Validation of HemoCue method and development of a revised hemoglobin cut-off to detect anemia in children aged 6-24-months

This study validates the HemoCue method against the gold-standard DCM method in children aged 6–24 months in Bangladesh and proposes a revised hemoglobin cut-off of 11.00 g/dL to correct the underestimation of anemia prevalence caused by HemoCue's tendency to overestimate hemoglobin levels.

The Gist

The Big Picture: A Case of the "Optimistic Scale"

Imagine you are trying to weigh yourself to see if you are healthy. You step on a scale, but this scale has a glitch: it always adds 2 pounds to your actual weight.

• The Reality: You actually weigh 150 lbs.
• The Glitchy Scale: It says you weigh 152 lbs.

If the doctor says, "You are underweight if you are under 150 lbs," your glitchy scale might tell you, "Don't worry, you're fine!" even though you actually need help. You might miss out on the vitamins or food you need because the scale lied to you.

This is exactly what happened in this study.

The researchers were looking at anemia (a condition where blood lacks healthy red blood cells) in young children (ages 6 months to 2 years) in Bangladesh. They wanted to see if a popular, portable tool called HemoCue was telling the truth about how much hemoglobin (the oxygen-carrying part of blood) the kids had.

The Two "Scales" in the Study

1. The Gold Standard (DCM): Think of this as the Master Scale in a high-end laboratory. It is heavy, expensive, requires electricity, and takes time. But it is the most accurate way to measure blood. The World Health Organization uses this as the "truth."
2. The Field Tool (HemoCue): Think of this as a Handheld Pocket Scale. It's small, cheap, battery-powered, and gives a result in 10 seconds. It's perfect for remote villages where there is no electricity or fancy labs.

The Problem: The "Optimistic" Pocket Scale

The researchers took blood from nearly 850 samples of children. They measured the blood twice: once with the Master Scale (DCM) and once with the Pocket Scale (HemoCue).

The Result?
The Pocket Scale (HemoCue) was being too optimistic. It consistently reported higher hemoglobin levels than the Master Scale.

• The Master Scale said: "35% of these kids are anemic."
• The Pocket Scale said: "Only 23% are anemic."

The Consequence: Because the Pocket Scale was lying and saying the kids were healthier than they really were, nearly one-third of the sick children were missed. They weren't getting the treatment they needed because the tool told the doctors, "Everything looks fine!"

The Investigation: Why Can't We Just "Calibrate" It?

The researchers tried to fix the Pocket Scale. They thought, "Maybe if we just subtract a little bit from every number the Pocket Scale gives us, it will match the Master Scale."

They tried to draw a straight line between the two tools to create a "correction formula." But it didn't work. The Pocket Scale was acting like a drunk driver—sometimes it was a little off, sometimes a lot off, and the pattern was too messy to predict with a simple math formula. You couldn't just say "subtract 0.5" to fix it.

The Solution: Changing the "Pass/Fail" Line

Since they couldn't fix the machine, they decided to fix the rules.

Imagine a teacher grading a test.

• Old Rule: If you get 10.5 out of 10, you pass.
• The Problem: The Pocket Scale is a "generous" grader. It gives everyone 1 extra point. So, a kid who actually got a 9.5 (failing) gets a 10.5 (passing) on the Pocket Scale.

The Fix: The researchers realized that to get the same "truth" as the Master Scale, they had to make the passing grade harder. Instead of passing at 10.5, they needed to pass at 11.0.

• New Rule: If the Pocket Scale says the child has less than 11.0, they are anemic.

The Result: Catching the Missing Kids

When they applied this new, stricter rule to the Pocket Scale data:

• Old Count: 23% of kids were flagged as anemic.
• New Count: 34% of kids were flagged as anemic.

This new number (34%) was almost identical to the "Truth" number (35%) from the Master Scale.

Why Does This Matter?

In the real world, you can't always take a child to a fancy lab with the "Master Scale." You have to use the "Pocket Scale" in the village.

• Before this study: Doctors using the Pocket Scale were missing 1 out of every 3 anemic children.
• After this study: If they use the new cut-off of 11.0, they will catch almost all the sick children, even while using the cheap, portable tool.

The Takeaway

The HemoCue device is a fantastic tool for poor areas, but it has a "generous" bias. It likes to make people look healthier than they are. This study didn't fix the machine, but it gave doctors a new rulebook. By raising the bar for what counts as "healthy," they can now use the cheap tool to find the sick children who actually need help.

In short: Don't trust the Pocket Scale's raw numbers blindly. If you are using it in Bangladesh for young kids, remember: If the number is below 11.0, treat it as anemia.

Technical Summary

1. Problem Statement

Accurate measurement of hemoglobin (Hb) is critical for determining the prevalence of anemia, a major public health issue affecting nearly 40% of children under five globally. The Direct Cyanmethemoglobin (DCM) method is the established gold standard but is often impractical for large-scale field studies in resource-poor settings due to cost, infrastructure requirements, and complexity. Consequently, the HemoCue 201 photometric device is widely used for its portability and speed.

However, existing literature suggests that HemoCue tends to overestimate Hb levels compared to DCM. This systematic bias leads to a significant underestimation of anemia prevalence when using the standard World Health Organization (WHO) cut-off of <10.5 g/dL. There is a lack of validated, adjusted cut-off values specifically for the HemoCue method in children under two years old in low-income settings, potentially resulting in missed diagnoses and inadequate public health interventions.

2. Methodology

Study Design & Population:

• Data Source: Secondary analysis of data from the MAL-ED study (Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition) conducted in Mirpur, Dhaka, Bangladesh (Feb 2010 – Feb 2017).
• Participants: 589 children under two years of age (including birth cohort and case-control sub-studies).
• Sample Size: 846 venous blood samples were analyzed.
• Data Split: The dataset was randomly divided into a training set (70%) and a testing set (30%) for validation.

Measurement Techniques:

• Gold Standard: Direct Cyanmethemoglobin (DCM) method (venous blood).
• Field Method: HemoCue 201 instrument (venous blood).
• Anemia Definition: Initially defined by WHO standards (<10.5 g/dL) for DCM.

Statistical Analysis:

• Agreement Analysis: Bland-Altman plots, Pearson correlation, Cohen's Kappa, and McNemar's test were used to compare the two methods.
• Correction Factor Attempt: Researchers attempted to derive a linear or quadratic correction factor to adjust HemoCue values to match DCM. Linearity assumptions were tested but failed due to significant dispersion in HemoCue readings.
• Cut-off Optimization: Receiver Operating Characteristic (ROC) curve analysis was performed. Youden's Index was used to identify the optimal cut-off, with a constraint to achieve at least 70% sensitivity and 70% specificity to ensure balanced diagnostic performance.
• Software: Analyses were conducted using R (version 4.5.2).

3. Key Contributions

• Validation of Bias: The study quantitatively confirms that HemoCue significantly overestimates Hb levels in children under two, leading to a ~34% underestimation of anemia prevalence when using standard cut-offs.
• Rejection of Correction Factors: The authors demonstrated that a simple mathematical correction factor (linear or quadratic) could not be reliably applied due to the non-linear dispersion of HemoCue data against the DCM standard.
• Proposed Revised Cut-off: The study proposes a new, empirically derived HemoCue cut-off value of 11.00 g/dL for diagnosing anemia in children aged 6–24 months, replacing the standard 10.5 g/dL.
• Validation Framework: The proposed cut-off was rigorously validated using a train-test split approach, demonstrating improved sensitivity and specificity compared to the WHO standard.

4. Key Results

Descriptive Statistics:

• Median Hb (DCM): 10.8 g/dL (IQR: 10.1–11.5).
• Median Hb (HemoCue): 11.6 g/dL (IQR: 10.5–12.5).
• Correlation: Strong positive correlation ($r = 0.72$), but with moderate agreement (Kappa = 0.51).

Anemia Prevalence Discrepancy:

• Using the standard 10.5 g/dL cut-off:
  • DCM detected 35.2% anemia.
  • HemoCue detected only 23.2% anemia.
  • Result: HemoCue missed over 12% of anemic cases (underestimation).

Performance of New Cut-off (11.00 g/dL):

• Training Set: Sensitivity 94.07%, Specificity 53.69%.
• Testing Set (Validation):
  • Sensitivity: 84.38%
  • Specificity: 69.47%
  • Balanced Accuracy: 77.77%
  • Anemia Prevalence: 34.2% (closely aligning with the DCM gold standard of 35.2%).
• Statistical Significance: McNemar's test showed no statistically significant disagreement between the new HemoCue cut-off and the DCM standard ($p = 0.538$), unlike the significant disagreement found with the 10.5 g/dL cut-off.

5. Significance and Implications

• Public Health Impact: Adopting the 11.00 g/dL cut-off for HemoCue devices in resource-poor settings will significantly reduce the underestimation of anemia prevalence. This ensures that a larger proportion of anemic children are identified and referred for treatment.
• Policy Guidance: The findings suggest that the standard WHO cut-off of 10.5 g/dL is inappropriate for HemoCue devices in this specific demographic (children <2 years). Programs relying on HemoCue should adjust their diagnostic thresholds to 11.0 g/dL to maintain diagnostic accuracy.
• Limitations & Future Work:
  • The study used venous blood, whereas HemoCue is often used with capillary blood in the field. The authors acknowledge that the new cut-off requires validation for capillary samples.
  • The study is specific to the Bangladesh context; factors like altitude and genetic hemoglobinopathies in other regions may require further validation.
  • Despite the device's tendency to overestimate, it remains the most feasible tool for large-scale surveys; therefore, adjusting the cut-off is a pragmatic solution until more accurate portable technology becomes available.

Conclusion:
The paper provides robust evidence that the HemoCue method overestimates hemoglobin in young children, leading to missed anemia diagnoses. By validating a revised cut-off of 11.00 g/dL, the authors offer a practical, data-driven solution to improve the accuracy of anemia surveillance and intervention in low-resource settings.