Do psychological and behavioral factors classified by the West Haven-Yale Multidimensional Pain Inventory (Swedish version) predict the early clinical course of low back pain in patients receiving chiropractic care?

Study design

In this prospective multicenter outcome study data were collected during the inclusion
phase of an ongoing randomized controlled clinical trial (RCT) in a multi-center setting
described in detail elsewhere 50].

Objectives

The objective of the study was to investigate if MPI-S subgroup (AC, ID, and DYS)
assignment at the 1
^st
visit among patients with recurrent and persistent LBP receiving chiropractic care
could predict the short-term clinical course using the following outcomes:

1. A definite improvement at the 4
^th
visit (primary outcome).

2. A clinically relevant reduction of pain intensity at the 4
^th
visit (secondary outcome).

Patients and setting

A total of 40 chiropractors were recruited to collect data on consecutive patients
seeking treatment for LBP. The clinicians were known to the research team as they
had successfully collected data in a previous research project 51] and were accustomed to integrating clinical research into their daily practice. The
clinics were located across Sweden. Patients of working age with recurrent (previous
episodes the past year) 52], 53] and persistent (more than 30 days of pain the past year) 54] LBP with no indication of serious spinal pathology were recruited. Patients who were
pregnant, did not pay for the treatments themselves or had chiropractic treatment
less than 3 months ago were excluded. See Table 1 for a summary of the inclusion and exclusion criteria.

Table 1. Criteria for eligibility

Data collection

The subjects were screened for eligibility during the 1
^st
visit and a questionnaire with demographic and baseline data (including MPI-S) was
administered by the clinician or the receptionist. As the data for this study were
collected during the screening phase of an RCT, only some demographic data were collected
at this stage (age was collected at the RCT inclusion visit and could therefore only
be specified for those subjects who were eligible for the RCT). A follow-up questionnaire
was administered (by the same clinician or receptionist) at the 4
^th
visit irrespective of when it occurred. If the patient reported a definite improvement
by the 2
^nd
or 3
^rd
visit, the follow-up questionnaire was administered at this visit. Only the patients
who reported a definite improvement went on to be included in the RCT. For this study
all subjects who responded to the 4
^th
visit questionnaire and had a complete MPI-S dataset were included. A detailed description
of the inclusion process of the RCT can be found in a recently published study protocol
50].

Dependent variables

The primary outcome measure was the self-rated improvement measured at the 4
^th
visit. The patients were asked to rate their perceived improvement on a five-step
ordinal scale (1?=?Definitely worse, 2?=?Probably worse, 3?=?Unchanged, 4?=?Probably improved, 5?=?Definitely
improved). Previous research has used the answer as a dichotomous variable defining improvement
only if the subject answered “definitely improved” to avoid overestimation of a positive
treatment outcome 41]–43]. Dichotomizing the variable in this way have been associated with clinically relevant
outcomes at 3 and 12 months post treatment and in this study the variable was used
in the same way to conform to previous research. As an alternative analysis the primary
outcome was also analyzed untransformed with the ordinal scale to capture the variation
within the material. The follow-up period was not pre-determined as it was up to the
chiropractors to schedule the visits according to their clinical judgment and the
patients’ preferences. However, in a previous study, the 4
^th
visit (during a course of chiropractic treatments) occurred within 2 weeks in 42 %
of the cases and between 2 and 4 weeks in 29 % of the cases 42]. Therefore it was likely that the follow-up period for the vast majority of the subjects
in this study would be within 4 weeks.

As a secondary outcome, the difference in pain intensity between the 1
^st
and 4
^th
visit on the NRS-11 scale was also analyzed. Pain intensity is a commonly reported
outcome in LBP research 55] and has been suggested as a standard outcome by the NIH Taskforce 56]. Change scores (difference between baseline and follow-up) of at least two points
or 30 % (for chronic pain patients) have been found to be clinically relevant 55], 57]. The data were thus analyzed as a dichotomous variable divided into a minimal improvement
of 30 % and less than 30 %. The main reason for choosing change score rather than
using an absolute pain score was to allow for comparison with the subjective global
improvement measure. Change scores also compensate for the likely differences in baseline
values of pain intensity of the subgroups inherent in the clustering procedure. Therefore
choosing 30 % change allows for a fair comparison of the clinical course between subgroups.
An alternative analysis was performed with pain intensity as a dichotomous variable
divided into an improvement of minimum 2 points and less than that.

Independent variable

Data regarding the subjects’ psychological and behavioral profile were collected during
the 1
^st
visit (as part of the screening questionnaire) using the MPI-S instrument. The MPI-S
is based on the cognitive-behavioral conceptualization of pain and is made up of 34
items resulting in 8 scales; 5 assessing psychological dimensions (Pain Severity (PS),
Interference (I), Life Control (LC), Affective Distress (AD), Support (S)) and 3 assessing
behavioral dimensions (Punishing Responses (PR), Solicitous Responses (SR), Distracting
Responses (DR)) 25]. From these eight scales three clusters or subgroups are derived, each with a particular
set of characteristics 27], 58], 59]. The AC subgroup is characterized by low pain severity, low interference with everyday
life due to pain, low life distress, high activity levels and high perception of life
control. This subgroup is considered to have the most favorable prognosis and the
least long-term sickness absence 30]. The DYS subgroup may be considered as the opposite end of the spectrum with high
pain severity, marked interference with everyday life due to pain, high affective
distress, low perception of life control and low activity levels. Thus, this is the
cluster with the worst prognosis with the most long-term sickness absence 30]. The ID subgroup has been characterized by deviating scores mainly in the behavioral
dimensions with low levels of social support, low levels of solicitous and distracting
responses from significant others and high scores on punishing responses compared
to the DYS and AC patients. These subgroups were used as the predictive variable in
the model.

Data analysis

The subgroups were formed from the eight original scales using centroid vectors from
a validated sample 27]. These centroid vectors have been used successfully in a previous publication to
characterize 4 different patient populations with LBP 60].

The dichotomized outcome for perceived improvement was analyzed using (robust) modified
Poisson regression to determine the chance (expressed as a relative risk) of a definite
improvement as well as a clinically relevant reduction of pain intensity at the 4
^th
visit stratified according to the three cluster structure from the initial screening
visit.

The patient’s expectation of improvement at the 1
^st
visit was thought to be a possible confounding factor. Expectation was measured during
the 1
^st
visit on an 11 point numerical rating scale asking “How great do you think the chance
is that your pain will improve considerably?” with answer options ranging from “No
chance (0)” to “Great chance (10)”. This single item question was modified from a
previously validated question used to measure the expectations of return to work 61].

Outcomes were analyzed without and with “patient’s expectation of improvement” from
the 1
^st
visit included in the regression model to control for confounding and interaction.

Ethics, consent and permissions

The study has been conducted according to good clinical research practice and the
guidelines of the Helsinki declaration. All subjects have signed an informed consent
form at the 1
^st
visit where they agreed to take part in the trial. The project has been approved by
the local ethical research committee at the Karolinska Institutet: 2007/1458-31/4.