Diagnostic and analytic performance of four point-of-care hemoglobin testing devices used in routine HIV and maternity care in Uganda

Abstract

Background

Hemoglobin estimation (Hb) is the most requested hematology test, especially among pregnant/postnatal women and people living with HIV (PLHIV). In Uganda, several point-of-care (POC) Hb testing devices are currently used and performance may be affected by multiple factors. This study evaluated the diagnostic and analytic performance of four Hb POC devices.

Methods

During October 2021–April 2022, four POC Hb devices (HemoCue Hb 301, DiaSpect Tm, Hemo Control, and VERI-Q Red) were evaluated against the Beckman Coulter as gold standard at four sites: Rakai Health Sciences Program clinic/Kalisizo General Hospital, Mbarara Municipal Council Health Center IV, Mpigi Health Center IV, and Atutur Hospital. Adult, consented participants aged ≥ 18 years, provided 4 mL of venous blood for concurrent Hb testing across the POC Hb devices. Anemia classification followed World Health Organization guidelines: <11.0 g/dl for pregnant women, < 12.0 g/dl for non-pregnant adult women, and < 13.0 g/dl for adult men. The Bland–Altman method was used to assess the agreement of Hb results (mean difference, standard deviation of differences, and limits of agreement). Diagnostic accuracy parameters (sensitivity, specificity, positive predictive value, negative predictive value, and accuracy) were calculated to evaluate the performance of each device compared to the reference machine.

Results

A total of 768 participants (192 from each site), comprising equal numbers of PLHIV and pregnant/postnatal women were recruited. Participants’ mean (SD) Hb values were 12.1 (± 2.2) g/dl. There were strong positive correlations (all r ≥ 0.9) between each POC Hb device and the Beckman Coulter. 

The mean difference between POC Hb devices and Coulter readings across sites was 0.06 for the HemoCue Hb 301, 0.16 for VERI-Q Red, 0.35 for Hemo Control, and 0.57 for Diaspect Tm, with the Diaspect Tm showing the broadest limits of agreement (−0.76,1.91). The DiaSpect Tm, Hemo Control, HemoCue Hb 301 and VERI-Q Red were at least 94% (79%), 88% (87%), 77% (91%) and 75% (76%) sensitive (specific), respectively, across the study sites.

Conclusion

All devices demonstrated clinically acceptable performance (within ± 1 g/dL of the reference hemoglobin), with the HemoCue Hb 301, DiaSpect Tm and Hemo Control performing most comparable to the standard, suggesting potential for interchangeable use.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-025-13354-9.

Background

Globally, anemia affects a significant portion of the population. Across all ages, the global prevalence of anemia was 24.3% in 2021, corresponding to 1.92 billion cases [1]. The most affected populations were children younger than 5 years, women, and countries in Sub-Saharan Africa and South Asia. In Uganda, anemia is a significant public health problem, particularly among children and women. It is caused by iron deficiency, parasitic infections, acute and chronic infections, and other nutritional deficiencies [2]. Hemoglobin (Hb) estimation is one of the most common tests in routine patient management at healthcare facilities, especially among pregnant/postnatal women and people living with HIV (PLHIV) who are more susceptible and vulnerable to clinical complications from anemia [3, 4]. In Uganda, Hb estimation is a routine test offered under the minimum healthcare package by the Ministry of Health (MOH) from Health Centre II to National Referral Hospital level and may be performed by either laboratory or clinical staff depending on test complexity. Several devices can be used for Hb estimation [5]but the standard recommended by the International Committee for Standardization in Hematology is the Cyanmethemoglobin (HiCN) method [6]. Although the HiCN tests almost all forms of Hb, it has some associated challenges i.e., it is highly manual, time-consuming, uses potassium cyanide which is toxic, the working solution is not stable for a prolonged period, and is liable to false readings due to errors in spectrophotometry [7, 8].

In routine settings, the World Health Organization (WHO) recommends use of the complete blood count (CBC) as the reference standard for assessing blood Hb concentration [9]. The Beckman Coulter ACT 5 Diff CP hematology analyzer is the gold standard for CBC testing, with excellent precision and accuracy when compared to the HiCN method [10]. However, the Coulter is expensive and only available at tertiary healthcare centers, large private healthcare facilities, and research centers in Uganda. Where CBC testing is not available, the WHO recommends onsite Hb testing with a Hb point-of-care (POC) device [9]. There are several devices on the market, some of which have not been validated in low- and middle-income settings. A few previous studies [11–16] suggest Hb POC devices are a reliable approach for Hb estimation. However, these findings may not be directly applicable to Uganda due to availability of additional un-validated devices in use and the fact that diagnostic performance may vary in different settings, as performance partly depends on disease prevalence [17]. Consequently, there is need for recommendations to guide POC Hb procurement for use in primary healthcare settings. This study determined the diagnostic and analytical performance, including operational suitability, of four commonly available POC Hb devices routinely used in HIV and maternity care in Uganda.

Methods

Study design and setting

This cross-sectional study recruited participants from four purposively selected sites: Rakai Health Sciences Program (RHSP)/Kalisizo General Hospital in southern Uganda, Mbarara Municipal Council Health Centre IV in the west, Atutur Hospital in the east, and Mpigi Health Centre IV in the central region. Selection of study sites was purposive to represent primary healthcare settings in different geographic regions of the country. Additionally, eligible sites had to have the reference device (Coulter) on site, since samples had to be analyzed within 24 h of collection.

Sourcing of the POC Hb devices

To obtain Hb POC devices for evaluation, two media adverts were run, two months apart, in two of the most read newspapers in the country (The New Vision and The Daily Monitor), inviting suppliers to submit their POC Hb devices for evaluation. In addition, known POC Hb device vendors were directly contacted by email. Four new POC device types were submitted and were returned to respective providers following the evaluation: HemoCue Hb 301 (HemoCue AB Kuvettgatan 1 SE-262 71 Ängelholm – Sweden), VERI-Q Red (MiCo BioMed Co., Ltd), DiaSpect Tm (DiaSpect Medical GmbH Ebendorfer Chaussee 3 39179 Barleben Germany) and Hemo Control (EKF-diagnostic GmbH Ebendorfer Chaussee 3 39179 Barleben Germany). All the devices were confirmed to have undergone factory calibration and had auto-calibration ability. Device technical specifications are summarized in Table 1.

Table 1.

POC Hb device technical specifications

	HemoCue Hb 301	DiaSpect Tm	Hemo Control	Veri -Q red
Dimensions (mm)	160 × 140 × 70	150 × 90 × 40	160 × 160 × 68	55 × 90 × 20
Weight (g)	500	180	700	51
Technology presentation	Handheld	Handheld	Tabletop	Handheld
Power	AC Power or 4 type AA batteries	Rechargeable lithium-ion battery	Rechargeable NiMH battery	2XAAA alkaline batteries
External equipment	Not required	Not required	Not required	Not required
Batching	Not possible	Not possible	Not possible	Not possible
Instrument connectivity	serial port connection to PC	USB and Bluetooth	USB and licensed Bluetooth	No connectivity ability
Data storage	No ability	500 results	4000 results	500 results
Reagent shelf life	> 12 months	> 12 months	> 12 months	> 12 months
IQC	Available	Available	Available	Available
Equipment cost	($760)	($427)	($577)	($86)
Reagent, consumable and control cost	<$2 per test AND <$100 per month for controls	<$2 per test AND <$100 per month for controls	<$2 per test AND <$100 per month for controls	<$2 per test AND <$100 per month for controls
Maintenance	By operator	By operator	By operator	By operator
Daily calibration	Auto-calibration	Auto-calibration	Auto-calibration	Calibration Code chip for specific strip Lot
Sample volume	~ 10 µl	~ 10 µl	8 µl	7 µl
Measurement range	0–25.6 g/dL	0.5–25.5 g/dL	0–25.6 g/dL	5–25 g/dL
Measurement time	< 10 s	60 s	< 60 s	5 s
Operating temperature	10–40˚C	10–35 °C	10–40˚C	10–40˚C
Operating relative humidity	5–90%	0-100%	10–85%	10–90%
Type of sample	Capillary/EDTA venous blood	Capillary/EDTA venous blood	Capillary/EDTA venous blood	Capillary/EDTA venous blood
Precise sample measurement	No manual pipetting of precise quantity required	No manual pipetting of precise quantity required	No manual pipetting of precise quantity required	Manual pipetting of precise quantity (7 µl) required

Reference machine

Two Beckman Coulter machine series were used as reference: ACT 5 Diff CP hematology analyzer at the RHSP/Kalisizo General Hospital study site and DxH 520 at the other three sites. Machine evaluation showed the Beckman Coulter to have excellent precision and accuracy for CBC parameters, with correlation coefficients > 0.95 for directly measured parameters, including hemoglobin [18]. Before the initiation of participant enrolment, we conducted an inter-machine performance comparison between the two series. Six College of American Pathologists (CAP) Proficiency Testing (PT) specimens were used. All were tested on the same day using the ACT 5 Diff CP and the DxH 520. Resultant Hb differences were all within the CAP stated allowable error (± 0.67). The same samples had also prior been tested by several laboratories participating in the CAP hematology PT scheme worldwide, with a summary report provided to each participating laboratory (including the RHSP lab). Some of the participating labs had used the ACT 5 Diff CP, others the DxH 520. From both a local and global perspective, the ACT 5 Diff CP consistently estimates hemoglobin with slightly higher values compared to the DxH 520, although the differences were all within the stated allowable error (± 0.67) (Table 2). Also, the differences between local Hb values using either machine and respective peer means were all within the stated allowable error (± 0.67). Thus far, both Beckman Coulter machine series perform comparably and can be used interchangeably. Therefore, we used both as reference for the evaluation of the POC Hb devices. Other reference machine checks carried out prior to participant enrolment included performing startups, reproducibility checks, checking for system logs (reagent, startup, error, and maintenance logs) and calibration. Additionally, room temperature charts were assessed, internal quality controls and external quality assessment (EQA) participation reviewed. Overall, all the facilities were generally compliant, with a few gaps which were addressed prior to initiating participant recruitment.

Table 2.

Performance comparison between Beckman Coulters ACT 5 diff CP and DxH 520

Sample ID#	Peer mean (Using ACT 5 Diff CP)	Local values (Using ACT 5 Diff CP)	Peer mean (Using DxH 520)	Local values (Using DxH 520)	Allowable Error
LN908R	1.08	1.1	0.92	0.98	± 0.67
LN909R	5.40	5.6	5.27	5.20	± 0.67
LN910R	9.39	9.7	9.46	9.07	± 0.67
LN911R	14.07	14.3	13.68	13.84	± 0.67
LN912R	17.67	17.8	17.41	17.53	± 0.67
LN913R	21.87	22.0	21.87	21.73	± 0.67

Participants, sample size determination, inclusion, and exclusion criteria

We included all interested pregnant or postnatal women, and people living with HIV aged 18 years and above who accessed care at the selected sites/facilities, and were able and willing to provide written informed consent. We excluded individuals who were too ill for a blood draw and those who could not provide written informed consent. An equal number of (i) pregnant/post-natal women (within 42 days after birth) aged ≥ 15 years and (ii) PLHIV aged ≥ 18 years were recruited from among individuals routinely accessing care at respective study sites. These participant categories were selected because anemia is common in these groups of individuals with accurate diagnosis and quantification of anemia crucial to guidance of treatment options. The sample size was determined based on a sensitivity/specificity accuracy index. We assumed each of the devices under evaluation to be at least 92% ±3% sensitive, 95% ± 2.5% specific, with an assay failure rate of < 10%, requiring 348 anemic, 334 non-anemic, and a total of 759 participants. Per study site, 48 HIV-positive anemic, 48 HIV-positive non-anemic, 48 pregnant/postnatal anemic, and 48 pregnant/postnatal non-anemic individuals were enrolled. Anemia status was ascertained by the Coulter device and categorization done basing on WHO ranges for pregnant women (< 11.0 g/dl), non-pregnant adult women (< 12.0 g/dl), and adult men (< 13.0 g/dl).

Enrolment at facilities

To minimize impact on staff time, the study was integrated within routine activities at respective recruitment sites. At each site, information on study eligibility was posted at both HIV antiretroviral therapy (ART) and maternity clinics. Facility staff (clinical officers, nurses, or midwives) who routinely attended to clients in respective clinics identified potential participants. Initially, participants with clinical signs/symptoms of anemia (identified at the client’s routine clinic visit) were prioritized for enrollment. Those with clinical signs/symptoms but found not to have anemia after the CBC test with the Beckman Coulter were eligible for the non-anemic arm. Once the target number of anemic cases was achieved, participants without clinical signs/symptoms of anemia were then enrolled. Symptoms of anemia included any of the following: easy tiredness (fatigue), palpitations, shortness of breath (dyspnea), headache, dizziness or light-headedness, cold hands and feet, leg cramps, or sleeplessness (insomnia). Clinical signs of anemia included pallor, tachycardia, cheilosis, icterus, hyperdynamic precordium, flow murmurs, splenomegaly, passage of fecal Blood, or peripheral neuropathy. Eligible individuals were approached for study participation, those willing provided written informed consent. Participants who declined to take part in the study continued to receive the standard of care as provided by the participating site. Enrolled participants responded to a brief questionnaire on sociodemographic characteristics and provided 4 ml of venous blood in Ethylene Diamine Tetra Acetic Acid (EDTA) vacuum tubes. The collected Blood specimens were transported to the site lab at room temperature in cool boxes without ice packs. Hb tests on respective blood specimens were concurrently performed across the four POC Hb devices within 24 h of collection following respective device manufacturer standard operating procedures. Existing standard of care and/or CBC results were used for treatment decisions, but not results from POC Hb devices. One hundred of the POC Hb device tests were repeated (25 at each study site) to assess the precision of each POC Hb device.

Data analysis

Statistical analyses were performed using STATA software version 15.0 (StataCorp LLC, College Station, TX, USA) and R 4.2.6. Normality of hemoglobin distribution was assessed using the Shapiro-Wilk test [19]whereas correlations between Hb values obtained using respective POC Hb devices (DiaSpect Tm, HemoCue Hb 301, Hemo Control, VERI-Q Red) and the Beckman Coulter used Pearson correlation coefficients (r). The Bland-Altman test [20] was performed to assess the level of agreement between respective POC Hb devices and Beckman Coulter readings. Means of differences (95%CI) in Hb (including limits of agreement) between devices under investigation and the reference machine were reported. Diagnostic accuracy parameters—sensitivity, specificity, Negative Predictive Value (NPV), Positive Predictive Value (PPV), and accuracy—were calculated to assess each device’s performance relative to the Beckman Coulter reference. The diagnostic accuracy of each device and site was also analyzed using receiver operating characteristic (ROC) curves.

Results

Participant characteristics

Between October 2021 and April 2022, 768 (192 per site) participants were enrolled with a mean (SD) age of 32.6 (12.7) years with the majority (55.5%) living with HIV. The majority were female (87.7%), 36.1% of whom were pregnant (Table 3). Overall, participants’ mean (SD) Hb was 12.1 (± 2.2), as determined using the Coulter reference machine, with a range of 2.7–18.5 g/dl (Table 3).

Table 3.

Study participants’ baseline characteristics

	ATUTUR	MBARARA	MPIGI	RAKAI	Overall
Age
Mean (SD)	36.1(16.5)	29.4(8.5)	32.8(11.5)	32.1(12.2)	32.6(12.7)
Median (IQR)	33	28	30	30	30(23–39)
Range (min, max)	74(15–89)	59(18–77)	62(16–78)	65(15–80)	74(15–89)
Sex
Female	158(82.7)	177(92.2)	171(89.1)	164(86.8)	670(87.7)
Male	33(17.3)	15(7.8)	21(10.9)	25(13.2)	94(12.3)
HIV status
Positive	110(57.6)	104(54.2)	107(55.7)	424(55.5)	424(55.5)
Negative	81(42.4)	88(45.8)	84(43.8)	339(44.4)	339(44.4)
Unknown	0(0)	0(0)	1(0.5)	1(0.1)	1(0.1)
Pregnancy status
Not, and did not give birth in last 42 days	62(39.2)	55(31.1)	70(40.9)	59(36)	246(36.7)
Not, gave birth within last 42 days	16(10.1)	33(18.6)	65(38)	59(36)	173(25.8)
Pregnant	80(50.6)	85(48)	34(19.9)	43(26.2)	242(36.1)
Don’t know or don’t want to say	0(0)	4(2.3)	2(1.2)	3(1.8)	9(1.3)
Hemoglobin measurements (Coulter) (g/dl)
Mean (SD)	11.8(2.02)	12.3(2.5)	12.2(2.1)	11.9(2.3)	12.1(2.2)
Median (IQR)	11.6	11.9	12.1	11.8	11.8
Range	10.5(6.07–16.6)	15.8(2.7–18.5)	12.4(6.1–18.5)	13.1(4.3–17.4)	15.8(2.7–18.5)

Device accuracy, precision, and distribution of differences in Hb estimates between the Beckman Coulter and POC Hb devices

Each measurement included in this set of comparisons is from an individual participant. All differences in Hb estimates between site Beckman Coulters and the respective POC Hb devices were normally distributed (Fig. 1). We observed very strong positive correlations between each POC Hb device and the Beckman Coulter (all Pearson correlation coefficients, r ≥ 0.9) (Table 4). Relatedly, areas under ROC curves for all devices across study sites (all AUC ≥ 0.9) were significantly high. At the Mbarara Municipal Council Health Center IV study site, four devices (HemoCue Hb 301-AUC = 0.966, Hemo Control-AUC = 0.962, DiaSpect Tm-AUC = 0.970 and VERI-Q Red-AUC = 0.954) exhibited comparable accuracy with almost similar observations at the RHSP (HemoCue Hb 301-AUC = 0.948, Hemo Control-AUC = 0.936, DiaSpect Tm-AUC = 0.953 and VERI-Q Red-AUC = 0.924) and Atutur Hospital (Fig. 2). Bland-Altman plots for Hb concentration comparisons between POC Hb devices and the Beckman Coulter are shown in Fig. 3A and E. Mean differences (95%CI) with limits of agreement between HemoCue Hb 301 and Coulter readings were 0.13 (−0.99, 1.24), −0.01 (−1.63, 1.61), 0.31 (−0.83, 1.46) and − 0.17 (−1.47, 1.13) at the Mbarara, Mpigi, RHSP and Atutur Hospital study sites, respectively, whereas for the DiaSpect Tm and Coulter readings, mean biases (limits of agreement) were 0.54 (−0.5, 1.58), 0.49 (−1.39, 2.36), 0.88 (−0.25, 2.02) and 0.43 (−0.74, 1.61) at the above respective study sites. The mean biases (limits of agreement) between Hemo Control and Coulter readings were 0.42 (−0.69, 1.53), 0.15 (−1.31, 1.61), 0.52 (−0.67, 1.72) and 0.32 (−0.9, 1.55) at the Mbarara, Mpigi, RHSP, and Atutur Hospital sites, respectively, whereas between VERI-Q Red and Coulter readings, mean differences (limits of agreement) were 0.0 (−1.66, 1.66), −0.12 (−2.02, 1.78), 0.81 (−0.74, 2.36) and − 0.03 (−1.83, 1.77) at the mentioned respective sites. With sites combined, all devices demonstrated excellent precision with insignificant mean differences. Respectively, standard errors were 0.024, 0.024, 0.043 and 0.034, using the HemoCue Hb 301, DiaSpect Tm, VERI-Q Red, and Hemo Control (Table 5).

Fig. 1.

Distribution of the differences in Hemoglobin estimates between site Beckman Coulters and respective POC Hb devices

Table 4.

Pearson correlation coefficients (r) for each POC Hb device across study sites

	Mbarara	RHSP	Mpigi	Atutur
HemoCue Hb 301	0.98	0.97	0.93	0.95
Hemo Control	0.98	0.97	0.94	0.96
DiaSpect Tm	0.98	0.97	0.91	0.96
VERI-Q Red	0.94	0.94	0.90	0.90

Fig. 2.

Device Non-parametric Receiver Operator Characteristic Curves across sites

Fig. 3.

Bland-Altman plots for POC Hb devices

Table 5.

Mean differences, and standard errors using respective POC Hb devices

	Mean difference (95% CI)	P-value	Standard Error
HemoCue Hb 301	−0.020(−0.069-0.028)	0.416	0.024
DiaSpect Tm	0.026(−0.021-0.073)	0.268	0.024
VERI-Q Red	0.020(−0.066-0.106)	0.643	0.043
Hemo Control	−0.003(−0.069-0.063)	0.925	0.034

POC Hb device validity

Table 6 presents the diagnostic accuracy parameters for anemia screening for each device compared to the Coulter at each site. At Atutur, the HemoCue Hb 301 demonstrated 77% sensitivity, 94% specificity, 92% PPV, and 81% NPV. The DiaSpect Tm showed 100% sensitivity, 81% specificity, 83% PPV, and 100% NPV. The Hemo Control had 97% sensitivity, 88% specificity, 88% PPV, and 97% NPV. The VERI-Q Red exhibited 78% sensitivity, 89% specificity, 87% PPV, and 81% NPV. Across all sites, the DiaSpect Tm had the highest sensitivity, while the HemoCue Hb 301 had the highest specificity.

Table 6.

Sensitivity, specificity, positive and negative predictive values of the POC Hb devices across the study sites

	DiaSpect Tm	Hemo Control	HemoCue Hb 301	VERI-Q Red
Atutur Hospital
Sensitivity (%)	100	97	77	78
Specificity (%)	81	88	94	89
Sensitivity + Specificity©	1.81	1.85	1.71	1.67
Positive Predictive Value (%) π	83	88	92	87
Negative Predictive Value (%) π	100	97	81	81
Mbarara Municipal Council Health Center IV
Sensitivity (%)	97	95	88	75
Specificity (%)	88	92	97	96
Sensitivity + Specificity©	1.85	1.87	1.85	1.71
Positive Predictive Value (%) π	88	91	96	94
Negative Predictive Value (%) π	97	96	90	82
Mpigi Health Center IV
Sensitivity (%)	94	88	82	80
Specificity (%)	79	94	94	94
Sensitivity + Specificity©	1.73	1.82	1.76	1.74
Positive Predictive Value (%) π	82	93	93	93
Negative Predictive Value (%) π	93	89	84	83
RHSP/Kalisizo General Hospital
Sensitivity (%)	100	99	99	98
Specificity (%)	80	87	91	76
Sensitivity + Specificity©	1.80	1.86	1.90	1.74
Positive Predictive Value (%) π	83	89	92	80
Negative Predictive Value (%) π	100	99	99	97

©For a test to be useful, sensitivity + specificity should be at least 1.5 (halfway between 1, which is not useful, and 2, which is perfect) [18]

^πThe positive and negative predictive values shown here are those for the enrolled study population, which was purposely selected

Discussion

A systematic review of anemia among Ugandan pregnant women reported a pooled prevalence of 30% with the majority of cases related to malaria and HIV [21]. Among children under five years of age, the prevalence is even higher at 50%[22]. Although Hb estimation is one of the tests recommended to be offered under the minimum healthcare package in Uganda, access is largely limited to big public and private healthcare facilities with automated hematology analyzers. The Ugandan MOH plans to expand Hb test access to the smallest healthcare facility unit through POC testing. There are several POC Hb devices on the Ugandan market. However, there is insufficient evidence on the most appropriate device(s) for use in Ugandan primary healthcare settings. We evaluated the performance of a range (four) of POC Hb devices (HemoCue Hb 301, DiaSpect Tm, Hemo Control and VERI-Q Red) in multiple geographical regions of Uganda using the Beckman Coulter as the gold standard method.

Device accuracy, precision, agreement, sensitivity, specificity, and positive and negative predictive values were determined. Overall, all POC Hb devices exhibited acceptable analytic and diagnostic performance characteristics. All four POC Hb devices are clinically useful for anemia testing with all sensitivity + specificity values across study sites significantly greater than 1.5[23]. The DiaSpect Tm demonstrated the highest diagnostic sensitivity (at least 94% sensitive) and the HemoCue Hb 301 the highest specificity (at least 91% specific) through all the sites (Table 6). Combined (sensitivity + specificity), the Hemo Control exhibited the best diagnostic performance characteristics with near-perfect values of 1.85, 1.87, 1.82, and 1.86 at Atutur Hospital, Mbarara Municipal Council Health Center IV, Mpigi Health Center IV, and RHSP/Kalisizo Hospital, respectively. Test strip failure (not quantified) occurred only with VERI-Q Red, requiring unexpected repeat testing.

All four POC Hb devices share several features including the use of both capillary and venous blood, do not require any external equipment to operate, internal quality controls are locally available, are compatible with commercial proficiency testing materials, and were factory calibrated against the HiCN method, with autocalibration ability. Other than the Hemo Control which is a table-top, all the other three devices are handheld. In addition to USB connection through a USB cable to a computer, the DiaSpect Tm and Hemo Control are Bluetooth enabled, although the Hemo Control’s Bluetooth requires a license to function. The VERI-Q Red is the cheapest, lightest, and least voluminous. However, it has no data connectivity ability, making data sharing impossible. It is also the only device that requires manual pipetting of precise specimen quantity (7 µl), making its Hb measurements error-prone, especially when used by inexperienced personnel. Among evaluated devices, the HemoCue Hb 301 has no data storage capacity, making data management challenging. The Hemo Control has the highest storage capacity of up to 4000 test results. It is, however the most voluminous, heaviest, and second most expensive (after HemoCue Hb 301).

This study had some limitations. First, we evaluated only four devices that were received following an advertisement and contact of known vendors. We acknowledge that there may be several more device brands on the market and in use. Second, we enrolled participants from only two population categories: pregnant/postnatal women and PLHIV. Other categories, such as children, were not included; therefore, our findings may not be generalizable to all subpopulations. Third, we did not cater for population differences in malaria endemicity across study sites, as we did not do any malaria testing. Fourth, we used two Beckman Coulter machine series as a reference at different sites. Much as differences in Hb readings using both machines were within allowable error during performance comparison, the measurement error between the two series adds to the measurement error in the evaluation. Fifth, we did not test for inter-operator variability, and lastly, we used EDTA venous blood; thus, the capillary mode, which is mostly used in routine POC testing, was not evaluated. This is important as some previous studies reported significant differences in hemoglobin concentration using venous versus capillary blood.

Conclusion

All evaluated devices exhibited clinically acceptable performance [24] characteristics with the HemoCue Hb 301, DiaSpect Tm, and Hemo Control performing comparably, suggesting potential for interchangeable use for anemia diagnosis at Ugandan primary healthcare facilities. The devices varied across a spectrum of features including equipment capital cost, dimensions, weight, connectivity, and data storage abilities among others. In the future, evaluation using capillary blood, in more diverse sub-population categories could be additive.

Abbreviations

ART

Antiretroviral therapy

CAP

College of american pathologists

CBC

Complete blood count

EQA

External quality assessment

Hb

Hemoglobin

HiCN

Cyanmethemoglobin

HIV

Human immunodeficiency virus

IQR

Interquartile range

NPV

Negative predictive value

PLHIV

People living with hiv

POC

Point of care

PPV

Positive predictive value

PT

Proficiency testing

RHSP

Rakai health sciences program

SD

Standard deviation

WHO

World Health Organization

Authors’ contributions

E.N.K, C.S, R.M, F.N, G.K, G.N, C.O, W.N, J.N, and S.N developed the protocol. C.S., R.M.G, S.J, R.M, J.K, C.O, J.N, C.M, M.N, G.K, and F.N implemented the study. C.S. and H.N. wrote the manuscript draft. All authors reviewed the manuscript.

Funding

This publication has been supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the U.S. Centers for Disease Control and Prevention (CDC) under the terms of 6 NU2GGH002009-04-05 and in part (SJR, TCQ) by the Division of Intramural Research, National Institute of Allergy, and Infectious Diseases.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Ethics approval and consent to participate

This study was reviewed and approved by Uganda Virus Research Institute, Research and Ethics Committee (Ref. GC/127/720), Uganda National Council for Science and Technology under # HS437ES, as well as CDC and was determined not to be research, and was conducted consistent with applicable federal law and CDC policy 45 C.F.R. part 46.102(l)(2), 21 C.F.R. part 56; 42 U.S.C. § 241(d); 5 U.S.C. § 552a; 44 U.S.C. § 3501 et seq.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.