A Capillary Glucose That Does Not Fit the Patient
A nurse asks you to perform a point-of-care glucose on a 68-year-old man in the intensive care unit. He is hypotensive, his hands are cool and mottled, and he has a fine tremor. A 20-gauge IV in his left forearm is infusing dextrose-containing fluid. You obtain a capillary sample from his right index finger; the meter reads 42 mg/dL. The nurse says, "Great, I'll draw up the D50." Twenty minutes earlier, a venous glucose sent to the laboratory on the same patient came back at 118 mg/dL. The meter's daily quality control that morning passed, and the strips on the cart are in date.
1. Do you report 42 mg/dL as a valid result? Walk through your reasoning.
Read the worked answer
No - not as a confirmed value. Report it as an unverified, discrepant reading and pursue a reliable number. Reasoning:
- Perfusion. Capillary glucose assumes capillary blood reflects arterial blood. In a hypotensive patient with cool, mottled extremities, peripheral flow is sluggish and capillary values can read falsely low.
- Magnitude of the discrepancy. 42 versus a laboratory venous value of 118 twenty minutes earlier is a 76 mg/dL gap. Real glucose does not move that far that fast without treatment; the two numbers cannot both be right.
- Meter-side checks that are cheap to rule out. QC passed and strips are in date, so a gross device failure is less likely - but check the strip lot, whether the meter is the model validated for this unit, whether the strip was expired at the individual level, and whether the sample was adequate.
Conclusion: the most probable explanation is the sampling site, not the patient. Confirm with a venous specimen sent to the laboratory. Do not report a critical value you cannot defend, and do not suppress it either - report it as discrepant with your concern attached.
2. If a repeat fingerstick is needed while you wait for the laboratory, what technique changes do you make and why?
Read the worked answer
Site selection:
- Use a finger with good color and warmth, on the lateral side of the fingertip.
- Avoid the arm with the dextrose infusion - that is the single most likely source of a wildly wrong number on this patient.
- Avoid edematous, cyanotic, callused, or previously punctured sites, and avoid the fingertips of a hand that has been used for a prior stick in the last few minutes.
Preparation:
- Warm the hand first - a warm pack or warm blanket for a few minutes improves perfusion and the reliability of the sample.
- Position the hand below heart level.
Collection:
- Cleanse and let it dry completely; residual alcohol dilutes the drop and can lower the reading.
- Wipe away the first drop and use the second, which reflects capillary blood rather than interstitial fluid.
- Do not milk or squeeze the finger - that expresses interstitial fluid into the sample.
- Make sure the strip is fully dosed and at room temperature.
Interpretation: even a technically perfect capillary sample in a poorly perfused patient is only an estimate. The venous laboratory glucose is the reference value here.
3. The nurse repeats that she is going to give D50 now. How do you handle that in the moment, and what do you document?
Read the worked answer
In the moment:
- Be direct and specific, not vague: "That reading is from a cold, mottled finger on a hypotensive patient, and the lab read 118 twenty minutes ago. I don't think 42 is real. I've sent a venous glucose stat."
- Do not physically obstruct care or make a treatment decision - that is not your role. Your role is to make sure the decision is not made on a number you believe is wrong, and to say so clearly.
- If the patient has clinical signs of hypoglycemia - sweating, confusion, seizure, decreased consciousness - a confirmatory lab should not delay treatment. The clinician may reasonably treat on clinical grounds. That is a different decision than treating a number.
- Escalate to the charge nurse or the provider if the disagreement persists.
Documentation:
- Meter ID, strip lot, and QC status.
- Site used, condition of the site (cool, mottled), and technique (first drop wiped, no milking).
- The 42 mg/dL value, flagged as unverified/discrepant.
- The prior laboratory value of 118 mg/dL and its time.
- Confirmatory specimen sent, time sent, who was notified, and what they said.
- Any repeat testing and its result.
Reasoning: the record must show that the discrepancy was recognized and acted on. A silent chart with a "42" in it and D50 given afterwards is a chart that cannot explain itself.