Skip to main content
eScholarship
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

Social Influences on Healthcare Outcomes in a Major Academic Medical Center

Abstract

Social scientists note that patients’ interactions with others can trigger changes in emotional state, behavior, attitudes, or opinions, and that these changes can have clinically significant outcomes.

Hospital room assignment, a largely random process, may create social relationships that influence individual social states. The nature of these social relationships, in turn, can impact clinical outcomes as social interactions, the placement, presence, or circumstance of social relations, and perhaps even the mere presence of another individual are key components of the care process. As such, it stands to reason that peer and social support effects may influence measurable clinical outcomes among hospitalized patients.

I test this thesis using a large, novel, retrospective dataset extracted from the clinical record of a major academic medical center, exploring the impact of chance socially and clinically concordant matches on a patient’s length of stay in the hospital, risk of readmission and mortality within thirty days of hospital discharge, and risk of mortality during hospitalization, employing a variant of the standard lagged linear-in-means peer influence model proposed by Manski (1993), where markers for clinical outcomes of the focus patient at time of discharge from the hospital are regressed on the concordance of peer group-average of markers for sociodemographic factors and clinical condition of the roommates with related factors and conditions of the focus patient, controlling for nosocomial conditions and relevant characteristics of the focus patient.

The analyses proceed in three stages. The first stage demonstrates the existence of a homophilic effect on outcomes of concordant pairings along sociodemographic and clinical lines occurring by chance in a population of hospitalized patients. The second stage demonstrates the existence of a social contagion or spillover effect of positive or negative clinical status on the health outcomes of another. The final stage of analysis examines homophily and contagion jointly, assessing the impact of concordance between roommates on the transference of benefit or threat between hospitalized patients.

I find that evidence exists for both peer effects on clinical outcomes and social concordance, and while both significant, social concordance may be more a more important determinant of clinical outcomes than peer effects. These findings lend support to the hypotheses that social concordance (or decreased social distance) between hospitalized roommates may be clinically beneficial, and that these benefits – manifested in shorter hospitalizations – may be mediated by information sharing, psychological comfort due to affiliation with a similar other, or by sharing clinically similar experiences.

These findings suggest that social influence on clinical outcomes exists, and that social and structural factors of the care process may impact patient outcomes in a statistically significant and clinically meaningful way.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View