Lumbosacral Transitional Vertebrae: Association with Low Back Pain

Population

OAI included 4796 participants who were recruited on the basis of their risk factors for knee osteoarthritis (OA) or mild to moderate knee OA. Of these participants, 4636 (97%) patients with diagnostic baseline pelvic radiographs were studied. The study population consisted of 1919 men (41%) between the ages of 45 and 79 years (mean age, 61 years ± 9 [standard deviation]) and 2717 women (58.5%) between the ages of 45 and 79 years (mean age, 61 years ± 9). Patient characteristics, including age, sex, race, and body mass index are shown in Table 1. The study’s participants represented a sample selected by OAI for the study of knee OA and were not recruited on the basis of whether they experienced back pain. Sixty-two percent (860 of 1390) of the participants who had knee OA and were in the progression subcohort of OAI had back pain, and 58% (1900 of 3289) of the participants in the incidence subcohort (patients who did not have knee OA but who were at risk for developing it) had back pain. However, this difference was not statistically significant (P = .08).

Table 1 Characteristics of 4636 Patients with and without LSTV

Note.—Unless otherwise stated, data are numbers of patients, with percentages in parentheses.

^* χ² for categorical variables and analysis of variance for continuous variables.

^† BMI = body mass index

Imaging Technique

Standardized standing pelvic radiographs were performed by using a dedicated foot-positioning mat with the toes internally rotated at a 5° angle, and the x-ray beam was positioned approximately 7 cm above the pubic symphysis. The images were obtained at five institutions: Ohio State University (Columbus, Ohio), University of Maryland School of Medicine (Baltimore, Md), Johns Hopkins University School of Medicine (Baltimore, Md), University of Pittsburgh School of Medicine (Pittsburgh, Pa), and Memorial Hospital of Rhode Island (Pawtucket, RI). Representative radiographs are shown in the Figure.

Figure a:

Download as PowerPoint

Figure b:

Download as PowerPoint

Figure c:

Download as PowerPoint

Figure d:

Download as PowerPoint

Image Analysis

All images were reviewed on picture archiving and communication system workstations (Agfa, Ridgefield Park, NJ). All of the 4796 pelvic anteroposterior radiographs were independently read by two radiologists (L.N., 5 years of experience; W.V., 8 years of experience). Initially, the pelvic radiographs were independently screened by these radiologists for image quality (ie, the ability to assess the relationship between the sacral ala and the lumbar transverse process), postsurgical changes obscuring transitional anatomy, and presence of LSTV. Out of 4796 radiographs, 160 were considered to be of poor quality and were excluded from the study.

Radiographs with adequate image quality were classified according to the presence of LSTV. The presence of an LSTV was determined manually by evaluating the craniocaudal width of the transverse process, with a threshold of greater than 19 mm as measured by using the digital caliper on a picture archiving and communication system, or by the presence of articulation or complete fusion of the transverse process with the sacrum. All discrepancies were settled by a third radiologist (T.M.L., 25 years of experience). LSTV cases were graded according to a radiographic classification system modified from that of Castellvi et al (9), hereafter, the Castellvi classification (Figure). Table 2 outlines the main characteristics of the Castellvi classification. To minimize the number of categories, we classified the cases independently from the bilateral or unilateral findings into four basic types (11,14). The hip joints were also assessed for OA according to the Osteoarthritis Research Society International classification (18). All of the hip images had been independently read 1 year earlier by both a radiologist (L.N.) and a rheumatologist (N.E.L., 30 years of experience). In cases where there was disagreement among the readers, a radiologist (T.M.L.) was also consulted.

Table 2 LSTV according to Castellvi Classification

Clinical Data

The OAI back pain and function questionnaire provided information about frequency, severity, and location of pain. Patients were asked about the frequency of their back pain during the past 30 days and graded it as never, rarely, some of the time, most of the time, or all of the time. Patients were also asked to evaluate the average severity of their back pain during the past 30 days and to classify it as none, mild, moderate, or severe. The location of the pain was assessed with four questions that varied according to the presence of upper back pain, middle back pain, LBP, or buttock pain within the past 30 days. A five-part variable was created for location of back pain. The categories were mutually exclusive and included LBP, buttock pain (without LBP), pain located in the middle or upper back (without LBP or buttock pain), and no back pain at any location.

Information was also provided regarding limitation of physical activity due to back pain and physical activity level by using the Physical Activity Scale for the Elderly (PASE) (19–22).

Reproducibility Analysis

Interobserver reproducibility was calculated from all studies by using the initial readings by the two radiologists. Intraobserver reproducibility was obtained by using 500 randomly selected studies that were independently analyzed by two radiologists on two separate occasions. Cohen κ values were calculated to assess intraobserver and interobserver agreement of the radiographic classification according to Castellvi classification.

Statistical Analysis

Statistical analysis was performed by using SAS 9.2 (SAS Institute, Cary, NC). Descriptive statistics for baseline demographic data were calculated for LSTV and non-LSTV groups, as well as for each LSTV subtype compared with non-LSTV. Continuous variables are presented as mean ± standard deviation and categorical variables as frequency (percentages). For bivariate comparisons, continuous variables were compared by using analysis of variance or ranked analysis of variance as required, and categorical variables were compared by using a χ² test.

Multivariable models were used to assess the relationship between back pain and LSTV, adjusted for age, sex, race, body mass index, and radiographic hip OA. Age, race, hip OA, and body mass index were included in the models because these are known risk factors for back pain. Back pain variables, measured as presence or absence of upper back pain, middle back pain, LBP, or buttock pain, were modeled by using a multinomial model. Odds ratios (ORs), 95% confidence intervals (CIs), and P values were reported for comparison of LSTV to non-LSTV, with each back pain location category (lower back, buttocks, and middle or upper back) versus no back pain. The LSTV subtypes were treated as a categorical variable, and separate ORs were obtained for each value. Back pain variables measured with ordinal scales used proportional odds models, reported as the OR, and were in a higher category on the ordered scale compared with the next lower category. The statistical significance of all calculations was defined as P < .05.

Patient Characteristics

LSTV was found in 18.1% (841 of 4636) of participants (Table 1)—28.1% (539 of 1919) of men and 11.1% (302 of 2717) of women (P < .001). No significant difference in the prevalence of LSTV was observed by race. The 841 patients with LSTV were placed into four modified Castellvi classification (9) subgroups; of these, 351 patients (41.7%) had LSTV type I, 349 (41.4%) had LSTV type II, 97 (11.5%) had LSTV type III, and 44 (5.2%) had LSTV type IV (Table 2).

LSTV and LBP

Overall, patients with LSTV were more likely to have reported LBP in the past 30 days than were participants without LSTV (OR: 1.42 [95% CI: 1.21, 1.68], P < .001) (Table 3). Four hundred eighty-four patients (10.4%) were diagnosed with Bertolotti syndrome (ie, LSTV with LBP). The subgroup analysis (Table 4) demonstrated that types II and IV were strongly associated with LBP. Compared to the non-LSTV group, types II and IV had respective ORs of 3.48 (95% CI: 2.62, 4.64) and 3.94 (95% CI: 1.63, 9.54). Types I and III had respective ORs of 0.78 (95% CI: 0.62, 0.99) and 0.69 (95% CI: 0.45, 1.07).

Table 3 ORs of Buttock Pain, LBP, and Middle/Upper Back Pain Compared within LSTV Patients and Subtypes Compared with Non-LSTV Patients

Note.—Results from a multinomial model.

^* OR, 95% CI, and P value from multinomial model. LBP, buttock pain, or middle/upper back pain compared to no back pain.

Table 4 Severity of Pain, Pain Frequency, and Limitation of Activity in LSTV Patients

Note.—Reference standard was patients without LSTV.

^* Proportional odds regression model adjusted for age, sex, race, body mass index, and hip OA was used.

^† Severity scale: 0 = no pain, 1 = mild pain, 2 = moderate pain, 3 = severe pain.

^‡ Frequency scale: 0 = rarely bothered by pain, 1 = bothered by pain some of the time, 2 = bothered by pain most of the time, 3 = bothered by pain all of the time.

^§ Past 30 days activity limited by back pain (yes or no). Logistic regression model adjusted for age, sex, race, body mass index, and hip OA was used.

LSTV and Buttock Pain

Individuals with LSTV were more likely to report buttock pain than were participants without LSTV (OR: 1.68 [95% CI: 1.09, 2.59], P = .019) (Table 3). Compared to the non-LSTV group, Types II and IV were associated with buttock pain, with respective ORs of 4.12 (95% CI: 2.31, 7.33) and 9.47 (95% CI: 2.62, 34.24) (Table 3). Types I and III were not associated with buttock pain.

LSTV and Middle and Upper Back Pain

Pain in the middle or upper back was not associated with the presence or absence of LSTV (P = .96) (Table 4). The relationships of LSTV subtypes with middle or upper back pain, however, were similar to those of LBP and buttock pain, with pain in type II (OR: 1.9 [95% CI: 1.04, 3.48], P = .037) and type IV (OR: 6.48 [95% CI: 1.95, 21.55], P = .002), and less pain in type I (OR: 0.44 [95% CI: 0.22, 0.88], P < .001).

Comparison between Back Pain and Other Variables

As presented in Table 4, back pain frequency (OR: 1.22 [95% CI: 1.01, 1.48], P = .042) and severity (OR: 1.40 [95% CI: 1.16, 1.69], P < .001) at any location were higher in the LSTV group; types I, II, and IV had a significant association with frequency and severity of back pain, but type III did not. Types II and IV were more likely to report higher severity or frequency, while type I was less likely to report higher severity or frequency of back pain. No significant differences in physical activity were observed between the LSTV and non-LSTV groups. However, the subgroup analysis demonstrated via PASE that types II and IV were associated with lower physical activity levels. Compared to the non-LSTV group, type II and IV had respective ORs of 1.67 (95% CI: 1.37, 2.05) and 2.39 (95% CI: 1.40, 4.06). Type I and III had no statistically significant associations with PASE.

No significant differences were found between LSTV and non-LSTV hip OA cohorts. No differences in presence, absence, or severity of hip OA were found among LSTV subgroups.

Reproducibility Analysis

The Cohen κ value for interobserver agreement on the presence or absence of LSTV in the entire cohort was 0.75, and intraobserver agreements for the two radiologists in a set of 500 randomly selected radiographs were 0.78 and 0.79. The Cohen κ value for interobserver agreement using the categorical Castellvi classification was 0.65, and the values for intraobserver agreement were 0.72 and 0.68. These values were classified as good reader agreement (23).