Risk factors for patients with social determinants of health not to follow up with community-based organizations to which they have been referred

A retrospective study of 342 patients at Long Island Jewish Medical Center found that while certain demographic and social factors appeared to correlate with follow-up rates after referrals to community-based organizations, none reached statistical significance, suggesting the need for larger studies to better understand and address social determinants of health in improving patient engagement.

The Gist

Imagine the Emergency Department (ED) as a busy train station. Every day, thousands of people arrive there, often in distress, needing help to get back on track. For a long time, the station only offered medical tickets (medicine and treatment). But doctors realized that many passengers were missing something else crucial: a place to sleep, money for food, or a job. These are the Social Determinants of Health (SDOH)—the real-life conditions that make us sick or keep us well.

To help, the hospital at Long Island Jewish (LIJ) Medical Center set up a "Concierge Desk" (called the EDDC). When a patient arrived, staff would ask, "Do you need help with housing, food, or transportation?" If the answer was yes, the staff would hand them a special map to a Community-Based Organization (CBO)—a local group that could actually provide those resources.

The Problem: The Map vs. The Journey
The study asked a simple question: Once the hospital gave the patient the map, did they actually use it to find the help they needed?

Think of it like this: The hospital is a generous chef who gives a hungry person a gift card to a grocery store. The gift card is the referral. But if the person doesn't have a car, doesn't know the way, or is too overwhelmed to go, they might never use the gift card. The study wanted to find out who was most likely to leave the gift card on the counter and walk away.

The Study: A Small Snapshot
The researchers looked at a specific group of 342 passengers (patients) who were given these "maps" between February and July 2023. They tried to see if certain characteristics—like your race, the language you speak, your age, or whether you have health insurance—made you more or less likely to use the map.

The Results: Trends, but No Smoking Gun
Here is the twist: The study found some interesting patterns, but none of them were statistically "proven" to be the main reason people didn't follow up. It's like looking at a small group of people and noticing that more women than men used the gift card, but the group was too small to say for sure if that's a rule for everyone.

However, the numbers showed some surprising trends:

• The "Minority" Advantage: Surprisingly, non-White patients and those who spoke languages other than English were more likely to use the maps than White patients or English speakers. It's as if the people who usually face the biggest barriers were actually the ones showing up to the grocery store more often in this specific group.
• The Older & Insured: Older adults (over 64) and people with insurance were also more likely to follow through.
• The "No Difference" Factors: Whether you were employed or unemployed, or male or female, didn't seem to make a huge difference in this small group.

Why Didn't the Results "Stick"?
The researchers admit the group they studied was too small to draw a hard line. Imagine trying to predict the weather by looking at the sky for only five minutes; you might see a cloud, but you can't be sure it's going to rain. Because the sample size was small, the math couldn't confirm that these trends were real rules rather than just a coincidence.

The Big Picture: Why This Matters
Even though the math wasn't perfect, the study highlights a massive issue: The system is broken in the middle.

1. The Gap: Hospitals are good at spotting problems (screening for hunger or homelessness), but they are bad at making sure the solution actually happens.
2. The Cost: When patients don't get help with their social needs (like housing or food), they get sicker. They end up back in the Emergency Department, which is expensive for everyone. It's like paying for a band-aid every day instead of fixing the cut.
3. The Future: The authors suggest we need bigger studies to understand the real reasons people drop off the map. They also suggest using technology (like AI) to better match patients with the right help and to follow up with them personally, rather than just handing them a piece of paper and hoping for the best.

In a Nutshell
This paper is a report card for a hospital's attempt to connect sick people with community help. It says, "We are trying to give people the tools they need, but many are still dropping them." While we couldn't pinpoint exactly why some people drop the tools and others don't (because our group was too small), we know that fixing this connection is the key to saving money and, more importantly, saving lives.

Technical Summary

1. Problem Statement

Social Determinants of Health (SDOH)—including factors like housing instability, food insecurity, and lack of transportation—significantly impact health outcomes and drive up healthcare costs by increasing reliance on Emergency Departments (EDs). While healthcare systems increasingly screen for SDOH and refer patients to Community-Based Organizations (CBOs) for support, a critical gap exists: low follow-up rates.

Specifically, the study focuses on the Emergency Department Discharge Center (EDDC) at Long Island Jewish (LIJ) Medical Center. Despite implementing a screening and referral system using the Unite Us® platform, internal reviews indicated a dismal 4.5% follow-up rate among referred patients. The problem is identifying which patient-level risk factors (demographic and social) correlate with a failure to engage with these vital community resources, thereby hindering the effectiveness of SDOH interventions.

2. Methodology

• Study Design: Retrospective cross-sectional analysis.
• Setting: Emergency Department of Long Island Jewish (LIJ) Medical Center, a 583-bed hospital serving a diverse urban/suburban population in Northwell Health's service area.
• Data Source: A database of patients screened for SDOH between February and July 2023.
• Sample Size: 342 patients who were screened and referred to CBOs.
• Data Collection:
  • Screening was conducted via paper-based surveys.
  • Referrals were processed through Unite Us® (formerly NowPow®).
  • Variables captured included: Race, Language, Age, Gender, Employment Status, Insurance Status, and Housing Status.
  • Outcome Variable: Whether the patient followed up with a referred CBO within two weeks.
• Statistical Analysis:
  • Descriptive statistics were used to calculate follow-up percentages.
  • Chi-squared tests (with $N-1$ degrees of freedom) were employed to determine associations between demographic/social factors and follow-up status.
  • Significance threshold was set at $p < 0.05$.
• Ethical Considerations: The study was deemed exempt by the Northwell Health IRB as a quality improvement project; all data were anonymized.

3. Key Contributions

• Identification of Non-Significant Trends: The study provides empirical data on follow-up behaviors in a real-world ED setting, highlighting that while certain demographic trends exist, they may not be statistically significant in smaller cohorts.
• Contextualizing Low Follow-Up Rates: It reinforces the challenge of translating SDOH screening into actual patient engagement, noting that even with established referral platforms (Unite Us®), the conversion rate remains low.
• Highlighting the "Sample Size" Limitation: The paper contributes to the literature by demonstrating how limited sample sizes in quality improvement projects can mask statistically significant relationships that likely exist in broader populations.

4. Results

Out of the 342 patients screened, follow-up rates were analyzed across several variables. Crucially, none of the variables reached statistical significance ($p > 0.05$). However, notable descriptive trends were observed:

Variable	Group	Follow-up Rate	Trend Observation
Race	Non-White	73.2% (41/56)	Higher than White patients (55.6%)
	White	55.6% (5/9)
Language	Non-English	81.8% (9/11)	Higher than English speakers (68.5%)
	English	68.5% (37/54)
Age	> 64 years	80.0% (8/10)	Higher than younger patients (69.1%)
	≤ 64 years	69.1% (38/55)
Gender	Female	76.5% (26/34)	Higher than males (64.5%)
	Male	64.5% (20/31)
Employment	Employed	71.4% (10/14)	Similar to unemployed (70.6%)
	Unemployed	70.6% (36/51)
Insurance	Insured	77.8% (21/27)	Higher than uninsured (65.8%)
	Uninsured	65.8% (25/38)

• Key Finding: Contrary to some literature suggesting language barriers or lack of insurance are primary deterrents, Non-English speakers and Insured patients in this specific cohort showed higher follow-up rates, though the sample sizes for some subgroups (e.g., Non-English speakers, $n=11$) were very small.

5. Significance and Discussion

• Statistical Power vs. Clinical Reality: The lack of statistical significance is attributed primarily to the limited sample size (342 patients screened, with even smaller numbers in specific subgroups). The authors argue that the observed trends (e.g., higher engagement by older adults and non-White patients) align with existing literature but require larger datasets to be confirmed.
• Complexity of SDOH: The results suggest that follow-up behavior is influenced by a complex interplay of factors (cultural, systemic, logistical) that cannot be fully captured by simple demographic binaries in small samples.
• Systemic Barriers: The low overall follow-up rates point to systemic issues beyond patient demographics, such as:
  • Inconsistent screening processes.
  • Gaps in referral communication.
  • Lack of patient motivation or support post-referral.
• Future Directions: The authors recommend:
  • Larger, longitudinal studies to increase statistical power.
  • Refining referral systems to include better tracking and feedback loops.
  • Investigating additional variables such as transportation access, mental health status, and specific housing instability metrics.
  • Developing tailored interventions for underserved populations to bridge the gap between ED discharge and community resource utilization.

Conclusion: While this study did not identify statistically significant risk factors due to sample constraints, it highlights the urgent need for more robust data collection to understand why patients fail to engage with SDOH resources. Improving these follow-up rates is essential for reducing health disparities, lowering ED utilization, and creating a more equitable healthcare system.