Reply to ‘Timing of the Baseline Assessment Confounds Efficacy Claims for a Post-Surgical Telehealth Mindfulness Intervention After Lumbar Surgery’

We appreciate the methodological concerns raised by Abudayeh et al in their letter regarding our published study on the feasibility, acceptability and adaptation of a postsurgical telehealth mindfulness-based intervention (MBI) aimed at enhancing pain reduction and other patient-reported outcomes following lumbar spine surgery. ¹ Our single-arm, pilot study was designed to adapt the MBI for this patient population and conduct an initial evaluation of its feasibility and acceptability in order to plan for a possible randomized controlled trial (RCT) evaluating whether this intervention enhances both short- and long-term outcomes of spine surgery when offered as part of the rehabilitation pathway. Additionally, we reported paired t-tests measuring changes in clinical and psychosocial outcomes from preoperative assessment to 3 months postoperatively, as well as the percentage of participants who achieved a minimal clinically important difference (MCID), to illustrate the potential clinical utility of the postoperative MBI. Descriptive statistics were also provided for measures at 2 weeks post-surgery prior to starting the MBI. In their letter, Abudayeh et al suggest that analyzing outcomes between 2 weeks and 3 months postoperatively would better isolate the effects of the MBI without inflating results from surgical improvements.

We acknowledge the importance of isolating intervention effects. It is important to note that the preoperative-to-postoperative design of our study aligns with established practices in research on perioperative behavioral interventions for lumbar spine surgery patients.^2,3 For example, in a meta-analysis of high-quality randomized controlled trials (RCTs) evaluating perioperative cognitive-behavioral interventions in spine surgery, all included studies utilized a preoperative to postoperative analysis, with postoperative time points ranging from 3 weeks to 3 years. ³ The specific effects of these interventions were isolated using random assignment and comparison groups. We utilized a similar baseline time point as these studies to plan for a future RCT that will expand upon the existing evidence on perioperative behavioral interventions. In addition, early postoperative assessments are confounded by significant pain from the surgical procedure and opioid medication. Postoperative complications and potential readmission as well as hospital discharge to a facility also occur making this early timeframe an unreliable baseline for isolating postoperative intervention effects.

To address the letter writers’ question as to whether analyzing outcomes from 2 weeks postoperative (pre-intervention) to 3 months postoperative (post-intervention) would yield different results compared with using the preoperative baseline, we conducted new paired t-tests using the postoperative time points. Results (shown in Table 1) indicated that there were significant improvements in back pain intensity, leg pain intensity, pain bothersomeness, pain interference, and disability, with large effect sizes (Hedge’s g), although effects sizes appeared smaller for back pain intensity and pain bothersomeness when using the 2-week postoperative baseline. There were also significant improvements in kinesiophobia, pain catastrophizing, and pain-related self-efficacy, with medium to large effect sizes. Only the effect size for pain catastrophizing was smaller than that in the original analysis. Unlike the original results, there were no longer significant reductions in depression and anxiety, and there were medium-sized improvements during this time frame compared with large effect sizes when analyzed from the preoperative baseline. Aligned with the previous results, there were no significant changes in perceived stress or dispositional mindfulness. Our original analyses using the preoperative baseline assessment can be found in our prior publication. These findings suggest that for most outcomes, there was continued, medium to large-sized improvements from pre-to post-intervention during the postoperative period. For some of these outcomes, effect sizes were smaller, likely due to improvement that occurred between the preoperative and 2-week postoperative time points encompassing the surgery. We emphasize that any effect sizes from a small pilot study are underpowered and unreliable and should be interpreted with caution. Our primary outcomes were metrics of feasibility including enrollment rate, retention, session attendance, and intervention satisfaction.

Table 1.

Within-Subject Changes in Pain, Disability, and Psychosocial Factors (N = 12)

Measure (possible score)	Mean w/in-subject diff (SD), [95% CI] 3 mos postop – 2 wks postop	Hedge’s g effect size [95% CI]
Back pain intensity (0-10)	−2.33 (2.78) [−4.10; −0.57]*	−0.78 [−1.39; −0.15]
Leg pain intensity (0-10)	−3.33 (3.14) [−5.31; −1.26]**	−1.00 [−1.65; −0.31]
Pain bothersomeness (0-4)	−1.17 (0.94) [−1.76; −0.57]**	−1.15 [−1.85; −0.43]
Pain interference a	−11.81 (10.24) [−18.32; −5.31]**	−1.07 [−1.75; −0.37]
ODI (0-100)	27.00 (14.00) [18.10; 35.90]***	1.79 [0.87; 2.69]
Depressiona	−2.89 (6.43) [−6.98; 1.20]	−0.42 [−.96; 0.14]
Anxiety a	−5.40 (9.96) [−11.73; 0.93]	−0.50 [−1.06; 0.07]
TSK-13 (13-52)	−6.20 (4.71) [−9.19; −3.21]***	−1.23 [−1.94; −0.48]
PCS (0-52)	−6.25 (8.56) [−11.69; −0.81]*	−0.68 [−1.26; −0.07]
PSEQ (0-60)	13.42 (13.13) [5.07; 21.76]**	0.95 [0.28; 1.59]
PSS-4 (0-16)	−0.25 (3.65) [−2.57; 2.07]	−0.06 [−0.59; 0.46]
FFMQ-15 (1-5)	0.16 (0.29) [−0.03; 0.34]	0.50 [−0.08; 1.05]

Abbreviations: Preop, preoperative; Postop, postoperative; ODI, Oswestry Disability Index; TSK, Tampa Scale of Kinesiophobia; PCS, Pain Catastrophizing Scale; PSEQ, Pain-related Self-efficacy Questionnaire; PSS, Perceived Stress Scale; FFMQ, Five Facet Mindfulness Questionnaire.

*P < 0.05; **P ≤ 0.01; ***P ≤ 0.001, paired samples t-test with two-sided significance test.

^aRaw scores of PROMIS measures are converted to normed T-scores based on a reference population with a mean of 50 and a standard deviation of 10.

In their letter, Abudayeh et al also asked whether recalculating MCID attainment in pain and disability outcomes using the 2-week baseline would alter the percentage of participants achieving clinically meaningful benefit. We believe that using established MCIDs for patient-reported outcome measures in lumbar spine surgeries is no longer valid if ascertained using the postoperative baseline. These MCIDs were established for changes from preoperative assessment to outcomes measured up to 1 year post-surgery using items assessing patients’ reported satisfaction with the outcome of their spine surgery. ⁴ Thus, the time frame over which these MCIDs are established encompasses patients’ perception of improvement from the time they had surgery through rehabilitation. However, it may be useful in future studies to utilize validated MCIDs for reductions in acute and subacute postoperative pain to ascertain the clinical benefit of interventions to manage postoperative pain. To our knowledge, these have not been established in spine surgery populations.

We acknowledge that our single-arm pilot study cannot definitively attribute clinical improvements to the MBI. As noted in our limitations, a fully powered RCT with an appropriate comparison group is necessary to evaluate the distinct effects of the MBI beyond the effects of surgery and other postoperative management received. It is our hope that the results of our study are viewed in light of their exploratory nature and used to conduct larger, high-quality studies. We recommend that if a future RCT is conducted, the assessment time points are carefully selected to adequately test the study’s hypotheses and to allow for interpretation of the results within the context of existing literature on perioperative behavioral interventions in spine surgery. We sincerely thank Abudayeh et al for their thoughtful critique and their commitment to scientific rigor.