presents a flowchart of study recruitment. One hundred and seventy patients were eligible to participate in this observational study. Of these, seven patients were not offered informed consent due to failure on the part of the recruiting physicians, and an additional three patients refused to participate. Of 160 consented patients, three patients experienced clinical improvement and did not go on to receive MRI, and three patients were excluded for having nerve root impingement due primarily to causes other than LDH. One hundred fifty four subjects met initial criteria, including 106 subjects in the younger group (age <60) and 48 subjects in the older group (age ≥ 60). 21 subjects went on to receive lumbar decompression surgery, and were excluded from this analysis of nonsurgical outcomes. Individuals who underwent surgery were younger than those who did not (48.4 ± 13.2 vs. 53.6 ± 13.5; p=0.05), were less likely to be retired, and were more likely to be disabled or unemployed. There were otherwise no demographic or clinical factors that were significantly associated with surgical treatment over the follow-up period (data not shown).
The study sample had a mean age ± SD of 53.6 ± 13.5 years, with 33% of participants of female gender, and 93.8% of white race. The age of older adults ranged from 60 to 87. 60% of older adults were age 60-69, 27% were age 70-79, and 13% were 80 or older. Patients who were eligible to participate but were missed or refused were not materially different from study participants with respect to demographic features. Baseline characteristics of the study sample by age group are presented in . Older adults had higher comorbidity burden (median [IQR] of 1 [0,3] vs. 4 [2,6]; p= <0.0001) and a shorter duration of symptoms (4.2 ± 3.4 vs. 5.2 ± 2.8; p=0.006) at clinical presentation as compared to younger adults. Employment status was significantly different in older adults (p<0.0001). Some physical examination and MRI characteristics differed by age group. A positive straight leg raise test (SLR) was significantly less common in older adults, and conversely, a positive femoral stretch test (FST) was significantly more common in older adults. Midlumbar disk herniation and foraminal disk herniations were more common in older adults. Baseline ODI scores and VAS leg pain were comparable in younger adults and older adults. Baseline VAS back pain, however, was slightly lower in older adults as compared to younger adults (4.2 vs. 5.4; p=0.07).
Baseline Characteristics of the Study Sample by Age Group
Associations between age group and outcomes of nonsurgical treatment at six-month follow-up are presented in . The proportion of missing data for the outcomes of change in the ODI, leg pain, and back pain was 8%, 4%, and 4%, respectively. There were no statistically significant bivariate associations between age group and the primary outcome of ODI change score at 6 months. In multivariate analysis including the covariates of gender, race, employment status, prior LBP, tobacco history, comorbidity (SACQ), duration of symptoms, baseline ODI, herniation level, herniation location, and herniation morphology, the association of age group with ODI remained nonsignificant. Up to 3% of data were missing for some covariates. Age group was not significantly associated with the secondary outcome of leg pain in bivariate analyses. In multivariate analysis including all covariates used in the full model for ODI described above (with adjustment for baseline leg pain), age group continued to not be associated with leg pain change scores. Older adults showed significantly less improvement in back pain as compared to younger adults (2.0 ± 4.1 vs. 3.2 ± 3.1; p = 0.04) in bivariate analysis. However, older adults had reported less back pain at baseline as compared to younger adults (see ). In multivariate analysis including all covariates used in the full models described above (with adjustment for baseline back pain), adjusted back pain improvement was not significantly different in older adults as compared to younger adults (2.4 vs. 2.7; p=0.69). When the outcomes of change in ODI, leg pain and back pain at 6 months were expressed instead as % change from baseline, age group was not significantly associated with any outcome in both bivariate and multivariate analyses (data not shown). When age was treated instead as a continuous variable in a secondary analysis, age remained not significantly associated with change in ODI, leg pain, or back pain (data not shown).
Associations between Age Group and Outcomes (Change Scores) at 6 month Follow-up
describes treatments utilized by study participants over the 6-month follow-up period. Oral corticosteroid tapers were utilized less frequently in older adults than in younger adults (16% vs. 29 %; p=0.09). Physical therapy was utilized more frequently in older adults than in younger adults (79.6% vs. 59.6 %; p=0.02), as were transforaminal ESIs (37.5% vs. 22.6 %; p=0.06). To account for the influence of treatments received, we conducted secondary analyses of the associations between age group and 6-month change scores for disability and pain, while adjusting for the utilization of oral corticosteroids, physical therapy, transforaminal ESI, and baseline covariates. Our findings were not materially changed by accounting for these treatments. No single treatment was significantly associated with outcomes for disability and pain (data not shown).
Treatments Utilized During the Follow-up Period
In longitudinal analyses, we examined the outcomes of disability and pain by age group at 1 month, 3 months, and 6 months, while adjusting for demographic and historical features which were significant in our primary analyses: comorbidity score, duration of symptoms, and work status. Specific herniation characteristics were not adjusted for, because we did not wish to ‘adjust out’ for age-related anatomic factors. depicts outcome scores over time, adjusted for comorbidity, duration of symptoms, and work status. Both groups demonstrated the largest improvements in ODI and pain scores over the first month of follow-up, with a slower rate of improvement thereafter. In longitudinal analyses, no age group*time interaction was found for disability on the ODI, indicating no differences in rates of improvement between older and younger individuals. Significant age group*time interactions were noted, however, for the outcomes of leg pain (p=0.02) and back pain (p=0.04). The meaning of this interaction can be easily appreciated by simple visual inspection of longitudinal trends for adjusted pain scores in . This figure demonstrates that trajectories of improvement between age groups were similar for the outcome of ODI. However, a greater amount of the total improvement in leg pain and back pain intensity in older adults was noted in the first month of follow-up, as compared to younger adults.