Documenting Diagnosis: Exploring the Impact of Electronic Health Records on Diagnostic Safety

Accurate and robust documentation affects the standard of care. Better access to clinical information provides important insights for the care team and important clues for diagnosis and treatment. Enhanced documentation enables well-organized availability of meaningful, accurate, and complete health records. Such records can improve the quality of care delivered, improve coordination and communication across care teams and with patients, and support the execution of integrated decision support.^31-33 However, these benefits are only achieved with documentation integrity, which requires an accurate and complete health record. 

EHR usability has been directly associated with diagnostic error issues^34,35 and has not supported the development of higher levels of situational awareness.³⁶ Challenges and opportunities arise in optimizing complex user interfaces, improving inefficient workflows, and optimizing interoperability by applying human factors and design principles, CDS, and personalization or customization (as appropriate). 

Systems designed to use medical terminology or international medical coding systems rather than free text can prevent inaccurate information.³⁷ Potential added functionality to assist with documentation includes templates, standard phrases and paragraphs, and automated object insertion to improve efficiency of data capture, timeliness, legibility, consistency, and completeness.³⁸ In addition, giving patients access to clinical notes has shown various advantages, such as heightened control over their health condition, increased involvement, better medication adherence, and heightened accountability among clinicians.^39,40

Diagnostic uncertainty, a concept that has yet to be adequately operationalized in medical practice, is a natural part of medicine and more common in primary care than any other specialty.^41-43 Providers may encounter diagnostic uncertainty, where the patient’s symptoms or clinical presentation do not clearly indicate a specific diagnosis. Notably, no diagnostic code exists for “I don’t know.” In such situations, providers may need to rely on provisional or working diagnoses, which can make it challenging to assign an accurate International Classification of Diseases, 10th Revision (ICD-10) code. 

ICD-10 codes are highly detailed and specific, often requiring providers to choose from a vast array of codes that correspond to different diagnoses, conditions, and symptoms. Without a definitive diagnosis, providers may struggle to accurately select the most appropriate code from the extensive list of options. 

A review of EHR documentation evaluated two signs of diagnostic uncertainty, identifying diagnostic uncertainty with moderate reliability. The first was the use of direct expression (e.g., use of question marks, differential diagnoses, and vocabulary such as “probably, maybe, likely”). The second was indirect inference (e.g., absence of documented diagnosis at the end of a visit, ordering of multiple consultations or diagnostic tests).⁴⁴ 

Approximately 80 percent of data within the EHR is unstructured text, including visual data (e.g., endoscopy, laparoscopy), biosensor data from monitors and devices, audio data, and clinical notes (e.g., progress notes, discharge summaries, diagnostic test reports).^45,46 Although much of the data in a patient’s EHR is coded, essential details about the patient’s care and management are often hidden in unstructured clinical notes. This practice makes it challenging and time consuming for physicians to review during their typical clinical workflow. 

The use of free text within clinical notes is integral to medical documentation as it enables clinicians to capture a comprehensive perspective of an individual, extending beyond structured data entry. Within clinical and progress notes, clinicians articulate their current evaluation, including their reasoning, and outline future steps in diagnosis or treatment. EHR-integrated interventions can target key diagnostic processes, including but not limited to:

• Dashboards to identify at-risk patients.⁴⁷ 
• Diagnostic timeouts for clinicians to reassess the working diagnosis.⁴⁷ 
• Patient-facing questionnaires to gather patient concerns.⁴⁷ 
• Initiatives that allow patients to review diagnoses and problems documented in the EHR for accuracy.^48,49
• More robust mechanisms for followup for tracking diagnostic information and communication.⁵⁰  
• Innovative ways for the healthcare team to communicate and collaborate on not only the initial encounter but also results of diagnostic tests and referrals.⁵¹ 

The goal of these initiatives is to transform the EHR from a billing and communication tool for clinicians to a central form of communication among clinicians, patients, and care partners. 

The concept of documentation integrity includes not just the content and information included but also information governance, authorship validation, amendments, and record corrections. Preserving documentation integrity is critical to maintain the highest levels of care and patient safety, reduce fraud and abuse, and reduce the risk of a malpractice lawsuit.^52,53 Design features such as template-driven drop-down boxes or lists provide rigid structures that support standardization that may prevent clinicians from telling a patient’s complete story. 

Research has found that clinicians experience incredible rates of stress and burnout as a result of the cognitive load required for adequate EHR documentation and record keeping.⁵⁴ Furthermore, because hospitals are reimbursed based on diagnosis-related groups, they face financial pressures within coding practices to maximize reimbursement or perceived performance.^55-57

Clinicians have adapted to navigating the requirements for adequate documentation to secure reimbursement. Physicians often resort to copying and pasting previous notes, making minor modifications, which may inadvertently contribute to the proliferation of unnecessary and irrelevant data. Future regulations must support clinicians in creating high-quality documentation while recognizing systematic “defensive medicine” and “return on investment” challenges. Existing quality and reimbursement programs must adjust their data collection and quality measurement practices to ensure that reported data and reimbursement accurately represent the patient population under treatment, rather than solely reflecting the completeness of coded data.