Validity of diagnostic coding for undernutrition in hospitals

Background & aims We examined the accuracy of ICD-10 diagnostic coding for undernutrition in Danish Hospitals, including the use of Nutritional Risk Screening 2002 guidelines. Methods We investigated a random sample of hospitalized patients registered in the Danish National Registry of Patients with a discharge diagnosis of undernutrition between 2002 and 2011 in the North Denmark Region. Based on medical record review we estimated the positive predictive value (PPV) of the undernutrition diagnosis. Stratification was made by calendar period, hospital type (local vs. university), gender, age, speciality and type of diagnosis code. Subsequently, we evaluated the use of Nutritional Risk Screening 2002 as recommended by the European Society of Clinical Nutrition and Metabolism and the Danish National Board of Health. Results We could retrieve the medical records of 172/200 sampled patients with undernutrition (86%). Nineteen patients were classified as being definite (screening-confirmed) cases and another 103 patients as probable (clinically-confirmed) cases of undernutrition, yielding a PPV of 11.0% (95% confidence interval [CI]: 6.8–16.7) for definite undernutrition and 70.9% (95% CI: 63.5–77.6) for any confirmed undernutrition. Only 26.2% of patients coded with undernutrition had been screened according to the Nutritional Risk Screening 2002. Conclusions This population-based study found modest agreement between ICD-10 codes for undernutrition compared to a standard method (Nutritional Risk Screening 2002) as documented in medical doctors' records in Danish hospitals. Diagnoses of undernutrition contained in hospital discharge registries should be used with caution.