Results

Table 1 reports the sample and its subsets’ demographic and clinical features. The sample included n=258 consecutive cases with mean age of 67.4 (±15.0; range 27-93) years; 67.4% (n=174) were females. The most frequent pain localizations, 67.3% (n=175), was ‘lumbar spine and sciatica’. The Median of V1-V2 visit-interval was 82 days. The sample subsets were similar for all demographic and clinical features except for mean age and PD-V1 outcomes. Indeed, the mean age in the BUAS+/SLART+ subset (64.5 ±15.7) was slightly lower than that of the BUAS+/SLART- subset (69.3 ±14.3); uncertain PD-V1-outcome was reported by 79% and 19% of the BUAS+/SLART- and BUAS+/SLART+ subsets, respectively; inverted percentages were reported for the positive PD-V1-outcome. Thirty patients (1.2%) reported initial dizziness and somnolence during the gabapentin titration period; however, no patient stopped the therapy as these side effects decreased over time.

PD outcomes

Table 2 shows the frequency distribution of PD-V1 and PD-V2 outcomes, and the mean (± SD) of the PD final scores, split by the two subsets of patients: the BUAS+/SLART-, and the BUAS+/SLART+ one. At V2, 77% (n=199) of the entire sample showed negative PD-outcome. In particular, at V2, all patients (100.0%) of the BUAS+/SLART- subset showed negative PD-outcome. Of the BUAS+/SLART+ subset, 43% showed negative PD-outcome at V2. In this subset, positive PD-outcome at V1 decreased at V2 by 60%, and uncertain PD-outcome increased roughly two folds. Finally, in the BUAS+/SLART+ subset, among patients who showed positive PD-outcome at V1, 37% and 23%, respectively, shifted at V2, to uncertain or negative PD-outcomes; 20% remained, at V2, with positive PD-outcome.
To note, we have compared, both in the sample and the study’s subsets, the mean of PD-V1 and PD-V2 final scores (Table 2) and found significant differences between them (t-test, p<0.001, respectively). At V2, there were no PD ‘positive’ or ‘uncertain’ outcomes in the BUAS+/SLART- subset. By contrast, in the BUAS+/SLART+ subset, mean PD-V1 final scores (vs PD-V2) decreased, respectively, from 23.8 to 19.0 for the ‘positive’ PD outcome, and from 18.0 to 15.9 for the ‘uncertain’ PD outcome. In both subsets, the mean of the ‘negative’ PD-outcome at V2 was 4.3 (BUAS+/SLART-), and 6.7 (BUAS+/SLART+), respectively. Interestingly, in the BUAS+/SLART+ subset, PD-V1 ‘positive’ and ‘uncertain’ final scores were roughly 5 and 3 points higher.

BPI outcomes

Fig. 2 shows, for the entire sample, the mean scores of the BPI items, at V1 and V2, and their V2-V1 absolute differences (solid line; V2-V1 values are anchored by a dashed line). For all BPI items, the V2-V1 differences of the BPI item mean scores were of >2 points, implying a clinically significant improvement in both pain control and the pain-interference with daily activities. All BPI items showed statistically significant differences between V2 and V1 (t-test , p<0.001, respectively). The same results were also found within the study’s subsets, respectively. For details view table 3, which shows the mean scores of the BPI items spilt by the BUAS+/SLART- and the BUAS+/SLART+ subsets, their V2-V1 differences (Δ and 95% CI), and the results of the t-test analyses. To note, the relief obtained with the therapy improved by roughly 40% in both subsets. All BPI items in both subsets showed statistically significant differences between V2 and V1 (t-test, p<0.001, respectively).

Associations and MLR

Table 4 reports the results of the association analyses (χ2-analysis) between the PD-V2 outcomes and the independent predictors. It also reports the post hoc analyses results as cell contribution for the reported association (the two most influential contributions). No significant associations were found between the PD-V2 outcomes and the independent predictors gender, age groups, and pain sites (χ2-analysis, p > 0.05). Statistically significant associations (χ2-analysis, p < 0.0001, respectively) were found between the PD-V2 outcomes and the independent predictors SLRT-V1 and PD-V1, respectively. In particular, post hoc analysis showed that the ‘positive’ and ‘negative’ SLRT-V1 outcomes were associated, respectively, with ‘uncertain’ and ‘negative’ PD-V2 outcomes. Finally, ‘positive’ and ‘uncertain’ PD-V1 outcomes were associated, respectively, with ‘uncertain’ and ‘negative’ PD-V2 outcomes.
For the MLR analysis, PD-V2 outcomes were the dependent variable (reference outcome class, ‘negative’) while SLRT-V1 and BUAS-test-V1 outcomes were the predictors. Other predictors were excluded, given the lack of associations observed with the χ2-analysis. As reported in table 5, The model was found to fit the data significantly (p=0.000), and the predictor SLRT-V1 had a significant overall effect on the outcome (Likelihood Ratio Tests, p=0.000). In particular, for the ‘positive’ and the ‘uncertain’ PD-V2 outcomes, the predictor ‘negative’ SLRT had adjusted odds-ratios (p-values, 95% CI) of 0.038, respectively (p=0.000, 0.012-0.115 and 0.016-0.090, respectively). Thus, for ‘negative’ SLRT cases, the risks for ‘positive’ or ‘uncertain’ PD-V2 outcomes to occur decrease versus the Reference outcome class PD-V2 ‘negative’.