Suffering Begets Suffering, and the Future of Primary Care
Suffering Begets Suffering, and the Future of Primary Care
Stefan G. Kertesz
0 University of Alabama at Birmingham School of Medicine , Birmingham, AL , USA
1 Birmingham VA Medical Center , Birmingham, AL , USA
T suffering, health care, and what policy advocates call the
wo papers in this issue delve into the interplay of illness,
social determinants of health (Fig. 1). Both deliver
wellcrafted evidentiary glimpses into how different forms of
human suffering collide, overlap, and reinforce one another.
Each tells a somewhat different tale, but leads to a similar
quandary about whether our attributional models remain
unequal to the task of understanding the suffering we health care
providers and policymakers are called upon to address.
The first, by Charkhchi et al., looks at 24,480 American
survey respondents with major medical conditions. All were
respondents to the Behavioral Risk Factor Surveillance
System (BRFSS) survey, with particular focus on respondents
who had one of four illnesses (cancer, stroke, cardiovascular
disease, and chronic lung disease).1 Secondarily, the authors
tested the data for plausible effects of housing or food
insecurity on health care access, and on health status itself.
The latter purpose aligns closely with a proliferation of
initiatives in which communities seek to address social
determinants as an upstream intervention to avert health
The BRFSS analyses revealed that housing insecurity
(37%) and food insecurity (31%) were somewhat more
common among persons who had major illnesses, compared to
persons who did not (33% for housing, and 24% for housing).
The adjusted odds of having food or housing insecurity
were more pronounced for cardiovascular and lung disease,
less pronounced for cancer and stroke. The authors suggest
that stroke often proves so debilitating that survivors receive a
surge of support, often moving in with family or into nursing
facilities. In this way, persons suffering stroke obtain
protection not consistently available for the other three illnesses.
In secondary analyses, housing and food insecurity were
associated with double the odds of reporting difficulty
accessing health care. Housing (but not food) insecurity was
associated with lower odds of having good health status. The
authors suggest that chronic illness affects both housing and
food security, and that both may affect access to care and,
possibly, health status.
An alternative explanation, one among many, would
suggest that some associations reflect causal vectors that sit
outside the available data. For example, perhaps lung and
cardiovascular disease are somewhat more robustly associated with
housing and food insecurity as a result of behavioral risks, like
tobacco. Such risks are more prevalent among Americans who
are poor. From this point of view, one could speculate that a
lack of economic power may, on average, predispose to (a)
risk behaviors and (b) long-term economic hardship, both of
which emerge in a statistical portrait of certain illnesses in the
BRFSS. Many other hypotheses could be proposed.
The challenge of multiple compelling explanations applies
with particular power to the present study of homeless adults
by Patanwala et al.3 The authors had a different objective,
which was to illustrate the burden of physical symptoms
among a cohort of persons over the age of 50 who were
currently or recently homeless (Bhomeless experienced^
persons). The resulting paper reflects an extraordinary data
collection process, in which the research team sought 350
currently homeless older adults living in the area around Oakland
California, interviewed them extensively, and then followed
them. It is rare to obtain so rich and rigorous a portrait of the
increasingly old and sick people experiencing homelessness
Analytically this cohort reflects a daunting range of social,
biological, environmental, and life history variables as
plausible contributors to the experience of symptoms as basic as pain
and difficulty sleeping.
Structuring their analysis is the Theory of Symptom
Management, a framework developed by scholars of nursing.4
The results will not necessarily surprise readers but they are
meaningful in more than one way. Descriptively, the
HOPEHOME sample is vulnerable. Over half had experienced
childhood abuse, and over half also experienced recent abuse.
Symptom burden was high (one third qualified for
moderatehigh physical symptom burden) and psychological symptoms,
loneliness, and regret were extremely prevalent.
Multivariable analyses of these cross-sectional data
suggested several variables were independently associated with
symptom burden (female sex, childhood abuse, cannabis use,
having ≥ 2 chronic conditions, anxiety, hallucinations, and
In reading both papers, a certain frustration is almost sure to
arise. Some part of the reader seeks clear, linear associations
between cardinal variables. This is, I suggest, our Binner
epidemiologist.^ And even in reading the methods section
Fig. 1 Depiction of independent variables that may be associated with physical symptom burden among homeless adults assessed by Patanwala
et al.3 optimized for graphic clarity and the convenience of the editorial writer. Arrows depicting potential bidirectional, direct, and indirect
associations among these variables are not shown.
for both papers, my inner epidemiologist started to sweat.
What should be seen as Bcause,^ I asked myself, and what is
really Beffect.^ In preparing this comment, I drafted a graphic
to help readers distinguish between dependent variable
(symptoms, at left) and hypothesized independent predictors. The
same exercise could have been done for the study by
Charkhchi et al. My graphic plan was to put the dependent
variable at far left, with Bproximal^ and Bdistal^ causes
splayed out from left to right, respectively. I was forced to
give up as competing theories ricocheted and collided in my
mind. For example:
Does substance use lead to job loss, which in turn
exacerbates health conditions and symptoms? Or, do pain symptoms,
intensified and rendered chronic by childhood abuse,
somehow lead to job loss and homelessness, which in turn impede
physical recovery? Or might homelessness induce substance
use as a form of self-treatment, resulting in alienation from
friends, loneliness, and an intensification of physical
symptoms? No single interpretation seemed more credible than the
Ultimately, I decided that the only way to portray this web
of suffering was to make sure the arrows from one variable did
not graphically overlap the text referencing another. As a
result, smaller text units like Bregret^ would necessarily be
positioned closer on the page to the dependent variable, so that
larger text blocks could get a clean shot at the same dependent
variable, from further away on the page. So much for theory.
Suffering begets suffering and this editorial is far from the first
The frustration of the inner epidemiologist does not reflect a
problem in either of these two papers. Rather, such frustration
reflects an inevitable shortfall in how statistical models can
account for the challenges afflicting populations we clinicians
and policymakers wish to serve.
At this moment, a range of specific causes of human misery
have captured some policymakers’ attention as prime targets
for fixing: lack of housing, loneliness, opioid prescriptions,
and many more.
Each such cause deserves a rigorous scholarship,
coupled to a practical policy response. But any response
built on a single-target solution is likely to fall short or
perhaps, prove counterproductive. A moral response to
human suffering amidst unaffordable rental markets calls
for expanded housing assistance,6 but that assistance
cannot assure a return to health.7 Food insecurity
hinders engagement in care,8 but so too does a failure to
address unremitting symptoms, like pain.9 We who work
in primary care are taught to think about whole people,
but the imprecations of our managers and regulators
undermine that, as they search for easy solutions to
Our future in general internal medicine, most certainly in
primary care, stands on the question of whether we will
confront the complexity of each human being’s suffering
with full appreciation for the physical, social, emotional,
and existential factors that differentiate one human
situation from another and which, paradoxically, bind us
Corresponding Author: Stefan G. Kertesz, MD, MSc; University of
Alabama at Birmingham School of Medicine, Birmingham, AL 35205,
USA (e-mail: ).
Compliance with ethical standards:
Conflict of Interest: The author declares that he has no conflict of
1. Charkhchi P , Dehkordy SF , Carlos RC . Housing and food insecurity, care access and health Status among the chronically ill: an analysis of the Behavioral Risk Factor Surveillance System . J Gen Intern Med . 2018 .
DOI:https://doi.org/10.1007/s11606-017-4255-z Koo D , O'Carroll PW , Harris A , DeSalvo KB . An environmental scan of Recent initiatives incorporating social determinants in public health . Prev Chronic Dis . 2016 ; 13 : E86 .
Physical , psychological, social, and existential symptoms in older homeless-experienced adults: an observational study of the HOPE HOME cohort . J Gen Intern Med . 2018 . DOI: https://doi.org/10.1007/s11606- 017-4229-1.
Noor M , Shnabel N , Halabi S , Nadler A . When suffering begets suffering: the psychology of competitive victimhood between adversarial groups in violent conflicts . Pers Soc Psychol Rev . 2012 ; 16 ( 4 ): 351 - 74 .
N Engl J Med . 2018 ; 378 ( 3 ): 211 - 3 .
Kertesz SG , Baggett TP , O'Connell JJ , Buck DS , Kushel MB . Permanent supportive housing for homeless people-reframing the debate . N Engl J Med . 2016 ; 375 ( 22 ): 2115 - 7 .
Kushel MB , Gupta R , Gee L , Haas JS . Housing instability and food insecurity as barriers to health care among low-income Americans . J Gen Intern Med . 2006 ; 21 ( 1 ): 71 - 7 .
Upshur CC , Bacigalupe G , Luckmann R. BThey don't want anything to do with you^: patient views of primary care management of chronic pain . Pain Med . 2010 ; 11 ( 12 ): 1791 - 8 .