Post-mortem urine dipstick analysis for hyperglycemia and ketoacidosis: observer agreement and diagnostic value.

Abstract

The post-mortem diagnosis of hyperglycemia and/or ketoacidosis is challenging and usually requires costly ancillary testing of vitreous humor or serum samples. A screening tool that would help to determine whether ancillary testing is needed is therefore desirable. The aim of this study was to add to the literature testing the validity and diagnostic utility of post-mortem dipstick urinalysis. More specifically, we determined inter-observer agreement of visual dipstick assessment, the correlation between glucose and ketone urine dipstick scores and formal laboratory testing results, and the diagnostic value of specific dipstick scores expressed with likelihood ratios. Results demonstrate almost perfect interobserver agreement for 108 glucose dipstick scores (Fleiss' kappa 0.914) but only moderate interobserver agreement for 96 ketones dipstick scores (Fleiss' kappa 0.467). Dipstick glucose scores correlated strongly with vitreous humor glucose levels (Spearman's rank correlation coefficient of 0.841, n = 107). Correlation between ketone dipstick scores and serum levels of beta-hydroxybutyrate (BHB) and blood acetone was also positive but much weaker (0.317, n = 91; and 0.411, n = 92, respectively). The diagnostic value of specific dipstick scores was determined by calculating likelihood ratios for substantial hyperglycemia (vitreous humor glucose > 10 mmol/L), substantial ketoacidosis (serum BHB > 2.50 mmol/L) and elevated blood acetone (> 20 mg/L). Our results suggest substantial screening potential of dipstick urinalysis for glucose, especially when scores are at the lower and higher end of the spectrum. Overall, dipstick analysis results for ketones must be interpreted with great caution. A sub analysis of the data showed that a serum BHB above 2.50 mmol/L was only seen in 1.8% of cases without demonstrable acetone (> 20 mg/L).