Article Text

Download PDFPDF

Variation in haemoglobin measurement across different HemoCue devices and device operators in rural Cambodia
  1. Aviva I Rappaport1,
  2. Susan I Barr1,
  3. Timothy J Green2,
  4. Crystal D Karakochuk1,3
  1. 1Department of Food, Nutrition and Health, University of British Columbia, Vancouver, British Columbia, Canada
  2. 2South Australian Health and Medical Research Institute, Adelaide, South Australia, Australia
  3. 3BC Children's Hospital Research Institute, Vancouver, British Columbia, Canada
  1. Correspondence to Dr Crystal Karakochuk, 216-2205 East Mall, Vancouver, British Columbia, Canada V6T 1Z4; crystal.karakochuk{at}ubc.ca

Abstract

Point-of-use haemoglobinometers, such as the HemoCue, are a common method to measure haemoglobin (Hb) concentration in field settings as the device is portable, requires only a small finger-prick capillary blood sample and computes an immediate Hb reading. The aim of this study was to compare Hb measurements across different HemoCue devices and across device operators using capillary blood samples collected from women during a trial in rural Cambodia. We compared mean±SD capillary Hb concentration (g/L) across n=12 different HemoCue Hb 301 devices and across n=9 device operators among 2846 Cambodian women. Significant variability in mean Hb concentration was observed across HemoCue devices (means ranged from 117 to 124 g/L) and across device operators (means ranged from 118 to 124 g/L). This variability is of particular concern when a single HemoCue device or device operator is used at different time points in surveys or research trials.

Trial registration number NCT02481375

  • diagnostic screening
  • ANALYTICAL METHODS
  • HAEMATOLOGY

Statistics from Altmetric.com

Introduction

Point-of-use haemoglobinometers, such as the HemoCue (HemoCue AB, Angelholm, Sweden), are a common method to measure haemoglobin (Hb) concentration in field settings as the device is portable, requires only a small finger-prick capillary blood sample and provides an immediate Hb reading.1 The HemoCue is often used to estimate the prevalence of anaemia in clinical, blood donation and field settings.

Factors such as humidity2 and blood collection technique3 ,4 have been reported to influence Hb concentration with the use of a haemoglobinometer in field settings. For example, milking the finger to stimulate blood flow can push extra interstitial fluid into the capillary sample, causing an underestimation of Hb concentration.5 Further, variability has been reported between different drops of capillary blood from the same sample site,5 ,6 suggesting that different drops of capillary blood could result in different estimations of Hb concentration in the same individual. The extent to which different devices or different device operators could influence the measurement of Hb concentration in the field setting is largely unknown; however, it is important, as it could influence the reliability of Hb concentration in small pre–post studies where measurements are conducted with a single device or device operator. This could lead to inaccurate conclusions about changes in Hb concentrations over time and/or across different surveys.

Our aim was to compare mean Hb concentrations across 12 different HemoCue 301 devices and nine different device operators using capillary blood samples collected from women in Cambodia. The findings have implications for the accurate interpretation of Hb concentrations.

Materials and methods

Study design and participants

We used data collected during screening as part of an iron supplementation trial in rural Cambodia. The study took place in 26 villages in the province of Kampong Chhnang, which is at sea level in central Cambodia. Women were recruited via convenience sampling. Ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board in Canada (H15-00933) and the National Ethics Committee for Health Research in Cambodia (110-NECHR). Details of the study have been published elsewhere (Karakochuk CD, Barker MK, Whitfield KC, et al. The effect of oral iron with or without multiple micronutrients on hemoglobin concentration and hemoglobin response among non-pregnant Cambodian women of reproductive age: a 2×2 double-blind randomized controlled supplementation trial. Forthcoming, 2017). The women in this study were apparently healthy non-pregnant women of reproductive age (18–45 years).

Blood sample collection

Research staff received two days of training on capillary blood collection procedures as per global guidelines, and staff members were supervised during the screening procedures in the field.3 ,7 A total of 12 HemoCue 301 devices were used for screening over five days. HemoCue devices were labelled #1–12. HemoCue devices #1–5 were previously used devices from Helen Keller International (Phnom Penh, Cambodia), devices #6–10 were newly purchased and used in the current study for the first time, and devices #11–12 were previously used devices on loan from the Institut de Recherche Pour Le Développement (Phnom Penh, Cambodia). The HemoCue devices were cleaned daily prior to use as per procedures outlined in the HemoCue 301 operating manual. Quality control (QC) tests were conducted on the HemoCue device using the recommended QC solution, Hb 301 Control (Level II) (Eurotrol BV, Ede, The Netherlands).8

Blood was collected between July and August 2015. A non-fasting capillary blood sample was obtained by a finger-prick from the index finger on the left hand. The finger was first cleaned with rubbing alcohol and allowed to dry before the capillary sample was collected using the HemoCue safety lancets. The first two drops of capillary blood were wiped away and the third drop was used to immediately fill the microcuvette.

Data and statistical analysis

Data were recorded for the total number of samples measured by each HemoCue device and operator, and the number of villages covered by each HemoCue device and operator. Hb concentration (g/L) is presented as mean±SD. One-way analysis of variance (ANOVA) was used to determine if there was significant variation associated with HemoCue device or device operator across groups. Two-sided p values <0.05 indicated statistical significance. Data were analysed in Stata software SE 13.1 for Mac (Stata, College Station, Texas, USA).

Results

Quality control

QC tests were performed with the recommended Hb 301 Control (Level II, Lot 42578, Ref 188.002.002)9 on all HemoCue devices daily prior to use. All reported values were within acceptable levels (target Hb 131±12 g/L (119–143) as defined by the value of the Hb 301 Control).

Participants

Capillary blood samples from 2846 non-pregnant women of reproductive age (18–45 years) in 26 villages in Kampong Chhnang province were used in this analysis. HemoCue devices were chosen for use at each village by each of the nine research staff; each HemoCue device (n=12) had between one and six QC tests conducted (depending on the number of days it was used in the screening). Device operators (n=9) were trained and performed the finger-prick capillary tests under supervision in the field. There were n=271 instances in which the HemoCue device number was not recorded; these Hb values were not included in the analysis to determine variability by HemoCue device.

Hb measurements across HemoCue devices (n=12)

Mean Hb concentrations were compared across the n=12 HemoCue devices (table 1). The total number of samples measured, the number of villages covered and the number of device operators included are also reported for each HemoCue device. Significant variability (p<0.05, ANOVA) in mean Hb concentrations was observed across HemoCue devices (means ranged from 117 to 124 g/L). Some devices measured Hb concentrations higher or lower than others. For example, HemoCue #8 had a mean Hb measurement of ∼122 g/L (based on n=219 samples in n=7 villages measured by n=1 operator), whereas HemoCue #1 had a mean Hb measurement of ∼119 g/L (based on n=270 samples in n=10 villages measured by n=1 operator).

Table 1

Mean±SD Hb concentration (g/L) across n=12 HemoCue (HC) devices*

Hb measurements across HemoCue device operators (n=9)

Mean Hb concentrations were compared across the n=9 HemoCue device operators (table 2). The total number of samples measured, the number of villages covered and the number of HemoCue devices used are also reported for each HemoCue device operator. Significant variability (p<0.05, ANOVA) in mean Hb concentrations was observed across HemoCue device operators (means ranged from 118 to 124 g/L). Some operators consistently measured Hb concentrations higher or lower than others. For example, HemoCue operator #7 had a mean Hb measurement of ∼119 g/L (based on n=270 samples in n=10 villages measured with n=1 device), whereas HemoCue operator #5 had a mean Hb measurement of ∼124 g/L (based on n=366 samples in n=13 villages measured with n=1 device). HemoCue operator #7 consistently used HemoCue device #8, and we cannot tell if the apparently lower value is a result of the device, the operator, or both.

Table 2

Mean±SD Hb concentration (g/L) across n=9 HemoCue device operators (DOs)*

Discussion

In this study in Cambodia, we found significant variability in mean Hb concentrations across n=12 HemoCue devices and across n=9 HemoCue device operators. The arithmetic average between HemoCue device #8 and #1 is approximately 0 (data not shown), suggesting that prevalence of anaemia at a population level is not affected by HemoCue device or operator. However, the issue of variability occurs when a single device or operator is used. Therefore, research or programmes that use one haemoglobinometer or device operator should optimally use the same device or operator over one or more measurements of Hb concentration to minimise variability.

There are limitations of our analysis that require acknowledgement. Despite a total of n=12 HemoCue devices and n=9 device operators included in this cross-sectional survey among a large sample of women (n=2846 women), many operators only used one HemoCue device. Therefore, it is difficult to determine if the differences observed in Hb were a result of the device, or a result of the collection method of the operator, or both. Measurement errors are common in point-of-care haemoglobinometers. Variability has been reported between drops of capillary blood.5 ,6 In a sample of women in central Honduras (n=87), Morris et al5 observed difference in Hb of 0.5 g/L in the left hand (131.0±14.9 g/L) and the right hand (130.5±14.9 g/L) of the same individual. Of note, only one device operator (nurse) conducted all of the sampling in the study of Morris et al.

The current literature suggests some solutions to the problem of variability in capillary blood. One recommendation is to collect multiple samples and calculate the average Hb concentration to minimise sampling error.3 ,6 Bond and Richards-Kortum6 suggest the collection of multiple successive drops of capillary blood into a tube containing anticoagulant prior to analysis to minimise the drop-to-drop variation. A laboratory-based study in Germany confirmed that using more microcuvettes and averaging the Hb measurements to determine Hb concentration improves the accuracy.9 This may be a potential solution for minimising the variability in Hb concentrations when using single haemoglobinometers and device operators.

HemoCue (HemoCue AB, Angelholm, Sweden) recommends the use of QC solutions to ensure the HemoCue device is accurately measuring Hb within an acceptable range. The company indicates the device should be within the acceptable range of Hb Control (Eurotrol BV, Ede, The Netherlands). In our QC value sheet, the target Hb concentration was 131±12 g/L (note, this target value varies by lot number). We argue that such a large range (119–143) exceeds the acceptable margin for sampling error. The suggested QC ranges for these devices warrant reconsideration. Further, a problem with the current models of the HemoCue devices is that real-time calibration is not possible. Therefore, if a device measures the QC solution outside of the acceptable range, the manual indicates that the device should simply not be used and to contact technical assistance. This is problematic if only one or few devices are available for use in remote field settings. Further, we query whether the recommendations for use of the HemoCue should stipulate that more than one device be used in population-level surveys to minimise the potential measurement bias as compared with the use of just one machine. As in this research, we found that inappropriately low values were counterbalanced by inappropriate high values, thus leading to a relatively accurate estimate.

We observed significant variability in Hb concentration across different HemoCue devices and across different device operators; however, from our study it is impossible to ascertain which of the two factors is the major cause of the variation, or the magnitude to which each of the two factors contribute to the variation. This is concerning as the HemoCue is often used in field settings to measure Hb concentration. In cross-sectional surveys to estimate anaemia prevalence, use of multiple devices and/or operators may help attenuate overall bias. In intervention trials where the primary outcome is change over time, use of a single device/operator for repeated measures over time could be preferable, as random error in change scores would be minimised. Therefore, inaccurate estimates could have serious implications for nutrition policy and programming. More work is needed to improve the accuracy of Hb measurement using HemoCue devices.

Take home messages

  • Point-of-use haemoglobinometers are commonly used to measure haemoglobin concentration in field settings.

  • We found significant variation in haemoglobin concentrations across different haemoglobinometer devices and device operators.

  • In cross-sectional surveys to estimate anaemia prevalence, use of multiple devices and/or operators may help attenuate overall bias.

  • In intervention trials where the primary outcome is change over time, use of a single device/operator for repeated measures over time could be preferable, as random error in change scores would be minimized.

  • Inaccurate measurement of haemoglobin concentration could influence anemia prevalence estimates and have serious implications for nutrition policy and programing.

Acknowledgments

The authors thank Keith Porter, Kroeun Hou, Ngik Rem and Sokhoing Ly, Helen Keller International, for their operational support; and Chanthan Am, The National Institute of Public Health Laboratory, for assistance with lab analyses.

References

Footnotes

  • Handling editor Mary Frances McMullin

  • Contributors CDK drafted the research protocol. CDK oversaw data collection in the field. AIR and CDK conducted statistical analyses. AIR and CDK drafted the manuscript. All authors contributed to the data interpretation and manuscript preparation. AIR and CDK had responsibility for the final content. All authors read and approved the final manuscript.

  • Funding Funding was provided by the International Development Research Centre, through the Department of Foreign Affairs, Trade and Development, Canada.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval University of British Columbia Clinical Research Ethics Board in Canada and the National Ethics Committee for Health Research in Cambodia.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.