Assessing ICD-10 Coding Accuracy: The Case of Chronic Subdural Haematoma in Hospital Records

The International Classification of Diseases 10th Edition (ICD-10) is a cornerstone in the healthcare system, acting as a universal language for reporting diseases and health conditions. While it is globally recognized, its accuracy in identifying specific diseases is an area of ongoing scrutiny. A recent study by Yordanov et al. (2023) investigated the accuracy of the ICD-10 framework in the identification of patients with Chronic Subdural Haematoma (CSDH) from hospital records. This paper seeks to unpack the findings of that study and discuss their implications for clinical coding in healthcare institutions.

The Underlying Challenge with CSDH and ICD-10 codes

CSDH is a prevalent neurosurgical pathology, with an escalating incidence. This condition is notorious for its high perioperative morbidity and approximately 10% mortality at one year. Case identification and ascertainment are crucial for developing, implementing, and evaluating care frameworks. However, as Yordanov et al. (2023) found, diagnostic ICD-10 codes may not always be reliable for this task.

Yordanov’s Method and Results

Yordanov et al.’s (2023) study focused on patients coded with S06.5 (traumatic subdural haematoma) or I62.0 (non-traumatic subdural haematoma) between 2014 and 2018. The analysis revealed that of 1861 patients, 10.2% had a non-traumatic SDH diagnosis (I620), and 89.8% had traumatic subdural haematomas (S065).

Importantly, certain demographic factors improved the identification of CSDH. Male sex, older age, and shorter length of hospital stay emerged as significant predictors of CSDH diagnosis. The best model, using these factors, achieved an impressive sensitivity of 88.4%, specificity of 84.5%, and a positive predictive value (PPV) of 87.9%.

Implications for Clinical Coding

This study illuminates the inherent complexity of clinical coding using the ICD-10 framework. On the one hand, the ICD-10 codes universal usage ensures a degree of consistency across healthcare institutions. However, as demonstrated in Yordanov et al.’s (2023) study, the ICD codes alone might not be adequate for accurate patient identification with specific conditions.

The results advocate for a more comprehensive approach to clinical coding, considering additional demographic data. While such an approach could potentially exclude a significant number of cases, the improved accuracy may outweigh this drawback, particularly for diseases with a high incidence and significant clinical impact such as CSDH.

Clinical coders, healthcare providers, and policymakers should consider these findings when designing and implementing clinical coding strategies. Further research into the impact of integrating demographic data into clinical coding frameworks could also be beneficial.

Looking Forward

With an increasingly digitised healthcare landscape, optimising clinical coding strategies will become ever more crucial. As this study by Yordanov et al. (2023) has shown, additional factors such as demographic data can significantly enhance the accuracy of patient identification using ICD-10 codes.

Ultimately, a balanced approach is needed. One that continues to harness the power of universal classification systems like the ICD-10, while also acknowledging their limitations and seeking ways to supplement them where necessary. This is not merely a question of efficient record-keeping. At its core, it’s about improving patient care – ensuring that each individual receives the most appropriate treatment for their specific needs.


Yordanov, S., Khan, S., Stubbs, D., Davies, B., Santarius, T., Hutchinson, P., & Joannides, A. (2023). Assessing the accuracy of the International Classification of Disease (ICD) framework in the identification of patients with chronic subdural haematoma from hospital records. Surge, DOI:

Related Articles


This site uses Akismet to reduce spam. Learn how your comment data is processed.