lifetime exposure to smoking was defined as the number of packs smoked during the patients' lifetime [22], disc degeneration at any lumbar level was dichotomized into Pfirrmann's grades I + II + III (“no/mild disc degeneration”) vs. IV + V (“severe disc degeneration”), Modic changes were dichotomized as “no” vs. “yes” (at any vertebral level, irrespective of type or location), and disc protrusion was dichotomized into “no” (normal disc contour or bulging) vs. “yes” (focal protrusion or extrusion). Absolute and relative frequencies were used to describe categorical variables. For continuous ones, medians and interquartile range, or mean and SD, were used depending on whether data were normally distributed. Chi-square test was used to compare the proportion of subjects showing severe DD between cases and controls. In order to control for potential confounders, a multivariate logistic regression was performed to estimate the association between severe DD and chronic LBP. Based on previous studies, the variables which were included in the maximal model were age (years), gender, body mass index (BMI; kilograms per square meter), extremely intense physical activity, Modic changes (yes/no), disc protrusion/herniation (yes/no), and spinal stenosis (yes/no) [3, 14, 24, 25]. A backward strategy was used; those variables which changed ≥10 % of the effect size when they were eliminated from the model were considered as confounders. The collinearity of the maximal models was evaluated using the criteria proposed by Belsley [26], and the significance level was set at 5 %. Stata (version 11.0; Stata Corp., College Station, TX, USA) program was used for analysis.
Results Among the 84 subjects without LBP who were screened, 17 reported having had ≥1 relevant LBP episode in their lives, and medical records showed that three more had requested medical care for this reason. Among the 287 LBP patients who were screened, 47 had pain lasting less than 90 days. Therefore, 304 subjects (64 controls and 240 cases) were included. A flow diagram showing screening, enrollment, and study participants is presented in Fig. 3, following STROBE guidelines [27].
Among controls, median age was 45 years and 45.3 % were women. Among cases, median age was 43 years, and 55.0 % were women. Table 1 details sample characteristics. There were some missing values due to questions that patients left unanswered in the questionnaires (Table 1). The variable with the highest rate of missing data was the date when pain began, which was left unanswered by 55 out of the 240 chronic LBP patients (Table 1). Characteristics of patients who answered and did not answer this question were very similar (Table 2). Nevertheless, it was hypothesized that differences between patients who did and did not answer this question might exist. Therefore, a sensitivity analysis was performed, in which these 55 cases were excluded. The number of discs showing each grade of disc degeneration in male and females is shown in Table 3. Severe disc degeneration at ≥1 level was present in 46.9 % out of the 64 subjects without LBP and in 65.8 % out of the 240 LBP patients (p =0.006; Table 1). Results were consistent in the sensitivity analysis which only included the 185 patients who provided data on the exact duration of pain (46.9 vs. 64.9 %, p =0.011). In the regression models, “Modic changes” and “disc protrusion/herniation” were identified as confounders of the association between severe DD and LBP. Before adjusting for these variables, the odds ratio (OR) (95 % confidence interval (CI)) for suffering from chronic low back pain when having severe DD was 2.06 (1.05; 4.06). After adjusting for them, the odds ratio was 1.81 (0.81; 4.05; Table 4). Results were very similar in the sensitivity analysis which only included the 185 patients who provided data on the exact duration of pain. In this analysis, the same variables were confounders and, once results were adjusted, OR was 1.71 (0.75; 3.89).
Discussion Results from this study suggest that persons with severe DD may have 81 % higher odds of suffering from chronic LBP, but the 95 % CI for this estimator is 0.81–4.05, indicating that this result is not significant. In fact, these results show that the association between severe DD and chronic LBP is not significant when Modic changes and disc contour abnormalities are considered. This study was powered to show a significant association between LBP and severe DD anticipating that the prevalence of the latter across cases and controls would be 60 vs. 30 %, respectively, while the prevalence actually found turned out to be 65.8 and 46.9 % (i.e., an OR of 1.8). Therefore, an alternative explanation could be that this study was underpowered, which impeded this association from reaching statistical significance. It is impossible to rule out that this possibility and, given these results, DD may or may not be a cause of chronic LBP. In fact, the 95 % CI for the odds ratio on
Neuroradiology Fig. 3 Flow diagram shows screening, enrollment, and study participant details
Patients referred for Lumbar MRI for chronic low back pain (cases), or Cranial MRI for headache, which turned out to be normal (controls)
Cases (LBP patients) Assessed for eligibility (n = 287)
Controls (neither LBP nor history of significant LBP) Assessed for eligibility (n = 84) Not eligible (n=20) History of LBP
Not eligible (n=47) LBP < 90 days
Were included and underwent a lumbar MRI (n=64)
Were included and underwent a lumbar MRI (n=240)
Provided exact date of pain appearance
Did not provide exact date of pain appearance
(n=185)
(n=65)
Total enrolled and analyzed (n =64)
Total enrolled and analyzed (n =240)
Analysis of data
Logistic regression including all subjects
Sensitivity analysis: Logistic regression including only the cases which provided data on exact duration of pain
(240 cases + 64 controls)
(185 cases + 64 controls)
Consistent results
this association is wide, and some previous case/control studies have found statistically significant ORs for reporting pain when having DD [9, 10]. However, should there be a significant association between severe DD and chronic LBP, the corresponding OR would be
DIAGNOSTIC NEURORADIOLOGY
Disc degeneration and chronic low back pain: an association which becomes nonsignificant when endplate changes and disc contour are taken into account Francisco M. Kovacs & Estanislao Arana & Ana Royuela & Ana Estremera & Guillermo Amengual & Beatriz Asenjo & Helena Sarasíbar & Isabel Galarraga & Ana Alonso & Carlos Casillas & Alfonso Muriel & Carmen Martínez & Víctor Abraira
Received: 24 August 2013 / Accepted: 14 October 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract Introduction The objective of this study was to assess the association between severe disc degeneration (DD) and low back pain (LBP). Methods A case–control study was conducted with 304 subjects, aged 35–50, recruited in routine clinical practice across six hospitals; 240 cases (chronic LBP patients with a median pain duration of 46 months) and 64 controls (asymptomatic subjects without any lifetime history of significant LBP). The following variables were assessed once, using previously validated methods: gender, age, body mass index (BMI), lifetime smoking exposure, degree of physical activity, severity of LBP, disability, and findings on magnetic
resonance (MRI) (disc degeneration, Modic changes (MC), disc protrusion/hernia, annular tears, spinal stenosis, and spondylolisthesis). Radiologists who interpreted MRI were blinded to the subjects' characteristics. A multivariate logistic regression model assessed the association between severe DD and chronic LBP, adjusting for gender, age, BMI, physical activity, MC, disc protrusion/hernia, and spinal stenosis. Results Severe DD at ≥1 level was found in 46.9 % of the controls and 65.8 % of the cases. Crude odds ratio (95 % CI), for suffering chronic LBP when having severe DD, was 2.06 (1.05; 4.06). After adjusting for “MC” and “disc protrusion/ hernia,” it was 1.81 (0.81; 4.05).
Electronic supplementary material The online version of this article (doi:10.1007/s00234-013-1294-y) contains supplementary material, which is available to authorized users. F. M. Kovacs Departamento Científico, Fundación Kovacs, Paseo de Mallorca 36, 07012 Palma de Mallorca, Spain
A. Estremera : G. Amengual : H. Sarasíbar : C. Martínez Hospital Son Llàtzer, Ctra. De Manacor, Km. 4, 07198 Palma de Mallorca, Spain
F. M. Kovacs : E. Arana : A. Estremera : G. Amengual : B. Asenjo : H. Sarasíbar : I. Galarraga : A. Alonso : C. Casillas : C. Martínez Spanish Back Pain Research Network, Fundación Kovacs, Paseo de Mallorca 36, 07012 Palma de Mallorca, Spain
B. Asenjo Hospital Carlos Haya, Avda Carlos Haya s/n, 29011 Málaga, Spain
E. Arana (*) Servicio de Radiología, Fundación Instituto Valenciano de Oncología, Prof Beltrán Báguena, 19, 46009 Valencia, Spain e-mail: [email protected] A. Royuela : A. Muriel : V. Abraira CIBER Epidemiología y Salud Pública (CIBERESP), Madrid, Spain A. Royuela : A. Muriel : V. Abraira Unidad de Bioestadística Clínica, IRYCIS, Hospital Ramón y Cajal, Ctra. Colmenar Km. 9.1, 28034 Madrid, Spain
I. Galarraga Hospital de Manacor, Ctra. De Manacor s/n, 07500 Manacor, Mallorca, Spain A. Alonso Fundación Jiménez Díaz, Avda. Reyes Católicos 2, 28040 Madrid, Spain C. Casillas Instituto de Traumatología Unión de Mutuas, Av. del Lledó, 67, 12004 Castellón, Spain
Neuroradiology
Conclusions The association between severe DD and LBP ceases to be significant when adjusted for MC and disc protrusion/hernia. These results do not support that DD as a major cause of chronic LBP.
by the institutional review boards of the participating hospitals.
Keywords Low back pain . Vertebral endplate changes . Disk degeneration . Magnetic resonance imaging . Lumbar spine
All subjects (cases and controls) were aged 30–50 years and had been prescribed an MRI. Cases were submitted to lumbar MRI for chronic LBP (i.e., lasting ≥90 days) [12], with or without leg pain. Controls had been prescribed a cranial MRI for headache which turned out to be normal, did not present LBP and, according to their statements and medical records, either had no history of LBP or had experienced only one episode in their life, lasting less than 7 days. Exclusion criteria were: ethnically not Spanish, history of spine surgery, current pregnancy, vertebral fractures, “red flags” for fractures or potential underlying systemic diseases, and signs suggesting cauda equina syndrome [8]. Sample size was established at 294 subjects (235 cases and 59 controls) based on the following assumptions: the prevalence of DD among chronic LBP patients would be 60 %, and 30 % among asymptomatic controls [13, 15], with an alpha error of 0.05, a statistical power of 95 %, a ratio of cases per controls of 4:1, and a percentage of subjects with missing data precluding inclusion in the regression models ≤25 %.
Abbreviations BMI Body mass index DD Disc degeneration LBP Low back pain P25 25th percentile P75 75th percentile RMQ Roland–Morris Questionnaire VAS Visual analog scale
Introduction Low back pain (LBP) has a lifetime prevalence of up to 84 % [1] and is the main cause of disability worldwide [2]. The process of disc degeneration disease (DD) is similar to aging, although at a faster pace [3, 4]. Some clinicians consider DD a potential “cause” of chronic LBP [5]. However, virtually all healthy subjects above a certain age show DD on magnetic resonance imaging (MRI) [3], and none of the currently available diagnostic procedures allow identifying the subjects in whom it may cause LBP [6]. In fact, controversy exists regarding the correlation between DD and current or future symptoms [3, 7, 8]. Inconsistencies in the definitions of DD and LBP, use of non-validated methods for assessing DD and LBP, and lack of adjustment for potential confounders, as well as other methodological shortcomings, impede drawing firm conclusions from the case–control and prospective studies which have assessed their association and led to inconsistent results [7]. Moreover, these studies were conducted with North American and Scandinavian subjects [9–12], while genetic factors influence DD and no study has assessed this association in Southern Europe. Therefore, the objective of this study was to assess the association between severe DD and chronic LBP in Spanish subjects, following standardized definitions, using previously validated methods and adjusting for potential confounders.
Methods This was a case–control study conducted in radiological departments at six hospitals across six cities in Spain, between June and December 2012. The study protocol was approved
Subjects
Procedure All subjects referred to the radiology departments of the participating hospitals for a lumbar MRI for LBP, or for a cranial MRI for headache, were screened. Those complying with the inclusion criteria were included in the study after signing the corresponding informed consent. Neither subjects nor physicians received compensation for participating in this study. Controls were invited to participate once the recruiting radiologist had established that the cranial MRI was normal. Once included, all cases and controls underwent a lumbar MRI and were assigned a number code, which substituted their name on the self-administered questionnaire and the images. Only the study coordinator (FMK), who did not interpret any images, had access to the name corresponding to each code. Radiologists read images at their own institutions and were blinded to patients' identities and clinical data. Data were introduced into a database at a central coordination office by two administrative assistants who separately double checked that the data introduced coincided with the questionnaires. At the analysis phase, the code assigned to each subject made it possible to pair each subject's clinical data with the MRI. When data analysis started, the statisticians were given the key, contained within the code, which made it possible to distinguish “cases” from “controls.” This procedure ensured that (a) radiologists were unaware of whether the images they were reading corresponded to a
Neuroradiology
case or a “control”; (b) they were blinded to subject's demographic and clinical characteristics; (c) the researchers who were aware of the code identifying each subject as a case or a control did not read the images; and (d) only the statisticians and the auxiliary personnel who handed the selfadministered questionnaire to the subjects had access to subjects' demographic and clinical data. Clinical assessment All subjects (cases and controls) underwent the same assessment. The self-administered questionnaire gathered data on subjects' age (date of birth), gender, height (centimeters), weight (kilogram), lifetime smoking exposure (pack-years smoked), duration of pain (asking separately whether pain had lasted for ≥90 days and date of appearance), degree of physical activity, current pain severity, and disability. Physical activity was assessed through a previously validated questionnaire, which gathered data on job, physical requirements during work, daily activities and leisure, and mean time per week spent on activities implying an increase in heart and breathing rates [16]. Pain intensity was measured with a 10-cm visual analog scale (VAS) and LBP-related disability with the validated Spanish version of the Roland– Morris questionnaire (RMQ) [17, 18]. Subjects completed all the self-administered questionnaires by themselves, in private. The only instructions they received were those included in the validated instruments measuring physical activity, pain intensity, and disability [16–18]. MRI evaluation All examinations were performed on six 1.5 T MR imaging systems with a six-channel phased-array spine coil and with patients in supine position, following standard imaging protocols (Supplementary Table) [12, 14]. Findings on MRI were reported using the previously validated Spanish version of the “Nordic Modic Consensus Group Classification,” which includes the standardized report of findings related to disc degeneration (Pfirrmann's grades “I,” “II,” “III,” “IV,” or “V”), vertebral endplate changes at each vertebral level (type, Modic types I, II, and III, intravertebral location, intravertebral volume, maximum height, endplate area, osteophytes, localized endplate defects, Schmorl's nodes, and irregular endplates), disc protrusion (“normal disc contour,” “bulging, ” “focal protrusion,” or extrusion”), spondylolisthesis (yes/ no), and spinal stenosis (yes/no) (Figs. 1 and 2) [19–21]. MRIs were interpreted by eight radiologists. Their postresidency experience as radiologists ranged from 13 to 19 years, and their experience interpreting spine imaging ranged between 8 and 13 years. They were trained in different institutions without fellowships. The mean (25th percentile (P25); 75th percentile (P75)) kappa values for their inter-
Fig. 1 Sagittal T1- (a) and T2- (b) weighted images from a symptomatic subject (“case”). Modic change types I and III, and disc bulging, are seen at the L1–L2 level. At the L5–S1 level, Modic change type II and a localized endplate defect are seen
observer agreement when assessing disc degeneration, disc contour abnormalities, and Modic changes using the methods implemented in this study have shown to be 0.478 (0.393;
Fig. 2 Sagittal T1- (a ) and T2- (b ) weighted images from an asymptomatic subject (control). Modic change type II, disc bulging, focal protrusion, irregular endplate, localized endplate defects, and osteophytes are seen at the L4–L5 level. At the L5–S1 level, disc bulging, Modic type II changes, and osteophytes are seen
Neuroradiology
0.553), 0.480 (0.401; 0.547), and 0.420 (0.325; 0.559), respectively [19], which is in line with the values reported in the literature irrespective of the classification system used [22, 23]. Data analysis At the analysis phase, the degree of physical activity was classified as “extremely intense” vs. “not extremely intense” [16], lifetime exposure to smoking was defined as the number of packs smoked during the patients' lifetime [22], disc degeneration at any lumbar level was dichotomized into Pfirrmann's grades I + II + III (“no/mild disc degeneration”) vs. IV + V (“severe disc degeneration”), Modic changes were dichotomized as “no” vs. “yes” (at any vertebral level, irrespective of type or location), and disc protrusion was dichotomized into “no” (normal disc contour or bulging) vs. “yes” (focal protrusion or extrusion). Absolute and relative frequencies were used to describe categorical variables. For continuous ones, medians and interquartile range, or mean and SD, were used depending on whether data were normally distributed. Chi-square test was used to compare the proportion of subjects showing severe DD between cases and controls. In order to control for potential confounders, a multivariate logistic regression was performed to estimate the association between severe DD and chronic LBP. Based on previous studies, the variables which were included in the maximal model were age (years), gender, body mass index (BMI; kilograms per square meter), extremely intense physical activity, Modic changes (yes/no), disc protrusion/herniation (yes/no), and spinal stenosis (yes/no) [3, 14, 24, 25]. A backward strategy was used; those variables which changed ≥10 % of the effect size when they were eliminated from the model were considered as confounders. The collinearity of the maximal models was evaluated using the criteria proposed by Belsley [26], and the significance level was set at 5 %. Stata (version 11.0; Stata Corp., College Station, TX, USA) program was used for analysis.
Results Among the 84 subjects without LBP who were screened, 17 reported having had ≥1 relevant LBP episode in their lives, and medical records showed that three more had requested medical care for this reason. Among the 287 LBP patients who were screened, 47 had pain lasting less than 90 days. Therefore, 304 subjects (64 controls and 240 cases) were included. A flow diagram showing screening, enrollment, and study participants is presented in Fig. 3, following STROBE guidelines [27].
Among controls, median age was 45 years and 45.3 % were women. Among cases, median age was 43 years, and 55.0 % were women. Table 1 details sample characteristics. There were some missing values due to questions that patients left unanswered in the questionnaires (Table 1). The variable with the highest rate of missing data was the date when pain began, which was left unanswered by 55 out of the 240 chronic LBP patients (Table 1). Characteristics of patients who answered and did not answer this question were very similar (Table 2). Nevertheless, it was hypothesized that differences between patients who did and did not answer this question might exist. Therefore, a sensitivity analysis was performed, in which these 55 cases were excluded. The number of discs showing each grade of disc degeneration in male and females is shown in Table 3. Severe disc degeneration at ≥1 level was present in 46.9 % out of the 64 subjects without LBP and in 65.8 % out of the 240 LBP patients (p =0.006; Table 1). Results were consistent in the sensitivity analysis which only included the 185 patients who provided data on the exact duration of pain (46.9 vs. 64.9 %, p =0.011). In the regression models, “Modic changes” and “disc protrusion/herniation” were identified as confounders of the association between severe DD and LBP. Before adjusting for these variables, the odds ratio (OR) (95 % confidence interval (CI)) for suffering from chronic low back pain when having severe DD was 2.06 (1.05; 4.06). After adjusting for them, the odds ratio was 1.81 (0.81; 4.05; Table 4). Results were very similar in the sensitivity analysis which only included the 185 patients who provided data on the exact duration of pain. In this analysis, the same variables were confounders and, once results were adjusted, OR was 1.71 (0.75; 3.89).
Discussion Results from this study suggest that persons with severe DD may have 81 % higher odds of suffering from chronic LBP, but the 95 % CI for this estimator is 0.81–4.05, indicating that this result is not significant. In fact, these results show that the association between severe DD and chronic LBP is not significant when Modic changes and disc contour abnormalities are considered. This study was powered to show a significant association between LBP and severe DD anticipating that the prevalence of the latter across cases and controls would be 60 vs. 30 %, respectively, while the prevalence actually found turned out to be 65.8 and 46.9 % (i.e., an OR of 1.8). Therefore, an alternative explanation could be that this study was underpowered, which impeded this association from reaching statistical significance. It is impossible to rule out that this possibility and, given these results, DD may or may not be a cause of chronic LBP. In fact, the 95 % CI for the odds ratio on
Neuroradiology Fig. 3 Flow diagram shows screening, enrollment, and study participant details
Patients referred for Lumbar MRI for chronic low back pain (cases), or Cranial MRI for headache, which turned out to be normal (controls)
Cases (LBP patients) Assessed for eligibility (n = 287)
Controls (neither LBP nor history of significant LBP) Assessed for eligibility (n = 84) Not eligible (n=20) History of LBP
Not eligible (n=47) LBP < 90 days
Were included and underwent a lumbar MRI (n=64)
Were included and underwent a lumbar MRI (n=240)
Provided exact date of pain appearance
Did not provide exact date of pain appearance
(n=185)
(n=65)
Total enrolled and analyzed (n =64)
Total enrolled and analyzed (n =240)
Analysis of data
Logistic regression including all subjects
Sensitivity analysis: Logistic regression including only the cases which provided data on exact duration of pain
(240 cases + 64 controls)
(185 cases + 64 controls)
Consistent results
this association is wide, and some previous case/control studies have found statistically significant ORs for reporting pain when having DD [9, 10]. However, should there be a significant association between severe DD and chronic LBP, the corresponding OR would be
