Effectiveness of Narrative Medicine on Pain Intensity and Quality of Life
Background and Significance: To our knowledge there are no randomized controlled trials
(RCTs) that evaluate the benefits of narrative medicine, but there has been research on the
effect of writing about traumatic events on health outcomes which resembles writing a
narrative. However, the populations included have been almost exclusively young healthy
students or workers, and the majority of studies have been performed by the same research
Study Population: We will recruit patients with cancer who report a pain intensity of at
least 5/10. We will include patients who are still able to read and write. To evaluate this
physical ability, we will use the Karnofsky scale. The minimum score a patient needs to have
will be 50%.
Study Design: This is a randomized single blind (evaluator) controlled trial. Patients will
sign an informed consent. Patients will be randomized into three groups.
1. Narrative group. Patients will write a story about their illness for at least 20 minutes
once a week for 3 weeks; 2. Writing-control group. Patients will fill a pain diary
(requiring approximately 20 minutes) once a week for three weeks; 3. Control group. Subjects
will not write/fill out anything.
Patients in the narrative group will be asked to write about their illness and how it has
affected their lives. We will not guide the narrative in any other way.
The number of office visits will be the same in the three groups; all patients will be seen
weekly up to a total of 8 weeks. In addition, the pain physician will fill a check list to
make sure that the presence and severity of symptoms important to patients, such as pain,
nausea, anorexia, constipation, and lack of sleep are addressed in all subjects.
We will evaluate the content of the narratives. For the evaluation of the narratives, we
will use a Likert scale that goes from no emotional disclosure to very much emotional
disclosure. We will also count the number of cognitive words and the number of positive and
negative emotion words. The quantitative analysis of vocabulary will be performed with a
text-analysis computer program.
Primary Outcome: Reduction of pain intensity is the primary outcome. Pain intensity will be
evaluated by a registered nurse, who is not aware of the group allocation, before
randomization and then weekly for 8 weeks. To evaluate pain, we will use the numerical
rating scale in which 0 is no pain and 10 is the worst pain imaginable.
Secondary Outcomes: Health related quality of life and the sense of well-being are secondary
outcomes. For the evaluation of health related quality of life, we will employ the treatment
outcomes of pain survey (TOPS) which is a modification of the SF-36 for patients with pain.
Patients will fill the questionnaire with the help of a registered nurse, who is not aware
of the group allocation, before randomization, and then at 4 weeks and 8 weeks.
For the evaluation of global well-being, we will use a Likert scale that goes from very bad
to excellent. Patients will be asked the global measure of well-being weekly.
In addition, we will evaluate how pain interferes with general activity, mood, work,
relation with others, sleep, and enjoyment of life using the interference factors of the
Brief Pain Inventory (BPI).
Adherence to Treatment: We will call all participants once a week to remind them about
writing the story, the pain dairy, or to come to the office visits.
Data Analysis: We will use an intention to treat analysis.
To analyze the effect of the treatment on pain intensity, we will employ an analysis of
repeated measures using generalized estimating equations (GEE).
To determine the effect of the exposure on quality of life, we will follow the same
procedure described above. The outcome of interest will be the score of each dimension
evaluated in the TOPS. As explanatory variables, we will include in the regression models
the treatment group, the time, and the interaction between treatment group and time.
To analyze the effect of the exposure on pain interference, we will employ GEE (similar to
the procedure we described above). The outcome of interest will be the BPI global score,
which is a sum of the scores of each one of the dimensions evaluated (general activity,
mood, work, relation with others, sleep, and enjoyment of life). If the BPI global score is
different among groups, we then proceed to evaluate each dimension separately to determine
which one is affected by narrative medicine. As explanatory variables, we will include in
the regression models the treatment group, the time, and the interaction between treatment
group and time.
To determine the effect of the exposure on the global measure of well-being, we will use a
similar approach described for pain intensity and quality of life. The outcome variable will
be the well-being score and the explanatory variables will be treatment group, the time, and
the interaction between treatment group and time.
To determine the impact of the quality of the narrative on each of the outcomes evaluated
(pain intensity, pain interference, quality of life, and well-being) we will include in the
corresponding regression model, in addition to the explanatory variables (treatment group,
the time, the interaction between treatment group and time), the emotional disclosure score.
Due to the subjectivity of the emotional disclosure evaluation, two physicians will perform
the evaluation (the pain physician and the psychiatrist). We will measure the agreement
between the two evaluators using the Kappa statistics for ordinal variables.
To identify and characterize those patients who best respond to narrative medicine, we will
include in all the regression models described above the gender, education level, age of the
subject, and the interaction between these variables and the treatment group. The evaluation
of the interactions will permit us to determine if the effect of narrative medicine depends
on the gender, education level, or age of the patients. In addition, we will use latent
growth curve analyses to explore the effect of these variables.
To establish the feasibility of narrative intervention, we will compare the number of
patients who adhere to the treatment in each group using a chi square test.
Study Period: The exposure will have a duration of 3 weeks and then we will continue the
follow up for 5 more weeks, for a total of 8 weeks. We estimate that in 12 months we can
finish the recruitment and follow up of the patients and that we need 3 more months to
analyze the data and to prepare the manuscript for publication.
Sample Size: We have shown that a difference of 2 units in a scale from 0 to 10 is
clinically important to patients. To detect such difference between the groups with 80%
power and an alpha error of 0.05, assuming that the baseline pain intensity in our patients
will be on average 6 ± 4, we estimated the need for 63 patients per group. To allow for a
10% loss of follow up, we will include a total of 70 patients per group.
To assure that patients with advanced cancer will be similarly distributed in the three
groups, we will use a stratified randomization. Our measure of disease severity will be the
Karnofsky scale. The cutoff will be 80%. A score of 80% or higher indicates that the patient
is able to work.
Allocation: Randomized, Endpoint Classification: Efficacy Study, Intervention Model: Parallel Assignment, Masking: Single Blind, Primary Purpose: Treatment
M. Soledad Cepeda, MD, PhD
Javeriana University School of Medicine
Colombia: Institutional Review Board