A comparison of international normalised ratio (INR) measurement in hospital and general practice settings: evidence for lack of standardisation
Previous reports of discrepancies in international normalised ratio (INR) measurement between centres have focused on hospital based methodologies.1–3 Previously, we have demonstrated differences in derived INR values for the same sample tested in primary care and in one of three different haematology laboratories.4 Our present study is an extension of the previous one, investigating comparative results based on contemporaneous samples measured in one primary care centre and in two hospital laboratories using a variety of techniques.
Venous blood was drawn from patients in one primary care centre over a three month period. The sample was tested on site for INR estimation using the Thrombotrak NPT and Thrombotest reagent. The remainder of the venous sample was placed in a citrated collection bottle and sent to two reference laboratories routinely used by the general practitioner to measure INR values (laboratories 1 and 2). Laboratory 1 determined INR values using three separate methods: a Thrombotrak using Thrombotest, an ACL machine using IL reagent, and a KC-10 machine using Manchester reagent. Laboratory 2 determined INR values using a KC-10 machine with Manchester reagent. Because laboratory 1 acts as the regional reference laboratory and uses the ACL/IL combination as its routine method of INR testing, the result obtained was taken as the gold standard. Samples were sent to the laboratory using routine transport with no samples tested more than 12 hours after venesection.
Fifty four separate venous samples from 26 patients were sent from the practice to the laboratories. The INR values obtained ranged from 1.0 to 6.1. Table 1 shows the mean difference in results from the various machines. There was a significant mean difference in the practice Thrombotrak results relative to the ACL and the KC-10 in both laboratories, but none between practice measurements and those obtained in laboratory 1 using the same technology. There were also significant differences between all hospital systems. Furthermore, there was a significant difference between ACL and KC-10 results from the same laboratory and between KC10 results from different laboratories.
Our results suggest that regular differences occur in INR measurements obtained on the same samples using different methodologies and draws attention to inherent problems associated with INR measurement in different settings. The clinical implication of these findings is that patients could receive different doses of warfarin depending upon which centre monitors their INR. Nevertheless, the best agreement to be found was between the practice derived INR and the laboratory derived INR using the same technology. This shows that primary care INR estimations are as reliable as laboratory estimations using the same combination of reagents and technology. Therefore, it follows that as long as continuity of INR estimation by location and method is maintained for individual patients, the rate of unnecessary warfarin dose adjustments will be reduced.
Mean difference (SEM) in INR between methods (n = 54)
