Non-Invasive Haemoglobinometer for Haemoglobin Screening in Blood Donors

NON-INVASIVE HAEMOGLOBINOMETER FOR HAEMOGLOBIN SCREENING IN BLOOD DONORS

HEALTH TECHNOLOGY ASSESSMENT SECTION (MaHTAS) HEALTH TECHNOLOGYMEDICAL DEVELOPMENT ASSESSMENT DIVISION SECTION

MEDICAL DEVELOPMENT DIVISION

MINISTRY OF HEALTH1 MALAYSIA 001/2018

DISCLAIMER Technology review is a brief report, prepared on an urgent basis, which draws on restricted reviews from analysis of pertinent literature, on expert opinion and / or regulatory status where appropriate. It has been subjected to an external review process. While effort has been made to do so, this document may not fully reflect all scientific research available. Additionally, other relevant scientific findings may have been reported since completion of this review.

Please contact: [email protected], if you would like further information.

Health Technology Assessment Section (MaHTAS), Medical Development Division Ministry of Health Malaysia Level 4, Block E1, Precinct 1 Government Office Complex 62590 Putrajaya

Tel: 603 88831229

Fax: 603 8883 1230

Available at the following website: http://www.moh.gov.my

Prepared by: Ros Aziah Binti Mohd Rashid Senior Assistant Director Health Technology Assessment Section (MaHTAS) Medical Development Division Ministry of Health Malaysia

Information Specialist: Tuan Abdul Hafiz bin Abdul Hamid Medical Officer Assistant Health Technology Assessment Section (MaHTAS) Medical Development Division Ministry of Health Malaysia

Reviewed by: Dr Junainah Binti Sabirin Public Health Physician Deputy Director Health Technology Assessment Section (MaHTAS) Medical Development Division Ministry of Health Malaysia

External reviewers: Dr Norasrina Binti Ishak Transfusion Medicine Specialist National Blood Centre Ministry of Health Malaysia

Dr Nik Rus Mazeni Bt Nik Yusoff Consultant Pathologist (Haematology) Hospital Kuala Lumpur

DISCLOSURE

The author of this report has no competing interest in this subject and the preparation of this report is totally funded by the Ministry of Health, Malaysia.

EXECUTIVE SUMMARY

Background Information provided by 164 countries to the World Health Organisation (WHO) Global Database on Blood Safety indicates that, worldwide, more than 92 million blood donations are collected annually. At least 13 million prospective donors are deferred from donating blood due to anaemia, existing medical conditions or the risk of infections that could be transmitted through transfusion.

The aim of haemoglobin (Hb) screening is to ensure that the prospective donor is not anaemic and also minimizes exclusion of healthy donors. Haemoglobin screening safeguards anaemic individuals from donating blood and also protects returning donors from donation-induced iron deficiency (DIID), the depletion of iron stores by repeated donations. Collecting a unit of blood from a donor with a normal Hb level also provides good quality blood components, with adequate and consistent Hb content in the collected blood.

The gold standard for pre-donation Hb screening is the measurement of venous whole blood obtained via venipuncture in a cell counter analyser. This method has a high sensitivity and specificity; however, it is invasive and time consuming and requires an additional venipuncture. Hence, point-of-care testing (POCT) methods are widely applied for rapid pre-donation Hb screening. The POCT devices for hb measurement use either capillary invasive or non-invasive methods. The capillary blood (also known as fingerstick/ fingerprick) is obtained after lancing a fingertip, is an invasive method and offers only minimal risks to the donor. However, there is still blood borne exposure between the person performing the test and the donor, the need to use alcohol pads, lancets and gloves and the risk of tenderness and bruising in the fingertips. All this make non- invasive methods an appealing options. However, whether the sensitivity and accuracy of the non-invasive method are comparable to the capillary is still questionable.

This technology review aims to evaluate evidence regarding non-invasive method following a request from the Medical Transfusion Specialist, Hospital Sultanah Nur Zahirah, Ministry of Health.

Objective/aim The objective of this technology review was to evaluate the effectiveness, safety, cost-effectiveness, organisational and social issues of the non-invasive haemoglobinometer for haemoglobin screening in blood donors.

Results and conclusions A total of 1383 records were identified through the Ovid interface and PubMed, and 18 were identified from other sources (references of retrieved articles). There were nine articles included in this review comprised of diagnostic accuracy studies. The studies were conducted in Korea, India, Italy, Spain, Netherlands, Belgium, German and Brazil. The effectiveness of the included studies is as shown in Table 1.

Table 1: Pertinent details of included diagnostic accuracy studies Study / population Cut-off value Non-invasive Invasive (capillary) Gold Brand name Diagnostic accuracy Brand name Diagnostic accuracy standard Kim MJ et al. 2013 Whole blood NBM 200 Sensitivity : 38.6% Hemocue Sensitivity : 42.7% Beckman N=506 ≥12.5g/dL Specificity : 93.6% Specificity : 98.6% Coulter Apheresis ICC : 0.69 ICC : 0.86 LH500 ≥12.0 g/dL R2 : 0.47 R2 : 0.76

Singh A et al. 2015 Whole blood NBM 200 Bias : -0.66g/L Diaspect Bias : -0.18g/L Sysmex N=534 ≥ 12.5g/dL Sensitivity : 71.7% Sensitivity : 98.1% KX-21 Specificity : 79.5% Specificity : 78.4% NPV : 95.8% NPV : 99.7% PPV : 30.2% PPV : 35.9% ICC : 0.43 ICC : 0.78 CV : 4.28 CV : 2.19

Hemocontrol Bias : -0.22g/L Sensitivity : 86.8% Specificity : 94.7% NPV : 98.3% PPV : 66.9% ICC : 0.77 CV : 2.51

Bellardinelli A et al. Male ≥ 13.5g/dL NBM 200 Sensitivity : 98% Hemocue Sensitivity : 99% Beckman 2013 Female ≥12.5g/dL Specificity : 97% Specificity : 99.5% Coulter N=1008 AcT-5 Pronto 7 Sensitivity : 93% Specificity : 83%

Herraiz ALP et al. NA NBM 200 Sensitivity : 94.34% Hemocue Sensitivity : 96.23% Beckman 2015 Specificity : 97.48% Specificity : 99.37% Coulter N=129 r : 0.648 r : 0.859 LH750 Pronto 7 Sensitivity : 91.19% Specificity : 98.74% r : 0.618 Haemospect Sensitivity : 93.08% Specificity : 98.74% r : 0.416 Baart AM et al. Male ≥ 13.5g/dL NBM 200 Mean difference Hemocue MD:0.24 Sysmex 2016 Female ≥12.5g/dL (MD): 0.10 201 correlation:0.77(m)/ XT-188i N=2589 correlation:0.42(m)/ 0.75(f) 0.36 (f) sensitivity:30.1% sensitivity: 12.1% specificity:99.8% specificity:98.6% PPV:94.4% PPV:43.8% NPV:94.1% NPV:92.4% Haemospect MD: -0.06 Hemocue MD:0.35 Correlation:0.03(m)/ 301 correlation:0.75(m)/ 0.16 (f) 0.71(f) sensitivity:3.5% sensitivity:27.9% specificity:95.9% specificity:99.4% PPV:7.1% PPV:79.4% NPV:91.8% NPV:94.1%

De Clippel D et al. Male ≥ 13.5g/dL Haemospect Bias: -2.3 g/dL Compolab Bias: 3.4g/dL Sysmex 2017 Female ≥12.5g/dL Classic Correlation: 0.48 Correlation: 0.78 XN-9000 N=1483 Haemospect Bias:-1.5g/dL alternative Correlation:0.46

Sunming A et al. Pronto 7 Bias : -5.9g/L Hemocue Bias: 4.3g/L Advia 2015 Sensitivity : 63.2% Sensitivity : 23.1% 2120i N=610 Specificity : 76.2% Specificity : 99.2% NPV : 94.3% NPV : 90.6% PPV : 24.8% PPV : 78.9% CV : 2.73 CV : 3.23 r : 0.59 r : 0.79 Pinto M et al. 2013 NA NBM 200 Median: 14.6g/dL Hemocue Median: 15.2g/dL Abbott N=205 (different: 0.2g/dL) (different: 0.8g/dL) CellDyn

Ardin S et al. 2015 Male ≥ 13.5g/dL NBM 200 Bias: -0.12 g/dL Compolab Bias: -0.53g/dL Sysmex N=351 Female ≥ Accordance: 91.7% Accordance:92.8% KX-21 12.5g/dL Pronto 7 Bias: -0.50g/dL Accordance: 86.4%

Haemospect Bias: -0.22 g/dL Accordance: 88.4% NA= not available

Effectiveness There was fair level of evidence retrieved to suggest that non-invasive haemoglobinometer had lower sensitivity, lower specificity and lower correlation with reference standard compared to invasive method. However, there was one study reported that sensitivity of non-invasive method was higher compared to invasive method [63.2% vs 23.1%]. The non-invasive haemoglobinometer also shown to have lower accordance or agreement with the gold standard compared to invasive method. The PPV was also lower compared to invasive method. However, the NPV for non-invasive haemoglobinometer was inconclusive.

Safety In terms of safety, there was no evidence retrieved on the adverse events. However, non-invasive haemoglobinometer had lower ability to detect low Hb levels in first time and regular donor compared to capillary invasive method as reported in one study.

Cost-effectiveness There was no retrievable evidence on the cost-effectiveness on non-invasive haemoglobinometer.

Organisational Issue There was limited fair level of evidence retrieved to suggest that non-invasive haemoglobinometer required less than 30 to 45 minutes to train new operator, had fast measurement (however, the whole process of pre-donation screening was time consuming due to inability of patients to move). The measurement was also interfered with the patients’ condition such as skin colour, cold finger, artificial fingernail, nail polish and etc.

Psychological/social There was limited fair level of evidence retrieved to suggest that non-invasive haemoglobinometer was preferred by the blood donors. Among the non-invasive

6 haemoglobinometer, Haemospect and Pronto 7 rated as the most usable and satisfied by operator.

Methods Electronic databases were searched through the Ovid interface: Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE® Daily and Ovid MEDLINE® 1946 to Present, EBM Reviews - Cochrane Central Register of Controlled Trials - October 2017, EBM Reviews - Cochrane Database of Systematic Reviews - 2005 to October 2017, EBM Reviews - Health Technology Assessment – 4th Quarter 2017, EBM Reviews – NHS Economic Evaluation Database 4th Quarter 2017. Searches were also run in PubMed database and U.S. Food and Drug Administration (USFDA) website. Google and Google Scholar was also used to search for additional web-based materials and information. Additional articles were identified from reviewing the references of retrieved articles. Last search was conducted on 13th March 2018.

NON-INVASIVE HAEMOGLOBINOMETER FOR HAEMOGLOBIN SCREENING IN BLOOD DONORS

1. BACKGROUND Information provided by 164 countries to the World Health Organisation (WHO) Global Database on Blood Safety indicates that, worldwide, more than 92 million blood donations are collected annually. At least 13 million prospective donors were deferred from donating blood due to anaemia, existing medical conditions or the risk of infections that could be transmitted through transfusion.1

A guideline by WHO recommended that Hb level of not less than 12.0 g/dl for females and not less than 13.0 g/dL for males as the threshold for whole blood donation. However, in Malaysia, the practice is Hb level must be between 13.5g/dL to 18.0g/dL for male and 12.5g/dL to 18.0g/dL for female. 2

The venous Hb level of a donor drops by 1 to 1.5g/dL after blood donation. Thus a screening test should be sensitive enough to detect donors with pre-existing anaemia or who may rendered anaemic after blood donation. Since every single blood matters for blood bank, the screening test should also be able to avoid unnecessary deferrals, should be smoothly incorporable into operational practices and should have a reasonable cost. Donor discomfort should be minimum, handling should be easy and should be portable so as to be applicable in mobile setting.3

The gold standard for pre-donation Hb screening is the measurement of venous whole blood obtained via venipuncture in a cell counter analyser. This method has high sensitivity and specificity; however, it is invasive and time consuming and requires an additional venipuncture. Hence, point-of- care testing (POCT) methods are widely applied for rapid pre-donation Hb screening.4 The POCT devices for hb measurement use either capillary invasive or non-invasive methods.5 The capillary blood (also known as fingerstick/ fingerprick) is obtained after lancing a fingertip, is an invasive method and offers only minimal risks to the donor. However, there is still blood borne exposure between the person performing the test and the donor, the need to use alcohol pads, lancets and gloves and the risk of

tenderness and bruising in the fingertips. All this make non-invasive methods an appealing options. However, whether the sensitivity and accuracy of the non-invasive method are comparable to the capillary is still questionable.6

In Malaysia, Hb screening in blood donors is performed using several invasive methods such as copper sulphate gravimetric and spectrophotometric method (eg: Hemocue, Diaspect). The copper sulphate test to determine Hb proved unreliable due to variability in the droplet of capillary blood obtained and has largely been replaced by a spectrophotometric test.7

This technology review aims to evaluate evidence regarding non-invasive method following a request from the Medical Transfusion Specialist, Hospital Sultanah Nur Zahirah, Ministry of Health.

2. OBJECTIVE / AIM The objective of this technology review was to evaluate the effectiveness, safety, cost-effectiveness, organisational and social issues of the non- invasive haemoglobinometer for haemoglobin screening in blood donors.

3. TECHNICAL FEATURES- NON INVASIVE METHOD Presently, there are three technologies that use spectrophotometry for Hb measurement, differing in the type of sensor. They are:

i. Occlusion spectroscopy A portable device operative via a ring-shaped sensor fitted on the donor’s finger. The pneumatic cuff applies pressure and temporarily stops the blood flow. It creates an optical signal and yield a high signal-to-noise ratio. Optical elements in multi-wavelength sensor perform a measurement of the light transmitted through the finger. The wavelengths used is between 600 and 1500 nm. The differential in light absorption before and after blood flow obstruction in the finger is calculated to determine the Hb level. Example of the occlusion spectrophotometry is NBM 200.3

Figure 1: A ring shaped sensor applied over the thumb occludes the local blood flow temporarily and the sensor measures the light transmitted through the finger spectrophotometrically for occlusion spectroscopy method. 9

ii. Pulse CO-oximetry The technology places a multi-wavelength pulse CO-oximetry sensor over the individual’s fingertip. It acquires blood constituent data based on light absorption through a finger probe. The basic operation principle of this instrument is the differential absorption of multiple wavelengths of visible light (using spectrophotometry) to distinguish between oxyhemoglobin (oxygenated blood), deoxyhemoglobin (non-oxygenated blood), carboxyhemoglobin (carbon monoxide blood), methemoglobin (blood with oxidized hemoglobin) and other constituents of blood plasma. The amount of blood in the tissue fluctuates with the patient's pulse (photo- plethysmography) and therefore changes the amount of light absorbed. Data are obtained by sending several infrared light beams (500 to 1300nm) through the capillary surface at the tip of the patient's finger and measuring the changes in light absorption during the pulsatile blood cycle. The maximum radiant power of the strongest light beam is ≤25mW. The detector receives the light and then turns it into an electronic signal and sends it to the instrument to perform the calculations. The signal from the sensor is transformed by Masimo Rainbow SET technology to calculate oxygen saturation, pulse and total Hb concentration (SpHb [g/dL]) in the patient. The SpHb measurement relies on a calibration equation of multiple wavelengths which quantifies the percentage of total blood Hb. Example of the occlusion spectrophotometry is Pronto-7 7

Figure 2: The Ponto 7 utilises a sensor with various light emitting diodes (LED) that pass light through the sites to multiple multiple photodiodes (detectors). iii. Transcutaneous reflection spectroscopy It is a handheld device that uses a button sensor that adheres to the palm side of the finger of the non-dominant hand. A sensor head placed on the skin and projects a white light into the underlying tissue. Some of the projected light is absorbed by the various components of tissue, while some of it is reflected back to the device. The spectrometer breaks the light down into its

separate wavelengths. An electronic evaluation unit connected to the system to analyse the quantitative Hb value. Example of the occlusion spectrophotometry is Haemospect3

Figure 3: Images of the Haemospect (based on transcutaneous reflection spectroscopy.

4. METHODS 4.1. Searching Electronic databases were searched through the Ovid interface: Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE® Daily and Ovid MEDLINE® 1946 to Present, EBM Reviews - Cochrane Central Register of Controlled Trials - October 2017, EBM Reviews - Cochrane Database of Systematic Reviews - 2005 to October 2017, EBM Reviews - Health Technology Assessment – 4th Quarter 2017, EBM Reviews – NHS Economic Evaluation Database 4th Quarter 2017. Searches were also run in PubMed database and U.S. Food and Drug Administration (USFDA) website. Google and Google Scholar was also used to search for additional web-based materials and information. Additional articles were identified from reviewing the references of retrieved articles. Last search was conducted on 13th March 2018.

Appendix 1 showed the detailed search strategies.

4.2. Selection A reviewer screened the titles and abstracts against the inclusion and exclusion criteria and then appraise the full text articles for final article selection.

The inclusion and exclusion criteria were:

Inclusion criteria Population Blood donors Interventions Non-invasive haemoglobinometer Comparators Invasive haemoglobinometer (using capillary/ fingerstick/ fingerprick blood), gold standard (venous blood on haematology analyser) Outcomes Sensitivity, specificity, positive predictive value (NPV), positive predictive value (PPV), correlation, agreement, bias, cost-effectiveness, safety, organisational issue and social issue Study design Diagnostic accuracy study, Systematic Review (SR), Health Technology Assessment (HTA), economic evaluation study English full text articles

Exclusion criteria Study Randomised Controlled Trial (RCT) or non-randomised design controlled trial, cohort study, pre- and post-intervention study , cross sectional study, case series, case report, studies conducted in animals, narrative reviews Non English full text articles

Relevant articles were critically appraised using Critical Appraisal Skills Programme (CASP) and graded according to the NHS Centre for Reviews and Dissemination (CRD) University of York, Report Number 4 (2nd Edition) (Appendix 2). Data were extracted and summarised in evidence table as in Appendix 3.

5. RESULTS AND DISCUSSION A total of 1383 records were identified through the Ovid interface and PubMed, and 18 were identified from other sources (references of retrieved articles). After removal of 364 irrelevant or duplicates, 1037 records were screened and 989 were excluded. Of these, 53 relevant abstracts were retrieved in full text. After reading, appraising and applying the inclusion and exclusion criteria to the 53 full text articles, nine full text articles were included and 44 full text articles were excluded. The articles were excluded due to irrelevant study design (n=31), irrelevant population (n=8), irrelevant outcome (n=5). Flow chart of study selection is shown in figure 2.

There were nine full text articles which comprised of diagnostic accuracy studies finally selected for this review. The studies were conducted in Korea, India, Italy, Spain, Netherlands, Belgium, German and Brazil

Number of records identified Number of additional records through electronic databases identified from other sources searching (n=1383) (n=18)

Number of records after duplicates removed (n= 1037)

Number of records Number of records screened (n= 1037 ) excluded (n=989)

Number of full-text Number of full-text articles assessed articles excluded for eligibility (n= 53) (n=44) with reasons: - Irrelevant study design (n=31) - I rrelevant population (n= 8 ) Number of full-text articles - Irrelevant outcome(n=5)

included in qualitative

synthesis (n=9)

Figure 4: Flow chart of study selection

Risk of bias One of the tools that are being used by MaHTAS to assess the risk of bias is the CAPS checklist which consist of eight critical appraisal tools designed for SR, RCT, cohort studies, case control studies, economic evaluations, diagnostic accuracy studies, qualitative studies and clinical prediction rule. This is achieved by answering a pre-specified question of those criteria assessed and assigning a judgement relating to the risk of bias as either ―Yes‖ indicates low risk of bias, ―No‖ indicates high risk of bias, and ―Can’t tell‖ indicates unclear or unknown risk of bias.

Figure 5 shows risk of bias of the nine diagnostic accuracy studies included. All studies were unclear regarding the blinding and disease status of the population.

–

Criteria assessed

hodsperforming the test

nappropriate reference

Was there comparison a with a standard? patients Didall get the diagnostictest andreference standard? Couldof result the test have influenced been by the results ofthe reference standard? blinding Disease statusof tested populationdescribed? clearly Met describedsufficient in detail? followed Protocol Kim MJ et al. 2015 + + ? ? + Singh A et al. 2015 + + ? ? + Bellardinelli A et al. 2015 + + ? ? + Herraiz ALP et al. 2015 + + ? ? + Baart AM et al. 2016 + + ? ? + De Clippel D et al. 2017 + + ? ? + Sunming A et al. 2015 + + ? ? + Ardin S et al. 2015 + + ? ? + Pinto M et al. 2013 + + ? ? +

Figure 5: Assessment of risk of bias of diagnostic study (CASP)

5.1. EFFECTIVENESS Kim MJ et al. conducted a diagnostic accuracy study on 506 prospective blood donors to evaluate the accuracy of the non-invasive Hb sensor and compared it’s measurement with in-used portable haemoglobinometer. The Hb was measured using a non-invasive Hb sensor (NBM 200), a capillary invasive (fingerprick) haemoglobinometer (Hemocue) and automated hematology analyser (LH 500). The correlations between Hb measurement taken by NBM 200 and Hemocue with those by an automated hematology analyser were assessed using intra class coefficients (ICC). The cut-off for haemoglobin value was ≥12.0 g/dL for apheresis and ≥12.5 g/dL for whole blood donation. Haemoglobin measurements were compared among three different Hb level groups. They reported that, mean Hb measurements for NBM 200 was 14.1g/dL (1.4 SD) and Hemocue was 14.3 g/dL (1.4 SD). Sensitivity (ineligible to donate whole blood) and specificity (eligible to donate whole blood) for NBM 200 was 38.6% (95% CI: 28.1, 50.3) and 93.6% (95% CI: 90.9, 95.5) respectively. Meanwhile sensitivity and specificity for Hemocue was 42.7% (95% CI: 31.6, 54.5) and 98.6% (95% CI: 97.0, 99.4) respectively. The ICC was lower for NBM 200 compared to Hemocue [ICC NBM 200 was

0.69 (95% CI: 0.64, 0.73) and Hemocue was 0.86 (95% CI: 0.79, 0.90)]. Coefficient of determination was also lower for NMB 200 compared to Hemocue [R2 for NBM 200 was 0.47 and R2 for Hemocue was 0.76]. The authors concluded that the possibility to judge to be eligible for donors who are ineligible to donate was substantial when using NBM-200. Even though the NBM-200 has the apparent advantage of non-invasiveness, its use in pre-screening should be given meticulous attention. Since pre- donation testing is crucial to protecting donors’ health, complete evaluation of the instrument should be performed prior to use.9, level 3

Another diagnostic accuracy study was conducted by Singh A et al. to evaluate the suitability of non-invasive and invasive POCT devices for blood donor Hb screening. A total of 534 prospective blood donors’ Hb was measured by a non-invasive spectrophotometric based method (NBM 200) and capillary invasive (fingerprick) methods (Diaspect and Hemocontrol). The cut-off for haemoglobin value was ≥12.5 g/dL for whole blood donation. Performance of the devices were compared with the reference method (KX-21). They found that the non-invasive method had lowest sensitivity compared to the other two capillary invasive methods [sensitivity for NBM 200, Haemocontrol and Diaspect was 71.7%, 86.8% and 98.1% respectively]. However, the specificity of NBM 200 was comparable to Diaspect but lower when compared to Hemocontrol [Specificity for non-invasive (NBM 200), capillary invasive (Diaspect and Hemocontrol) was 79.5%, 78.4% and 94.7% respectively]. Negative predictive value (NPV) were lowest for NBM 200 (30.2%) followed by Diaspect (35.9%) and Hemocontrol (66.9%). Meanwhile positive predictive value (PPV) for NBM 200 (95.8%) was comparable to the other invasive methods; Diaspect (99.7%) and Hemocontrol (98.3%). The ICC against reference method for non-invasive was lowest among all [ICC for NBM 200, Hemocontrol and Diaspect was 0.43, 0.77 and 0.78 respectively]. The non-invasive also had least precision among all [coefficient of variation (CV) for NBM 200, Hemocontrol and Diaspect was 4.28, 2.51 and 2.19 respectively]. The authors concluded that the non-invasive NBM- 200 device has the advantage of inflicting less pain and hence being more acceptable to donors, but it lacks the desired sensitivity. This test is not effective in excluding ineligible donors which may pose a threat to donors' health. DiaSpect gives fast and accurate results and suited for screening donors in blood banks.10, level 3

Belardinelli A et al. conducted a diagnostic accuracy study to evaluate the accuracy of rapid, non-invasive methods for Hb screening in blood donors and to assess whether they had the potential to enhance recruitment and retention of blood donors. Consecutive blood donors were tested before routine donation procedure by non-invasive device in separate occasions [NBM 200 (n=445) and Masimo Pronto 7 (n=463)]. Both trials were compared to capillary (fingerstick) sample testing (Hemocue). Gold

15 standard used was venous sample testing on cell counter analyser (Beckman Coulter AcT-5). The cut-off for haemoglobin value was ≥12.5 g/dL for female and ≥13.5 g/dL for male. A survey questionnaire about compliance with the non-invasive pre-donation screening methods and their preferences was also administered to 350 donors. From the study they found that bias of measurement for Hemocue, NBM 200 and Pronto 7 was 0.83 g/dl [0.70 SD; (95% limit of agreement (LOA):-0.54, 2.20)], 0.29 g/dl [0.98 SD; (95% LOA:-1.64, 2.21)] and -0.53 g/dl [1.04 SD; (95% LOA:- 2.57, 1.51)] respectively. The sensitivity and specificity for Hemocue was highest among all (99% and 99.5% respectively) compared to NBM 200 (98% and 97% respectively) and Pronto 7 (93% and 83% respectively). The authors concluded that analysis of fingerstick or by non-invasive method did not replicate the results of cell counter analysis of venous sample. Compared with fingerstick, non-invasive haemoglobin test eliminate pain and reduce stress, but have lower level of sensitivity and specificity.11, level 3

Herraiz ALP et al. conducted a diagnostic accuracy study to analyse three non-invasive methods (NBM 200, Haemospect and Pronto 7) and to compare the methods with invasive capillary invasive (fingerstick) (Hemocue 301). Reference standard used was venous sample on haematology analyser (Coulter LH750). Total of 129 blood donors were recruited in the study. Hemoglobin was measured with capillary invasive test and all three non-invasive devices. Venous blood sample was also taken to be tested on reference standard. Questionnaire was distributed to the operators who used the non-invasive method to evaluate the satisfaction [scale 0 (unsatisfactory) and 5 (excellent)]. From the study they found that sensitivity and specificity for invasive capillary was higher (96.23% and 99.37% respectively) compared to non-invasive methods (NBM 200: 94.34% and 97.48% respectively, Pronto 7: 91.19% and 98.74% respectively, Haemospect: 93.08% and 98.74% respectively). The correlation index for capillary invasive was also higher (r=0.859) compared to non-invasive methods [NBM 200 (r=0.648), Pronto 7 (r=0.618) and Haemospect (r= 0.416)]. The authors concluded that the value of Hb with Hemocue showed the best correlation with LH 750 and lower for the rest, although acceptable. Especially with pronto 7 and NBM 200. The three devices could be used as screening test for blood. However, it’s cut off should be analysed to avoid the high percentage of false rejection.12, level 3

Another diagnostic accuracy study was conducted by Baart et al. to assess the precision and practicability of non-invasive devices and compared them with measurements in invasive capillary (fingerstick) and venous samples. A total of 2589 blood donors were included in the study. Haemoglobin levels were measured with the non-invasive devices (NBM 200 and Haemospect) and compared with values obtained with the invasive devices (Hemocue 201 and Hemocue 301) and venous values.

The cut-off for haemoglobin value was ≥12.5 g/dL for female and ≥13.5 g/dL for male. Precision was evaluated in terms of sensitivity and specificity for detection Hemoglobin levels below the donation cut-off level. From the study they found that Hb level measured with all (both invasive and non-invasive devices) differed significantly from venous measurement (p<0.05) [mean difference with venous Hb (mmol/L ±SD) for Hemocue 201, Hemocue 301, NBM 200 and Haemospect was 0.24 (±0.42), 0.35 (±0.44), 0.10 (±0.60) and -0.06 (±0.73) respectively]. Correlation with the reference standard for invasive devices was also higher compared to non- invasive devices [correlation for Hemocue 201, Hemocue 301, NBM 200 and Haemospect was r=0.77, r=0.75, r=0.42 and r=0.03 respectively for men and r=0.75, r=0.71, r=0.36 and r=0.16 respectively for women]. Sensitivity and specificity for invasive methods were higher (Hemocue 201: 30.1% and 99.8% respectively, Hemocue 301: 27.9% and 99.4% respectively) compared to non-invasive methods (NBM 200: 12.1% and 98.6% respectively, Haemospect: 3.5 % and 95.9% respectively). Positive and negative predictive value was also higher for invasive method (Hemocue 201: 94.4% and 94.1% respectively, Hemocue 301: 79.2% and 94.1% respectively) compared to the non-invasive method (NBM 200: 43.8% and 92.4% respectively, Haemospect: 7.1 % and 91.8% respectively). However, the significance in the values between those methods were not mentioned. The authors concluded that measurements with invasive devices showed stronger correlation and better agreement with venous values than non-invasive devices. Sensitivity was low for all measurement devices, particularly the non-invasive ones. 13, level 3

De Clippel D et al. conducted a diagnostic accuracy study to assess the ability of two point of care testing (POCT) devices to detect low Hb levels in new donors and to identify recovered Hb levels in regular donors with decreased levels at the previous donation. In each study arm, the Hb results measured with POCT device (invasive capillary and non-invasive) were compared with the venous Hb complete blood count (CBC) as a gold standard. The cut-off for haemoglobin value was ≥12.5 g/dL for female and ≥13.5 g/dL for male. Both a capillary invasive Hb (Compolab) and a non-invasive spectroscopy Hb (Haemospect) were tested. A total of 1483 blood donors were enrolled in three study arms [Arm 1: Compolab versus venous, Arm 2: Haemospect (classic) versus venous, Arm 3: Hemospect (alternative) versus venous]. In Haemospect (classic) group Digiclip was positioned at the middle finger to ensure that the fiber optics were in the center of the terminal phalanx. Meanwhile, in Haemospect (alternative) group Digiclip was positioned more proximally over the terminal phalanx with the fiber optics close to the distal inter phalangeal fold. They found that, the non-invasive Haemospect device showed the smallest fixed bias {Compolab : bias of 3.4 [9.2 g/L SD,( 95% LOA: -21.4 , 14.6 g/L)], Haemospect Classic : bias of -2.3 [12.1 g/L SD, (95% LOA: -26.0 , 21.4 g/L)] and Haemospect Alternative: bias of -1.5 g/L [12.9 g/L SD, (95%

LOA: -26.7, 23.7 g/L)]}. The highest correlation was reported in the capillary invasive Compolab study arm (r=0.78) compared to non-invasive Haemospect (classic) study arm (r=0.48) and non-invasive Haemospect (alternative) study arm (r=0.46). Capillary invasive Compolab had highest true pass rate, true deferral rate and accuracy compared to Haemospect. In terms of donor safety, capillary invasive (Compolab) had higher ability to detect low Hb levels in first time and regular donor compared to non- invasive method as shown in figure 6. The authors concluded that there was a stronger correlation between invasive devices with venous values compared to values obtained with non-invasive devices. 14, level 3

Figure 6: Test performance of different POCTs (%)

Sunming A et al. conducted a diagnostic accuracy study to assess the accuracy of the non-invasive hemoglobinometer (Masimo Pronto 7) and the capillary invasive haemoglobinometer (Hemocue 301) in comparison to the gold standard (Advia 2120i) and to analyse precision of all three methods. A total of 610 consecutive blood donors were recruited. However, only 553 donors were analysed at the end of the study. All measurements were performed in parallel and in duplicate. Sensitivity was defined as percentage of donors correctly identified with Hb below the cut- off for blood donation among all donors with a venous Hb below the cut- off. Meanwhile, specificity was the percentage of donors correctly identified with Hb value above the cut-off for blood donation among all donors with a venous Hb above the cut-off. Cut off value was 135 g/L for male and 125 g/L for female. From the study they found that the non- invasive method underestimated (mean bias; -5.9 g/L (95%CI: -25.47, 13.88) and capillary method overestimated Hb values (mean bias; 4.3 g/L

(95%CI: -8.13, 16.71). The capillary invasive method was less sensitive (23.1%) in identifying donors having Hb below the cut off value for blood donation, but it was high in specificity (99.2%). In contrast, the non- invasive method was more sensitive (63.2%) but less specific (76.2%). The PPV and NPV for the non-invasive method was 24.8% and 94.3% respectively. Meanwhile, PPV and NPV for the capillary invasive method was 78.9% and 90.6% respectively. Intra-class correlation (ICC) was lower in non-invasive method (r=0.59; 95% CI: 0.52, 0.65) compared to capillary invasive method (r=0.74; 95% CI: 0.70, 0.78). However, non- invasive method was more precise (CV=2.73) compared to capillary invasive method (CV=3.23). The author’s concluded that the non-invasive Hb measurement is a reasonable first-line screening of blood donors but a second method should be available to retest those not testable with the non-invasive device or with Hb values below the cut offs. 15, level 3

Ardin S et al. conducted a diagnostic accuracy study to evaluate the use of three non-invasive methods (NBM 200, Pronto 7 and Haemospect) for Hb screening in healthy blood donors in comparison to an invasive capillary method (Compolab) and venous blood sample as reference (Sysmex). The evaluation was conducted in three independent studies with total of 351 participants. The non-invasive NBM 200 was evaluated in Phase 1 (n=120), while Pronto 7 was evaluated in phase 2 (n=110) and Haemospect was evaluated in phase 3 (n=121) of the study. The cut-off for haemoglobin value was ≥12.5 g/dL for female and ≥13.5 g/dL for male. Statistical analysis was performed and the usability was assessed by operators using a specifically design questionnaire. From the study they found that, bias (difference of mean values) was lower in non-invasive method [bias of -0.12 ± 1.14 SD (95% LOA; -2.35, 2.11) for NBM 200 , - 0.22g/dL ±12.4 SD (95% LOA ; -2.64, 2.21) for Haemospect, and -0.50 ± 0.99 SD (95% LOA; -2.44, 1.43) for Pronto 7 respectively] compared to the invasive method [ bias of -0.53± 0.81 SD (95% LOA; -2.13, 1.06) for Compolab]. Using the venous Hb level as reference standard, the percentage of donors correctly screened (accordance of acceptable and unacceptable donors between Sysmex and test method) was 88.4% by Haemospect, 91.7% by NBM 200, 86.4% by Pronto 7 and 92.8% by Compolab. The authors concluded that, non-invasive Hb tests represent an attractive alternative by eliminating pain and reducing risks of blood contamination. The main problem for generating reliable results seems to be pre-analytical variability in sampling. Despite the sensitivity to environmental stress, all methods are suitable for Hb measurement.16, level 3

Pinto M et al. conducted a diagnostic accuracy study to analyse the feasibility of non-invasive (occlusion spectroscopy method) in the blood banks, as well as to compare the method with capillary invasive (fingerstick) haemoglobinometer and an automatic blood analyser. A total

of 205 consecutive blood donors were enrolled into the study. Samples were collected by a capillary invasive (Hemocue 201) and venous puncture (automatic blood analyser) to determine Hb level and compared with the non-invasive determination by occlusion spectroscopy using NMB 200. The percentage errors of capillary invasive and non-invasive method of all donors stratified by sex, weight and age levels were compared to automatic blood analyser using Wilcoxon signed rank test. From the study they found that there was significantly higher percentage error in Hb value between capillary invasive method and automated analyser in the whole group [median Hb for capillary invasive method was 15.2 g/dL compared to median for automated analyser value of 14.4 g/dL (p< 0.001)], as well as in the stratified results by age and sex. Meanwhile, they found there was no different on the percentage error between non-invasive and automated analyser in whole group analysis [median Hb for non-invasive was 14.6 g/dL compared to median for automated analyser value of 14.4 g/dL]. However, the significant different for non-invasive method was reported in female group [median Hb for non-invasive was 12.9 g/dL compared to median for automated analyser value of 13.3 g/dL, (p=0.026)] and weight level of 61 to 70 kg [p=0.034, data was not reported). 17, level 3

5.2. SAFETY There was no retrievable evidence on the adverse event of non-invasive haemoglobinometer. Some of the non-invasive haemoglobinometer in the market (such as Orsense and Masimo) had received 510 (k) from USFDA.18

5.3. COST-EFFECTIVENESS There was no retrievable evidence on the cost-effectiveness on non- invasive haemoglobinometer. The estimated price for non-invasive haemoglobinometer is approximately USD 700 (RM 2854) per unit.19

5.4. ORGANISATIONAL Training There were two studies which reported that non-invasive haemoglobinometer required less than 30 to 45 minutes to train new operator. It also did not require any expertise. 11, 12, level 3

Measurement time Baart et al. reported that, staff at the collection centres mentioned the non- invasive measurement itself was fast. However, the whole process of pre- donation screening was time consuming due to inability of patients to move to fill the donation forms and to have their blood pressure measured. The device also need to be reset every three to four hours of measurement. 13, level 3

Interference of testing Baart et al. reported that non-invasive measurement was affected by the skin colour, smoking and finger abnormalities such as cold fingers or callus among the donors. 13, level 3

Sunming et al. reported that there were 38 (7%) of donors that the non- invasive devices could not detect a pulse curve for various reasons such as nail polish [11 (2%)], artificial fingernails [11 (2%)], cold fingers [7 (1%)], callus of the skin [3 (0.5%)] and unknown cause [6 (1%)].15, level 3

Ardin S et al. reported that there were technical errors which occurred in the Pronto 7 because of donor characteristics like skin colour and nail polish. 16, level 3

5.5. PSYCHOLOGICAL/SOCIAL Preference Belardinelli A et al. reported that from 350 donors administered with questionnaires, majority of the donors like non-invasive but did not like capillary fingerstick test or double venous puncture (57%), followed by no preference (24%), like capillary test but did not like non-invasive or double venous puncture (10%) and like both capillary fingerstick and non-invasive but did not like double venous puncture (9%). 11, level 3

Baart et al. found that majority of the donors (n=1754; 68%) preferred the non-invasive measurement, 262 (10%) preferred the invasive measurement while 573 (22%) had no preference. 13, level 3

Satisfaction Herraiz et al. reported that the mean of satisfaction score [using scale from 0 (unsatisfactory) to 5 (excellent)] among the operators were highest for capillary invasive Hemocue (4.21), followed by non-invasive method [Pronto 7 (3.93), Haemospect (3.89) and NBM 200 (3.21)]. 12, level 3

Usability Ardin et al. conducted usability test to evaluate visual appearance, handling, cleaning, visualization of result, time to result and donor’s feedback among operators using questionnaire scale 1-6 (1 very good, 6 unsatisfactory). They reported that Haemospect was ranked as good (mean 1.9), Pronto 7 as satisfactory (mean 3.2) and NBM 200 as unsatisfactory (mean 4.1). 16, level 3

5.6. LIMITATIONS This technology review has several limitations. The selection of studies was done by one reviewer. Although there was no restriction in language during the search but only English full text articles were included in this review.

6. CONCLUSION Effectiveness There was fair level of evidence retrieved to suggest that non-invasive haemoglobinometer had lower sensitivity, lower specificity and lower correlation with reference standard compared to invasive method. However, there was one study reported that sensitivity of non-invasive method was higher compared to invasive method [63.2% vs 23.1%]. The non-invasive haemoglobinometer also shown to have lower accordance or agreement with the gold standard compared to invasive method. The PPV was also lower compared to invasive method. However, the NPV for non- invasive haemoglobinometer was inconclusive.

Cost-effectiveness There was no retrievable evidence on the cost-effectiveness on non- invasive haemoglobinometer.

Organisational Issue There was limited fair level of evidence retrieved to suggest that non- invasive haemoglobinometer required less than 30 to 45 minutes to train new operator, had fast measurement (however, the whole process of pre- donation screening was time consuming due to inability of patients to move). The measurement was also interfered with the patients’ condition such as skin colour, cold finger, artificial fingernail, nail polish and etc.

Psychological/social There was limited fair level of evidence retrieved to suggest that non- invasive haemoglobinometer was preferred by the blood donors. Among the non-invasive haemoglobinometer, Haemospect and Pronto 7 rated as the most usable and satisfied by operator.

7. REFERENCES 1. Guideline on assessing Donor suitability for blood donation. Available at http://www.who.int/bloodsafety/publications/bts_guideline_donor_suitabilit y/en/ . Accessed on 16 August 2017

2. Transfusion practice guidelines for clinical and laboratory personnel. National Blood Centre. Ministry of health Malaysia. 4th edition 2016

3. Chaudhary R, Dubey A and Sonker A. Techniques used for the screening of hemoglobin levels in blood donors: current insights and future directions. J Blood Med. 2017; 8: 75-88

4. Patel AJ, Wesley R, Leitman SF, et al. Capillary versus venous haemoglobin determination in the assessment of healthy blood donors. Vox Sang. 2013; 104:317–323

5. Briggs C, Kimber S and Green L. Where are we at with point- of- care testing in haematology?. Br J Haematol. 2012; 158: 679–690

6. Hemoglobin screening using noninvasive methods. Available at https://www.mlo-online.com/hemoglobin-screening-using-noninvasive- methods.php. Accessed on 17 October 2017

7. Pagliaro P, Belardinelli A, Boko V et al. A non-invasive strategy for haemoglobin screening of blood donors. Blood Transfus. 2014; 12: 458- 463

8. Levy TS, Gomez-Humaran IM, Ruan M CM et al. Validation of Masimo Pronto 7 and HemoCue 201 for hemoglobin determination in children from 1 to 5 years of age. PloS ONE. 2017;12 (2): e0170990

9. Kim MJ, Park Q, Kim MH et al. Comparison of the Accuracy of Noninvasive Hemoglobin Sensor (NBM-200) and Portable Hemoglobinometer (HemoCue) with an Automated Hematology Analyzer (LH500) in Blood Donor Screening.Ann Lab Med. 2013;33: 261-267

10. Singh A, Dubey A, Sonker A et al. Evaluation of various methods of point- of-care testing of haemoglobin concentration in blood donors. Blood transfuse. 2015; 13: 233-239

11. Bellardinelli A, Benni M, Tazzari PL et al. Noninvasive methods for haemoglobin screening in prospective blood donors. Int Soc Blood Trans. 2013; 105: 110-120

12. Herraiz ALP, Gambarte JDR, Lopez BE et al. A Comparative Study of Three Non-invasive Systems for Measurement of Hemoglobin with HemoCue System Having Coulter LH750 as Reference Value. Hematol Transfus Int J. 2015; 1 (3):1-10

13. Baart AM, Kort WL, Hurk K et al. Hemoglobin assessment: precision and practicability evaluated in the Netherlands—the HAPPEN study. Transfusion. 2016; 56 (8) :1984-1993

14. De Clippel D, Van Heddegem L, Vandewalle G et al. Hemoglobin screening in blood donors: a prospective study assessing the value of an invasive and a noninvasive point-of-care device for donor safety. Transfusion. 2017;57(4):938-945

15. Sunming A, Hron G, Westphal A et al. The impact of noninvasive, capillary, and venous haemoglobin screening on donor deferrals and the haemoglobin content of red blood cells concentrates: a prospective study. Transfusion. 2015; 55: 2847-2854

16. Ardin S, Störmer M, Radojska S. Comparison of three noninvasive methods for hemoglobin screening of blood donors. Transfusion. 2015; 55: 379-387

17. Pinto M, Barjas-Castro ML, Nascimento S et al. The new noninvasive occlusion spectroscopy hemoglobin measurement method: a reliable and easy anemia screening test for blood donors. Transfusion. 2013; 53: 766- 769

18. NBM 200 Pulse Oxymetry Device. Available at https://www.accessdata.fda.gov/cdrh_docs/pdf14/k142209.pdf. Accessed on 7 October 2017

19. Non-invasive haemoglobinometer. Available at www.cliawaived.com/cf.inventory.htm?action=showinvone&invid=1622 . Accessed on 7 October 2017

8. APPENDIX

8.1. Appendix 1: LITERATURE SEARCH STRATEGY

Epub Ahead of Print, In-Process & Other Non-Indexed Citations, Ovid MEDLINE®Daily and Ovid MEDLINE®1946 to Present

1 BLOOD DONORS/ (19644) 2 blood don*.tw. (24041) 3 1 or 2 (33269) 4 HEMOGLOBINS/ (69209) 5 ferrous hemoglobin.tw. (44) 6 h?moglobin*.tw. (119209) 7 OXIMETRY/ (12509) 8 oximetr*.tw. (9943) 9 pulse oximetr*.tw. (6219) 10 HEMOGLOBINOMETRY/ (6794) 11 h?moglobinometr*.tw. (105) 12 hemoglobinometr*.tw. (105) 13 spHb.tw. (78) 14 non invasive hemoglobinometer.tw. (0) 15 non-invasive hemoglobinomet*.tw. (0) 16 non invasive h?moglobinomet*.tw. (0) 17 non-invasive h?moglobin.tw. (20) 18 noninvasive hemoglobin*.tw. (74) 19 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 (173701) 20 3 and 19 (1217) 21 limit 20 to humans (1139)

OTHER DATABASES EBM Reviews - Cochrane Central Register of Controlled Trials EBM Reviews - Cochrane database of systematic reviews EBM Reviews - Health Same MeSH, keywords, limits used as Technology Assessment per MEDLINE search EBM Reviews – NHS Economic Evaluation Database EBM Reviews – Database of Abstract of Review of Effects

PubMeD Search (((BLOOD DONORS[Title/Abstract]) OR blood don* [Title/Abstract])) AND(((((((((((((((HEMOGLOBINS[Title/Abstract]) OR ferrous haemoglobin [Title/Abstract]) OR h?moglobin* [Title/Abstract]) OR OXIMETRY [Title/Abstract]) OR oximetr* [Title/Abstract]) OR pulse oximetr*[Title/Abstract]) OR HEMOGLOBINOMETRY[Title/Abstract]) OR h?moglobinometr* [Title/Abstract]) OR hemoglobinometr* [Title/Abstract]) OR spHb [Title/Abstract]) OR non invasive hemoglobinometer [Title/Abstract]) OR non-invasive hemoglobinomet* [Title/Abstract]) OR non invasive h?moglobinomet* [Title/Abstract]) OR non- invasive h?moglobin [Title/Abstract]) OR noninvasive hemoglobin* [Title/Abstract]) OR hemoglobinometer[Title/Abstract]

8.2. Appendix 2

HIERARCHY OF EVIDENCE FOR TEST ACCURACY STUDIES

Level Description

1. A blind comparison with reference standard among an appropriate sample of consecutive patients 2. Any one of the following Narrow population spectrum 3. Any two of the following Differential use of reference standard 4. Any three or more of the following Reference standard not blind Case control study 5. Expert opinion with no explicit critical appraisal, based on physiology, bench research or first principles

SOURCE: NHS Centre for Reviews and Dissemination (CRD) University of York, Report Number 4 (2nd Edition)

8.3. Appendix 3 Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors? Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 1.Kim MJ, Study design 3 506 donors Non- Capillary Hemoglobin measurements (mean ± Park Q, Kim Diagnostic accuracy study invasive blood SD) MH et al. M=291 , (NBM 200) (fingerstick)  NBM 200: 14.1±1.4, Median Comparison Objective F=215 -Hemocue (IQR) 14.3 (13.2, 15.2) of the To evaluate the accuracy of  Hemocue : 14.3±1.4, Median accuracy of the non-invasive Hb sensor (IQR) 14.2 (13.2, 15.4) non-invasive and to compare its hemoglobin measurements with those of a Intraclass correlation coefficient Sensor currently used portable (ICC) (NBM-200) hemoglobinometer  NBM 200: 0.69 (95% CI: 0.64, and Portable 0.73) Hemoglobino Methods Study conducted in 2 blood  Hemocue: 0.86 ((95% CI: 0.79, meter 0.90) (Hemocue) donation sites with an 5ml venous sample collected Automated for testing on LH500 Hematology (automated hematology) Analyzer (LH500) in Capillary blood (fingerstick) Blood Donor was measured using Hemocue Screening. Ann Lab Non-invasive (NBM 200) was Med. 2013; placed at the thumb for 33: 261-267 measurement

Korea Hb measurement obtained

from NBM 200 and Hemocue were compared with LH500

Data were split into the

following 3 groups since the Of the 70 donors who were deemed eligibility criteria for an apheresis donation is ≥12.0 ineligible to donate blood by the LH500 (Hb <12.5 g/dL): g/dL and that for a whole blood  43 donors (61.4%) were eligible donation is ≥12.5 g/dL. to donate whole blood by the 27

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) NBM-200 Sensitivity were calculated as  39 donors (55.7%) were eligible the percentage of ineligible to donate whole blood by the donors who were correctly HemoCue identified as ineligible to Correlation of determination (R2) donate blood  NBM 200: 0.47

 Hemocue: 0.76 Specificity were calculated as the percentage of eligible Sensitivity and specificity donors who were correctly identified as eligible to donate NBM 200 blood Whole blood The degree of homogeneity Sensitivity 38.6 (95% CI: 28.1-50.3) between the Hb Specificity 93.6 (95% CI: 90.9-95.5) measurements obtained using the 2 POCT devices (NBM- Apheresis 200, HemoCue) and the Sensitivity 37.8 (95% CI: 24.1-53.9) reference value obtained using Specificity 94.0 (95% CI: 91.5-95.8) the LH500 was evaluated by utilizing the intraclass Hemocue correlation coefficient (ICC) Whole blood Sensitivity 42.7 (95% CI: 31.6-54.5) ICCs were interpreted as: Specificity 98.6 (95% CI: 97.0-99.4) <0 : poor agreement 0.01-0.20:slight agreement Apheresis 0.21-0.40: fair agreement Sensitivity 33.3 (95% CI: 21.0-48.5) 0.41-0.6 :moderate agreement Specificity 99.4 (95% CI: 98.1-99.8) 0.61-0.80:substantial agreement 0.81-1.00:almost perfect agreement

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 2.Singh A, Study design 3 N=534 Non- Invasive Dubey A, Diagnostic accuracy study prospective invasive  DiaSpect Sonker A. blood donors NBM-200  HemoControl Evaluation of Objective various To evaluate the suitability of (n= 485 one methods of non-invasive and invasive time testing) point-of-care POCT devices for blood 456 M, 29 F testing of donor Hb screening haemoglobin (n=49 concentration Methods duplicate in blood The donors were tested by testing) donors. DiaSpect and HemoControl Blood using capillary blood obtained Transfus. from pricking the index finger 2015;13: of the right hand 233-9 Simultaneously, Hb testing by NBM-200 was done on the India thumb of other hand

Blood donors with the pre- CV: coefficient of variation defined level of Hb for blood donation by HemoControl were Author’s conclusion accepted for blood donation POCT devices which are becoming increasingly popular for screening Hb 2 ml blood sample was in blood banks are time-saving and collected in an EDTA vial from easy to operate but lack reliability in the diversion pouch of the many circumstances. It is very blood bag and subjected to Hb important to check the precision and automated cell analyser KX-21 accuracy of a device before its use blood banks. The non-invasive NBM- Blood samples of the donors 200 device has the advantage of who were deferred from blood inflicting less pain and hence being donation on the basis of low more acceptable to donors, but it lacks

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) Hb level by HemoControl were the desired sensitivity. This test is not also collected and tested on effective in excluding ineligible donors the KX-21 which may pose a threat to donors' health. DiaSpect gives fast and Data were splitted into 3 accurate results and is, therefore, group: suited for screening donors in blood group 1: Hb≤11.0 g/dL banks group 2: Hb 11.1-12.4 g/dL group 3: Hb ≥12.5 g/dL

The sensitivity, specificity, positive predictive value and negative predictive value of each method were calculated

ICC were interpreted as follows: <0: poor agreement 0.01-0.20: slight agreement 0.21-0.40: fair agreement 0.41-0.60:moderate agreement 0.61-0.80:substantial agreement 0.81-1.00:almost perfect agreement

Precision of the three methods was analysed by coefficient of variation (CV) on repeat testing by a particular method.

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors? Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up (If Comments Patient Applicable) Characteristic 3.Belardinelli Study design 3 NBM 200 trial Non- Capillary Capillary (Hemocue, n=908) A, Benni M, Diagnostic accuracy study (n=445) invasive fingerstick  Bias : 0.83 g/dl [0.70 SD; 95% limit Tazzari PL et M=296 ,F= haemoglobi sample of agreement (LOA):-0.54, 2.20) al. Objective 149 nometer: testing  Sensitivity: 99% Noninvasive  To evaluate rapid, non-  NBM 200 (Hemocue)  Specificity: 99.5% methods for invasive methods for hb  Masimo haemoglobin screening in blood donors for Pronto 7 Non–invasive (NBM 200) screening accuracy Masimo  Bias : 0.29 g/dl [0.98 SD; 95% limit in Pronto 7 trial of agreement (LOA):-1.64, 2.21) prospective  To assess whether they had (n=463)  Sensitivity: 98% the potential to enhance blood M=231, F=132  Specificity: 97% donors. Int recruitment and retention of Society blood donors Non-invasive (Masimo Pronto 7)

Blood  Bias : -0.53 g/dl [1.04 SD; 95% limit Methods Transfus of agreement (LOA):-2.57, 1.51) (Vox Consecutive blood donors  Sensitivity: 93% Sanguinis). were tested before routine  Specificity: 83% 2013; 105 : donation procedure by non-

116-120 invasive device (NBM 200 and Preference Masimo Pronto 7) Italy  Donors who like noninvasive test, Both trials (in separate but do not like capillary fingerstick occasion) were compared to test or double venous puncture capillary sample testing (57%) (Hemocue)  Donors who like capillary fingerstick and noninvasive test, but do not like Gold standard was venous double venous puncture (9%) sample testing on cell counter  Donors who like capillary fingerstick analyser (Beckman Coulter test, but do not like noninvasive test AcT-5) or double venous puncture (10%)  No preferences (24%) The hemoglobin determination

on the donors was first Training performed by non-invasive  Less than 30 min to train method followed by capillary fingerstick  Did not require special expertise

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up (If Comments Patient Applicable) Characteristic At the beginning of donation, a Author’s conclusion venous sample was taken Analysis of finger pulp blood by either direct sampling by fingerstick A survey questionnaire about (Hemocue), or by noninvasive compliance with the non- haemoglobin tests does not replicate invasive pre-donation the results of cell counter analysis of screening methods and their venous samples. Compared with preferences was administered fingerstick, non-invasive haemoglobin to 350 donors tests eliminate pain and reduce stress, but have a lower level of specificity and Analysis was performed on sensitivity Analyse-it software

Agreement of the methods was analysed using Bland and Altman approach

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 4.Pajares- Study design 3 Blood donors Non- Invasive Invasive capillary (Hemocue) Herraiz AL, Diagnostic accuracy study (n=129) invasive capillary  Sensitivity: 96.23% Rodriguez- hemoglobin fingerstick:  Specificity: 99.37% Gambarte JD, Objective M=102, F=57 ometer:  Correlation index (r) : 0.859 Equiga-Lopez  To analyze three non-  Hemocue B et al. A invasive methods:  NBM200 Non-invasive (NBM 200) Comparative NBM200, Haemospect  Haemosp  Sensitivity: 94.34% Study of Three and Pronto7 ect  Specificity: 97.48% Non-invasive  To compare these  Pronto7  Correlation index (r) : 0.648 Systems for methods with the Measurement currently used HemoCue Non-invasive (Pronto 7) of Hemoglobin 301 (invasive fingerstick  Sensitivity: 91.19% with HemoCue system) and a venous Hb  Specificity: 98.74% System pre-donation sample  Correlation index (r) : 0.618 Having analysed in a hematology Coulter LH750 analyser (Coulter LH750) Non-invasive (Haemospect) as Reference as a reference value  Sensitivity: 93.08%

Value.  Specificity: 98.74% Methods Hematol  Correlation index (r) : 0.416 Transfus Int J. All blood donors underwent 2015; 1(3): an interview and medical Training 00016 examination according to  The use of non-invasive methods

protocols, including heart required no more than 45 minutes

rate, blood pressure and Hb and did not require any special Spain measure with HemoCue. expertise.

Hb was also measured with Time to measure the three non-invasive  HemoCue=35 seconds devices  NBM200=100 seconds  Haemospect = 20 seconds Venous blood sample was  Pronto7= 30 seconds taken to be tested on hematology analyser (gold Satisfaction standard)  HemoCue (mean 4.21)  Pronto 7 (mean 3.93)

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) Questionnaire was distributed  Haemospect (mean 3.89) to the operators who used  NBM 200 (mean 3.21) non-invasive method to evaluate the usability, speed of the method or time to Author’s conclusion readiness, cleaning and The value of hemoglobin with HemoCue maintenance, intrinsic errors, shows the best correlation with the sensitivity to ambient light LH750, being lower for the rest, although and handling. acceptable, especially with P7 and NBM200. The three devices could be The survey was made with a used as screening test for blood. scale and evaluated from 0 However, it’s cutoff should be analysed to (unsatisfactory) to 5 avoid the high percentage of false (excellent) rejection.

Bland-Altman method and the Pearson correlation coefficient were used for analysis

Pearson’s values: ≥ 0.75 (good correlation) 0.40 - 0.75 (favorable correlation) < 0.40 (poor correlation)

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 5.Baart AM, Study design 3 N=2589 Non- Invasive: - NBM 200 Kort WL, Diagnostic accuracy study invasive:  Hemocue Mean difference (MD): 0.10mmol/L (± Hurk K et al. 1370 male,  NMB 201 0.60 SD) male / 0.19 mmol/L (± 0.65 Hemoglobin Objective 1219 female 200  Hemocue SD) female assessment: To assess whether  Haemo 301 correlation:0.42(male )/0.36 (female) precision noninvasive Hb measurement spect sensitivity: 12.1% and provides similar precision and specificity:98.6% practicability bias as the invasive capillary PPV:43.8% evaluated in Hb measurement when NPV:92.4% the compared to measurement in Netherlands venous samples. Haemospect —the MD: -0.06 mmol/L (± 0.73 SD) male / - HAPPEN Methods 0.06 mmol/L (± 0.64 SD) female study. Whole blood donors who Correlation:0.03(male)/ 0.16 (female) Transfusion. visited one of three sensitivity:3.5% 2016; 56 (8) participating blood collection specificity:95.9% :1984-1993 centers in the Netherlands PPV:7.1% between June and October NPV:91.8% 2014 were invited to participate Netherlands in the study Hemocue 201 MD:0.24 mmol/L (± 0.42 SD) male / All donors Hb level were 0.18 mmol/L (± 0.41 SD) female measured using fingerstick correlation:0.77(male)/ 0.75(female) capillary on HemoCue 201 sensitivity: 30.1% specificity:99.8% In a random subset, Hb levels PPV:94.4% were then measured with the NPV:94.1% HemoCue 301 Hemocue 301 Then, Hb levels were MD:0.35 mmol/L (± 0.44 SD) male / measured with one of two non- 0.33 mmol/L (± 0.44 SD) female invasive measurement correlation:0.75(male)/0.71(female) devices: either the NBM 200 sensitivity:27.9% (OrSense) or the Haemospect specificity:99.4%

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) (MBR). PPV:79.4% NPV:94.1% Allocation to one of the non- invasive devices was randomly Preference: assigned.  1754 donors (68%) preferred the non-invasive measurement A questionnaire on satisfaction  262 donors (10%) preferred the with the invasive and the non- invasive measurement invasive measurements as well  573 donors (22%) had no as on possible side effects that preference were experienced during or after the measurements were Interference: distributed to the donors  Skin colour: The difference decreased on average 0.1 Possible influencing factors mmol/L when the darkness such as skin colour, hand or increased with one point (range, finger abnormalities, and 0-18) smoking were registered.  Smoking: Smoking resulted on average in a 0.1 mmol/L smaller A venous sample was difference (p values ranged collected from the donors (both from less than 0.01 to 0.039) eligible and deferred)  Hand or finger abnormality (such as cold fingers or callus): Hb levels in venous samples Significant effect on differences were measured on the next between measurements with day on a hematology analyser both the HemoCue 201 and the venous measurement and the Mean values and standard HemoCue 30 and the venous deviations (SDs) of Hb levels measurement in women (p = measured with the different Hb 0.043 and p = 0.036, measurement devices respectively). were calculated The staff at the collection centres Pearson correlation mentioned that, the advantages were: coefficients were calculated for  noninvasive devices were easy measurements with the test to use device and the venous  the measurement itself is fast measurements, and Bland-

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) Altman plots were created Disadvantages:  time to measure the size of the finger for the right cuff size (Haemospect),  some fingers being too big leading to an occasionally damaged cuff (NBM 200)  donors should not move during the noninvasive measurement (could not fill out questionnaires or have blood pressure measured simultaneously)  Device had to be reset after every three to four measurements (Hemospect) and was time-consuming.

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 6.De Clippel Study design 3 1483 donors Non-invasive Capillary Bias D, Van Diagnostic accuracy study  Haemospect invasive Compolab : bias of 3.4 [9.2 g/L SD,( Heddegem Arm 1 (n=494;  Compolab 95% LOA: -21.4 , 14.6 g/L)] st L, Objective 313 1 time Haemospect Classic : bias of -2.3 Vandewalle To assess the ability of two donors, 181 [12.1 g/L SD, (95% LOA: -26.0 , 21.4 G et al. POCT devices to detect low regular g/L)] Hemoglobin Hb levels in new donors and to donors) Haemospect Alternative: bias of -1.5 screening in identify recovered Hb levels in g/L [12.9 g/L SD, (95% LOA: -26.7, blood regular donors with decreased Arm 2 (n=503; 23.7 g/L)] st donors: a levels at the previous donation 321 1 time prospective donors, 182 Correlation study Methods regular Compolab (r= 0.78) assessing Whole blood donors were donors) Haemospect Classic (r=0.48) the value of enrolled in three study arms: Haemospect Alternative (r= 0.46) an invasive  Arm 1 : Compolab vs Arm 3 (n= 486; st and a venous (CBC) 29 1 time noninvasive  Arm 2: Haemospect donors, 196 point-of-care (classic) vs venous(CBC) regular device for  Arm 3: Hemospect donors) donor safety. (alternative) vs Transfusion. venous(CBC) 2017;57(4):9 38-945 Haemospect (classic) defined as measurements using the Belgium Haemospect device were conducted according to the working instructions of the manufacturer. Digiclip was positioned at middle finger to ensure that the fiber optics were in the center of the terminal phalanx. Subsequently the correct sex was chosen on the instrument

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) Haemospect (alternative) defined as Digiclip was positioned in an alternative way—in accordance with a variation of instructions for use of the Haemospect device. Digiclip was positioned more proximally over the terminal phalanx with the fiber optics close to the distal inter phalangeal fold

Each study arm measured Hb (with a point-of-care testing device) were compared with the venous Hb results

Two different application methods for the positioning of the Digiclip were assessed

The correlation between the POCT result and the venous (CBC) result was plotted, the correlation coefficient (Spearman) was calculated and Bland-Altman (BA) plots were created within each study arm

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 7.Sumning A, Study design 3 610 Non- Invasive  Capillary and venous Hb values were Hron G, Diagnostic accuracy study consecutive invasive capillary obtained from all donors, and Westphal A. donors (Pronto 7) (Hemocue) noninvasive Hb values were obtained The impact Objective from 515 donors of To assess the accuracy of the 7 refused to noninvasive, non-invasive and the capillary participate  In 38 (7%) of donors, the noninvasive capillary, and test in comparison to the gold device could not detect a pulse curve venous standard and to analyse 50 low Hb for different reasons: hemoglobin precision of all three methods  nail polish,11 (2%) screening on 553 included  artificial fingernails, 11 (2%) donor To assess sensitivity, in analysis  cold fingers, 7 (1%) deferrals and specificity, negative and  callus of skin, 3 (0.5%) the positive predictive values for  unknown cause, 6 (1%) hemoglobin the non-invasive and the content of capillary test in comparison to red blood the venous reference cells concentrates: To assess, the correlation of a prospective predonation Hb values study. obtained by each test system Transfusion. with the final Hb content 2015;55: 2847–2854 Methods Consecutive, consenting German donors presenting for a voluntary whole blood donation were tested in parallel with a capillary POC method, a non- invasive method (Pronto-7, Masimo) and a venous method as gold standard (Advia 2120)

All tests were performed in duplicate

Both non-invasive tests were performed on the same finger

Accuracy was determined by Bland-Altman analysis of bias and 95% limits of agreement (LOA)

Coefficients of variation (CV) for duplicate measurements were used to evaluate precision

Sensitivity was defined as percentage of donors correctly identified with a Hb below the cut-off for blood donation among all donors with a venous Hb below the cut-off

Specificity was the percentage of donors correctly identified with a Hb value above the cut- off for blood donation among all donors with a venous Hb above the cut-off

Cut off value:  Male 135 g/L  Females 125 g/L

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

German Blood donors were tested in three consecutive trials by Haemospect , NBM 200 or Pronto-7 in Usability comparison to the conventionally Haemospect = good grade used invasive method using the (mean, 1.9) CompoLab Hb to measure Hb Pronto-7 = satisfactory grade levels on capillary samples (mean, 3.2) NBM 200 = obtained by a fingerstick lancing unsatisfactory grade (mean, 4.1) technique as shown below: Pronto-7 Technical errors occurred due to donor characteristics such as skin colour, nail polish, and ambient light

NBM 200 Showed high sensitivity to ambient light and needed the longest time to result by up to 115

seconds.

Usability test was evaluated among operators using questionnaire scale 1-6 (1 very good, 6 unsatisfactory) to visual appearance, handling, cleaning, visualization of result, time to result and donor’s feedback

Descriptive statistics (mean and standard deviation [SD]) as well as Bland-Altman plots were used for analysis

Evidence Table : Effectiveness Question : How accurate is non-invasive haemoglobinometer for screening of blood donors?

Bibliographic Study L Number of Intervention Comparison Length of Outcome Measures/Effect Size General citation Type/Methods E Patients & Follow Up Comments Patient (If Characteristic Applicable) 9.Pinto M, Study design II- 205 donors Non- Invasive - Barjas-Castro Diagnostic accuracy study 3 invasive: Fingerstick: ML, NBM 200 Hemocue Nascimento S Objective 201 et al. The new To analyse the feasibility of noninvasive occlusion spectroscopy occlusion method in blood banks and to spectroscopy compare the method with hemoglobin fingerstick hemoglobinometer measurement and hemoglobin (Hb) method: a determination on an reliable and automatic blood analyzer easy anemia screening test Methods for blood Consecutive volunteer blood donors. donors of both sexes were Transfusion.2 enrolled during August 2011 013; 53:766- after informed consent was 769 obtained Abbreviation: Hb-V: venous method Brazil Blood samples of each donor Hb-NI: non-invasive method were obtained by fingerstick Hb-F: fingerstick method for Hb testing using (Hemocue 201) Authors conclusion: Non-invasive method seems to be a Venous blood samples were good method in terms of precision and obtained from donors for Hb feasibility for anemia screening of blood determination in an donors as well as being much more automated hematologic comfortable for donors. analyzer

The non-invasive occlusion spectroscopy Hb determination

The results of Hb determination by the three methods were tabulated on a data sheet containing donor sex, age, and weight, which could impact the results