Accepted Manuscript Title: Allo-antibody Identification: A software approach! Author: Aseem Kumar Tiwari, Ravi C Dara, Dinesh Arora, Geet Aggarwal, Ganesh Rawat, Subhasis Mitra, Pandey K Prashant, Vimarsh Raina PII: DOI: Reference:
S1473-0502(14)00159-1 http://dx.doi.org/doi:10.1016/j.transci.2014.08.020 TRASCI 1727
To appear in:
Transfusion and Apheresis Science
Received date: Revised date: Accepted date:
29-5-2014 2-8-2014 25-8-2014
Please cite this article as: Aseem Kumar Tiwari, Ravi C Dara, Dinesh Arora, Geet Aggarwal, Ganesh Rawat, Subhasis Mitra, Pandey K Prashant, Vimarsh Raina, Allo-antibody Identification: A software approach!, Transfusion and Apheresis Science (2014), http://dx.doi.org/doi:10.1016/j.transci.2014.08.020. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
TITLE: Allo-antibody Identification: A software approach! RUNNING TITLE: Allo-antibody Identification: A software approach! AUTHORS: Aseem Kumar Tiwari, MD Senior Consultant, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Ravi C Dara, MD Senior Resident, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Dinesh Arora, DCP Associate Consultant, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Geet Aggarwal, MBBS Junior Resident, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Ganesh Rawat, DMLT Senior Technologist, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail:[email protected] Subhasis Mitra, Intern, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Pandey K Prashant, MD Consultant, Department of Transfusion Medicine Jaypee Hospital, Sector-128, Noida-201304 [email protected] Vimarsh Raina, MD Director, Laboratory Services and Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] CORRESPONDING AUTHOR Aseem K Tiwari, MD Senior Consultant, Department of Transfusion Medicine 1 Page 1 of 16
Medanta-The Medicity, Sector-38, Gurgaon-122001 Email: [email protected] Fax: +91-124-483411 Phone No: +91-124-4411441 Ext-2701, 2718
2 Page 2 of 16
Abstract Background: Unexpected allo-antibody identification is difficult serological test requiring in-depth knowledge of antibody behavior, identification rules, knowledge of zygosity of antigens and dosage phenomenon. Software which uses an algorithm based on characteristics of antibodies is now available to interpret specificity of allo-antibody. A study was undertaken to evaluate the effectiveness of one such software (Resolvigen) for antibody identification compared to manual antibody identification method. Materials and Methods: The study was a retrospective observational study where 238 allo-antibody results were reevaluated using Resolvigen software (Ortho clinical diagnostics, Johnson and Johnson, USA) and agreement between manual and software approaches was studied. Resolvigen software was also evaluated for usefulness, ease of use and predicted future usage by administering investigators a questionnaire with Likert scale. Results: Agreement between the results of manual and automated methods ranged from 98.6% for single antibody to 65% for two antibodies (p=0.000). Resolvigen software came out as very useful, easy to use, and with high predicted future usage. Conclusion: This study concludes that Resolvigen can either replace manual method or be used as adjuvant to routine manual method.
Keywords: Allo-antibody, Resolvigen, Allo-immunization, Irregular antibody
3 Page 3 of 16
1. Introduction Antibody screening is done to detect red cells antibodies other than expected anti-A and anti-B. These are therefore called ‘unexpected’ antibodies. Only 0.3 to 1.0% of general population have unexpected antibodies1 and the incidence is higher in women due to the pregnancy and patients receiving multiple transfusions.[1] The unexpected alloantibodies are immune antibodies formed by exposure to red cells due to pregnancy or blood transfusion. Methods such as Conventional Tube Technique (CTT), Column Agglutination Technique (CAT) and Solid Phase Red Cell Adherence (SPRCA) are usually used to detect unexpected antibodies during the pre-transfusion testing. The use of antibody screening using column agglutination technology (CAT) is on the rise in blood transfusion services in India. In CAT, sephadex gel or glass-beads in the column are used to capture agglutinates. CAT is significantly more sensitive than CTT for both antibody detection and antigen testing. Advantages of the CAT includes standardized pipetting of reagents and specimens, macroscopic agglutination reactions, stability of results up to 24 hours, reduced inter-operator variability and significantly reduced specimen volume. However, even with CAT, antibody identification is one of the difficult tasks in serological testing as it requires in-depth knowledge of behavior of all antibodies, antibody identification rules, knowledge of zygosity(homozygous or heterozygous) of antigens and dosage phenomenon. In some cases it is comparatively simpler to identify (single alloantibody) while in others it may require great skill-set and experience (multiple antibodies).Some cases 4 Page 4 of 16
(complex multiple antibodies) creates a challenge even for a skilled and experienced technologist. Mis-interpretation in such cases may even lead to hemolytic reaction in patient. Thus proper antibody identification after reliable antibody screening technique is the “need of the hour”. Automation is happening in every arena of Transfusion Medicine and now we also have an automated system (software) to interpret specificity of allo-antibody. Such software use an algorithm based on behavior and characteristics of antibodies. Antibody identification software can support the technicians in identifying antibodies and can also guide to carry out supplementary tests to rule out other antibodies. It may be used as a supplement or may eventually replace manual method of antibody identification. The aim of this study was to evaluate the effectiveness of one such software (Resolvigen) for antibody identification compared to manual antibody identification method. There is hardly any published report on this subject. 2. Materials and methods 2.1
Study place and duration
The study was conducted at the department of Transfusion Medicine in a tertiary health-care centre in National Capital Region (NCR), India in the month of February and March 2014. 2.2
Study design
This was a retrospective observational study. 2.3
Data Collection and Manual Evaluation
Data of 238 in-patients, out-patients and apparently healthy blood donors who were alloimmunized in immuno-hematology work up records during the previous two years (2012-
5 Page 5 of 16
2013) was collected. Nine were healthy blood donors while the rest of 229 cases were patients. All cases were diagnosed to be allo- immunized by using three cell antibody screening cells (Surgiscreen, Ortho clinical diagnostics, Johnson and Johnson, USA). Antibody specificity was determined by using either eleven (Resolve A, Ortho clinical diagnostics, Johnson and Johnson, USA) or twenty two reagent cell antibody identification panels (Resolve A and B, Ortho clinical diagnostics, Johnson and Johnson, USA).Alloimmunized cases were grouped into three categories based on antibody number (single, two and multiple unexpected antibodies). 2.4
Data re-evaluation using Resolvigen software
Agglutination reaction with each identification panel cell for every allo-immunized case was re-evaluated using Resolvigen software (Ortho clinical diagnostics, Johnson and Johnson, USA). As per package insert, [2] Resolvigen is expert software that supports the antibody identification in the lab. This software gets easily integrated with automated immunohematological systems like Othro AutoVue Innova and Ultra. It easily interprets antibody identification results, suggests additional tests to confirm or exclude other antibodies, suggests supplemental methods to enhance the agglutination reaction and has a provision for online support in complex cases. Total of 238 allo-immunized cases (in-patients, out-patients and blood donors) confirmed manually (using Biovue system and manual interpretation of antigen-gram) with single, two or multiple antibody specificity were re-evaluated using Resolvigen software. Resolvigen software was evaluated for the agreement of the responses related to antibody specificity. 2.5
Evaluation of software - for usefulness, ease of use and future usage
To investigate whether this software holds good for usefulness in antibody identification, we performed a controlled experiment. Within the experiment, 41 investigators (technicians, 6 Page 6 of 16
residents and consultants) were trained to use the Resolvigen software. Thereafter, they used Resolvigen software for antibody identification in randomly chosen five cases each. Once they had completed the evaluation, they answered 11 questions of a questionnaire made for evaluating usefulness, ease of use and predicted future usage (adapted from Davis et al). According to Davis et al [3] these concepts were fundamental determinants of user acceptance, which is an important requirement for software evaluation. For measuring usefulness and ease of use, we applied the Likert scale. [4] Each investigator was asked to respond to each question in terms of answer between likely and unlikely with one of seven responses: extremely likely, quite likely, slightly likely, neither, slightly unlikely, quite unlikely, extremely unlikely. 2.6
Statistical analysis
Agreement analysis was performed to compare the results of antibody identification by two methods (manual and software based). Each response of the questionnaire was assigned a score and analyzed. Reliability of the questionnaire was assessed by using Cronbach’s alpha- measure of reliability. To study correlation between usefulness, ease of use and future usage, Pearson product correlation coefficient was used. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS), version 20.0 (SPSS Inc., Chicago, IL, USA). 2.7
Ethical clearance
Ethics committee approval was not needed since it was a retrospective observational study. 3. Results
7 Page 7 of 16
238 donors and patients (in-patients and out-patients) with single, two or multiple antibody specificity were re-evaluated using Resolvigen software. Out of 238 cases, 210 cases were allo-immunized due to single alloantibody, 28 cases were having two and 9 cases were having more than two allo-antibodies. Among the 210 single allo-immunized cases, one case exhibited discordant result when reevaluated using Resolvigen software; manually diagnosed as having single anti-E alloantibody but the software suggested that case is allo-immunized due to three alloantibodies (Anti-E, anti-c and anti-Lea). In this case, software result was found to be correct. In allo-immunized cases with two antibodies, three cases exhibited discordant results while in cases with multiple allo-antibodies out of nine cases three cases were discordant. In two and multiple allo-immunized cases results of manual method were considered decisive (Table 1). 3.1
Agreement analysis
Table 2 presents the comparison between the two methods of antibody identification (manual and software based). Overall agreement between the results of two methods ranges from 98.6% for cases allo-immunized with single antibody and 65% with two antibodies. Kappa values for single and two allo-immunized cases were 0.98 and 0.57 with p value of 0.00 respectively suggesting almost perfect agreement for single and moderate agreement for two allo-immunized cases.[5,6] For allo-immunized cases with multiple allo-antibodies there was no agreement between the two methods. 3.2
Reliability of questionnaire
Reliability is the extent to which one would obtain the same result if the same measures are repeated under the same conditions. The reliability of the questionnaire was assessed using Cronbach’s alpha which was 0.709 (highly reliable).
8 Page 8 of 16
3.3
Usefulness of Resolvigen
Figure no 1 presents the results of questions 1 to 4. Our investigators considered the Resolvigen software to be very useful as all 41investigators answered the questions as extremely likely, quite likely or slightly likely with the mean value of 6.6. Moreover, all investigators avowed that software makes their job easier and quicker (mean value is 6.6 and 6.3 respectively). 3.4
Ease of use of Resolvigen
Figure no 2 presents the results of questions 5 to 9. Our investigators considered the software easy to use. All investigators found that the software was easy to learn (mean of 6.6), easy to use, skilful and easy to remember (mean of 6.6, 6.6 and 6.8 respectively). We found that one item (Clearness) was having the slightly lower mean value of 6. 3.5
Correlating usefulness, ease of use and self-predicted future usage
Considering the Resolvigen useful and easy to use we correlated the results of usefulness and ease of use to self-predicted future usage. Table 3 shows Pearson product correlation coefficients. Both usefulness and ease of use are positively correlated to self-predicted future usage but correlation coefficient between the ease of use and future usage is much higher (0.50) than correlation coefficient between the usefulness and future usage (0.24).“p” values for the correlation coefficient between the usefulness and future usage is 0.127 while for correlation coefficient between the ease of use and future usage is 0.001 which is highly significant. 4. Discussion Our study re-evaluated 238 manually confirmed allo-immunized cases by using Resolvigen software. Out of 238 cases, only seven cases were showing discordant results. Agreement 9 Page 9 of 16
analysis has shown that there is significant agreement in interpretation of results between two methods in case of single and two allo-immunized cases. In cases with allo-immunization due to multiple allo-antibodies there was no agreement, possibly because of complexity of cases (mixture of many allo-antibodies with different behavior and characteristics) or because of smaller sample size of cases (n=9) with multiple allo-antibodies. Auto antibodies commonly interfere with these types of cases making it even more complex. Resolvigen software works well when used for evaluation of cases allo-immunized with single and two allo-antibodies. In this study, one case was manually diagnosed as having single anti-E alloantibody which on re-evaluation with Resolvigen found out to be having anti-E, c, Lea. In this case, software result was found to be correct. Anti-c and anti- Lea were missed during the manual interpretation. We checked the previous records and found that AHG compatible units were issued to the patient. This underlines the usefulness of the software in antibody interpretation. In cases with two allo-immunization three cases showed discordant results; in the first one anti-C was not diagnosed by software but was clearly present on re-confirmation by manual method. In other two cases, out of two antibodies identified by manual method one was not matching as per software interpretation. Both the cases were rechecked using manual method by stored serum samples and here results of manual method were considered decisive. To investigate whether the discrepant results were due to transcription error (as results were entered manually) every discordant sample was rechecked. Agreement analysis also shows 98.6% agreement for cases allo-immunized with single antibody while only 65% agreement between the two methods for antibodies with two antibodies allo-immunization but it was significant (p=0.000). Thus it can be easily said that Resolvigen can be easily incorporated for antibody identification in the system as most of the allo-immunized cases are due to single or two allo-antibodies. Availability of computerized help for antibody identification makes the identification faster and easier mainly for beginners. Even for the experienced 10 Page 10 of 16
investigators Resolvigen acts as a guide for resolving complex cases by giving valuable steps according to the antibody behavior. Our study also evaluates usefulness, ease of use and predicted future usage. On evaluating usefulness investigators considered the Resolvigen to be very useful (mean of 6.6). Several reasons are responsible for this high mean value: as interpretation by using this software makes their job easier as they have to just enter the findings in the software and results are available within seconds. There was no interpretation variability from one investigator to another. Many of the investigators also said that with use of this software obviated the need to remember the behavior of each allo-antibody. Hence our users confirmed that Resolvigen was very useful for facilitating their work. Considering the ease of use of software our investigators found learning to use Resolvigen was very easy (mean of 6.6). It was easy to remember steps of using the software for all investigators (mean of 6.8). All investigators found the software competent enough to identify allo-antibodies quickly (mean of 6.6). Only one item (clear) had slightly lower mean rating of 6.02. To find out how users would accept this software for their future usage we used Pearson product correlation coefficients. Correlation coefficient between the ease of use and future usage was much higher (0.50) than between usefulness and future usage (0.24). Figure 3 also shows mean of items used for evaluating future usage was less (mean of 6) in comparison with ease of use and usefulness (mean of 6.5 and 6.6 respectively). This was because users would want to use the software in future because of ease but at the same time they did not want to be over dependent on it and were worried that they run the risk of forgetting the conventional manual method.
11 Page 11 of 16
When comparing with other studies; Krishnan et al [7] from France showed that there was no significant difference in the results of antibody identification by manual method and software using Resolvigen. Vitali and co-workers [8] also found the Resolvigen software useful and concluded that software acts an aid to antibody identification process. 5. Conclusion This paper reiterates the fact that Resolvigen software can either replace manual method or can be used as an adjuvant to routine manual method of antibody identification. Resolvigen not only works well for antibody identification in single or two allo-immunized cases but also provides help in complex cases. We measured the usefulness, ease of use and predicted future usage in the form of questionnaire and investigators considered the Resolvigen very useful and easy to use. We will have to add that, Resolvigen not only acts as an aid to routine antibody identification but also helps us to store data electronically avoiding misuse, loss and significantly decreasing our efforts in data retrieval. 6.
Authors’ Contribution
Aseem Kumar Tiwari and Ravi C Dara conceptualized and designed the study; Dinesh Arora Geet Aggarwal, Ganesh Rawat, Subhasis Mitra and Prashant Pandey helped in compiling and analyzing data; data was interpreted and manuscript was prepared by Ravi C Dara and Aseem Kumar Tiwari; Prashant Pandey and Vimarsh Raina revised the manuscript.
7.
Conflict of Interest
The authors do not have any conflict of interest.
12 Page 12 of 16
8. References [1] Rosse WF, Gallagher D, Kinney TR, et al. Transfusion and allo-immunization in sickle cell disease. The cooperative study of sickle cell disease. Blood. 1990;76:14311437. [2] Resolvigen [package insert].USA: Ortho clinical diagnostics; 2005. [3] Davis FD. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quaterly.1989; 319-340. [4] McIver JP, Carmines EG. Unidimensional Scaling. Sage publications; 1991. [5] George D, Mallery P. SPSS for Windows Step By Step: A Simple Guide And Reference. 4thed. Boston: Allyn & Bacon; 2003. [6] Kline P. The Handbook of Psychological Testing. 2nded. London: Routledge; 2000 [7] Krishnan K, Casina T: Comparison of antibody identification results by Resolvigen 3 antibody identification software and manual methodology [abstract]. Asian J Tansfus Sci. 2008; 2: 40. [8]Vitali E, Erminia C, Parolo R et al. Evaluation of different concentrations of red blood cell suspensions and a dedicated software for the identification of antibodies to red blood cells. Blood Transfus. 2006; 4:151-157.
13 Page 13 of 16
Figure 1: Figure showing usefulness of Resolvigen Figure 2: Figure showing ease of use of Resolvigen Figure 3: Figure comparing usefulness, ease of use and self-predicted future usage
Table 1: Discordant results between the two methods (manual and software) of antibody identification
No of cases 1 1 2 3 1 2 3
Manual work up Resolvigen Single alloantibody Anti E Anti E, c, Lea Two alloantibody Anti D, C Anti D Anti Jka, S Anti C, S, M Anti M, S Anti Lea, M, P1 Multiple alloantibodies Anti c, E, Jka Anti c, N (not excluded E, K) Anti c, E, Jka Anti c, N (not excluded E, K) Anti c, E, Jka Anti c, N (not excluded E, K)
14 Page 14 of 16
Table 2: Percentage agreement analysis between the two methods of antibody (manual and software) identification
Alloimmunisation
Agreement between two groups (%)
Single
98.6%
TwoTwo
65%
Multiple
Kappa
0.98
0.57
SE (%)
Confidence Interval (95%)
pvalue
LCI
UCI
21.0
89.8
98.1
0.000
11.1
35.4
79.0
0.000
No agreement
15 Page 15 of 16
Table 3: Correlating usefulness, ease of use and self-predicted future usage of Resolvigen
Usefulness Ease of use Usefulness 1 .10 P value 0.526 Ease of use .10 1 P value 0.52 Self-predicted future usage .24 .50** P value 0.127 0.001 ** Correlation is significant at the 0.01 level (2-tailed).
Self-predicted future usage .24 0.127 .50** 0.001 1
16 Page 16 of 16
S1473-0502(14)00159-1 http://dx.doi.org/doi:10.1016/j.transci.2014.08.020 TRASCI 1727
To appear in:
Transfusion and Apheresis Science
Received date: Revised date: Accepted date:
29-5-2014 2-8-2014 25-8-2014
Please cite this article as: Aseem Kumar Tiwari, Ravi C Dara, Dinesh Arora, Geet Aggarwal, Ganesh Rawat, Subhasis Mitra, Pandey K Prashant, Vimarsh Raina, Allo-antibody Identification: A software approach!, Transfusion and Apheresis Science (2014), http://dx.doi.org/doi:10.1016/j.transci.2014.08.020. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
TITLE: Allo-antibody Identification: A software approach! RUNNING TITLE: Allo-antibody Identification: A software approach! AUTHORS: Aseem Kumar Tiwari, MD Senior Consultant, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Ravi C Dara, MD Senior Resident, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Dinesh Arora, DCP Associate Consultant, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Geet Aggarwal, MBBS Junior Resident, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Ganesh Rawat, DMLT Senior Technologist, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail:[email protected] Subhasis Mitra, Intern, Department of Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] Pandey K Prashant, MD Consultant, Department of Transfusion Medicine Jaypee Hospital, Sector-128, Noida-201304 [email protected] Vimarsh Raina, MD Director, Laboratory Services and Transfusion Medicine Medanta-The Medicity, Sector-38, Gurgaon-122001 e-mail: [email protected] CORRESPONDING AUTHOR Aseem K Tiwari, MD Senior Consultant, Department of Transfusion Medicine 1 Page 1 of 16
Medanta-The Medicity, Sector-38, Gurgaon-122001 Email: [email protected] Fax: +91-124-483411 Phone No: +91-124-4411441 Ext-2701, 2718
2 Page 2 of 16
Abstract Background: Unexpected allo-antibody identification is difficult serological test requiring in-depth knowledge of antibody behavior, identification rules, knowledge of zygosity of antigens and dosage phenomenon. Software which uses an algorithm based on characteristics of antibodies is now available to interpret specificity of allo-antibody. A study was undertaken to evaluate the effectiveness of one such software (Resolvigen) for antibody identification compared to manual antibody identification method. Materials and Methods: The study was a retrospective observational study where 238 allo-antibody results were reevaluated using Resolvigen software (Ortho clinical diagnostics, Johnson and Johnson, USA) and agreement between manual and software approaches was studied. Resolvigen software was also evaluated for usefulness, ease of use and predicted future usage by administering investigators a questionnaire with Likert scale. Results: Agreement between the results of manual and automated methods ranged from 98.6% for single antibody to 65% for two antibodies (p=0.000). Resolvigen software came out as very useful, easy to use, and with high predicted future usage. Conclusion: This study concludes that Resolvigen can either replace manual method or be used as adjuvant to routine manual method.
Keywords: Allo-antibody, Resolvigen, Allo-immunization, Irregular antibody
3 Page 3 of 16
1. Introduction Antibody screening is done to detect red cells antibodies other than expected anti-A and anti-B. These are therefore called ‘unexpected’ antibodies. Only 0.3 to 1.0% of general population have unexpected antibodies1 and the incidence is higher in women due to the pregnancy and patients receiving multiple transfusions.[1] The unexpected alloantibodies are immune antibodies formed by exposure to red cells due to pregnancy or blood transfusion. Methods such as Conventional Tube Technique (CTT), Column Agglutination Technique (CAT) and Solid Phase Red Cell Adherence (SPRCA) are usually used to detect unexpected antibodies during the pre-transfusion testing. The use of antibody screening using column agglutination technology (CAT) is on the rise in blood transfusion services in India. In CAT, sephadex gel or glass-beads in the column are used to capture agglutinates. CAT is significantly more sensitive than CTT for both antibody detection and antigen testing. Advantages of the CAT includes standardized pipetting of reagents and specimens, macroscopic agglutination reactions, stability of results up to 24 hours, reduced inter-operator variability and significantly reduced specimen volume. However, even with CAT, antibody identification is one of the difficult tasks in serological testing as it requires in-depth knowledge of behavior of all antibodies, antibody identification rules, knowledge of zygosity(homozygous or heterozygous) of antigens and dosage phenomenon. In some cases it is comparatively simpler to identify (single alloantibody) while in others it may require great skill-set and experience (multiple antibodies).Some cases 4 Page 4 of 16
(complex multiple antibodies) creates a challenge even for a skilled and experienced technologist. Mis-interpretation in such cases may even lead to hemolytic reaction in patient. Thus proper antibody identification after reliable antibody screening technique is the “need of the hour”. Automation is happening in every arena of Transfusion Medicine and now we also have an automated system (software) to interpret specificity of allo-antibody. Such software use an algorithm based on behavior and characteristics of antibodies. Antibody identification software can support the technicians in identifying antibodies and can also guide to carry out supplementary tests to rule out other antibodies. It may be used as a supplement or may eventually replace manual method of antibody identification. The aim of this study was to evaluate the effectiveness of one such software (Resolvigen) for antibody identification compared to manual antibody identification method. There is hardly any published report on this subject. 2. Materials and methods 2.1
Study place and duration
The study was conducted at the department of Transfusion Medicine in a tertiary health-care centre in National Capital Region (NCR), India in the month of February and March 2014. 2.2
Study design
This was a retrospective observational study. 2.3
Data Collection and Manual Evaluation
Data of 238 in-patients, out-patients and apparently healthy blood donors who were alloimmunized in immuno-hematology work up records during the previous two years (2012-
5 Page 5 of 16
2013) was collected. Nine were healthy blood donors while the rest of 229 cases were patients. All cases were diagnosed to be allo- immunized by using three cell antibody screening cells (Surgiscreen, Ortho clinical diagnostics, Johnson and Johnson, USA). Antibody specificity was determined by using either eleven (Resolve A, Ortho clinical diagnostics, Johnson and Johnson, USA) or twenty two reagent cell antibody identification panels (Resolve A and B, Ortho clinical diagnostics, Johnson and Johnson, USA).Alloimmunized cases were grouped into three categories based on antibody number (single, two and multiple unexpected antibodies). 2.4
Data re-evaluation using Resolvigen software
Agglutination reaction with each identification panel cell for every allo-immunized case was re-evaluated using Resolvigen software (Ortho clinical diagnostics, Johnson and Johnson, USA). As per package insert, [2] Resolvigen is expert software that supports the antibody identification in the lab. This software gets easily integrated with automated immunohematological systems like Othro AutoVue Innova and Ultra. It easily interprets antibody identification results, suggests additional tests to confirm or exclude other antibodies, suggests supplemental methods to enhance the agglutination reaction and has a provision for online support in complex cases. Total of 238 allo-immunized cases (in-patients, out-patients and blood donors) confirmed manually (using Biovue system and manual interpretation of antigen-gram) with single, two or multiple antibody specificity were re-evaluated using Resolvigen software. Resolvigen software was evaluated for the agreement of the responses related to antibody specificity. 2.5
Evaluation of software - for usefulness, ease of use and future usage
To investigate whether this software holds good for usefulness in antibody identification, we performed a controlled experiment. Within the experiment, 41 investigators (technicians, 6 Page 6 of 16
residents and consultants) were trained to use the Resolvigen software. Thereafter, they used Resolvigen software for antibody identification in randomly chosen five cases each. Once they had completed the evaluation, they answered 11 questions of a questionnaire made for evaluating usefulness, ease of use and predicted future usage (adapted from Davis et al). According to Davis et al [3] these concepts were fundamental determinants of user acceptance, which is an important requirement for software evaluation. For measuring usefulness and ease of use, we applied the Likert scale. [4] Each investigator was asked to respond to each question in terms of answer between likely and unlikely with one of seven responses: extremely likely, quite likely, slightly likely, neither, slightly unlikely, quite unlikely, extremely unlikely. 2.6
Statistical analysis
Agreement analysis was performed to compare the results of antibody identification by two methods (manual and software based). Each response of the questionnaire was assigned a score and analyzed. Reliability of the questionnaire was assessed by using Cronbach’s alpha- measure of reliability. To study correlation between usefulness, ease of use and future usage, Pearson product correlation coefficient was used. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS), version 20.0 (SPSS Inc., Chicago, IL, USA). 2.7
Ethical clearance
Ethics committee approval was not needed since it was a retrospective observational study. 3. Results
7 Page 7 of 16
238 donors and patients (in-patients and out-patients) with single, two or multiple antibody specificity were re-evaluated using Resolvigen software. Out of 238 cases, 210 cases were allo-immunized due to single alloantibody, 28 cases were having two and 9 cases were having more than two allo-antibodies. Among the 210 single allo-immunized cases, one case exhibited discordant result when reevaluated using Resolvigen software; manually diagnosed as having single anti-E alloantibody but the software suggested that case is allo-immunized due to three alloantibodies (Anti-E, anti-c and anti-Lea). In this case, software result was found to be correct. In allo-immunized cases with two antibodies, three cases exhibited discordant results while in cases with multiple allo-antibodies out of nine cases three cases were discordant. In two and multiple allo-immunized cases results of manual method were considered decisive (Table 1). 3.1
Agreement analysis
Table 2 presents the comparison between the two methods of antibody identification (manual and software based). Overall agreement between the results of two methods ranges from 98.6% for cases allo-immunized with single antibody and 65% with two antibodies. Kappa values for single and two allo-immunized cases were 0.98 and 0.57 with p value of 0.00 respectively suggesting almost perfect agreement for single and moderate agreement for two allo-immunized cases.[5,6] For allo-immunized cases with multiple allo-antibodies there was no agreement between the two methods. 3.2
Reliability of questionnaire
Reliability is the extent to which one would obtain the same result if the same measures are repeated under the same conditions. The reliability of the questionnaire was assessed using Cronbach’s alpha which was 0.709 (highly reliable).
8 Page 8 of 16
3.3
Usefulness of Resolvigen
Figure no 1 presents the results of questions 1 to 4. Our investigators considered the Resolvigen software to be very useful as all 41investigators answered the questions as extremely likely, quite likely or slightly likely with the mean value of 6.6. Moreover, all investigators avowed that software makes their job easier and quicker (mean value is 6.6 and 6.3 respectively). 3.4
Ease of use of Resolvigen
Figure no 2 presents the results of questions 5 to 9. Our investigators considered the software easy to use. All investigators found that the software was easy to learn (mean of 6.6), easy to use, skilful and easy to remember (mean of 6.6, 6.6 and 6.8 respectively). We found that one item (Clearness) was having the slightly lower mean value of 6. 3.5
Correlating usefulness, ease of use and self-predicted future usage
Considering the Resolvigen useful and easy to use we correlated the results of usefulness and ease of use to self-predicted future usage. Table 3 shows Pearson product correlation coefficients. Both usefulness and ease of use are positively correlated to self-predicted future usage but correlation coefficient between the ease of use and future usage is much higher (0.50) than correlation coefficient between the usefulness and future usage (0.24).“p” values for the correlation coefficient between the usefulness and future usage is 0.127 while for correlation coefficient between the ease of use and future usage is 0.001 which is highly significant. 4. Discussion Our study re-evaluated 238 manually confirmed allo-immunized cases by using Resolvigen software. Out of 238 cases, only seven cases were showing discordant results. Agreement 9 Page 9 of 16
analysis has shown that there is significant agreement in interpretation of results between two methods in case of single and two allo-immunized cases. In cases with allo-immunization due to multiple allo-antibodies there was no agreement, possibly because of complexity of cases (mixture of many allo-antibodies with different behavior and characteristics) or because of smaller sample size of cases (n=9) with multiple allo-antibodies. Auto antibodies commonly interfere with these types of cases making it even more complex. Resolvigen software works well when used for evaluation of cases allo-immunized with single and two allo-antibodies. In this study, one case was manually diagnosed as having single anti-E alloantibody which on re-evaluation with Resolvigen found out to be having anti-E, c, Lea. In this case, software result was found to be correct. Anti-c and anti- Lea were missed during the manual interpretation. We checked the previous records and found that AHG compatible units were issued to the patient. This underlines the usefulness of the software in antibody interpretation. In cases with two allo-immunization three cases showed discordant results; in the first one anti-C was not diagnosed by software but was clearly present on re-confirmation by manual method. In other two cases, out of two antibodies identified by manual method one was not matching as per software interpretation. Both the cases were rechecked using manual method by stored serum samples and here results of manual method were considered decisive. To investigate whether the discrepant results were due to transcription error (as results were entered manually) every discordant sample was rechecked. Agreement analysis also shows 98.6% agreement for cases allo-immunized with single antibody while only 65% agreement between the two methods for antibodies with two antibodies allo-immunization but it was significant (p=0.000). Thus it can be easily said that Resolvigen can be easily incorporated for antibody identification in the system as most of the allo-immunized cases are due to single or two allo-antibodies. Availability of computerized help for antibody identification makes the identification faster and easier mainly for beginners. Even for the experienced 10 Page 10 of 16
investigators Resolvigen acts as a guide for resolving complex cases by giving valuable steps according to the antibody behavior. Our study also evaluates usefulness, ease of use and predicted future usage. On evaluating usefulness investigators considered the Resolvigen to be very useful (mean of 6.6). Several reasons are responsible for this high mean value: as interpretation by using this software makes their job easier as they have to just enter the findings in the software and results are available within seconds. There was no interpretation variability from one investigator to another. Many of the investigators also said that with use of this software obviated the need to remember the behavior of each allo-antibody. Hence our users confirmed that Resolvigen was very useful for facilitating their work. Considering the ease of use of software our investigators found learning to use Resolvigen was very easy (mean of 6.6). It was easy to remember steps of using the software for all investigators (mean of 6.8). All investigators found the software competent enough to identify allo-antibodies quickly (mean of 6.6). Only one item (clear) had slightly lower mean rating of 6.02. To find out how users would accept this software for their future usage we used Pearson product correlation coefficients. Correlation coefficient between the ease of use and future usage was much higher (0.50) than between usefulness and future usage (0.24). Figure 3 also shows mean of items used for evaluating future usage was less (mean of 6) in comparison with ease of use and usefulness (mean of 6.5 and 6.6 respectively). This was because users would want to use the software in future because of ease but at the same time they did not want to be over dependent on it and were worried that they run the risk of forgetting the conventional manual method.
11 Page 11 of 16
When comparing with other studies; Krishnan et al [7] from France showed that there was no significant difference in the results of antibody identification by manual method and software using Resolvigen. Vitali and co-workers [8] also found the Resolvigen software useful and concluded that software acts an aid to antibody identification process. 5. Conclusion This paper reiterates the fact that Resolvigen software can either replace manual method or can be used as an adjuvant to routine manual method of antibody identification. Resolvigen not only works well for antibody identification in single or two allo-immunized cases but also provides help in complex cases. We measured the usefulness, ease of use and predicted future usage in the form of questionnaire and investigators considered the Resolvigen very useful and easy to use. We will have to add that, Resolvigen not only acts as an aid to routine antibody identification but also helps us to store data electronically avoiding misuse, loss and significantly decreasing our efforts in data retrieval. 6.
Authors’ Contribution
Aseem Kumar Tiwari and Ravi C Dara conceptualized and designed the study; Dinesh Arora Geet Aggarwal, Ganesh Rawat, Subhasis Mitra and Prashant Pandey helped in compiling and analyzing data; data was interpreted and manuscript was prepared by Ravi C Dara and Aseem Kumar Tiwari; Prashant Pandey and Vimarsh Raina revised the manuscript.
7.
Conflict of Interest
The authors do not have any conflict of interest.
12 Page 12 of 16
8. References [1] Rosse WF, Gallagher D, Kinney TR, et al. Transfusion and allo-immunization in sickle cell disease. The cooperative study of sickle cell disease. Blood. 1990;76:14311437. [2] Resolvigen [package insert].USA: Ortho clinical diagnostics; 2005. [3] Davis FD. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quaterly.1989; 319-340. [4] McIver JP, Carmines EG. Unidimensional Scaling. Sage publications; 1991. [5] George D, Mallery P. SPSS for Windows Step By Step: A Simple Guide And Reference. 4thed. Boston: Allyn & Bacon; 2003. [6] Kline P. The Handbook of Psychological Testing. 2nded. London: Routledge; 2000 [7] Krishnan K, Casina T: Comparison of antibody identification results by Resolvigen 3 antibody identification software and manual methodology [abstract]. Asian J Tansfus Sci. 2008; 2: 40. [8]Vitali E, Erminia C, Parolo R et al. Evaluation of different concentrations of red blood cell suspensions and a dedicated software for the identification of antibodies to red blood cells. Blood Transfus. 2006; 4:151-157.
13 Page 13 of 16
Figure 1: Figure showing usefulness of Resolvigen Figure 2: Figure showing ease of use of Resolvigen Figure 3: Figure comparing usefulness, ease of use and self-predicted future usage
Table 1: Discordant results between the two methods (manual and software) of antibody identification
No of cases 1 1 2 3 1 2 3
Manual work up Resolvigen Single alloantibody Anti E Anti E, c, Lea Two alloantibody Anti D, C Anti D Anti Jka, S Anti C, S, M Anti M, S Anti Lea, M, P1 Multiple alloantibodies Anti c, E, Jka Anti c, N (not excluded E, K) Anti c, E, Jka Anti c, N (not excluded E, K) Anti c, E, Jka Anti c, N (not excluded E, K)
14 Page 14 of 16
Table 2: Percentage agreement analysis between the two methods of antibody (manual and software) identification
Alloimmunisation
Agreement between two groups (%)
Single
98.6%
TwoTwo
65%
Multiple
Kappa
0.98
0.57
SE (%)
Confidence Interval (95%)
pvalue
LCI
UCI
21.0
89.8
98.1
0.000
11.1
35.4
79.0
0.000
No agreement
15 Page 15 of 16
Table 3: Correlating usefulness, ease of use and self-predicted future usage of Resolvigen
Usefulness Ease of use Usefulness 1 .10 P value 0.526 Ease of use .10 1 P value 0.52 Self-predicted future usage .24 .50** P value 0.127 0.001 ** Correlation is significant at the 0.01 level (2-tailed).
Self-predicted future usage .24 0.127 .50** 0.001 1
16 Page 16 of 16
