Near Infrared Spectroscopy for Blood Glucose Measurement in Critical Care Settings

Maintenance of tight glycemic control in critically ill patients has been demonstrated to significantly reduce morbidity and mortality in critical care patients in the OR and ICU settings. Numerous (>20) peer-reviewed publications have demonstrated the benefits of tight glycemic control. For example, tight glycemic control has been shown to reduce surgical site infections by 60% in cardiothoracic surgery patients (Furnary et al., 1999), and has been shown to reduce overall ICU mortality by 40% with significant reductions in ICU morbidity and length of stay (Van den Berghe, 2001).

Historically, caregivers treated hyperglycemia (high blood glucose) only when glucose levels exceeded 220 mg/dL. However, based upon these recent clinical findings, experts now recommend controlling blood glucose to within the normoglycemic range (80-110 mg/dL). Adherence to such strict glucose control regimens requires near-continuous monitoring of blood glucose and frequent adjustment of insulin infusion to achieve normoglycemia while avoiding risk of hypoglycemia (low blood glucose). Surveys indicate that, in response to this demonstrated clinical need, more than 50% of US hospitals have now adopted tight glycemic control protocols for some critical care patients, with an additional 23% expected to adopt such protocols within the next 12 months.

Currently, these TGC protocols require intravenous administration of insulin to achieve and manage glucose control to within the normoglycemic range. The protocols also mandate frequent monitoring of patient blood glucose levels by nursing staff. With existing technology, each test involves removal of a blood sample and testing on handheld meters or blood gas analyzers. This process is cumbersome, labor intensive, and often disruptive to the patient. Although hospitals are responding to the identified clinical need, current technology has limited adoption due to two principal reasons:

1. Fear of hypoglycemia: administering insulin to maintain the patient in the target glucose range of 80-110 mg/dL exposes the patient to increased risk clinical hypoglycemia (blood glucose less than 50 mg/dL). Critical care patients are often sedated and do not exhibit typical signs of hypoglycemia, exacerbating the risk of hypoglycemia. The need for frequent monitoring and concern about risk of hypoglycemia increase caregiver stress and can limit adherence to tight glycemic control protocols.
2. Glucose monitoring by conventional methods represents a burdensome procedure: most tight glycemic control protocols require frequent glucose monitoring and insulin adjustment at 30 minute to 2 hour intervals (typically hourly) to avoid hypoglycemia while maintaining normoglycemia. Glucose determination with current technology (including instrument preparation, blood sample acquisition and instrument analysis) typically takes 5 minutes per measurement.

The Sponsor is developing an automated patient-attached blood glucose monitor that will automatically make frequent blood glucose measurements. Frequent measurements will allow better tracking of patient glucose status and will provide blood glucose trend information. The system will alert caregivers to impending or actual hypoglycemia conditions. Automation will reduce the labor burden associated with existing technology. The device is intended to help caregivers to better manage tight glycemic control in their patients.

The device measures glucose in whole blood by flowing a patient blood sample through an optical flow cell while transmitting near infrared light through the sample. The blood constituents, including glucose, absorb a portion of the near-infrared energy in characteristic fashion. The instrument measures the resulting transmission spectrum and analyzes the spectrum to determine the concentration of glucose in the blood sample.

Though the proposed study will represent the initial measurements made by the device in ICU patients, the Sponsor has collected data in volunteer studies conducted at InLight's facility. In one study, we collected blood from volunteer donors and created individual blood samples with variable glucose (glucose range 50-550 mg/dL) by spiking the samples with concentrated glucose solution. We also induced hematocrit variation between 30-50% in these samples by controlling the red cell to plasma ratio. Spectral measurements collected with the prototype device were mathematically correlated with simultaneous blood glucose measurements of the same samples obtained with a Yellow Springs Instruments YSI 2700 Select glucose analyzer (an industry-standard laboratory glucose analyzer) to generate a spectral glucose model and to provide baseline near-infrared glucose measurement results. This evaluation demonstrated that the device's glucose measurements tracked the YSI 2700 glucose measurements with an accuracy of 8.9 mg/dL (1 s.d.). See Figure 1.

In another study, we collected serial blood samples from 5 volunteer subjects with diabetes undergoing dynamic glucose changes induced by carbohydrate ingestion and insulin administration. We obtained spectral glucose measurements of the blood samples using the prototype near-infrared device and compared them with simultaneous blood glucose measurements of the same samples obtained with the YSI 2700 Select glucose analyzer. The Sponsor's results compared favorably with the YSI 2700 results, demonstrating a root-mean-squared error (1 standard deviation) of 5.9 mg/dL (Figure 2). This accuracy level is comparable to current handheld meter technology.

We anticipate that the OR and ICU environments will present new physiological and pharmacological challenges for our instrument. The proposed study is essentially a screening study that will give the Sponsor insight into these challenges. Access to the de-identified ICU charts will allow the Sponsor to correlate the glucose measurements and measurement errors with specific pharmacologic therapies and physiological events. In this manner, the proposed study is in effect a screening study.