Special issue: Original articles Responsible writing in science Analysis of laboratory critical values at a referral Spanish tertiary university hospital Ariadna Arbiol-Roca*1, Sofía Corral-Comesaña2, Ruth Cano-Corres3, María José Castro-Castro1, Macarena Dastis-Arias1, Dolors Dot-Bach1 1Laboratori Clínic, Hospital Universitari de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain 2Synlab Diagnósticos Globales SA, Esplugues de Llobregat, Barcelona, Spain 3Clinical Laboratory, Biochemistry Department, Parc Taulí Hospital Universitari, Institut d’Investigació i Innovació Parc Taulí I3PT, Universitat Autònoma de Barcelona, Sabadell, Spain The first two authors contributed equally to this work. *Corresponding author: ariadna.arbiol@bellvitgehospital.cat Abstract Introduction: The aim of this study was to analyse critical value data from our laboratory and compare our critical value reporting policy with others in the literature. Materials and methods: Analysis of critical values was performed on data obtained over a 6-month period in a tertiary university hospital. Results: We identified 5723 critical values, of which approximately 80% came from STAT testing (4577), 15% from routine inpatients testing (884) and 5% from routine outpatients testing (262). The highest proportion of critical values corresponded to oxygen partial pressure (17.7%), followed by potassium ion (17.6%) concentrations. The parameters associated with the highest critical value notification percentage in emergency patients were pH, haematocrit, glucose, potassium ion and haemoglobin concentrations. In inpatients, these parameters were glucose, phosphate, haemo- globin, sodium ion and potassium ion concentrations. In outpatients, they were calcium and potassium concentrations. Conclusions: The analysis of critical values in our hospital is in accordance with that reported in the literature. Our findings demonstrate the impor- tance of incorporating improvement actions not only in critical value notification, but especially in the registration of this activity. Keywords: critical values limits; critical values notification; patient safety Received: May 31, 2018 Accepted: November 20, 2017 Introduction A critical value is defined as “a laboratory test re- The literature indicates that there is no consensus sult that represents a pathophysiologic state at on the biomarkers which are more likely to result such variance with normal as to be life-threaten- in critical values, nor on what the critical value lim- ing unless something is done promptly and for its should be. A consensus must be reached which some corrective action could be taken” (1). among physicians on the list of biomarkers and The laboratory accrediting agencies have made the critical value limits of each, which should be critical value reporting part of the requirements established by each laboratory (6-9). Several or- for accreditation (2,3). Furthermore, the immediate ganizations have published guidelines for the re- notification of a critical value as a special requisite porting of critical results, one of which is issued by has been recognised and implemented worldwide the British Royal College of Pathologists (10). Fur- through the International Organization for Stand- thermore, independent lists are needed for differ- ardization (ISO) 15189:2012, and has been adopted ent study populations, as critical values will differ as a standard of Good Laboratory Practice (4,5). between neonatal, paediatric and adult care pa- tients (11). https://doi.org/10.11613/BM.2019.010704 Biochem Med (Zagreb) 2019;29(1):010704 1 Arbiol-Roca A. et al. Analysis of laboratory critical values Instant communication of these critical values tory performs 3.5 million tests, of which 30% are leads to faster diagnostic processes and a rapid urgent requests. change in patient management. Clinical laborato- This retrospective study, conducted over a ries must establish optimal critical value limits in 6-month period (January to June 2017), was devel- order to achieve a balance between informing oped to evaluate the implemented critical value clinical staff and overloading them with results communication process in a Spanish tertiary uni- that do not require urgent action (12-14). versity hospital. All data was obtained from re- Critical values are usually detected by the labora- ports generated from the LIS (Omega3000: Roche tory technical staff. It is necessary to ensure the va- Diagnostics SL, Spain). Data were exported from lidity of critical values from the beginning to en- the LIS into Microsoft Excel (Microsoft Office 2003®, sure there are no potential sources of interference Alburquerque, EEUU). The parameters recorded in- in order to ensure patient safety (15). Once a criti- cluded calcium (II), calcium ion, carbon dioxide cal value has been validated, it is necessary to no- partial pressure, carboxyhaemoglobin, erythro- tify the physician. Although electronic patient re- cytes, glucose, haemoglobin, oxygen partial pres- cords, laboratory information system (LIS) and sure, phosphate, pH, potassium ion, sodium ion hospital information system (HIS) software have and thrombocytes. been widely introduced in hospitals, the most Laboratory staff is notified of the presence of a common reporting system is still telephone com- critical value by the LIS, which is configured in munication directly by laboratory staff or via call such a way that a red warning flag automatically centres (16). However, some laboratories have shows up when a critical value appears. The result started employing automated systems (14,17). is then checked for analytical reliability, and once Communication of critical values is an important the potential margin for error has been eliminated issue; however, the guidelines do not identify who and the result validated, the person responsible should be responsible for receiving the critical val- for the patient’s healthcare is notified by the labo- ues notification (head physician, registered nurse, ratory technician. With regard to critical values, attending physician, etc.). This notification should our laboratory has established two separate pro- be registered, as well as confirmation that the car- cesses for inpatients and outpatients. For inpa- egiver has accepted responsibility for follow up. tients, either the nurse or the medical department Thus, the aforementioned critical value notifica- are notified of critical values by phone. For outpa- tion procedure of clinical laboratories must be tients, however, the laboratory staff gets in touch carefully considered, as it affects patient safety via the hospital’s call centre, who call the attend- and the efficiency and quality of clinical care (18). ing physician as well as send them an email. The In the present study, we analysed 6-month period call centre then sends confirmation to the labora- of critical value data and reporting practices to tory that the above has taken place. All critical val- compare our results with those reported in the lit- ues reported for both inpatients and outpatients erature. The aim of our study was to determine the must be collected and registered in the laborato- situation of our laboratory for the notification of ry’s internal archive by the laboratory technician critical values in order to detect strengths and who advised the value. Laboratory technicians weaknesses, which would enable us to improve register information about the communication the whole process. (the critical value, the technician who communi- cated the information and the person who re- ceived the information). The laboratory medical Materials and methods specialist also adds a comment to the critical result The clinical laboratory of University Hospital of in the LIS, such as “critical value notified by phone”, Bellvitge has been accredited by the ISO 15189 so that this information is recorded in the patient’s since February 2008. Every year, the clinical labora- clinical history. Biochem Med (Zagreb) 2019;29(1):010704 https://doi.org/10.11613/BM.2019.010704 2 Arbiol-Roca A. et al. Analysis of laboratory critical values The list of critical value limits used in our laborato- cal values notified (N critical values notified) / total ry at the time of the study is presented in Table 1. number of critical values (N critical values) × 100. In our laboratory, these limits were established All data were calculated using Microsoft Excel (Mi- based on literature sources, and were subsequent- crosoft Office 2003®, Alburquerque, EEUU). ly revised and approved by laboratory medical specialists and clinicians (19). In some cases, spe- cial critical values (serum calcium for intensive care Results unit (ICU) patients or potassium for nephrology During the 6-month study period, our clinical lab- outpatients) have been adopted. These critical val- oratory performed more than 1.4 million routine ues were derived from agreements with medical tests and approximately half a million STAT tests. services. For patients in the ICU, a calcium lower In the same period, the number of critical values than 1.75 mmol/L is not considered a critical value detected was 5723 (0.4%), approximately 80% of unless the serum albumin concentration is higher which were from STAT testing (4577), 15% from than 25 g/L. On the another hand, if the outpatient routine inpatient testing (884) and 5% from rou- belongs to the Nephrology Service, a serum po- tine outpatient testing (262). The prevalence of tassium concentration up to 6.2 mmol/L is consid- critical values was 0.9% for STAT testing and 0.08% ered a critical value, but is not necessary to report for routine testing. The frequency of critical values under an agreement between the Nephrology was calculated for each parameter, as shown in Ta- Service and our laboratory. Re-evaluation of these ble 2. The parameter with the highest percentage critical values is performed together with the clini- of all critical values was oxygen partial pressure cians when the need to modify a critical value aris- (17.7%), followed by potassium ion (17.6%), throm- es, such as following the publication of new data, bocytes (14%), carbon dioxide partial pressure new accords with clinical services, or technological (13.2%), calcium (II) (12.3%) and pH (11.2%). improvements. For this study, we defined critical values as those which exceed the critical value limits established Table 1. Critical values limits list for biochemical and haemato- for a single parameter for the same patient within logical parameters a 24-hour period. Lower Upper Parameters critical critical Statistical analysis values limit values limit P-Calcium (II), mmol/L < 1.75 > 3.20 The percentage of critical values was calculated as: total number of critical values detected for each P-Calcium ion, mmol/L < 0.85 > 1.60 parameter (N critical values)/number of test results Gas(B)-Carbon dioxide, mmHg < 20 > 70 reported for each parameter (N test results) × 100. Hb(B)-Carboxihaemoglobin, % - > 10 The percentage of critical values relative to all criti- B-Erythrocytes (Haematocrit) < 0.15 - cal values was calculated as: N critical values/total P-Glucose, mmol/L < 2.5 > 27 number of critical values × 100. B-Haemoglobin, g/L < 50 - For this study, the patient’s results were separated into two groups: STAT laboratory test results and Gas(B)-Oxygen, mmHg < 40 - routine laboratory results. The routine laboratory P-Phosphate, mmol/L < 0.39 - results were further subdivided into inpatients Pt (B)-Plasma; pH < 7.2 > 7.6 and outpatients. P-Potassium ion, mmol/L < 2.8 > 6.2 The percentage of critical values notified for emer- P-Sodium ion, mmol/L < 120 > 160 gency patients, inpatients and outpatients were B-Thrombocytes, x109/L < 30 - calculated for each parameter as: number of criti- B – blood. Hb – haemoglobin. P – plasma. Pt – patient. https://doi.org/10.11613/BM.2019.010704 Biochem Med (Zagreb) 2019;29(1):010704 3 Arbiol-Roca A. et al. Analysis of laboratory critical values Table 2. Parameters and number of test results reported for each parameter Parameters N test results N critical values % critical values % of all critical values Gas(B)-Oxygen 22,223 1011 4.5 17.7 P-Potassium ion 118,386 1007 0.9 17.6 B-Thrombocytes 112,015 802 0.7 14 Gas(B)-Carbon dioxide 44,144 756 3.4 13.2 P-Calcium (II) 44,420 703 1.6 12.3 Pt (B)-Plasma; pH 44,168 642 2.9 11.2 P-Glucose 122,836 337 0.3 5.9 P-Sodium ion 115,925 206 0.2 3.6 B-Haemoglobin 112,958 104 0.1 1.8 B-Erythrocytes (haematocrit) 112,141 80 0.1 1.4 P-Calcium ion 3144 40 1.3 0.7 P-Phosphate 19,249 34 0.2 0.6 Hb(B)-Carboxihaemoglobin 25 1 4.0 0 B – blood. Hb – haemoglobin. P – plasma. Pt – patient. N test results – number of total test results. N critical values - total number of critical values detected. % critical values - percentages of critical values with respect to number of test results for each parameter. % of all critical values - percentages of critical values with respect to all critical values. Tables 3 and 4 show the number of critical values, of parameters resulting in critical values be- the number of critical values notified and the longed to the patients in the emergency depart- percentage of notified critical values observed ment. For routine laboratory tests, the highest for STAT tests, inpatients and outpatients for each percentage of critical values was found for inpa- parameter. For STAT laboratory tests, the majority tients in the ICU. Table 3. Percentages of critical values notification for STAT patients STAT TESTING Parameters N critical values N critical values notified % critical values notification Gas(B)-Oxygen 1011 40 4 Gas(B)-Carbon dioxide 756 102 13.5 P-Potassium ion 652 418 64.1 Pt (B)-Plasma; pH 642 447 69.6 B-Thrombocytes 538 210 39.0 P-Calcium (II) 420 167 39.8 P-Glucose 229 157 68.6 P-Sodium ion 168 71 42.3 B-Haemoglobin 91 55 60.4 B-Erythrocytes (haematocrit) 69 48 69.6 Hb(B)-Carboxihaemoglobin 1 0 0 B – blood. Hb – haemoglobin. P – plasma. Pt – patient. N critical values - total number of critical values detected. N critical values notified - number of critical values notified. % critical values notification - percentage of critical values notification. Biochem Med (Zagreb) 2019;29(1):010704 https://doi.org/10.11613/BM.2019.010704 4 Arbiol-Roca A. et al. Analysis of laboratory critical values Table 4. Percentages of critical values notification for inpatients and outpatients INPATIENTS OUTPATIENTS LABORATORY TESTING LABORATORY TESTING N critical % Critical N critical % Critical Parameters N critical values values N critical values values values notified notification values notified notification P-Potassium ion 242 194 80 113 70 62 B-Thrombocytes 209 4 2 55 2 4 P-Calcium (II) 285 166 58 38 25 66 P-Glucose 74 68 92 34 20 59 P-Sodium ion 21 17 81 17 4 24 B-Haemoglobin 11 9 82 2 1 50 B-Erythrocytes (haematocrit) 9 7 78 2 1 50 P- Phosphate 33 30 91 1 0 0 B – blood. Hb – haemoglobin. P – plasma. Pt – patient. N critical values - total number of critical values detected. N critical values notified - number of critical values notified. % critical values notification - percentage of critical values notification. The parameters with the highest notification per- lowed by potassium ion concentration (17.6%). The centage for critical values in STAT laboratory tests literature shows that potassium, sodium and glu- were pH (69.6%), haematocrit (69.6%), glucose cose are responsible for the majority of critical re- (68.6%), potassium ion (64.1%) and haemoglobin sults (23). (60.4%). For inpatients, the parameters with the In STAT laboratory and routine laboratory tests for most notifications were glucose (91.9%), phos- inpatients, the parameters with the highest per- phate (90.9%), haemoglobin (81.8%), sodium ion centage of critical value reporting are pH, haemo- (81%) and potassium ion (80.2%). In outpatients globin, glucose and potassium ion concentrations they were calcium (65.8%) and potassium (61.9 %). (20). In our study, haematocrit, phosphate and so- dium ion concentrations were found to have the Discussion same notification percentages. For outpatients, calcium and potassium were the parameters with In this study, we analysed all data collected over a the highest proportion of critical value notifica- 6-month period to obtain an overview of how our tions. In the literature calcium and potassium system works with regard to the detection and re- showed lower levels of notification (19,20). porting of critical values. On reviewing the litera- For serum calcium concentration, our findings for ture, only few studies on critical value notification inpatients in the ICU differed from those reported have been performed in our country, and our re- in the literature (24). These differences are due to sults are in accordance with published data (20). an agreement with clinicians at our hospital, Comparing the obtained percentages of critical whereby, it is not necessary to notify clinicians if values in our laboratory with results published in the serum calcium concentration is lower than 1.75 the literature, the percentages for almost all pa- mmol/L unless the patient’s serum albumin con- rameters were similar (21-23). As in our study, the centration is higher than 25 g/L. vast majority of critical results are related to pa- tients in the ICU and emergency department (17). When routine laboratory tests were sorted by in- The highest proportion of critical values corre- patients and outpatients, the critical value notifi- sponded to oxygen partial pressure (17.7%), fol- cation percentage was significantly higher for in- https://doi.org/10.11613/BM.2019.010704 Biochem Med (Zagreb) 2019;29(1):010704 5 Arbiol-Roca A. et al. Analysis of laboratory critical values patients than for outpatients (except for thrombo- previous studies have demonstrated better timing cytes and calcium). This indicates that the notifica- of reports for automated alerts than traditional tion system for outpatients is weaker, and more systems. Another important issue is who should effort should be made to improve it. Other studies receive the result. In some studies, the vast major- have also shown a lower notification percentages ity of clinical laboratories consider the appropriate for outpatients (15). personnel to receive the critical value notification Laboratory staff should be educated about the re- to be the physician who requested the test, any sults and learn the entire list of critical values, rein- physician responsible for the patient or the nurse forcing the importance of communicating all of responsible for the patient (28). them. An indicator of the quality of this process may Our study is one of few studies performed in a ter- be the critical value reporting rate, as the failure to tiary hospital, which also employs different meth- report these values, estimated at 0.1% to 10%, could ods of critical value notification. As two main be indicative of the operational efficiency of labora- points, the obtained results demonstrate the im- tories (14). Different studies have demonstrated that portance of different critical value definitions for the notification of critical values varies between special situations (e.g., serum calcium for ICU pa- tests, with percentages varying from 0.7% to 9%, tients), as well as the importance of registration of similar to that found in our study (15). critical value communication. It is necessary to in- This study has been useful in enabling us to ob- corporate actions to improve not only critical val- serve the critical value notification percentage, ue notification, but especially the registration of and we were able to detect flaws in the annotation this activity. Therefore, we believe this study will system of the communication process. Indeed, be of interest to other clinical laboratories, and these values were reported to clinicians, but were recommend that they undertake this kind of study not always registered in the laboratory’s internal as it highlights the weaknesses and strengths of a archive. These values are initially communicated to system. This study shows that notification system clinicians, and once they are advised a comment is for outpatients is weaker than for inpatients. In an- registered in the LIS. This second step was forgot- other hand, we detected that most of the critical ten in several cases. This explains why the commu- values although they were communicated, it were nication percentage was much lower than expect- not registered in laboratory LIS. This evaluation ed. Due to the obtained results, a reminder was helps us to educate to all laboratory staff in order made to all laboratory staff to ensure this second to register all notifications. step is carried out. A limitation of our study is that it was conducted Most guidelines demonstrate that direct calling is retrospectively, and only evaluates data collected still the most used method for communicating during a 6-month period. A prospective study could critical values and, as such, all communication be designed to verify the obtained data and evalu- should be fully documented. Recent studies com- ate the results after improvement of the process. pared automated alert systems to telephone noti- Another limitation is that we did not analyse the fications from a call centre (17). Automated notifi- time between receipt of the samples and commu- cation systems are automated alert systems or nication of the critical values, which would be an computerised reminders via mobile phones, pag- interesting aspect to investigate in future studies. ers, email or other personal electronic devices, Acknowledgments used to alert clinicians of critical value laboratory test results (14,25-27). Upon receipt of an automat- We thank the anonymous reviewers whose com- ed notification, the responsible personnel ac- ments and suggestions helped improve and clari- knowledge the critical value and confirm receipt fy this manuscript. of the alert. If the alert is not acknowledged within a specified timeframe, these systems typically re- Conflict of interest vert to a manual notification system. Findings of None declared. Biochem Med (Zagreb) 2019;29(1):010704 https://doi.org/10.11613/BM.2019.010704 6 Arbiol-Roca A. et al. Analysis of laboratory critical values References 1. Lundberg GD. When to Panic over Abnormal Values. Med 16. Campbell C, Horvath A. Towards harmonisation of criti- Lab Observer 1972;4:47-54. cal laboratory result management - review of the literatu- 2. College of American Pathologists. Laboratory Accreditation re and survey of australasian practices. Clin Biochem Rev. checklist. Available at: http://www.cap.org/apps/cap.por- 2012;33:149-60. tal. Accessed January 18th 2018. 17. Piva E, Pelloso M, Penello L, Plebani M. Laboratory critical 3. Joint Commission on Accreditation of Healthcare Organiza- values: automated notification supports effective clinical tions. International Patient Safety Goals 2007. Available at: decision making. Clin Biochem. 2014;47:1163-8. https://doi. http://www.jointcommission.org/PatientSafety/National- org/10.1016/j.clinbiochem.2014.05.056 PatientSafetyGoals. Accessed January 8th 2018. 18. Don-Wauchope AC, Chetty VT. Laboratory defined critical 4. International Organization for Standardization (ISO). ISO value limits: how do hospital physicians perceive laboratory 15189:2012. Medical laboratories – particular requirements based critical values? Clin Biochem. 2009;42:766-70. https:// for quality and competence. Geneva, Switzerland: Internati- doi.org/10.1016/j.clinbiochem.2009.02.016 onal Organization for Standardization; 2012. 19. Wagar EA, Friedberg RC, Souers R, Stankovic AK. Critical va- 5. Clinical and Laboratory Standards Institute (CLSI). Manage- lues comparison: a College of American Pathologists Q-Pro- ment of Critical- and Significant-Risk Result. CLSI guideline bes survey of 163 clinical laboratories. Arch Pathol Lab Med. GP47 1st Edition. Wayne, PA: Clinical and Laboratory Stan- 2007;131:1769-75. dards Institute; 2015. 20. Llopis-Díaz MA, Gómez-Rioja R, Álvarez-Funes V, Martínez- 6. Kost GJ. Critical limits for emergency clinician notification at Brú C, Cortés -Rius M, Barba-Meseguer N, et al. Critical va- United States children’s hospitals. Pediatrics. 1991;88:597- lues reporting: Results of a Spanish laboratories survey. Rev 603. Lab Clin. 2010;3:177–82. 7. Kost GJ. Critical limits for urgent clinician notification at 21. Lum G. Evaluation of a laboratory critical limit (alert va- US medical centers. JAMA. 1990;263:704-7. https://doi. lue) policy for hypercalcemia. Arch Pathol Lab Med. org/10.1001/jama.1990.03440050098042 1996;120:633-6. 8. Fine RH. Laboratory critical limits. JAMA. 1990;264:334. 22. Howanitz JH, Howanitz PJ. Evaluation of serum and whole https://doi.org/10.1001/jama.1990.03450030050019 blood sodium critical values. Am J Clin Pathol. 2007;127:56- 9. https://doi.org/10.1309/Q3Y27QQLEL19340A 9. Lundberg GD, Iverson C, Radulescu G. Now read this: the SI units are here. JAMA. 1986;255:2329-39. https://doi. 23. Tillman J, Barth JH; ACB National Audit Group. A sur- org/10.1001/jama.1986.03370170093043 vey of laboratory ‘critical (alert) limits’ in the UK. Ann Clin Biochem. 2003;40:181-4. https://doi. 10. Croal B. The communication of critical and unexpected pat- org/10.1258/000456303763046148 hology results. The Royal College of Pathologists. Available at: https://www.rcpath.org/resourceLibrary/the-communi- 24. Howanitz JH, Howanitz PJ. Evaluation of total serum calci- cation-of-critical-and-unexpected-pathology-results-pdf. um critical values. Arch Pathol Lab Med. 2006;130:828-30. html. Accessed January 8th 2018. 25. Park HI, Min WK, Lee W, Park H, Park CJ, Chi HS, Chun S.. Eva- 11. Lippi G, Mattiuzzi C. Critical laboratory values communica- luating the short message service alerting system for criti- tion: summary recommendations from available guideli- cal value notification via PDA telephones. Ann Clin Lab Sci. nes. Ann Transl Med. 2016;4:400. https://doi.org/10.21037/ 2008;38:149-56. atm.2016.09.36 26. Kuperman GJ, Teich JM, Tanasijevic MJ, Ma’Luf N, Ritte- 12. Dighe AS, Rao A, Coakley AB, Lewandrowski KB. Analysis of nberg E, Jha A, et al. Improving response to critical labora- laboratory critical value reporting at a large academic me- tory results with automation: results of a randomized con- dical center. Am J Clin Pathol. 2006;125:758-64. https://doi. trolled trial. J Am Med Inform Assoc. 1999;6:512-22. https:// org/10.1309/R53XVC2U5CH6TNG8 doi.org/10.1136/jamia.1999.0060512 13. Plebani M, Sciacovelli L, Aita A. Quality Indicators for the To- 27. Etchells E, Adhikari NK, Cheung C, Fowler R, Kiss A, Quan tal Testing Process. Clin Lab Med. 2017;37:187-205. https:// S, et al. Real-time clinical alerting: effect of an automated doi.org/10.1016/j.cll.2016.09.015 paging system on response time to critical laboratory va- lues--a randomised controlled trial. Qual Saf Health Care. 14. Piva E, Sciacovelli L, Zaninotto M, Laposata M, Pleba- 2010;19:99-102. https://doi.org/10.1136/qshc.2008.028407 ni M. Evaluation of effectiveness of a computerized no- tification system for reporting critical values. Am J 28. Liebow EB, Derzon JH, Fontanesi J, Favoretto AM, Baetz Clin Pathol. 2009;131:432-41. https://doi.org/10.1309/ RA, Shaw C, et al. Effectiveness of automated notificati- AJCPYS80BUCBXTUH on and customer service call centers for timely and accu- rate reporting of critical values: a laboratory medicine best 15. Agarwal R, Chhillar N, Tripathi CB. Study of variables affec- practices systematic review and meta-analysis. Clin Bio- ting critical value notification in a laboratory catering to chem. 2012;45:979-87. https://doi.org/10.1016/j.clinbio- tertiary care hospital. Indian J Clin Biochem. 2015;30:89-93. chem.2012.06.023 https://doi.org/10.1007/s12291-013-0409-x https://doi.org/10.11613/BM.2019.010704 Biochem Med (Zagreb) 2019;29(1):010704 7