123 Hispanics/Latinos with chronic spine pain will be identified through consecutive
sampling of patients at a local FQHC who were referred by their primary care physician to
physical therapy (PT) for neck or low back pain. Participants who are deemed eligible and
enroll in the study will be scheduled for the first study assessment, where they will
provide written informed consent for participation. Participation includes 4 study
assessments and 8-weeks of treatment with either GOALS or Usual Care PT. Participants
will be stratified by sex (male/female) and diagnosis (low back pain/neck pain) and
randomized into either the GOALS or the Usual Care group following completion of the
Baseline assessment. Study assessments will be conducted 1-week before (Baseline) and
1-week (Post), 3-months (3-mo Post), and 6-months (6-mo Post) after completing 8 weeks of
GOALS or Usual Care PT.
The projected sample size for GOALS changed from N=138 to N=123 after adding an
eligibility criterion requiring participants to attend the PT evaluation. Our power
analysis indicated N=98 participants is the smallest sample size needed to detect the
effect of the GOALS intervention. Before beginning the trial, we estimated a 41%
attrition rate given data from the partnering clinic, suggesting we needed to recruit
N=138 to meet our N=98 sample size. However, we added an inclusion criterion that
participants need to attend the initial PT evaluation in order to be enrolled in the
study. We retained an estimated attrition rate of 20% of enrolled participants, requiring
an initial sample size of N=123 to meet the N=98 minimum sample size.
The primary aim of the study is to determine the efficacy of a culturally adapted
cognitive behavioral based PT tele-rehabilitation program compared to usual care in
Hispanics/Latinos with chronic spine pain, based on improvement in the pain interference
scale in the Brief Pain Inventory after treatment (primary outcome). The 1-week
post-intervention assessment will be analyzed using a linear regression model. The time
by condition (GOALS vs. Usual Care) interaction term will be added to the models to
assess for intervention effects. If the interaction is not significant, the main effect
for time and condition will be assessed. For all outcomes, the models will be adjusted
for sociodemographic and clinical characteristics that may affect treatment response.
Secondary outcomes of pain interference at other time-points (3- and 6-mo Post) and
patient reported outcomes and physical performance measures at all time points will be
evaluated in the same manner. It is hypothesize that participation in the GOALS
intervention compared to Usual Care will result in greater improvements in pain
interference (primary outcome), as well as secondary patient reported outcomes and
physical performance measures.
Pre-planned secondary analyses will explore biological/physical, psychological, and
environmental moderators and mediators of treatment response. The dependent variable for
this analysis will be treatment response, whereby participants will be dichotomized into
responder and non-responder groups based on a ≥30% reduction in pain interference as
recommended by IMMPACT guidelines. An exploratory, backward selection approach using
separate multivariate logistic regression models will be used to identify predictors that
are most strongly related to treatment response. These variables will then be entered
into a single combined multivariate logistic regression model adjusted for
sociodemographic covariates of treatment response. The study will also explore potential
mediators of treatment response with structural equation modeling (SEM) using maximum
likelihood estimation methods. Parameter estimates will be evaluated and indirect
effects, reflecting mediation, will be calculated and examined for statistical
significance and clinical meaningfulness. It is hypothesized that baseline indicators for
high fear of movement and sedentary behavior will predict a positive response to the
GOALS intervention. However, the study will also collect and analyze other candidate
predictors of treatment response based on a conceptual model of the pain experience in
Hispanics/Latinos.