I
am Richard Francoeur, Associate Professor of Social Work, at Adelphi
University. As a recipient of a CHI
Summer Scholars Award to participate in a course last summer on SAS programming
offered by the Epidemiology and Population Health Summer Institute at Columbia
University, I was invited to submit this blog this month to share about how the
course relates to my health scholarship and research endeavors. The SAS programming course provides a basic
overview about statistical syntax coding and analysis in a major software
program. I took this course to expand my
knowledge of this software program, which I may use in future work. The remainder of this posting will focus on my
work in comorbidity, an area informed by statistical thinking.
I
conduct research and scholarship to advance our knowledge of older,
middle-aged, and underserved adults with chronic physical illness. My publications, grants, presentations, and
current investigations emphasize 1) Older and middle-aged adults coping with
medical conditions or related physical symptoms who present with depression;
and 2) Hidden or emerging clinical issues in older and underserved populations
with chronic illness conditions, especially during palliative care.
An
earlier vein of my research produced a series of pioneering conceptual and
empirical publications on age-related financial stress-strain relationships and
underinsurance in outpatients receiving palliative radiation for recurrent
cancer. This research focused on patient
age as a buffering or magnifying influence on the relationship between
objective financial stress incurred by the patient and their family and the
patient's subjective perceptions of various aspects of financial strain that
they were experiencing. Financial stress
may have different impacts on health care access and health seeking behavior
depending upon the level of financial strain experienced. Thus, these factors constitute distinct
comorbidity influences.
For
the last several years, I have become interested in issues of comorbidity more
generally. Generally speaking, co-occurring conditions, disease markers, pain
and symptom clusters, economic contexts, and/or psychosocial factors may
interact to magnify or buffer relationships to health or mental health
outcomes. One type of focus in my work on comorbidity pertains to investigating
pain and physical symptoms in cancer that occur in pairs or clusters and are
related to depressive symptoms, which may consist of "sickness
malaise" as well as mental health reactions to physical symptoms. This area of research is important because
detecting co-occurring physical symptoms and understanding their influence on
sickness malaise and mental health can provide valuable clues for proactive
assessment in subgroups of patients, as well as about interventions with
"cross-over" effects (i.e., one intervention could relieve multiple
symptoms). It is practical to
investigate the interactions among pain and physical symptoms as they predict
mental health symptoms. Since social
workers and other mental health providers may be more aware of mental health
symptoms, such as those of depression or anxiety, they may become more likely
to screen for related physical symptoms as they become more aware of their
synergistic influences, which may have deleterious effects on mental health.
I
also innovate statistical methods and models I use in my research to improve
detection and interpretation of these synergistic influences. In the first statistical innovation I
published, I extended a non-graphical follow-up algorithm for interpreting
two-way interactions into a more comprehensive procedure that can also
interpret curvilinear interactions and three-way interactions in multiple
regression. Other investigators in
discussion forums for leading statistical software (Stata, SPSS) have noted
this innovation. I consider a second
statistical innovation I published as a kind of "breakthrough"
because it overcomes low sensitivity in multiple regression to detect terms
that involve interactions among predictor variables, a vexing challenge to
researchers ever since computer software to conduct multiple regression became
available in the 1960s. I have advocated
for the use of this statistical innovation in a few ResearchGate discussion
forums in which investigators posed questions regarding how to deal with
multicollinearity in interaction terms.
Incidentally, many of my publications can be downloaded from
ResearchGate (https://www.researchgate.net/profile/Richard_Francoeur). Currently, both statistical innovations are
being developed into an app that will be hosted by Adelphi, and this app should
be of interest beyond issues of comorbidity to research situations in many
fields in which there is a need to detect and interpret interactions among
variables.
A
second type of focus in my work on comorbidity pertains to detecting
psychometric profiles within subgroups.
I am currently working on publishing a new modeling specification
strategy that links multiple regression fully and without bias to confirmatory
factor analysis by making it possible to estimate all causal paths to a latent
construct and its observed items. I will
discuss this new strategy in the context of articles I am developing that will
report psychometric profiles of depression in subgroups characterized by
progressive cerebrovascular disease (hypertension, silent cerebrovascular
disease, stroke, post-stroke cognitive impairment, vascular cognitive
impairment), and some of these cerebrovascular subgroups will be qualified
further by co-occurring excess weight and diabetes. As with the first type of focus in my work on
comorbidity, for any of these subgroups, the psychometric profiles of
depression items (the 20 items of the CES-D Depression Inventory) will also
constitute a "symptom cluster," however it will differ in that the
depression items are not tested to be mutually interactive or synergistic in
their effects. Rather, they constitute
several psychometric items that constitute a psychometric profile for a
particular subgroup of patients characterized by progressive cerebrovascular
disease, and in some cases, by excess weight and diabetes as well.
My
publications are influential, particularly in the multidisciplinary areas of
cancer pain and symptom clusters, palliative care, and financial burden. I am pleased to report that I was recognized
this past year as a Fellow of the Social Research, Policy, and Practice section
of The Gerontological Society of America.
Selection as a Fellow is an acknowledgment by professional colleagues of
outstanding and continuing work in the field of gerontology and represents the
highest class of membership in the Sociey.
My work has also attracted much interest in high-quality open-access
journals as well as social networking sites, including ResearchGate. Two of my open-access articles are highly
accessed: 1) a recent report of my statistical innovation that improves
detection of interaction effects in moderated regression; and 2) a commentary
article on novel community programming strategies to ensure safe access to
medication for palliative care while preventing prescription drug abuse. I recently had the opportunity to share this
commentary article, as well as my publications on pain and symptom clusters in
palliative care, with the CEO of the Gerontological Society of America (GSA),
James Appleby, who had solicited ideas, recent research, and resources
regarding how to ensure people with pain receive the care they need while also
countering abuse, misuse, and diversion of prescription medicine. James Appleby seeks this information to help
inform his advocacy and policy conversations with the Alliance for Balanced
Pain Management (AfBPM), a group in which GSA has recently begun a partnership
or collaboration.
