0
Clinical Investigations in Critical Care |

Agreement Between Bedside Blood and Plasma Glucose Measurement in the ICU Setting* FREE TO VIEW

Javier Daniel Finkielman, MD; Lance J. Oyen, PharmD; Bekele Afessa, MD, FCCP
Author and Funding Information

*From the Division of Pulmonary and Critical Care Medicine (Drs. Finkielman and Afessa), Department of Internal Medicine, and Hospital Pharmacy Services (Dr. Oyen), Mayo Clinic College of Medicine, Rochester, MN.

Correspondence to: Bekele Afessa, MD, FCCP, Mayo Clinic, 200 First St SW, Rochester, MN 55905; e-mail: afessa.bekele@mayo.edu



Chest. 2005;127(5):1749-1751. doi:10.1378/chest.127.5.1749
Text Size: A A A
Published online

Study objective: To assess the agreement between bedside glucose (bGlu) measurements and laboratory plasma glucose (pGlu) measurements in the ICU setting.

Design: Retrospective study.

Setting: ICU of a tertiary academic medical center.

Patients: One hundred ninety-seven patients admitted to the ICU from January through December 2002 who underwent 816 simultaneous bGlu and pGlu measurements.

Interventions: None.

Measurements and results: To calculate the agreement between the two methods of glucose measurement, the mean difference was obtained, and the limits of the agreement were calculated as the mean difference ± 2 SDs. On 767 occasions, the mean bGlu was 159 mg/dL and the mean pGlu was 151 mg/dL (p < 0.001). The mean difference between the two techniques was 7.9 mg/dL (SD, 17.6 mg/dL), and the limits of agreement were + 43.1 and –27.2. On 31 occasions, the bGlu was reported as > 400 mg/dL, and in these cases the mean pGlu was 423 mg/dL (range, 300 to 900 mg/dL). On 18 occasions, the bGlu was reported as < 50 mg/dL, and in these cases the mean pGlu was 66.9 mg/dL (range, 13 to 198 mg/dL).

Conclusions: On average, bGlu provides a reasonable estimate for pGlu. However, for the individual patient, bGlu gives an unreliable estimate for pGlu. All of those taking care of critically ill patients should be aware of the limitations of bedside glucometry.

Figures in this Article

Tight control of glucose levels has been shown to improve outcome in critically ill patients.1To achieve such control, frequent glucose measurements are required. Due to the quick response time, bedside glucometry is often used for glucose measurements. Bedside glucometry is an accepted method for estimating blood glucose levels among ambulatory and hospital ward patients.2 However, among critically ill patients, bedside glucometry has been evaluated in only a few studies25 with small sample sizes and in very special and controlled conditions that diverged from day-to-day critical care practice. In some of these studies, the glucose samples were obtained by dedicated staff,5 there was a special handling of the samples,25 and the measurements were performed by the same investigator5 in a dedicated machine.3 The objective of this study was to assess the agreement between bedside and laboratory plasma glucose measurements in the ICU setting under routine clinical conditions.

In this retrospective study, the electronic records of all 1,192 consecutive admissions to the ICU of Rochester Methodist Hospital, Mayo Clinic, from January through December 2002 were identified from the acute physiology and chronic health evaluation (APACHE) III database. Patients were included in this study if simultaneous bedside glucose (bGlu) and plasma glucose (pGlu) measurements were available. If the patient had an arterial or central vein catheter, both measurements were performed from the same blood sample that was obtained from the catheter. When no catheter was available, two concurrent samples were obtained from a peripheral vein and a fingerstick for the measurements of pGlu and bGlu, respectively. The source of the samples for glucose measurements was not recorded. bGlu levels were measured at the bedside (SureStepFlexx; LifeScan, Inc; New Brunswick, NJ), and pGlu levels were measured either at the hospital clinical laboratory (Glucose Analyzer 2; Beckman Coulter; Fullerton, CA) or at the Central Clinical Laboratory (Hitachi 747–200 Analyzer; Roche; Indianapolis, IN; or Modular P800; Roche Diagnostics; Indianapolis, IN). The Mayo Foundation Institutional Review Board approved the study, and a waiver of informed consent was granted. Patients who did not authorize their medical records to be reviewed for research were excluded from the study. The data collected included demographics, bGlu, pGlu, mean arterial pressure (MAP), and use of vasopressors at the time of obtaining the blood sample for glucose measurement. Descriptive data were summarized as the mean (SD) or percentages. bGlu and pGlu levels were compared using the paired Student t test. To assess the agreement between the two methods, the mean difference was obtained, and the limits of the agreement were calculated as the mean difference ± 2 SDs.6 We performed subgroup analyses to determine the impact of abnormally elevated pGlu level, hypotension, and vasopressor use on the agreement between measurements made using the two methods. We defined an abnormally elevated pGlu level as a pGlu level > 110 mg/dL. Hypotension was defined as a MAP of < 60 mm Hg. Vasopressor use was defined as the use of dopamine, norepinephrine, epinephrine, or phenylephrine at the time of blood sampling for glucose measurement. A p value of < 0.05 was considered to be statistically significant. All glucose levels are presented in milligrams per deciliter.

After excluding 31 patients who did not authorize their medical records to be reviewed for research, 816 simultaneous bGlu and pGlu measurements were found for 197 patients. The mean age of these patients was 56 years (SD, 15.8 years), 95.4% were white, and 51.8% were women. The blood samples had been obtained on four occasions when patients were receiving continuous IV insulin treatment.

On 18 occasions, the bGlu level was reported as < 50 mg/dL. In these cases, the mean pGlu level was 66.9 mg/dL (SD, 50.4 mg/dL; range, 13 to 198 mg/dL). On 31 occasions, the bGlu level was reported as > 400 mg/dL. In these cases, the mean pGlu level was 423 mg/dL (SD, 110 mg/dL; range, 300 to 900 mg/dL). On 767 occasions, the mean bGlu level was 159 mg/dL (SD, 56.4 mg/dL), and the mean pGlu level was 151 mg/dL (SD, 54.3 mg/dL; p < 0.001). The mean difference between measurements made using the two techniques was 7.9 mg/dL (SD, 17.6 mg/dL), and the limits of agreement were + 43.1 and –27.2 (Fig 1 ). The pGlu level was elevated on 589 occasions and was nonelevated on 178 occasions, when the pGlu level was between 50 and 400 mg/dL. The mean differences in pGlu and bGlu levels were not statistically significant between the elevated and nonelevated bGlu groups (p = 0.331). In the elevated pGlu group, the mean difference between measurements made using the two techniques was 7.6 mg/dL (SD, 17.1 mg/dL), and the limits of agreement were + 41.8 and –26.6. In the nonelevated pGlu group, the mean difference between measurements made using the two techniques was 9.0 mg/dL (SD, 19.1 mg/dL), and the limits of agreement were + 47.3 and –29.2.

MAP values were available on 741 occasions when pGlu was between 50 and 400 mg/dL, and the patients were hypotensive on 23 occasions (3.1%). The mean differences between pGlu and bGlu measurements were not statistically significant between the two MAP groups (p = 0.809). Without hypotension, the mean difference between the two techniques was 8.0 mg/dL (SD, 17.9 mg/dL), and the limits of agreement were + 43.7 and –27.7. When hypotension was present, the mean difference between measurements made using the two techniques was 8.9 mg/dL (SD, 14.4 mg/dL), and the limits of agreement were + 37.6 and –19.8. Information about vasopressor use was available on 740 occasions when the pGlu level was between 50 and 400 mg/dL. Patients were receiving vasopressor therapy on 180 occasions and were not receiving vasopressors on 560 occasions. The mean differences in pGlu and bGlu levels were not statistically significant between the occasions on which vasopressors had and had not been received (p = 0.399). When patients had received vasopressors, the mean difference between measurements using the two techniques was 9.0 mg/dL (SD, 10.6 mg/dL), and the limits of agreement were + 30.2 and –12.2. Without vasopressor use, the mean difference between measurements made using the two techniques was 7.8 mg/dL (SD, 19.5 mg/dL), and the limits of agreement were + 46.7 and –31.2.

Our results show that for the whole cohort, bGlu measurement provides, on average, a reasonable estimation of pGlu level, but for the individual patient bGlu measurement gives an unreliable estimate of pGlu level. The presence of hypotension and the use of vasopressors do not appear to affect the difference between pGlu and bGlu.

Our results are similar to those of four small studies,25 despite differences in the study designs and the glucose meter used. The relevance of these findings is not minor, particularly at lower glucose concentrations. For example, in our cohort, considering our limits of agreement, a patient with a bGlu level of 80 mg/dL might have a pGlu level of 51 mg/dL. Additionally, a fingerstick bGlu measurement has been shown to be inaccurate in patients who are in shock,3low and high hematocrit values could overestimate or underestimate bGlu levels,4and arterial whole-blood sampling overestimates bGlu level.5 In our study, the difference between bGlu and pGlu levels was not influenced by the presence of hypotension or the use of vasopressors. However, since < 5% of the blood samples were obtained when the patients were hypotensive, we cannot exclude an association between the patients’ hemodynamic status, and difference between bGlu and pGlu levels.

Our study has several limitations. The study was performed in a single tertiary medical center and had a retrospective design. Although we highlighted the discrepancies between bGlu and pGlu measurements, we did not have adequate data to address the safety and cost savings of the different bedside glucometry devices. The small number of patients with hypotension and the unavailability of hemoglobin concentrations also weakened our findings. Prospective studies may help to address some of these issues. In the meantime, every ICU should evaluate the discrepancy between bGlu and pGlu measurements, and the clinical implication of this discrepancy.

Abbreviations: APACHE = acute physiology and chronic health evaluation; bGlu = bedside glucose; MAP = mean arterial pressure; pGlu = laboratory plasma glucose

Figure Jump LinkFigure 1. Comparison of bGlu and pGlu measurements according to the method of Bland and Altman.6 The difference between measurements made using the two techniques was plotted against their mean.Grahic Jump Location
Van den Berghe, G, Wouters, P, Weekers, F, et al (2001) Intensive insulin therapy in critically ill patients.N Engl J Med345,1359-1367. [CrossRef] [PubMed]
 
Ray, JG, Hamielec, C, Mastracci, T Pilot study of the accuracy of bedside glucometry in the intensive care unit.Crit Care Med2001;29,2205-2207. [CrossRef] [PubMed]
 
Atkin, SH, Dasmahapatra, A, Jaker, MA, et al Fingerstick glucose determination in shock.Ann Intern Med1991;114,1020-1024. [PubMed]
 
Louie, RF, Tang, Z, Sutton, DV, et al Point-of-care glucose testing: effects of critical care variables, influence of reference instruments, and a modular glucose meter design.Arch Pathol Lab Med2000;124,257-266. [PubMed]
 
Maser, RE, Butler, MA, DeCherney, GS Use of arterial blood with bedside glucose reflectance meters in an intensive care unit: are they accurate?Crit Care Med1994;22,595-599. [CrossRef] [PubMed]
 
Bland, JM, Altman, DG Statistical methods for assessing agreement between two methods of clinical measurement.Lancet1986;1,307-310. [PubMed]
 

Figures

Figure Jump LinkFigure 1. Comparison of bGlu and pGlu measurements according to the method of Bland and Altman.6 The difference between measurements made using the two techniques was plotted against their mean.Grahic Jump Location

Tables

References

Van den Berghe, G, Wouters, P, Weekers, F, et al (2001) Intensive insulin therapy in critically ill patients.N Engl J Med345,1359-1367. [CrossRef] [PubMed]
 
Ray, JG, Hamielec, C, Mastracci, T Pilot study of the accuracy of bedside glucometry in the intensive care unit.Crit Care Med2001;29,2205-2207. [CrossRef] [PubMed]
 
Atkin, SH, Dasmahapatra, A, Jaker, MA, et al Fingerstick glucose determination in shock.Ann Intern Med1991;114,1020-1024. [PubMed]
 
Louie, RF, Tang, Z, Sutton, DV, et al Point-of-care glucose testing: effects of critical care variables, influence of reference instruments, and a modular glucose meter design.Arch Pathol Lab Med2000;124,257-266. [PubMed]
 
Maser, RE, Butler, MA, DeCherney, GS Use of arterial blood with bedside glucose reflectance meters in an intensive care unit: are they accurate?Crit Care Med1994;22,595-599. [CrossRef] [PubMed]
 
Bland, JM, Altman, DG Statistical methods for assessing agreement between two methods of clinical measurement.Lancet1986;1,307-310. [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543