INTERINSTITUTIONAL COMPARISON OF BEDSIDE BLOOD-GLUCOSE MONITORING PROGRAM CHARACTERISTICS, ACCURACY PERFORMANCE, AND QUALITY-CONTROL DOCUMENTATION - A COLLEGE-OF-AMERICAN-PATHOLOGISTS Q-PROBES STUDY OF BEDSIDEBLOOD-GLUCOSE MONITORING PERFORMED IN 226 SMALL HOSPITALS

Objectives.-To assess the accuracy of bedside blood glucose monitoring (BCM) in small hospitals, to assess the compliance with which hospita l workers performing bedside BCM adhere to quality control (QC) proced ures, and to identify those practice characteristics in small hospital s that are associated with better BCM accuracy and with better perform ance of BCM QC. Design.-Over a 1-month period in 1996, voluntary parti cipants in the College of American Pathologists Q-Probes laboratory qu ality improvement program prospectively compared glucose results of 30 split samples run on BCM instruments with those performed on laborato ry glucose analyzers, collected quality control data on up to five inp atient BCM instruments, and completed questionnaires profiling BCM pra ctice characteristics in their institutions. Setting and Participants. -Two hundred twenty-six hospitals with 200 or fewer occupied beds. Mai n Outcome Measures.-The percentages of glucose determinations performe d on BGM instruments differing by more than 10%, 15%, and 20% from tho se split-sample results performed on laboratory glucose analyzers; the percent of BGM QC determinations required by institutions' BGM QC pro grams that BCM operators actually performed; and the percent of patien t values reported when BCM QC was documented to be out of range and un corrected, or reported when BCM QC was not performed at all. Results.- Of 6095 split-specimen glucose results that participants simultaneousl y performed on BGM instruments and on laboratory glucose analyzers, 45 .6% differed from each other by more than 10%, approximately 25% diffe red from each other by more than 15%, and almost 14% differed from eac h other by more than 20%. Of 216 laboratories that performed at least 30 QC events during the study period, sightly over a third completed 1 00% of their required QC determinations, and 10% completed, at most, 7 7% of their required BCM QC determinations. Of 115973 BCM determinatio ns that participants reported on hospitalized patients, 3.3% were repo rted when QC was either out of range or when there was no documentatio n that QC had been performed at all. Better accuracy and/ or better QC performance was associated with laboratory personnel rather than nurs ing personnel both supervising institutions' BGM QC programs and runni ng institutions' daily routine BCM QC; with BCM operators both routine ly running three, rather than two, levels of QC analytes; with BCM ope rators regularly comparing BGM results with laboratory analyzer glucos e results; and with institutions participating in external proficiency programs. Institutions that completed all required BGM QC tasks tende d to perform better on the BGM accuracy study than did those instituti ons that completed, at most, 77% of their required Qc. Conclusions.-We found the rates of BCM accuracy and of QC performance adequacy achiev ed in small hospitals to be similar to those determined in previous Q- Probes studies conducted in large institutions. A significant amount o f institutional bedside testing does not meet current standards for ac curacy or for quality control. Some institutions may improve their acc uracy and/or QC performances by having laboratory personnel intimately involved in their institution's BGM QC program, by routinely comparin g BCM results with those performed using glucose analyzers in the clin ical laboratory, by routinely running three rather than two glucose QC control levels, by participating in external proficiency programs, an d by strictly adhering to institutional QC protocols.