AJCP / Original Article
Mobile Computing Platform With Decision Support Modules for Hemotherapy Richard S. P. Huang, MD,1 Elena Nedelcu, MD,1 Yu Bai, MD,1 Amer Wahed, MD,1 Kimberly Klein, MD,1 Igor Gregoric, MD,2 Manish Patel, MD,2 Biswajit Kar, MD,2 Pranav Loyalka, MD,2 Sriram Nathan, MD,2 Paul Loubser, MD,3 Phillip A. Weeks, PharmD,4 Rajko Radovancevic, MD,2 and Andy N. D. Nguyen, MD1 From the 1Department of Pathology and Laboratory Medicine, 2Center for Advanced Heart Failure, and 3Cardiovascular Anesthesia, The University of Texas Health Science Center at Houston, Houston; and 4Texas Medical Center, Memorial Hermann Hospital, Houston. Key Words: Hemotherapy; Coagulation; Mobile computing platform; Decision support modules; Algorithms Am J Clin Pathol June 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
ABSTRACT Objectives: We describe the development of a mobile computing platform (MCP) with a decision support module (DSM) for patients in our coagulation-based hemotherapy service. Methods: The core of our MCP consists of a Microsoft Excel spreadsheet template used to gather and compute data on cardiopulmonary bypass (CPB) patients intraoperatively. The DSM is embedded into the Excel file, where the user would enter in laboratory results, and through our 45 embedded algorithms, recommendations for transfusion products would be displayed in the Excel file. Results: The DSM has helped decrease the time it takes to come to a transfusion recommendation, helps double-check recommendations, and is an excellent tool for teaching. Furthermore, the problems that occur with a paper system have been eliminated, and we are now able to access this information easily and reliably. Conclusions: The development and implementation of our MCP system has greatly increased the productivity and efficiency of our hemotherapy service.
834 834
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
Our pathology department started a coagulation-based hemotherapy service in May 2012. This is a pathology consultative service on coagulation issues that occur perioperatively and postoperatively in patients with heart failure. These patients undergo procedures such as aortocoronary bypass, cardiac valve repair or replacement, implantation of a left ventricular assist device (LVAD), and heart transplant. During almost all of these procedures, the patient is placed on cardiopulmonary bypass (CPB), where an extracorporeal mechanical pump temporarily takes over the function of the heart and lungs. Anesthesiologists and surgeons traditionally manage the coagulation issues of these surgical patients; however, due to the complexity and time restraints of CPB cases, pathologists at our institution are consulted to help manage these patients. We perform a preoperative evaluation for risk of bleeding prior to CPB, monitor the patient with consultation on transfusion during the operation, and assess the patient postoperatively with consultation on transfusion.1 These consultations are available 24 hours a day, 7 days a week. At our institute, one hemotherapy physician is on service for 1 week at a time, 24 hours a day. Surgeons usually consult the hemotherapy service physician during the surgery as well as for postoperative follow-up and management. However, cardiologists also often consult our service during postoperative bleeding. We write a consultation report for the surgery and progress notes on the patients postoperatively. Usually, we would follow the patients until the patient’s bleeding is under control or the patient’s coagulation status is optimized. Since the surgeons and cardiologists delegate transfusion support of patients to our service, we directly order laboratory tests and transfusions. The hospital’s Advanced Heart Failure Clinical Committee © American Society for Clinical Pathology
AJCP / Original Article
Materials and Methods
has approved this process. In this article, we focus mainly on the intraoperative aspect of our hemotherapy service. With no gold standard in determining the overall coagulation status of a patient, we use various laboratory tests to assess the coagulation status of the CBP patients. The standard tests we run on these patients include complete blood count (CBC), a disseminated intravascular coagulation (DIC) panel (prothrombin time [PT], partial thromboplastin time [PTT], fibrinogen, thrombin time, and d-dimer), antithrombin (AT), VerifyNowP2Y12 (VFN-P) for patients on clopidogrel or other adenosine diphosphate P2Y12 receptor antagonists, VerifyNow-ASA (VFN-A) for patients on aspirin, thromboelastography (TEG), and thromboelastography with heparinase (hTEG). Due to the sheer volume of data from these tests, we developed a mobile computing platform (MCP) to assimilate the laboratory data into a comprehensive data collection system that may be viewed by the pathologist in an easy and efficient manner. Furthermore, electronic medical records (EMRs) have been shown to support evidence-based decisions,2 and we have embedded decision support modules (DSMs) within our MCP to do precisely that. Our DSM enables the consultant pathologist to make accurate and quick recommendations to the surgeons and anesthesiologists.
Our MCP is composed of a software component and a hardware component. We describe the software component first. The software component centers on a Microsoft Excel (Microsoft, Redmond, WA) spreadsheet template. The template is used intraoperatively during CPB. In the template, the pathologist starts with data entry fields (DEFs) to enter pertinent information about the patient, including general demographic identifiers, medications, preoperative laboratory results, and coagulation risk assessment results ❚Image 1A❚. Dropdown menus are available in certain fields, such as names of the attending physicians, to increase the speed of data entry into the spreadsheet. In the next section, there are DEFs in which the pathologist can record data for the aforementioned set of tests done on the patient ❚Image 1B❚. The test result fields would be automatically highlighted if the values were abnormal (green for low values and red for high values). Here the values entered include reaction time (R) (from the TEG); heparinase reaction time (h-R), heparinase angle (h-A), heparinase maximum amplitude (h-MA), heparinase estimated percent lysis (h-EPL), heparinase lysis at 30 minutes (h-Ly30), and heparinase coagulation index (h-CI) (from the hTEG); hemoglobin (Hgb) and platelet count (from
A
B
C
D
❚Image 1❚ Sample of the intraoperative Excel (Microsoft, Redmond, WA) template. A, Data entry field (DEF) to enter pertinent information about the patient. B, DEF to record coagulation laboratory values perioperatively. C, Fields displaying summary of suggestions. D, DEF to record actual blood usage, time on pump, and chest tube drainage. For abbreviations, see Table 1.
© American Society for Clinical Pathology
835
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
835 835
Huang et al / Mobile Computing Platform for Hemotherapy
the CBC); PT/PTT, fibrinogen, thrombin time, and d-dimer (from the DIC screen); AT; VFN-A; and VFN-P. These tests would be sequentially performed and entered in the Excel spreadsheet during different phases of CPB, which include baseline, hemoconcentration/rewarming phase (just prior to termination of CPB), post-CPB phase, and during the potential postoperative bleeding phase. Data validation for input entry is also provided. If the input is not in an acceptable range or in the incorrect format, the data entry would be stopped with a warning message. As the user enters laboratory results into the MCP, the values would be automatically routed through the DSM in the Excel file. The DSM comprises 45 algorithms that are translated to Excel functions. The hemotherapy physicians at our institute, consisting of hematopathologists and blood bankers, developed the algorithms for the DSM by reviewing the literature for algorithms used during CPB surgery; hence, the algorithms are based on standard transfusion practice.3,4 Our group prepared a comprehensive hemotherapy guide, which includes all the algorithms, and the current version of this guide can be reviewed on our website at http://hemepathreview.com (go to Item 14, then Guide). Criteria for transfusion are listed in the References section of this guide. Furthermore, our group reviewed the algorithms, which are continuously revised based on experience learned on our service. The guide and the algorithms within them are also annually reviewed and approved by the hospital’s Advanced Heart Failure Clinical Committee. The algorithms consist of 36 level 1 algorithms ❚Table 1❚ and 9 level 2 algorithms ❚Table 2❚. Level 1 algorithms derive transfusion needs based on specific sets of laboratory tests. Level 2 algorithms combine results from level 1 algorithms for final transfusion suggestions. These recommendations would appear under “Suggestion Summary” and consist of the following fields: fresh-frozen plasma (FFP; single units), cryoprecipitate (doses), platelets (doses or apheresis units), RBCs (units), tranexamic acid (grams), protamine (milligrams), alert for use of factor VIIa (FVIIa), alert for use of AT concentrate, and alert for possible protamine overdose ❚Image 1C❚. The 36 level 1 algorithms can be divided into 10 different product/recommendation categories: (1) RBC transfusion for low Hgb; (2) FFP transfusion for low clotting factors, low AT, or low fibrinogen; (3) cryoprecipitate transfusion for very low fibrinogen or uremia; (4) AT concentrate transfusion for very low AT; (5) no FFP transfusion for prolonged PTT secondary to heparin; (6) platelet transfusion for low platelet count or medication effect; (7) tranexamic acid for primary fibrinolysis; (8) protamine for excess heparin after neutralization; (9) FVIIa contraindicated in secondary fibrinolysis; and (10) alert for possible protamine overdose. Of note, we currently do not have plans to include prothrombin complex concentrate since the four-factor 836 836
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
prothrombin complex concentrate has not been made available at our hospital. Also, fibrinogen concentrate has not been used by our service since we find cryoprecipitate adequate to restore fibrinogen level. Last, no DSM has been developed for the new anticoagulants dabigatran, rivaroxaban, and apixaban, even though we find these anticoagulants problematic for emergency cardiac surgery. Instead, we have guidelines for monitoring these medications and alleviating bleeding in our online guide. For illustration, let us look at a clinical scenario at the time of off-pump with the following laboratory results: Hgb, 6.7 g/dL; platelets, 55 ×103/mL; PT, 28.4 seconds; PTT, 210 seconds; fibrinogen, 186 mg/dL; and thrombin time, 110 seconds. When the user enters data into the DEF, the system would route those values to the DSM and consider the following level 1 algorithms: 1. Three RBCs to increase Hgb from 6.7 g/dL to above 10 g/dL 5. Two FFPs for low fibrinogen at 186 mg/dL 10. Four FFPs for prolonged PT at 28.4 seconds 13. No FFP for prolonged PTT at 210 seconds (this is due to the heparin effect with thrombin time at 110 seconds) 16. Two doses of platelets for platelet count at 55 ×103/mL 29. Protamine (50 mg) for excess heparin (prolonged PTT and thrombin time) Level 2 algorithms are then activated to combine results from level 1 algorithms for the actual number of blood components needed: I. Number of FFPs: maximum value from (5), (10) = maximum of 2 and 4 = 4 III. Number of platelets: from (16) = 2 IV. Number of RBCs: – To correct for FFP transfusion = 0.5 × FFPs = 0.5 × 4 = 2 – To correct for platelet transfusion = platelets = 2 – To correct for low Hgb: from (1) = 3 – Total RBCs = 2 + 2 + 3 = 7 VI. Protamine: from (29) = 50 mg The final results from level 2 algorithms are displayed in the “Suggestion Summary” (Image 1C). Note that fields such as Cryoprecipitate and Tranexamic Acid are left blank since their associated algorithms (both level 1 and level 2) have not been activated with the given laboratory results. As the laboratory data of the patient are entered, the pathologist can review the recommendations produced by the DSM and combine them with his or her clinical judgment to make a final recommendation to the surgeon/ anesthesiologist. It cannot be overemphasized that clinical judgment is critical for a final transfusion decision based on the clinical situation. In particular, the number of blood components should be based on the degree of microvascular © American Society for Clinical Pathology
AJCP / Original Article
bleeding and the patient’s bodyweight. The dosage recommendation from our DSM can be used as a starting point for further adjustment. The suggested transfusion is most applicable for the postpump phase, when most patients will exhibit some degree of microvascular bleeding. For suggested transfusion shown in other phases (baseline, hemoconcentration), transfusion typically is not needed. The suggestions for these phases just need to be kept in mind for later management. Also, there are DEFs on the template for
the pathologist to enter actual blood component usage, time on pump, and chest tube drain output ❚Image 1D❚ for future reference if needed. The hardware component of the MCP centers on Ultrabooks (Dell, Round Rock, TX), which have an encrypted hard drive that is password protected. The Excel program and the DSM templates are preloaded onto the Ultrabooks, so once the user securely logs onto the laptop, he or she can access the templates. Also, the templates can be downloaded
❚Table 1❚ Thirty-Six Level 1 Algorithms and Their Corresponding Excel Functionsa Level
Algorithm
Representative Excel Formulab
1. RBCs for anemia If Hgb =10,”0”, to 10.0 g/dL (1 RBC increased Hgb by 1.0 g/dL) ROUND(10-C33,0))) 2. FFPs for low clotting factors If h-R 10-15 min → 2 units of FFP =IF(AND(C2710),”2”,”0”) 3. FFPs for low clotting factors If h-R 15-20 min → 4 units of FFP =IF(AND(C2715),”4”,”0”) 4. FFPs for low clotting factors If h-R >20 min → 6 units of FFP =IF(C27>20,”6”,”0”) 5. FFPs for low fibrinogen If fibrinogen 150-200 mg/dL → 2 units of FFP =IF(AND(C37150),”2”,”0”) 6. Cryoprecipitate for very low fibrinogen If fibrinogen 0,C3750 mm → 2 units of FFP =IF(AND(C2820, C29>50),”2”,”0”) 8. Cryoprecipitate for very low fibrinogen If h-A 50 mm → 1 dose =IF(AND(C280,C29>50),”1”,”0”) of cryoprecipitate 9. FFPs for low clotting factors If PT 20-25 s → 2 units of FFP =IF(AND(C35>20,C3525 s → 4 units of FFP =IF(C35>25, “4”,”0”) 11. FFPs for low clotting factors If PTT 45-50 s → 2 units of FFP (however, if due to the =IF(C68=”hep effect”,”0”,(IF(AND(C36>45, heparin effect in algorithm 13, then disregard value), C3650 s → 4 units of FFP (however, if due to heparin =IF(C68=”hep effect”,”0”,(IF(C36>50, effect in algorithm 13, then disregard value), not ”4”,”0”))) applicable to on-pump time 13. No FFP due to heparin effect If PTT >45 s, TT >25 s → due to heparin effect =IF(AND(C36>45,C38>25), “hep effect”, “xxxx”) 14. FFPs for low ATIII If ATIII 35%-50% → 2 units FFP, applicable to baseline only =IF(AND(C40=35),”2”,”0”) 15. ATIII concentration for very low ATIII If ATIII 0,C40 50,C340,C34 70 mm → no FVIIa =IF(AND(C30>15,C29>70),”No FVIIa!”,”0”) 31. Secondary fibrinolysis, no FVIIa If h-EPL >15%, h-CI >3 → no FVIIa =IF(AND(C30>15,C32>3),”No FVIIa!”,”0”) 32. Secondary fibrinolysis, no FVIIa If h-Ly30 >8%, h-MA >70 mm → no FVIIa =IF(AND(C31>8,C29>70),”No FVIIa!”,”0”) 33. Secondary fibrinolysis, no FVIIa If h-Ly30 >8%, h-CI >3 → no FVIIa =IF(AND(C31>8,C32>3),”No FVIIa!”,”0”) 34. Possible protamine overdose If h-R >20 s, h-A 0, F280, pump and after, no bleeding seen) → yes to possible F290, C41, “xxxx”) AA, arachidonic acid; ADP, adenosine diphosphate; ARU, aspirin reaction units; ATIII, antithrombin III; CRF, chronic renal failure; FEU, fibrogen equivalent units; FFP, freshfrozen plasma; FVIIa, factor VIIa; h-A, heparinase angle; h-CI, heparinase coagulation index; h-EPL, heparinase estimated percent lysis; Hgb, hemoglobin; h-Ly30, heparinase lysis at 30 minutes; h-MA, heparinase maximum amplitude; h-R, heparinase reaction time; PRU, P2Y12 reaction units; PT, prothrombin time; PTT, partial thromboplastin time; TT, thrombin time; VFN, VerifyNow; VFN-A, VerifyNow-aspirin; VFN-P, VerifyNow-P2Y12. a Note that the use of recombinant FVIIa is not based on test results, only for severe bleeding not responding to any treatment. b See Excel file cell location for coagulation test.
© American Society for Clinical Pathology
837
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
837 837
Huang et al / Mobile Computing Platform for Hemotherapy
(from www.hemepathreview.com) and accessed on any desktop PC/laptop with MS Office (Microsoft) installed. Perioperatively, these files can be shared in real time between different faculty members and trainees through the University of Texas (UT) cloud-based file server (X-files). Through our individual UT account login, we can upload the data sheets for current cases onto this secure server in predesignated folders that can be accessed by the pathology team. Also, at the completion of each case, each data sheet is stored on our departmental encrypted hard drive under a predesignated folder and can be accessed by the UT faculty and trainees for clinical or research purposes.
Results We have now fully integrated the MCP system into our hemotherapy service workflow. Since integration of the MCP system in May 2012, we have accumulated spreadsheets for 35 LVAD cases, 27 heart transplant cases, and four total artificial heart cases (as of August 2013). The original intent of creating a system to collect data easily and efficiently for patients with heart failure undergoing surgery has been fulfilled. The use of this MCP system has eliminated the problems of illegibility, lost records, and inaccessibility that occur with a paper system. More important, the DSM has improved the perioperative workflow. By having a systematic set of algorithms, the pathologist can quickly have a summary of the laboratory data and come to a transfusion recommendation at a faster pace. This system also has been used to teach trainees the logic and principles behind many of the transfusion recommendations through the DSM algorithms.
Physicians on the hemotherapy service routinely and widely use the MCP and DMS. It is currently used only by the hemotherapy service team because the MCP is designed specifically for CPB surgery and the hemotherapy physicians. The anesthesiologist could also use the MCP during CPB surgery; however, they are often preoccupied with monitoring other aspects of surgery. Our transfusion recommendations have been well received by surgeons and anesthesia attending physicians, mainly because we have timely test results on the spot to direct transfusion. The recommended dose is mostly approximate given the fast pace of activities in the operating room (ongoing transfusion, ongoing bleeding, etc). We do not expect to have the dosage being followed exactly by surgeons and anesthesia attending physicians. The goal is to know what hemostatic defects the patient has to give the right blood components in an appropriate dosage depending on the severity of bleeding and how abnormal the coagulation test results are. With this mobile system, benefits such as accessibility and centralization of data have been realized. The patient data can be shared through the X-files between pathologists and their trainees whether they are in the operating room, office, or even at home. It becomes an important component when two providers are communicating about a patient through the phone, as they can visually see the data in front of them. Also, the centralization of these records into an electronic format on our departmental secure server has enabled our department to reliably access the data from these patients for clinical research and quality assurance/control.
❚Table 2❚ Nine Level 2 Algorithms to Combine Results From Level 1 Algorithms for Treatment and Their Corresponding Excel Functions Level
Summary of Components
Rules to Combine Algorithm Steps
Representative Excel Formulaa
I FFPs (single units) Maximum value from algorithms 2-5, 7, 9-12, =MAX(INT(C57),INT(C58),INT(C59),INT(C60), and 14 INT(C62),INT(C64),INT(C65),INT(C66), INT(C67),INT(C69)) II Cryoprecipitate (dose) Maximum value from algorithms 6, 8, and 35 =MAX(INT(C61),INT(C63),INT(C90)) III Platelets (apheresis units) Maximum value from algorithms 16-23 =MAX(INT(C71),INT(C72),INT(C73),INT(C74), INT(C75),INT(C76),INT(C77), INT(C78)) IV RBCs (units) RBCs to increase Hgb to 10 g/dL: A RBCs to correct Hgb for FFP: B = 0.5 × FFPs RBCs to correct Hgb for platelets: C = platelets If Hgb – (A + B) >10 g/dL, no RBCs needed Otherwise, RBCs = A + B + C =IF(C33-0.5*C44-C46>10,”0”,C56 + 0.5*C44 + C46) V EACA (g) Maximum value from algorithms 24-28 =MAX(INT(C79),INT(C80),INT(C81),INT(C82), INT(C83)) VI Tranexamic acid (g) Value from algorithm 29 =E84 VII ALERT for use of FVIIa Any positive value from algorithms 30-33 =IF(AND(C85=”0”, C86=”0”,C87=”0”, (contraindicated) C88=”0”),”---”,”No FVIIa!”) VIII ALERT for AT concentration Value from algorithm 15 =C70 IX ALERT for protamine overdose Value from algorithm 34 =E89 AT, antithrombin; EACA, e-aminocaproic acid; FFP, fresh-frozen plasma; FVIIa, factor VIIa; Hgb, hemoglobin. a See Excel file cell location for coagulation test.
838 838
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
© American Society for Clinical Pathology
AJCP / Original Article
Discussion It has been demonstrated that algorithms positively influence patient outcome.5-7 More specifically, simple computerized algorithms and protocols have been shown to reduce clinical decision errors.8 In addition, according to Goodnough and Despotis,9 there are limitations to retrospective chart audits in transfusion, and they suggest the use of prospective algorithms in combination with clinical information to make recommendations in transfusion. In our current model, we are using the embedded Excel algorithms to give the pathologist a set of recommendations to consider based on laboratory data. By combining that with clinical information, the pathologist can quickly come to a transfusion recommendation. The utility of our embedded algorithms in this scenario can be compared with a very simple example of a computer-calculated hematocrit we commonly find in the EMR. The hematocrit is calculated from RBCs and mean corpuscular volume (MCV): hematocrit = RBCs (cells/L) × MCV (L/cell).10 Time, as well as error rate in calculation, is reduced when this is automatically calculated and entered as a value into the EMR, which the clinician can easily and reliably use to base his or her clinical decision on. With our algorithms in the MCP, at a much higher level of complexity, we assimilate a plethora of laboratory data and run them through our computerized formulas, giving the pathologist recommendations to consider. Excel is ubiquitous and available on many mobile devices; its interface is familiar to many clinicians; programming skill is not needed to encode algorithms due to the rich set of formulas built into Excel; and it is relatively simply to create, modify, and maintain.11 These are the reasons we chose Excel as the software platform to create our MCP with DSM. In terms of the hardware selection, we wanted a hardware product that was portable, compatible, and secure. Since we used Excel as the software program to design the program, it was decided that a Windows-based system would be most ideal for our purposes. Essentially, any portable device that can run Excel would have been appropriate. Security is an issue when collecting and storing patient information, with two important concepts of authentication and confidentiality. Authentication is the process of verifying the identity of the communicating parties.12 Confidentiality refers to the act of preventing unauthorized access to confidential information.13 Here, we have ensured that only authorized personnel can access the data, and we have also prevented unauthorized access through appropriate measures. Our mobile devices (Dell Ultrabooks) contain an encrypted hard drive that is password protected and can be accessed only by relevant pathology personnel in our department. Furthermore, we took additional measures by using a secure cloud sharing system (X-files) instead of using nonsecure systems available on the market. Only authorized pathology personnel will have © American Society for Clinical Pathology
access to the system through their university password with X-files. Also, all the data we have collected are stored in an encrypted departmental file server that is password protected. An EMR has many well-established advantages over a paper record, such as efficient and indefinite storage, ease of transfer of data between facilities and providers, reduction in the number of lost records, and facilitating analysis of practice patterns and research activities.14 These are some of the reasons we decided to develop an electronic record system for our hemotherapy service. Initially, we had a paper system where the data are written onto a paper chart, but we encountered many problems with this method. After the implementation of our MCP, we have eliminated these problems. Last, it has been more than a year since the introduction of the MCP, and this tool has helped us decrease blood transfusion requirements and improve patient outcomes in CPB surgery. We have preliminary data showing that for our LVAD patients, we achieved a 59% reduction in blood product usage15 and a 65% reduction in complications in terms of emergency sternotomy (reoperation for excessive postoperative bleeding) at our institution16 compared with data published from other institutions. Besides contributing to a more positive clinical outcome seen in the data, our MCP has also aided us in collecting and analyzing the data more efficiently.
Conclusions The development and implementation of our MCP system has greatly increased the productivity and efficiency of our coagulation-based hemotherapy service. We have met our goal of creating a user-friendly electronic data collection system with DSM that the pathologist can use to help formulate transfusion recommendations and also use reliably in the future for clinical research. The DSM is an excellent teaching tool for our residents to interpret laboratory tests for transfusion purposes. Due to the time constraints to create and implement a mobile computing platform for our new hemotherapy service, we did not fully explore the feasibility of integrating the MCP with our hospital’s EMR. However, as our coagulation-based hemotherapy service grows, we are continuing to refine our MCP with DSM, and if feasible, we plan to integrate our system into the hospital’s EMR in the future. Address reprint requests to Dr Nguyen: Dept of Pathology and Laboratory Medicine, The University of Texas Health Science Center at Houston, 6431 Fannin St, MSB 2.292, Houston, TX 77030; [email protected]. This article has been presented in part at the meeting of the 133rd Association of Clinical Scientists; May 25, 2013; Boston, MA.
839
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
839 839
Huang et al / Mobile Computing Platform for Hemotherapy
References 1. Nguyen A. Introduction to coagulation-based hemotherapy. 2012. Available at www.hemepathreview.com. Accessed April 12, 2014. 2. Colesca SE, Zgodavova K. Transforming healthcare quality through information technology. Economia Seria Manage. 2008;11:21-39. 3. American Red Cross. Practice Guidelines for Blood Transfusion: A Compilation From Recent Peer-Reviewed Literature. 2nd ed. Washington, DC: American National Red Cross; 2007. 4. Jerrold L, Kenichi T. Management of surgical hemostasis: systemic agents. Vascular. 2008;16:S14-S21. 5. Grimm RH, Shimoni K, Harlan WR, et al. Evaluation of patient-care protocol use by various providers. N Engl J Med. 1975;292:507-511. 6. Wirtschafter DD, Scalise M, Henke C, et al. Do information systems improve the quality of clinical research? Results of a randomized trial in a cooperative multi institutional cancer group. Comput Biomed Res. 1981;4:78-90. 7. Pestotnik S, Classen D, Evans R, et al. Implementing antibiotic practice guidelines through computer-assisted decision support: clinical and financial outcomes. Ann Intern Med. 1996;24:884-890. 8. McDonald CJ. Protocol-based computer reminders, the quality of care and the nonperfectability of man. N Engl J Med. 1976;295:1351-1355. 9. Goodnough L, Despotis G. Future directions in utilization review: the role of transfusion algorithms. Transfus Sci. 1998;19:97-105.
840 840
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
10. Hanson CA. Peripheral blood and bone marrow: morphology, counts and differentials, and reactive disorders. In: McClatchey K, ed. Clinical Laboratory Medicine. Philadelphia, PA: Lippincott Williams and Wilkins; 2002:806. 11. Iyengar M, Svirbely J. The Medical Algorithms Project. Proc Eur Spreadsheet Risks Int Grp (EuSpRIG). 2009:113-118. 12. Needham RM, Schroeder MD. Using encryption for authentication in large networks of computers. Commun ACM. 1978;21:993-999. 13. Carney PA, Geller BM. Current medicolegal and confidentiality issues in large, multicenter research programs. Am J Epidemiol. 2000;152:371-378. 14. Sanbar S. Medical records: paper and electronic. Legal Med. 2007;34:347-356. 15. Nedelcu E, Gregoric I, Welsh K, et al. Intraoperative coagulation-based hemotherapy protocol decreases overall blood utilization with left ventricular assist device implantation. Paper presented at: American Society of Artificial Organs Conference; June 12-15, 2013; Chicago, IL. 16. Castillo B, Nedelcu E, Wahed A, et al. Intraoperative coagulation-based hemotherapy protocol decreases reoperation due to postoperative bleeding complications with left ventricular assist device implantation. Paper presented at: AABB Annual Meeting and CTTXPO; October 12-15, 2013; Denver, CO.
© American Society for Clinical Pathology
Mobile Computing Platform With Decision Support Modules for Hemotherapy Richard S. P. Huang, MD,1 Elena Nedelcu, MD,1 Yu Bai, MD,1 Amer Wahed, MD,1 Kimberly Klein, MD,1 Igor Gregoric, MD,2 Manish Patel, MD,2 Biswajit Kar, MD,2 Pranav Loyalka, MD,2 Sriram Nathan, MD,2 Paul Loubser, MD,3 Phillip A. Weeks, PharmD,4 Rajko Radovancevic, MD,2 and Andy N. D. Nguyen, MD1 From the 1Department of Pathology and Laboratory Medicine, 2Center for Advanced Heart Failure, and 3Cardiovascular Anesthesia, The University of Texas Health Science Center at Houston, Houston; and 4Texas Medical Center, Memorial Hermann Hospital, Houston. Key Words: Hemotherapy; Coagulation; Mobile computing platform; Decision support modules; Algorithms Am J Clin Pathol June 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
ABSTRACT Objectives: We describe the development of a mobile computing platform (MCP) with a decision support module (DSM) for patients in our coagulation-based hemotherapy service. Methods: The core of our MCP consists of a Microsoft Excel spreadsheet template used to gather and compute data on cardiopulmonary bypass (CPB) patients intraoperatively. The DSM is embedded into the Excel file, where the user would enter in laboratory results, and through our 45 embedded algorithms, recommendations for transfusion products would be displayed in the Excel file. Results: The DSM has helped decrease the time it takes to come to a transfusion recommendation, helps double-check recommendations, and is an excellent tool for teaching. Furthermore, the problems that occur with a paper system have been eliminated, and we are now able to access this information easily and reliably. Conclusions: The development and implementation of our MCP system has greatly increased the productivity and efficiency of our hemotherapy service.
834 834
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
Our pathology department started a coagulation-based hemotherapy service in May 2012. This is a pathology consultative service on coagulation issues that occur perioperatively and postoperatively in patients with heart failure. These patients undergo procedures such as aortocoronary bypass, cardiac valve repair or replacement, implantation of a left ventricular assist device (LVAD), and heart transplant. During almost all of these procedures, the patient is placed on cardiopulmonary bypass (CPB), where an extracorporeal mechanical pump temporarily takes over the function of the heart and lungs. Anesthesiologists and surgeons traditionally manage the coagulation issues of these surgical patients; however, due to the complexity and time restraints of CPB cases, pathologists at our institution are consulted to help manage these patients. We perform a preoperative evaluation for risk of bleeding prior to CPB, monitor the patient with consultation on transfusion during the operation, and assess the patient postoperatively with consultation on transfusion.1 These consultations are available 24 hours a day, 7 days a week. At our institute, one hemotherapy physician is on service for 1 week at a time, 24 hours a day. Surgeons usually consult the hemotherapy service physician during the surgery as well as for postoperative follow-up and management. However, cardiologists also often consult our service during postoperative bleeding. We write a consultation report for the surgery and progress notes on the patients postoperatively. Usually, we would follow the patients until the patient’s bleeding is under control or the patient’s coagulation status is optimized. Since the surgeons and cardiologists delegate transfusion support of patients to our service, we directly order laboratory tests and transfusions. The hospital’s Advanced Heart Failure Clinical Committee © American Society for Clinical Pathology
AJCP / Original Article
Materials and Methods
has approved this process. In this article, we focus mainly on the intraoperative aspect of our hemotherapy service. With no gold standard in determining the overall coagulation status of a patient, we use various laboratory tests to assess the coagulation status of the CBP patients. The standard tests we run on these patients include complete blood count (CBC), a disseminated intravascular coagulation (DIC) panel (prothrombin time [PT], partial thromboplastin time [PTT], fibrinogen, thrombin time, and d-dimer), antithrombin (AT), VerifyNowP2Y12 (VFN-P) for patients on clopidogrel or other adenosine diphosphate P2Y12 receptor antagonists, VerifyNow-ASA (VFN-A) for patients on aspirin, thromboelastography (TEG), and thromboelastography with heparinase (hTEG). Due to the sheer volume of data from these tests, we developed a mobile computing platform (MCP) to assimilate the laboratory data into a comprehensive data collection system that may be viewed by the pathologist in an easy and efficient manner. Furthermore, electronic medical records (EMRs) have been shown to support evidence-based decisions,2 and we have embedded decision support modules (DSMs) within our MCP to do precisely that. Our DSM enables the consultant pathologist to make accurate and quick recommendations to the surgeons and anesthesiologists.
Our MCP is composed of a software component and a hardware component. We describe the software component first. The software component centers on a Microsoft Excel (Microsoft, Redmond, WA) spreadsheet template. The template is used intraoperatively during CPB. In the template, the pathologist starts with data entry fields (DEFs) to enter pertinent information about the patient, including general demographic identifiers, medications, preoperative laboratory results, and coagulation risk assessment results ❚Image 1A❚. Dropdown menus are available in certain fields, such as names of the attending physicians, to increase the speed of data entry into the spreadsheet. In the next section, there are DEFs in which the pathologist can record data for the aforementioned set of tests done on the patient ❚Image 1B❚. The test result fields would be automatically highlighted if the values were abnormal (green for low values and red for high values). Here the values entered include reaction time (R) (from the TEG); heparinase reaction time (h-R), heparinase angle (h-A), heparinase maximum amplitude (h-MA), heparinase estimated percent lysis (h-EPL), heparinase lysis at 30 minutes (h-Ly30), and heparinase coagulation index (h-CI) (from the hTEG); hemoglobin (Hgb) and platelet count (from
A
B
C
D
❚Image 1❚ Sample of the intraoperative Excel (Microsoft, Redmond, WA) template. A, Data entry field (DEF) to enter pertinent information about the patient. B, DEF to record coagulation laboratory values perioperatively. C, Fields displaying summary of suggestions. D, DEF to record actual blood usage, time on pump, and chest tube drainage. For abbreviations, see Table 1.
© American Society for Clinical Pathology
835
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
835 835
Huang et al / Mobile Computing Platform for Hemotherapy
the CBC); PT/PTT, fibrinogen, thrombin time, and d-dimer (from the DIC screen); AT; VFN-A; and VFN-P. These tests would be sequentially performed and entered in the Excel spreadsheet during different phases of CPB, which include baseline, hemoconcentration/rewarming phase (just prior to termination of CPB), post-CPB phase, and during the potential postoperative bleeding phase. Data validation for input entry is also provided. If the input is not in an acceptable range or in the incorrect format, the data entry would be stopped with a warning message. As the user enters laboratory results into the MCP, the values would be automatically routed through the DSM in the Excel file. The DSM comprises 45 algorithms that are translated to Excel functions. The hemotherapy physicians at our institute, consisting of hematopathologists and blood bankers, developed the algorithms for the DSM by reviewing the literature for algorithms used during CPB surgery; hence, the algorithms are based on standard transfusion practice.3,4 Our group prepared a comprehensive hemotherapy guide, which includes all the algorithms, and the current version of this guide can be reviewed on our website at http://hemepathreview.com (go to Item 14, then Guide). Criteria for transfusion are listed in the References section of this guide. Furthermore, our group reviewed the algorithms, which are continuously revised based on experience learned on our service. The guide and the algorithms within them are also annually reviewed and approved by the hospital’s Advanced Heart Failure Clinical Committee. The algorithms consist of 36 level 1 algorithms ❚Table 1❚ and 9 level 2 algorithms ❚Table 2❚. Level 1 algorithms derive transfusion needs based on specific sets of laboratory tests. Level 2 algorithms combine results from level 1 algorithms for final transfusion suggestions. These recommendations would appear under “Suggestion Summary” and consist of the following fields: fresh-frozen plasma (FFP; single units), cryoprecipitate (doses), platelets (doses or apheresis units), RBCs (units), tranexamic acid (grams), protamine (milligrams), alert for use of factor VIIa (FVIIa), alert for use of AT concentrate, and alert for possible protamine overdose ❚Image 1C❚. The 36 level 1 algorithms can be divided into 10 different product/recommendation categories: (1) RBC transfusion for low Hgb; (2) FFP transfusion for low clotting factors, low AT, or low fibrinogen; (3) cryoprecipitate transfusion for very low fibrinogen or uremia; (4) AT concentrate transfusion for very low AT; (5) no FFP transfusion for prolonged PTT secondary to heparin; (6) platelet transfusion for low platelet count or medication effect; (7) tranexamic acid for primary fibrinolysis; (8) protamine for excess heparin after neutralization; (9) FVIIa contraindicated in secondary fibrinolysis; and (10) alert for possible protamine overdose. Of note, we currently do not have plans to include prothrombin complex concentrate since the four-factor 836 836
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
prothrombin complex concentrate has not been made available at our hospital. Also, fibrinogen concentrate has not been used by our service since we find cryoprecipitate adequate to restore fibrinogen level. Last, no DSM has been developed for the new anticoagulants dabigatran, rivaroxaban, and apixaban, even though we find these anticoagulants problematic for emergency cardiac surgery. Instead, we have guidelines for monitoring these medications and alleviating bleeding in our online guide. For illustration, let us look at a clinical scenario at the time of off-pump with the following laboratory results: Hgb, 6.7 g/dL; platelets, 55 ×103/mL; PT, 28.4 seconds; PTT, 210 seconds; fibrinogen, 186 mg/dL; and thrombin time, 110 seconds. When the user enters data into the DEF, the system would route those values to the DSM and consider the following level 1 algorithms: 1. Three RBCs to increase Hgb from 6.7 g/dL to above 10 g/dL 5. Two FFPs for low fibrinogen at 186 mg/dL 10. Four FFPs for prolonged PT at 28.4 seconds 13. No FFP for prolonged PTT at 210 seconds (this is due to the heparin effect with thrombin time at 110 seconds) 16. Two doses of platelets for platelet count at 55 ×103/mL 29. Protamine (50 mg) for excess heparin (prolonged PTT and thrombin time) Level 2 algorithms are then activated to combine results from level 1 algorithms for the actual number of blood components needed: I. Number of FFPs: maximum value from (5), (10) = maximum of 2 and 4 = 4 III. Number of platelets: from (16) = 2 IV. Number of RBCs: – To correct for FFP transfusion = 0.5 × FFPs = 0.5 × 4 = 2 – To correct for platelet transfusion = platelets = 2 – To correct for low Hgb: from (1) = 3 – Total RBCs = 2 + 2 + 3 = 7 VI. Protamine: from (29) = 50 mg The final results from level 2 algorithms are displayed in the “Suggestion Summary” (Image 1C). Note that fields such as Cryoprecipitate and Tranexamic Acid are left blank since their associated algorithms (both level 1 and level 2) have not been activated with the given laboratory results. As the laboratory data of the patient are entered, the pathologist can review the recommendations produced by the DSM and combine them with his or her clinical judgment to make a final recommendation to the surgeon/ anesthesiologist. It cannot be overemphasized that clinical judgment is critical for a final transfusion decision based on the clinical situation. In particular, the number of blood components should be based on the degree of microvascular © American Society for Clinical Pathology
AJCP / Original Article
bleeding and the patient’s bodyweight. The dosage recommendation from our DSM can be used as a starting point for further adjustment. The suggested transfusion is most applicable for the postpump phase, when most patients will exhibit some degree of microvascular bleeding. For suggested transfusion shown in other phases (baseline, hemoconcentration), transfusion typically is not needed. The suggestions for these phases just need to be kept in mind for later management. Also, there are DEFs on the template for
the pathologist to enter actual blood component usage, time on pump, and chest tube drain output ❚Image 1D❚ for future reference if needed. The hardware component of the MCP centers on Ultrabooks (Dell, Round Rock, TX), which have an encrypted hard drive that is password protected. The Excel program and the DSM templates are preloaded onto the Ultrabooks, so once the user securely logs onto the laptop, he or she can access the templates. Also, the templates can be downloaded
❚Table 1❚ Thirty-Six Level 1 Algorithms and Their Corresponding Excel Functionsa Level
Algorithm
Representative Excel Formulab
1. RBCs for anemia If Hgb =10,”0”, to 10.0 g/dL (1 RBC increased Hgb by 1.0 g/dL) ROUND(10-C33,0))) 2. FFPs for low clotting factors If h-R 10-15 min → 2 units of FFP =IF(AND(C2710),”2”,”0”) 3. FFPs for low clotting factors If h-R 15-20 min → 4 units of FFP =IF(AND(C2715),”4”,”0”) 4. FFPs for low clotting factors If h-R >20 min → 6 units of FFP =IF(C27>20,”6”,”0”) 5. FFPs for low fibrinogen If fibrinogen 150-200 mg/dL → 2 units of FFP =IF(AND(C37150),”2”,”0”) 6. Cryoprecipitate for very low fibrinogen If fibrinogen 0,C3750 mm → 2 units of FFP =IF(AND(C2820, C29>50),”2”,”0”) 8. Cryoprecipitate for very low fibrinogen If h-A 50 mm → 1 dose =IF(AND(C280,C29>50),”1”,”0”) of cryoprecipitate 9. FFPs for low clotting factors If PT 20-25 s → 2 units of FFP =IF(AND(C35>20,C3525 s → 4 units of FFP =IF(C35>25, “4”,”0”) 11. FFPs for low clotting factors If PTT 45-50 s → 2 units of FFP (however, if due to the =IF(C68=”hep effect”,”0”,(IF(AND(C36>45, heparin effect in algorithm 13, then disregard value), C3650 s → 4 units of FFP (however, if due to heparin =IF(C68=”hep effect”,”0”,(IF(C36>50, effect in algorithm 13, then disregard value), not ”4”,”0”))) applicable to on-pump time 13. No FFP due to heparin effect If PTT >45 s, TT >25 s → due to heparin effect =IF(AND(C36>45,C38>25), “hep effect”, “xxxx”) 14. FFPs for low ATIII If ATIII 35%-50% → 2 units FFP, applicable to baseline only =IF(AND(C40=35),”2”,”0”) 15. ATIII concentration for very low ATIII If ATIII 0,C40 50,C340,C34 70 mm → no FVIIa =IF(AND(C30>15,C29>70),”No FVIIa!”,”0”) 31. Secondary fibrinolysis, no FVIIa If h-EPL >15%, h-CI >3 → no FVIIa =IF(AND(C30>15,C32>3),”No FVIIa!”,”0”) 32. Secondary fibrinolysis, no FVIIa If h-Ly30 >8%, h-MA >70 mm → no FVIIa =IF(AND(C31>8,C29>70),”No FVIIa!”,”0”) 33. Secondary fibrinolysis, no FVIIa If h-Ly30 >8%, h-CI >3 → no FVIIa =IF(AND(C31>8,C32>3),”No FVIIa!”,”0”) 34. Possible protamine overdose If h-R >20 s, h-A 0, F280, pump and after, no bleeding seen) → yes to possible F290, C41, “xxxx”) AA, arachidonic acid; ADP, adenosine diphosphate; ARU, aspirin reaction units; ATIII, antithrombin III; CRF, chronic renal failure; FEU, fibrogen equivalent units; FFP, freshfrozen plasma; FVIIa, factor VIIa; h-A, heparinase angle; h-CI, heparinase coagulation index; h-EPL, heparinase estimated percent lysis; Hgb, hemoglobin; h-Ly30, heparinase lysis at 30 minutes; h-MA, heparinase maximum amplitude; h-R, heparinase reaction time; PRU, P2Y12 reaction units; PT, prothrombin time; PTT, partial thromboplastin time; TT, thrombin time; VFN, VerifyNow; VFN-A, VerifyNow-aspirin; VFN-P, VerifyNow-P2Y12. a Note that the use of recombinant FVIIa is not based on test results, only for severe bleeding not responding to any treatment. b See Excel file cell location for coagulation test.
© American Society for Clinical Pathology
837
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
837 837
Huang et al / Mobile Computing Platform for Hemotherapy
(from www.hemepathreview.com) and accessed on any desktop PC/laptop with MS Office (Microsoft) installed. Perioperatively, these files can be shared in real time between different faculty members and trainees through the University of Texas (UT) cloud-based file server (X-files). Through our individual UT account login, we can upload the data sheets for current cases onto this secure server in predesignated folders that can be accessed by the pathology team. Also, at the completion of each case, each data sheet is stored on our departmental encrypted hard drive under a predesignated folder and can be accessed by the UT faculty and trainees for clinical or research purposes.
Results We have now fully integrated the MCP system into our hemotherapy service workflow. Since integration of the MCP system in May 2012, we have accumulated spreadsheets for 35 LVAD cases, 27 heart transplant cases, and four total artificial heart cases (as of August 2013). The original intent of creating a system to collect data easily and efficiently for patients with heart failure undergoing surgery has been fulfilled. The use of this MCP system has eliminated the problems of illegibility, lost records, and inaccessibility that occur with a paper system. More important, the DSM has improved the perioperative workflow. By having a systematic set of algorithms, the pathologist can quickly have a summary of the laboratory data and come to a transfusion recommendation at a faster pace. This system also has been used to teach trainees the logic and principles behind many of the transfusion recommendations through the DSM algorithms.
Physicians on the hemotherapy service routinely and widely use the MCP and DMS. It is currently used only by the hemotherapy service team because the MCP is designed specifically for CPB surgery and the hemotherapy physicians. The anesthesiologist could also use the MCP during CPB surgery; however, they are often preoccupied with monitoring other aspects of surgery. Our transfusion recommendations have been well received by surgeons and anesthesia attending physicians, mainly because we have timely test results on the spot to direct transfusion. The recommended dose is mostly approximate given the fast pace of activities in the operating room (ongoing transfusion, ongoing bleeding, etc). We do not expect to have the dosage being followed exactly by surgeons and anesthesia attending physicians. The goal is to know what hemostatic defects the patient has to give the right blood components in an appropriate dosage depending on the severity of bleeding and how abnormal the coagulation test results are. With this mobile system, benefits such as accessibility and centralization of data have been realized. The patient data can be shared through the X-files between pathologists and their trainees whether they are in the operating room, office, or even at home. It becomes an important component when two providers are communicating about a patient through the phone, as they can visually see the data in front of them. Also, the centralization of these records into an electronic format on our departmental secure server has enabled our department to reliably access the data from these patients for clinical research and quality assurance/control.
❚Table 2❚ Nine Level 2 Algorithms to Combine Results From Level 1 Algorithms for Treatment and Their Corresponding Excel Functions Level
Summary of Components
Rules to Combine Algorithm Steps
Representative Excel Formulaa
I FFPs (single units) Maximum value from algorithms 2-5, 7, 9-12, =MAX(INT(C57),INT(C58),INT(C59),INT(C60), and 14 INT(C62),INT(C64),INT(C65),INT(C66), INT(C67),INT(C69)) II Cryoprecipitate (dose) Maximum value from algorithms 6, 8, and 35 =MAX(INT(C61),INT(C63),INT(C90)) III Platelets (apheresis units) Maximum value from algorithms 16-23 =MAX(INT(C71),INT(C72),INT(C73),INT(C74), INT(C75),INT(C76),INT(C77), INT(C78)) IV RBCs (units) RBCs to increase Hgb to 10 g/dL: A RBCs to correct Hgb for FFP: B = 0.5 × FFPs RBCs to correct Hgb for platelets: C = platelets If Hgb – (A + B) >10 g/dL, no RBCs needed Otherwise, RBCs = A + B + C =IF(C33-0.5*C44-C46>10,”0”,C56 + 0.5*C44 + C46) V EACA (g) Maximum value from algorithms 24-28 =MAX(INT(C79),INT(C80),INT(C81),INT(C82), INT(C83)) VI Tranexamic acid (g) Value from algorithm 29 =E84 VII ALERT for use of FVIIa Any positive value from algorithms 30-33 =IF(AND(C85=”0”, C86=”0”,C87=”0”, (contraindicated) C88=”0”),”---”,”No FVIIa!”) VIII ALERT for AT concentration Value from algorithm 15 =C70 IX ALERT for protamine overdose Value from algorithm 34 =E89 AT, antithrombin; EACA, e-aminocaproic acid; FFP, fresh-frozen plasma; FVIIa, factor VIIa; Hgb, hemoglobin. a See Excel file cell location for coagulation test.
838 838
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
© American Society for Clinical Pathology
AJCP / Original Article
Discussion It has been demonstrated that algorithms positively influence patient outcome.5-7 More specifically, simple computerized algorithms and protocols have been shown to reduce clinical decision errors.8 In addition, according to Goodnough and Despotis,9 there are limitations to retrospective chart audits in transfusion, and they suggest the use of prospective algorithms in combination with clinical information to make recommendations in transfusion. In our current model, we are using the embedded Excel algorithms to give the pathologist a set of recommendations to consider based on laboratory data. By combining that with clinical information, the pathologist can quickly come to a transfusion recommendation. The utility of our embedded algorithms in this scenario can be compared with a very simple example of a computer-calculated hematocrit we commonly find in the EMR. The hematocrit is calculated from RBCs and mean corpuscular volume (MCV): hematocrit = RBCs (cells/L) × MCV (L/cell).10 Time, as well as error rate in calculation, is reduced when this is automatically calculated and entered as a value into the EMR, which the clinician can easily and reliably use to base his or her clinical decision on. With our algorithms in the MCP, at a much higher level of complexity, we assimilate a plethora of laboratory data and run them through our computerized formulas, giving the pathologist recommendations to consider. Excel is ubiquitous and available on many mobile devices; its interface is familiar to many clinicians; programming skill is not needed to encode algorithms due to the rich set of formulas built into Excel; and it is relatively simply to create, modify, and maintain.11 These are the reasons we chose Excel as the software platform to create our MCP with DSM. In terms of the hardware selection, we wanted a hardware product that was portable, compatible, and secure. Since we used Excel as the software program to design the program, it was decided that a Windows-based system would be most ideal for our purposes. Essentially, any portable device that can run Excel would have been appropriate. Security is an issue when collecting and storing patient information, with two important concepts of authentication and confidentiality. Authentication is the process of verifying the identity of the communicating parties.12 Confidentiality refers to the act of preventing unauthorized access to confidential information.13 Here, we have ensured that only authorized personnel can access the data, and we have also prevented unauthorized access through appropriate measures. Our mobile devices (Dell Ultrabooks) contain an encrypted hard drive that is password protected and can be accessed only by relevant pathology personnel in our department. Furthermore, we took additional measures by using a secure cloud sharing system (X-files) instead of using nonsecure systems available on the market. Only authorized pathology personnel will have © American Society for Clinical Pathology
access to the system through their university password with X-files. Also, all the data we have collected are stored in an encrypted departmental file server that is password protected. An EMR has many well-established advantages over a paper record, such as efficient and indefinite storage, ease of transfer of data between facilities and providers, reduction in the number of lost records, and facilitating analysis of practice patterns and research activities.14 These are some of the reasons we decided to develop an electronic record system for our hemotherapy service. Initially, we had a paper system where the data are written onto a paper chart, but we encountered many problems with this method. After the implementation of our MCP, we have eliminated these problems. Last, it has been more than a year since the introduction of the MCP, and this tool has helped us decrease blood transfusion requirements and improve patient outcomes in CPB surgery. We have preliminary data showing that for our LVAD patients, we achieved a 59% reduction in blood product usage15 and a 65% reduction in complications in terms of emergency sternotomy (reoperation for excessive postoperative bleeding) at our institution16 compared with data published from other institutions. Besides contributing to a more positive clinical outcome seen in the data, our MCP has also aided us in collecting and analyzing the data more efficiently.
Conclusions The development and implementation of our MCP system has greatly increased the productivity and efficiency of our coagulation-based hemotherapy service. We have met our goal of creating a user-friendly electronic data collection system with DSM that the pathologist can use to help formulate transfusion recommendations and also use reliably in the future for clinical research. The DSM is an excellent teaching tool for our residents to interpret laboratory tests for transfusion purposes. Due to the time constraints to create and implement a mobile computing platform for our new hemotherapy service, we did not fully explore the feasibility of integrating the MCP with our hospital’s EMR. However, as our coagulation-based hemotherapy service grows, we are continuing to refine our MCP with DSM, and if feasible, we plan to integrate our system into the hospital’s EMR in the future. Address reprint requests to Dr Nguyen: Dept of Pathology and Laboratory Medicine, The University of Texas Health Science Center at Houston, 6431 Fannin St, MSB 2.292, Houston, TX 77030; [email protected]. This article has been presented in part at the meeting of the 133rd Association of Clinical Scientists; May 25, 2013; Boston, MA.
839
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
839 839
Huang et al / Mobile Computing Platform for Hemotherapy
References 1. Nguyen A. Introduction to coagulation-based hemotherapy. 2012. Available at www.hemepathreview.com. Accessed April 12, 2014. 2. Colesca SE, Zgodavova K. Transforming healthcare quality through information technology. Economia Seria Manage. 2008;11:21-39. 3. American Red Cross. Practice Guidelines for Blood Transfusion: A Compilation From Recent Peer-Reviewed Literature. 2nd ed. Washington, DC: American National Red Cross; 2007. 4. Jerrold L, Kenichi T. Management of surgical hemostasis: systemic agents. Vascular. 2008;16:S14-S21. 5. Grimm RH, Shimoni K, Harlan WR, et al. Evaluation of patient-care protocol use by various providers. N Engl J Med. 1975;292:507-511. 6. Wirtschafter DD, Scalise M, Henke C, et al. Do information systems improve the quality of clinical research? Results of a randomized trial in a cooperative multi institutional cancer group. Comput Biomed Res. 1981;4:78-90. 7. Pestotnik S, Classen D, Evans R, et al. Implementing antibiotic practice guidelines through computer-assisted decision support: clinical and financial outcomes. Ann Intern Med. 1996;24:884-890. 8. McDonald CJ. Protocol-based computer reminders, the quality of care and the nonperfectability of man. N Engl J Med. 1976;295:1351-1355. 9. Goodnough L, Despotis G. Future directions in utilization review: the role of transfusion algorithms. Transfus Sci. 1998;19:97-105.
840 840
Am J Clin Pathol 2014;141:834-840 DOI: 10.1309/AJCPRG5LYWL6DXMX
10. Hanson CA. Peripheral blood and bone marrow: morphology, counts and differentials, and reactive disorders. In: McClatchey K, ed. Clinical Laboratory Medicine. Philadelphia, PA: Lippincott Williams and Wilkins; 2002:806. 11. Iyengar M, Svirbely J. The Medical Algorithms Project. Proc Eur Spreadsheet Risks Int Grp (EuSpRIG). 2009:113-118. 12. Needham RM, Schroeder MD. Using encryption for authentication in large networks of computers. Commun ACM. 1978;21:993-999. 13. Carney PA, Geller BM. Current medicolegal and confidentiality issues in large, multicenter research programs. Am J Epidemiol. 2000;152:371-378. 14. Sanbar S. Medical records: paper and electronic. Legal Med. 2007;34:347-356. 15. Nedelcu E, Gregoric I, Welsh K, et al. Intraoperative coagulation-based hemotherapy protocol decreases overall blood utilization with left ventricular assist device implantation. Paper presented at: American Society of Artificial Organs Conference; June 12-15, 2013; Chicago, IL. 16. Castillo B, Nedelcu E, Wahed A, et al. Intraoperative coagulation-based hemotherapy protocol decreases reoperation due to postoperative bleeding complications with left ventricular assist device implantation. Paper presented at: AABB Annual Meeting and CTTXPO; October 12-15, 2013; Denver, CO.
© American Society for Clinical Pathology
