The purpose of this study is to analyze variation in outpatient disease coding for the same patients by different healthcare institutions and the factors affecting disagreement in disease coding. For this analysis, we analyze the Health Insurance Review and Assessment Service NPS data set in 2014 that contains 9,976,826 patient data records.
The main results of this study are as follows: first, as for the rate of disagreement in disease coding by the patients’ characteristics, we find that it is generally higher for males and varied by age group and insurance type. As for the rate of disagreement by the characteristics of healthcare institutions, institutions with fewer beds generally show a higher rate of disagreement. Also, public corporations and health centers show a higher rate of disagreement, though there is a variation by department. The rate is generally high in regions other than Seoul, Gyeonggi Province, and Metropolitan Cities.
Second, as for the rate of disagreement by patients’ course of movement between healthcare institutions, the rate is higher with the movement from a tertiary hospital to a health center. In case of movement to the same department, our two-unit comparison shows that the departments of orthopedics, pediatrics, and family medicine have a higher rate of disagreement. In case of movement to a different department, the two-unit comparison show that the departments of general medicine, medicine, anesthesiology and pain management, pediatrics, ophthalmology, otorhinolaryngology, and dental care have higher rates of disagreement. For high frequency of disagreement, we analyze up to three-unit agreement pairs of disease codes and find that the frequent diagnoses include benign hypertension, acute bronchitis, allergic rhinitis, acute gastritis, allergic urticaria, bilateral primary gonarthrosis, back pain accompanied by sciatica, and frozen shoulders. In addition, subsequent disease coding varied in distribution by preceding disease coding. As for the variation in disagreement among healthcare institutions, tertiary and general hospitals show higher CV in two- and three-unit agreement pairs. Clinics, Oriental medical centers, and health centers show higher SCV on a consistent basis while tertiary and general hospitals show relatively lower SCV. Furthermore, we find disagreement varies among healthcare institutions by primary disease code type.
Third, we perform the logistic regression analysis to find factors affecting the disagreement rate by disease coding unit for one-unit agreement pairs. We find that female patients and older patients, and smaller health centers with a fewer beds are more likely to have disagreements in coding. The departments of medicine, otorhinolaryngology, pneumonia, and Oriental medicine are more likely to encounter disagreement in coding. For two-unit agreement pairs, we find no clear variation by gender or age; smaller-sized health centers are more likely to encounter disagreement in coding. No significant variation is found in the disagreement rate by the number of beds or healthcare institution establishment. The departments of neurosurgery, pediatry, otorhinolaryngology, and dental care are more likely to show disagreements in coding. The analysis of three-unit agreement pairs shows that those who are female and who are older are more likely to encounter disagreement in coding but no significant variation is found in the disagreement rate by hospital size. Also, no significant variation is found in the disagreement rate by the number of beds or healthcare institution establishment type. The departments of psychiatry, industrial medicine, and preventive medicine are more likely to encounter disagreement in coding.
It is necessary to conduct further research on the factors found to affect the rate of disagreement in disease coding among healthcare institutions and its variation in this study and need more detailed review in South Korean medical practice. Efforts should be made to decrease the rate of disagreement in disease coding among healthcare institutions and its variation continuously for the same patient through the systematic efforts to make disease coding more accurate in healthcare institutions.