For clinicians, trainees, and health system leaders, the referral process is supposed to be a bridge connecting patients to the specialty care they need. But in everyday practice, this bridge often functions more like a maze. Many of us, whether as physicians in training or practicing providers, have watched patients become lost in a cycle of delays, miscommunications, and inappropriate specialist visits that worsen outcomes rather than improve them.

The consequences of a broken system

In my own experience as a medical student, I have seen the consequences firsthand. One patient with persistent numbness was shuttled between a neurologist and an orthopedic surgeon for months without clarity or progress. Other patients with chronic, poorly defined symptoms were sent to cardiologists, nephrologists, nutritionists, or physical therapists, not because their conditions clearly required these specialists, but because the pathway to resolving their complaint was unclear.

Unsurprisingly, many never completed those referrals in the first place, deterred by long wait times, insurance hurdles, transportation barriers, or difficulty contacting the specialist. Studies have shown that these factors, such as incomplete and incorrect referrals, as well as long wait times and unclear referral processes lead to substantial gaps in care, with one study showing up to half of specialty referrals not being followed through by older adults, and another where patients cited the difficulty of the referral process as the main reason why they did not follow through with them.

Even when referrals were completed, they were often inappropriate. For example, I often witnessed scenarios where the provider’s specialty was not one that could address the patient’s health concern. In those situations, neither could the patient explain why they had been referred there, nor was the provider informed why the patient was referred to them. One study found that over one-third of referrals do not match the provider’s expertise or scope of practice. This dysfunction drains time, contributes to burnout on both sides of the exam table, and delays diagnoses that could meaningfully alter a patient’s trajectory. Every inappropriate referral is a missed opportunity not only for timely care but for efficient use of a system already stretched thin.

Why the process fails

These failures are not random. One issue lies in systemic educational and structural shortcomings that shape how referrals are generated. Residents and early-career physicians were shown to consistently report inadequate training on referral decision-making, such as what to rule out beforehand, how to select the correct specialty, and how much information a specialist needs.

A second, equally critical issue is the quality of the referrals themselves. Multiple studies demonstrate that referral notes frequently lack essential clinical details, contain incomplete histories, or misrepresent the question being asked. Referring clinicians often believe they have provided sufficient information; specialists often report the opposite. The result is a broken communication loop that leaves patients bouncing between clinics while their underlying condition remains unaddressed.

A technological solution

Given the scale of the problem, solutions must be multifaceted. Yet one practical intervention could make an immediate difference: embedding a concise, evidence-based referral support tool directly within the electronic health record (EHR). Physicians interact with the EHR constantly, making it a natural point for improving decision-making without relying on memory or extensive extra training.

Such a tool could take two forms:

• A simple referral checklist prompting primary care providers to confirm that key diagnostic steps or basic evaluations have been completed before referring a patient.
• An intelligent alert system that flags when a constellation of symptoms or review-of-systems findings strongly suggests one specialty over another.

In an era of rapidly advancing AI, these prompts could be built from existing evidence-based guidelines or dynamically updated through machine learning tools trained on referral outcomes.

To be clear, integrating such a system is not without challenges. Providers already experience significant alert fatigue, and additional clicks could slow an already demanding workflow. Implementation costs for EHR vendors are also nontrivial. However, if designed with brevity and precision, these tools can significantly reduce provider burden by preventing wasted visits, unnecessary testing, and months of diagnostic delay. The long-term savings in cost, time, and patient suffering will far outweigh the relatively shorter-term inconvenience of building better referral infrastructure.

The path forward

The referral problem is too substantial to ignore, but it is not insurmountable. Evidence-based referral decision support embedded within the EHR brings us closer to a health care system where patients do not fall through the cracks simply because the path between primary care and specialty care is unclear. The responsibility lies with health systems, policymakers, and EHR vendors such as Epic, Cerner, and others to act. As physicians in training and future leaders, we must demand a referral process worthy of the patients who depend on it.

Abhijay Mudigonda is a medical student.