Page 133 - AIH-1-2

Artificial Intelligence in Health SDoH in clinical narratives

Figure 6. Adjusted odds ratios for the probability of mentioning homelessness/housing based on clinical case type, journal specialty, and journal’s
geographic region. The figure was plotted with Matplotlib.

3.4.6. Spiritual beliefs 550/10,000 case reports. This rate has persisted until 2022,
indicating a lasting change in awareness or reporting about
Mentions of spiritual beliefs were strongly correlated with race/ethnicity. Nonetheless, further studies are warranted
authors from sub-Saharan Africa (AOR: 9.17; 95% CI: 2.84
– 29.64) and the Indian subcontinent (AOR: 4.09; 95% to investigate the reason for the observed trend.
CI: 1.83 – 9.15), journals in the field of psychiatry (AOR: Homelessness mentions displayed a slight increase, but
7.61; 95% CI: 2.93–19.79), publications from the Middle the rate was only 1.29/10,000 case reports, contrasting with
East (AOR: 5.05; 95% CI: 1.99 – 12.85), and clinical cases the estimated US 1-year homelessness prevalence – about
related to endocrine system diseases (AOR: 3.47; 95% CI: 100 times higher. 36
1.38 – 8.68) and mental disorders (AOR: 3.05; 95% CI: Collectively, the data revealed no consistent longitudinal
1.27 – 7.31). In contrast, journals in the field of surgery SDoH reporting trends. Observable shifts were sporadic,
(AOR: 0.23; 95% CI: 0.06 – 0.96) and clinical cases related brief, or tied to specific periods, highlighting the variability
to neoplasms (AOR: 0.20; 95% CI: 0.08 – 0.50) were of SDoH in the medical literature.
associated with lower probabilities of mentioning patients’
spiritual beliefs. Further information on spiritual belief is 4.2. Risk of biases in the social determinants of
included in Figure 7 and Table S6. health

4. Discussion Our findings reported that diagnosis significantly affects
SDoH mentions. Both individual cultural norms (reflected
4.1. Low prevalence of social determinants of health by the author’s origins) and institutional policies (indicated
mentions by the journal’s origins and specialties) impacted
Our analysis revealed an uneven distribution of SDoH SDoH mention frequency. Notably, individual regional
factors, such that three SDoH factors did not display contexts exhibited distinct patterns when contrasted with
a clear time-dependent trend. Regarding sexual institutional regional contexts represented by journals. In
orientation (Figure S3), a brief increase in mentions addition, a journal’s specialty influences SDoH mentions.
occurred in the 1980s, peaking at 40/10,000 case reports. Specifically, journals on psychiatry, general medicine,
However, the mentions of sexual orientation sharply and medical specialties tend to mention SDoH more than
decreased in the 2000s, leveling at 5/10,000 case reports. surgical specialty journals. These findings emphasized
We theorized that this surge was associated with the the need for a standardized approach to SDoH reporting
AIDS/human immunodeficiency virus (HIV) outbreak across varied geographies and specialties.
in that period. Notably, our data revealed potential biases in SDoH

There was little variation in race/ethnicity mentions with reporting in the medical literature. Certain SDoH
time (until 2011), depicting steadiness at approximately reports, such as sexual orientation with infectious
300/10,000 case reports (Figure S4). However, between diseases or homelessness with mental disorders, are
2011 and 2013, race/ethnicity mentions surged to nearly overemphasized, potentially reinforcing stereotypes or

Volume 1 Issue 2 (2024) 127 doi: 10.36922/aih.2737