Accuracy of point-of-care capillary blood sugar measurements in critically ill patients: An observational study

ABSTRACT

Background:

Accurately monitoring blood glucose levels is vital for critically ill individuals. Point-of-care (POC) glucose meters are commonly used in local intensive care units (ICUs). This study aimed to assess the precision of POC glucose meter readings in critically ill individuals with specific evaluation in patients with and without shock against the reference standard of venous blood glucose measurements.

Methods:

An observational study was done on adult patients admitted in the ICU at a teaching institution. Capillary blood samples were collected from the patient’s fingertip using lancet device with aseptic measures. The sample was analyzed using the GlucoCare Sense Glucometer (RMD Mediaids Limited, Taiwan). At the same time, 2 ml of blood was drawn from the patient’s peripheral veins and analyzed by glucose oxidase-peroxidase method as reference.

Results:

POC glucose measurements averaged 140 ± 20.23 mg/dl, while laboratory values were recorded as 116.10 ± 17.13 mg/dl. The difference between the two methods was 24.34 ± 12.01 mg/dl. A strong correlation (r = 0.805) was found between capillary and laboratory blood glucose levels, indicating a significant association (P < 0.0001). Twenty-two (44%) patients were in shock during the study. The mean difference between laboratory and POC blood glucose levels was higher in patients with circulatory shock (36.82 ± 4.84 mg/dl) than those without shock (14.61 ± 4.49 mg/dl), P < 0.05.

Conclusion:

POC glucose meters may lead to underdetection of hypoglycemia in critically ill patients, as their values are higher than laboratory values. Moreover, the results showed that POC glucometers are inaccurate for monitoring glucose in hypotensive patients in shock. Standard venous glucose monitoring methods may be more appropriate for these patients.

INTRODUCTION

There has been an established correlation between the blood glucose levels of critically ill patients and their outcomes, as uncontrolled blood sugars manifesting as either hyperglycemia or hypoglycemia are linked with poor outcomes.[1,2] Patients with acute myocardial infarction and stroke exhibit similar associations between blood glucose levels and their outcomes, and this association is influenced by chronic hyperglycemia before the onset of acute illness. Patients with preexisting diabetes, especially those with poorly controlled diabetes, experience weakened and imbalanced relationships with higher glycemic levels. Patients without preexisting diabetes are more likely to experience poor outcomes when they develop hyperglycemia during acute critical illness.[3] Undetected diabetes mellitus may be present in a significant number of these patients and should be considered.[4]

Accurately monitoring blood glucose levels is vital for critically ill individuals as identifying symptoms of both hypo- and hyperglycemia can be challenging due to patient sedation or poor sensorium due to the underlying illness. As a result, implementing precise monitoring techniques is essential for early detection and treatment of any abnormality in blood glucose levels. There are various standards for blood glucose levels that different organizations recommend for patients in intensive care units (ICUs).[5,6]

Point-of-care (POC) glucose meters are commonly used in local ICUs and hospital wards to help achieve these targets. These meters are portable, easy to use, and provide quick results, all of which are crucial for patients who require strict glycemic control. In some cases, central laboratory results may be delayed, which can make it difficult to adjust treatment promptly. While POC glucose meters are accurate for stable outpatients, they can be affected by several factors in critically ill patients, such as hemoglobin level, temperature, blood PH, oxygenation, and peripheral perfusion.[7,8]

When dealing with shock, there are various possible explanations for the variations in blood sugar readings measured through various modalities. During hypoperfusion, where there is peripheral vasoconstriction, tissue glucose extraction increases due to low capillary flow. This can cause capillary blood glucose measurements to be falsely low.[9,10]

The objective of this research was to assess the precision of POC glucose meter readings in critically ill individuals with specific evaluation in patients with and without shock. The findings of this study could be useful in developing guidelines for blood sugar monitoring and insulin therapeutic protocols in the ICU and in establishing appropriate intervention thresholds, especially for patients experiencing shock.

METHODS

An observational study was conducted for 18 months (December 2020 until June 2022) at an ICU of a teaching institution in Delhi, India, after receiving approval from the institutional ethics committee.

The inclusion criteria were critically ill adult patients (age ≥18 years) that stayed in the ICU for more than 24 h. Exclusion criteria consisted of individuals who were pregnant, with bleeding disorders, or a platelet count lower than 25,000/μL or a hematocrit lower than 20%. Furthermore, individuals who had been subjected to substances that have an impact on the POC glucose meter measurements, such as abatacept, icodextrin, maltose, or intravenous immunoglobulins, were not considered in the analysis. Finally, if individuals did not provide consent, they were excluded from the study.

A sample size of 50 patients was required to ascertain the percentage of POC blood glucose values that coincide with the laboratory whole-blood glucose method with a margin of error of ±10% and power of 90%. The sample size was calculated in reference to the study by Critchell et al. who reported the mean difference between the two methods was 8.6 ± 18.6 mg/dl.[11]

The patient’s fingertip was pricked with a lancet device in a sterile manner to obtain capillary blood. A reagent strip containing glucose oxidase (20 IU), potassium ferricyanide (0.12 mg), and nonreactive ingredients (1.8 mg) was used to place a single drop of each sample. The sample was analyzed using the GlucoCare Sense glucometer (RMD Mediaids Limited, Taiwan). The glucometer system met the ISO 15197: 2013 standard and complies with CE certification, but has not been tested in critically ill patients. Test strips for each test were calibrated with a plasma calibration with a test range of 20–600 mg/dl.

The collection of 2 ml of blood from the patients’ peripheral veins was conducted with aseptic precautions from the arm without an intravenous line to ensure the accuracy of the results. A sterile blood tube containing fluoride was used to collect the blood, which was then sent to the central laboratory for analysis. The laboratory utilizes commercially available kits and ADVIA^® 2400 Chemistry System (Siemens Medical Solutions, Inc., Malvern, USA). Venous blood glucose measurements by glucose oxidase-peroxidase method were utilized as the reference standard. POC glucose and phlebotomy for the laboratory glucose testing were performed at exactly the same time and simultaneous estimations of glucose by capillary and venous blood samples were done, and the results were compared. No sampling was done during a cardiac arrest or within an hour of the event.

Statistical analysis

Data were analyzed using Microsoft Excel and Statistical Package for the Social Sciences (SPSS), IBM manufacturer, Chicago, IL, USA, version 21.0. Categorical variables were displayed as percentages and numbers, and continuous variables were presented through mean and median values. Statistical tests were used to compare POC and laboratory glucose levels, including the Pearson correlation coefficient and Fisher’s exact test. The Bland–Altman test was also utilized to compare blood glucose measurements. P < 0.05 was considered significant.

RESULTS

Fifty patients were recruited to this study having a mean age of 38.6 ± 15.4 years, including 17 (34%) females. During the sampling, 22 (44%) patients were in shock, with septic shock being the most common underlying cause (17, 34%). Table 1 provides the baseline clinical and demographic details. Among the patients, 12 (24%) were admitted to the ICU after undergoing emergency exploratory laparotomy for acute abdomen, which was the most common reason for ICU admission, followed by community-acquired pneumonia in 9 (18%). As for comorbidities, 15 patients had preexisting diabetes.

Table 1.

Clinical characteristics of included patients

	Shock (n=22), n (%)	No shock (n=28), n (%)	P
Gender
Male	15	18	>0.05
Female	7	10
Mean age (years)	38.31±15.69	38.82±15.53	>0.05
Diagnosis
Acute abdominal pain	8	4	N/A
Poisoning	2	3
GBS	1	4
Pneumonia	6	3
Others	5	14
Preexisting diabetes
Yes	6	9	>0.05
No	16	19
Hemodialysis
Yes	12	8	<0.001*
No	10	20
Mechanical ventilation
Yes	18	13	<0.001*
No	4	15
Urine output (mL/h)	46.73	84.32	<0.001*
Temperature (°F)	99.3	98.9	>0.05
Heart rate	113.6	88.2	<0.05*
SBP (mm Hg)	92.5	118.7	<0.001*
DBP (mm Hg)	50.3	74.6	<0.001*
Capillary refill (s)	4.1	1.5	<0.001*
eGFR (mL/min)	34.91	59.11	<0.001*
Hematocrit	33.5	34.75	>0.05
Hemoglobin	9.2	9.4	>0.05
WBC	13.7	12.4	>0.05
PH	7.37	7.41	>0.05
PO2	74.6	75.3	>0.05
Lactate	4.31	1.72	<0.001*
SOFA score	9.41	7.68	<0.05*
APACHE II score	26.3	21.1	<0.05*

*Significant (P<0.05). APACHE: Acute physiology, age, and chronic health evaluation, DBP: Diastolic blood pressure, eGFR: Estimated glomerular filtration rate, GBS: Guillain–Barré syndrome, SBP: Systolic blood pressure, SOFA: Sequential organ failure assessment, WBC: White blood cells, N/A; Not available

The mean POC capillary glucose measurement was found to be 140 ± 20.23 mg/dl, while the mean of laboratory blood glucose values was recorded as 116.10 ± 17.13 mg/dl. Upon comparison of the two methods, the mean difference in blood glucose measurements was calculated as 24.34 ± 12.01 mg/dl (95% confidence interval [CI]: 20.92–27.75 mg/dl). There was a strong correlation between the capillary blood glucose levels obtained through POC testing and the laboratory blood glucose levels [Figure 1]. The Pearson correlation coefficient value of 0.805 (95% CI: 0.680–0.885) suggests a significant association between the two variables (P < 0.0001).

Figure 1.

Pearson coefficient plot: Scattered plot showing positive correlation between POC glucose and venous blood glucose. POC: Point of care

Figure 2 displays a Bland–Altman plot, indicating that the upper limit of agreement (LOA) for the mean difference was 47.88 mg/dL, while the lower LOA was 0.79 mg/dL.

Figure 2.

Bland–Altman difference plot of paired glucose measurements (point-of-care glucose and venous blood glucose) in the overall study population (n = 50). LOA: Limits of agreement, POC: Point of care, SD: Standard deviation

We compared glucose measurements in patients of shock and those who were normotensive. The results revealed that the mean difference between laboratory blood glucose levels and POC blood glucose testing was 14.61 ± 4.49 mg/dl in patients who were not in shock during the sampling. However, in patients with circulatory shock, the mean difference was 36.82 ± 4.84 mg/dl [Figure 3].

Figure 3.

Bland–Altman difference plot of paired glucose measurements (point-of-care glucose and venous blood glucose) in patients with shock (n = 22). LOA: Limits of agreement, POC: Point of care, SD: Standard deviation

According to the ISO 15197:2013 standard, a significant requirement is that over 95% of blood glucose measurements should not deviate more than ±15 mg/dl from laboratory results for levels under 100 mg/dl and should remain between ±20% for concentrations of >100 mg/dl.[12] Our investigation revealed that among patients not in shock, 96.43% of POC measurements were within the agreement limits. However, for patients in the shock group, only 17.24% of measurements were within the target limits.

DISCUSSION

It is common for patients who are admitted to the ICU to experience fluctuations in their blood sugar levels. This can happen due to preexisting diabetes, stress-induced hyperglycemia, or irregular nutrition and fluid therapy. Patients with diabetes or neurological pathologies may require strict protocol-based glycemic control. Moreover, in sedated or unconscious patients, symptoms of hypoglycemia may not be noticeable. Therefore, it is important to have a protocol for glucose monitoring, using a rapid and accurate method to measure blood glucose levels. The study compared capillary blood sugar levels measured by a POC glucose meter to venous blood sugar levels measured by laboratory analysis.

Based on our observational research, though there appears to be a significant positive correlation between POC and laboratory blood glucose measurements in critically ill patients; after conducting a Bland–Altman agreement analysis, we discovered that the average POC blood glucose levels of ICU patients were higher than the venous glucose values (mean difference: 24.34 ± 12.01 mg/dl). This is concerning because it suggests that the POC capillary blood glucose measurements may overestimate the reference standard. This could potentially result in missed episodes of hypoglycemia in patients who are following a strict glycemic control protocol. Similarly, among the subgroup of ICU patients in shock, the mean POC capillary glucose levels were higher than the mean venous glucose, and the mean difference between the two measurements was higher in the shock group (36.82 ± 4.84 mg/dl), as compared to those who were normotensive (14.61 ± 4.49 mg/dl). However, the confidence limit [Figures 2 and 3] was narrower for patients in shock.

Observations from previous studies analyzing hypotensive subgroups indicate that patients who receive vasopressor support exhibit a broader range of confidence limits in comparison to nonshocked patients.[13,14] It is challenging to suggest a consistent “correction factor” for POC glucose measurements due to the absence of a predictable directional inclination. As a result, establishing adjusted intervention thresholds that are safe remains a challenge.

In line with the present study, various investigators have reported a low accuracy of POC glucose meters in ICU patients.[10,11,15,16,17] However, only a couple of researchers have reported POC measurements being higher than laboratory measurements,[11,17] as most research showed POC measurements being falsely low.[15,16] In contrast, several studies have reported good concordance between the two methods.[7,18,19] A study conducted recently analyzed the measurements of blood glucose in arterial, peripheral venous, and central venous samples taken from ICU patients, in comparison to the corresponding laboratory values. The study found that the two methods of estimating blood sugars were accurate, with no significant difference between the measurement values, when the source of blood sampling was consistent, except for arterial samples.[7] Although the manufacturers of POC glucose meters recommend using capillary blood and the test strips too are calibrated for the same, further research using this methodology can lead to the development of specialized glucose meters and test strips for critically ill patients, which can be used for arterial or venous blood samples.

Although in the present study, the glucose meter utilized for POC testing fulfilled the accuracy requirements for the normotensive group as per ISO 15197:2013 specifications;[12] however, the overall agreement was lower for the shock group, with just 17.24% of readings meeting the accuracy criteria. This highlights the difficulties in obtaining precise glucose measurements, both technically and clinically, for critically ill patients in circulatory shock. The results must be taken into account while dealing with critically ill patients at a primary health-care center.

There are certain limitations of the study. First, various glucometers were not compared. Second, critically ill patients are subject to various complications which may affect glucose values, and such values could not be adjusted as confounding factors. Third, follow-up of the patients was not done.

CONCLUSION

POC capillary blood sugar measurement for critically ill patients correlates with venous laboratory measurements, but the POC values are higher than laboratory values for patients with or without shock; therefore, POC measurements may lead to underdetection of hypoglycemia. According to our research, POC glucose meters are inaccurate in measuring blood sugar levels according to ISO 2003:15197 criteria for hypotensive, critically ill patients in shock compared to normotensive patients in the ICU. Therefore, alternative glucose monitoring methods such as central catheter blood POC measurements, automated blood/gas analyzers, or other advanced techniques may be more appropriate for patients in shock.

Research quality and ethics statement

This study was approved by the Institutional Review Board/Ethics Committee at Vardhman Mahavir Medical College, and Safdarjung Hospital, New Delhi (Approval number: IEC/VMMC/SJH/Thesis/034-20; Approval date: November 18, 2021). The authors followed the applicable EQUATOR Network (http://www.equator-network.org/) guidelines, specifically the “The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement:” Guidelines for reporting observational studies, during the conduct of this research project.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.