Medication-related Adverse Outcomes and Contributing Factors ...

3 downloads 164 Views 2MB Size Report
Dec 19, 2014 - I would also like to sincerely thank MSc Marita Halonen for ... express my profound gratitude to PhD Taina Pitkäaho, PhD Tarja Välimäki, PhD ...
Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients Patients’ medication process is a complex multi-stage and multiprofessional process and an important part of nurses’ daily work. In this study, medicationrelated adverse outcomes were analysed using medicationrelated incident reports, patients’ records using the Global Trigger

dissertations | 260 | MARJA HÄRKÄNEN

MARJA HÄRKÄNEN

MARJA HÄRKÄNEN

Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients

Tool method, and observations of medication administrations. The study demonstrated that medication-related adverse outcomes are common and three methods produced different information about these outcomes.

Publications of the University of Eastern Finland Dissertations in Health Sciences

Publications of the University of Eastern Finland Dissertations in Health Sciences isbn 15461+61654

Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients

MARJA HÄRKÄNEN

Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients An Analysis Using Hospital’s Incident Reports, the Global Trigger Tool Method, and Observations with Record Reviews

To be presented by permission of the Faculty of Health Sciences, University of Eastern Finland for public examination in Mediteknia MD100 Auditorium, Kuopio, on Friday, December 19th 2014, at 12 noon

Publications of the University of Eastern Finland Dissertations in Health Sciences 260

Department of Nursing Science Faculty of Health Sciences University of Eastern Finland Kuopio 2014

Juvenes Print ─ Suomen Yliopistopaino Oy Tampere, 2014 Series Editors: Professor Veli-Matti Kosma, M.D., Ph.D. Institute of Clinical Medicine, Pathology Faculty of Health Sciences Professor Hannele Turunen, Ph.D. Department of Nursing Science Faculty of Health Sciences Professor Olli Gröhn, Ph.D. A.I. Virtanen Institute for Molecular Sciences Faculty of Health Sciences Professor Kai Kaarniranta, M.D., Ph.D. Institute of Clinical Medicine, Ophthalmology Faculty of Health Sciences Lecturer Veli-Pekka Ranta, Ph.D. (pharmacy) School of Pharmacy Faculty of Health Sciences Distributor: University of Eastern Finland Kuopio Campus Library P.O.Box 1627 FI-70211 Kuopio, Finland http://www.uef.fi/kirjasto ISBN (print): 978-952-61-1635-8 ISBN (pdf): 978-952-61-1636-5 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706

III

Author’s address:

Department of Nursing Science University of Eastern Finland KUOPIO FINLAND

Supervisors:

Professor Katri Vehviläinen-Julkunen, Ph.D. Department of Nursing Science University of Eastern Finland Kuopio University Hospital KUOPIO FINLAND Professor Hannele Turunen, Ph.D. Department of Nursing Science University of Eastern Finland Kuopio University Hospital KUOPIO FINLAND

Reviewers:

Professor Walter Sermeus, Ph.D. Department of Public Health & Primary Care Catholic University Leuven LEUVEN BELGIUM Docent Anna Liisa Aho, Ph.D. School of Health Sciences University of Tampere TAMPERE FINLAND

Opponent:

Professor Marja Kaunonen, Ph.D. School of Health Sciences University of Tampere TAMPERE FINLAND

IV

V

Härkänen, Marja Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients - an Analysis Using Hospital’s Incident Reports, the Global Trigger Tool Method, and Observations with Record Reviews University of Eastern Finland, Faculty of Health Sciences Publications of the University of Eastern Finland. Dissertations in Health Sciences 260. 2014. 63 p. ISBN (print): 978-952-61-1635-8 ISBN (pdf): 978-952-61-1636-5 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706

ABSTRACT The purpose of this study is to present a comprehensive and valid estimate of the problems that arise in the medication process in hospitals. Specifically the study aims to examine medication-related adverse outcomes and contributing factors of hospital patients, to study the associations between adverse outcomes and contributing factors, and to compare differences between the detection methods. This study was conducted in one university hospital in Finland. Three types of data sets were analysed statistically including retrospectively collected medication-related incident reports (n=671) from the year 2010, retrospectively collected randomly selected patients’ records (n=463) from the year 2011 using the Global Trigger Tool (GTT) method, and observations (n=1058) of medication administrations by nurses’ with record reviews (n=122) during April to May 2012. In addition, secondary analysis of medication administration errors (n=453) detected by three methods was conducted. A total of (n=1059) medication errors and (n=311) adverse drug events were detected. Harm to patients was caused in 48% of detected medication errors in GTT data, 18% in incident reports, and 3% in observational data. Most of the detected errors were administration or documenting errors. The most common types of medication errors were wrong dose, omission, and wrong administration technique. There were differences between the detection methods when the information of the medication errors stages, types, and severities were compared. The most important work environmental factors contributing to errors were rush, lack of training, problems in the communication systems, in the electronic records, or in the common policies and procedures. Omission of double-checking, problems in communication and flow of information were the most common among the team factors contributing to errors. Of the employee-related factors performance deficit, stress/high volume workload, miscalculation of dosage or infusion rate, and knowledge deficit were the most common. The most important patient-specific factors were the amount of drugs, length of hospital stay, coronary artery disease, and co-morbidity. The most common drugrelated factors contributing to errors were other than p.o administration and specific drugs. This study demonstrated that medication-related adverse outcomes are common and incident reports, GTT, and observation methods produce different information about the problems in the medication process. Understanding the complex reality of the hospital environment and the medication process can be limited by using only one detection method, because each detection methods had its limitations. Thus, combining the methods revealed more diverse information regarding medication-related problems in hospital that can be used to increase safety in the medication process. National Library of Medicine Classification: WX 185; QV 56 Medical Subject Headings: Medication Errors; Drug-Related Side Effects and Adverse Reactions; Drug Therapy - adverse effects; Patient Safety; Hospitals; Nursing; Finland

VI

VII

Härkänen, Marja Lääkehoidon vaaratapahtumat ja niihin myötävaikuttavat tekijät sairaalapotilailla – analyysi käyttäen vaaratapahtumaraportteja, Global Trigger Tool menetelmää ja havainnointia yhdistettynä potilaskertomusanalyysiin Itä-Suomen yliopisto, terveystieteiden tiedekunta Publications of the University of Eastern Finland. Dissertations in Health Sciences 260. 2014. 63 s. ISBN (print): 978-952-61-1635-8 ISBN (pdf): 978-952-61-1636-5 ISSN (print): 1798-5706 ISSN (pdf): 1798-5714 ISSN-L: 1798-5706

TIIVISTELMÄ Tämän tutkimuksen tarkoituksena on esittää kattava ja luotettava kuvaus sairaalan lääkehoitoprosessissa esiintyvistä ongelmista. Tutkimuksen tavoitteena on erityisesti tutkia sairaalapotilaiden lääkehoidon vaaratapahtumia ja niihin myötävaikuttavia tekijöitä, tutkia lääkehoidon vaaratapahtumien ja niihin myötävaikuttavien tekijöiden yhteyttä sekä vertailla eroja eri tutkimusmenetelmien välillä. Tutkimus toteutettiin yhdessä suomalaisessa yliopistosairaalassa. Neljässä eri osatutkimuksessa analysoitiin tilastollisin menetelmin retrospektiivisesti kerättyjä sairaalan lääkehoitoon liittyviä vaaratapahtumaraportteja (n=671) vuodelta 2010, retrospektiivisesti kerättyjä satunnaisesti valittujen potilaiden hoitokertomuksia (n=463) vuodelta 2011 käyttämällä Global Trigger Tool (GTT) menetelmää, ja lääkkeiden antamisen havainnointia (n=1058) yhdistettynä potilaskertomusanalyysiin (n=122) huhti-toukokuussa 2012. Lisäksi sekundaarianalyysissä analysoitiin kaikki kolmella edellä mainituilla menetelmillä havaitut lääkkeiden antamisvirheet (n=453). Yhteensä havaittiin 1059 lääkitysvirhettä ja 311 lääkehoidon haittatapahtumaa. GTT menetelmän avulla havaituista lääkitysvirheistä potilaille haittaa aiheutui 48%:ssa, 18%:ssa vaaratapahtumaraportteihin raportoiduista lääkitysvirheistä ja 3%:ssa havainnoiduista lääkitysvirheistä. Suurin osa lääkitysvirheistä oli lääkkeiden antovirheitä tai kirjaamisvirheitä. Yleisimmät virhetyypit olivat väärä annos, lääke saamatta tai väärä lääkkeenantotekniikka. Tärkeimmät työympäristön myötävaikuttavat tekijät virheisiin olivat kiire, koulutuksen puute, tai ongelmat kommunikaatiosysteemeissä, sähköisissä sairauskertomuksissa tai yhteisissä ohjeissa ja toimintatavoissa. Kaksoistarkastuksen laiminlyönti, ongelmat kommunikaatiossa ja tiedonkulussa olivat yleisimmät tiimiin liittyvistä myötävaikuttavista tekijöistä. Työntekijään liittyvistä myötävaikuttavista tekijöistä yleisimpiä olivat työntekijän suorituksen puutteellisuus, stressi/työnkuorma, virheellinen lääkeannoksen tai infuusionopeuden laskeminen ja tiedon/osaamisen puutteellisuus. Potilaaseen yhteydessä olevista tekijöistä lääkkeiden määrä, sairaalahoidon pituus, koronaaritauti ja sairauksien määrä myötävaikuttivat lääkehoidon haitallisiin tapahtumiin. Lääkkeisiin liittyviä myötävaikuttavia tekijöitä olivat muu kuin lääkkeen suunkautta annostelu ja tietyt lääkevalmisteet. Tutkimus osoitti että lääkehoitoon liittyvät vaaratapahtumat ovat yleisiä. Erilaiset tutkimusmenetelmät tuottivat erilaista tietoa lääkehoitoprosessissa olevista ongelmista ja niihin myötävaikuttavista tekijöistä. Koska jokaisella tutkimusmenetelmällä on omat rajoituksensa, sairaalan kompleksisen lääkehoitoprosessin ymmärtäminen ja siten turvallisuuden parantaminen voi olla rajoittunutta käyttämällä vain yhtä tutkimusmenetelmää. Luokitus: WX 185; QV 56 Yleinen suomalainen asiasanasto: lääkehoito; virheet; vaaratilanteet; potilasturvallisuus; sairaalat; hoitotyö; Suomi

VIII

IX

“The very first requirement in a hospital is that it should do the sick no harm.” Florence Nightingale (Notes on Nursing: What It Is, and What It Is Not 1858)

X

XI

Acknowledgements This study was conducted at the University of Eastern Finland, Department of Nursing Science, in Doctoral programme of Nursing Science and At Safe –project both led by Professor Katri Vehviläinen-Julkunen. I wish to express my sincere gratitude to several people for providing help, guidance and support in different ways during my doctoral studies. First and foremost, I would like to show my deepest gratitude to my principal supervisor, Professor Katri Vehviläinen-Julkunen. You have provided me guidance and support with great wisdom and recognized my strengths and limitations. You have been there to share with me moments of joy and success, but have also closely witnessed a process of growth that has naturally also included some moments of growing pains. I would also like to express my warmest gratitude to my second supervisor, Professor Hannele Turunen. You have also been available whenever I have wanted to ask questions or discuss any issues. You have encouraged me in so many ways. I owe my heartfelt gratitude to both of you for your first-rate guidance and the responsibilities given me outside my thesis work, which have been important for honing my skills as a researcher. I would like to warmly thank the pre-examiners of my thesis, Professor Walter Sermeus and Docent Anna-Liisa Aho, for outstanding review process that provided me with one more learning opportunity. I would also like to thank for the encouraging evaluation I received from you. I have shared this process with many others with whom I have had the privilege to work. I would like to thank Pharmacist Susanna Saano, PhD, for your enthusiasm and excellent perspectives and for giving me concrete help during this process. I sincerely thank Pharmacist Jouni Ahonen, PhD, and Physician Marjo Kervinen, MD, for your notable expertise and the countless working hours you were able to put in the analysis and, later, writing of the publications. I would also like to sincerely thank MSc Marita Halonen for your invaluable help and your immeasurable work input, especially during data collection, but also for your friendship, support and encouragement throughout the years. I also owe my gratitude to Docent Ari Voutilainen for offering me your expertise, particularly on statistical analysis. I am also thankful for you sharing your views on science and on the work of the researcher. These conversations have helped me expand my own views. I am grateful that I have been able to conduct my thesis work as a part of a genuine scientific community. I would like to sincerely thank each one of my colleagues at the Department of Nursing Science for giving me the opportunity to acquire researcher skills beside you. I have particularly felt close to and helped by those who have graduated with a doctorate during my doctoral thesis process or briefly before it. I would particularly like to express my profound gratitude to PhD Taina Pitkäaho, PhD Tarja Välimäki, PhD Anne Vaajoki and PhD Marjorita Sormunen for having shared your experiences and giving me an opportunity to learn through our discussions. It takes a person in the same situation to best understand the joys and challenges of the task. I am grateful for having been surrounded by many post-graduate students throughout my doctoral thesis process. I would particularly like to thank MSc Reeta Lamminpää for sharing with me the everyday life of a doctoral student. I would also like to thank those who have given me practical help in this process. My gratitude goes to Petra Isotalo for taking care of practical matters related to my doctoral thesis, MSc Elisa Wulff for editing and proofreading my thesis and MSc MarjaLeena Lamidi for your help in conducting statistical analyses. I have been privileged to have had the opportunity to conduct full-time work on my doctoral dissertation with the funding of the Finnish Doctoral Programme in Nursing Science. However, this financial support is only one of the benefits that the programme has offered me. I am grateful for having been able to get to know faculty and doctoral students

XII

from different universities. I would particularly like to thank MSc Maaret Vuorenmaa, MSc Anne Oikarinen, MSc Sanna Koskinen, MSc Hannakaisa Niela-Vilén and MSc Linda Nyholm for walking beside me through this years-long journey and sharing your experiences along the way. I would like to show my warm gratitude to the director of the programme, Professor Helena Leino-Kilpi and Coordinator Heli Virtanen. The programme helped me to proceed with a firm sense of direction and also gave me many opportunities and competences in learning important researcher skills. I took part in an international researcher exchange at the Centre for Quality, Region of Southern Denmark, where I had the opportunity to deepen my understanding of patient safety research. I especially owe my sincere gratitude to Professor Erik Hollnagel, Jeanette Hounsgaard and Monika Kring. You made the exchange an unforgettable experience. Completing a doctoral thesis is always a major effort. I am grateful for those close to me who have supported throughout this journey or who have reminded me of life outside work. I am deeply grateful for my mother and stepfather, Anja and Viljo Tikkanen, for all the support and encouragement you have given me and for your firm belief in my skills. I would also like to thank you for all of help you have offered me in looking after my children, dogs or home whenever I have asked for it. I would also like to profoundly thank my in-laws, Kaisu and Kari Härkänen, and the grandparents of my husband, Tauno and Kerttu Härkänen, for all the help, support and reassurance you have given me. I would also like to thank my friends, especially Annakaisa Koljonen and Hanna and Jarkko Kaartinen. Thank you for the countless walks, gym visits, concerts and trips together and for the fact that you have always paid me visits without a need for a formal invitation. I would also like to thank you in supporting and accompanying me on my way to ”the big world” when I took part in my first international conference in London in 2012. My husband Mika, you have walked beside me for the last 15 years. You have always supported me and I know you are sincerely happy for me and proud of this accomplishment. Thank you for your love. I dedicate this doctoral thesis to our dear children, Roosa, Saara and Veera. I hope that in the future this thesis serves as an indication and reminder for you that you may and must have dreams and set goals in life, and that it is possible to make them come true by being brave and working persistently. Even though it is always challenging to combine family and work, I would like to thank my family for never making me feel guilty for having had to prioritise work at times. I hope that even in the moments when I have not been physically or emotionally present, you have never had to doubt what really matters in my life. Finally, I owe my deep gratitude to all who have financially supported my work: the Finnish Doctoral Education Network in Nursing Science (previous Finnish Doctoral Programme in Nursing Science), At Safe research project (strategic funding from the University of Eastern Finland), Faculty of Health Sciences (University of Eastern Finland), Kuopio University Hospital (EVO funding), the Finnish Nurses Association, Finnish Concordia Fund, and Center for Quality, Region of Southern Denmark (a scientific visit grant). In Savonlinna, at the first fall of snow, October 2014

Marja Härkänen

XIII

List of the original publications

This dissertation is based on the following original publications:

I

II

Härkänen M, Turunen H, Saano S and Vehviläinen-Julkunen K. Detecting medication errors: Analysis based on a hospital's incident reports. International Journal of Nursing Practice, 2013 (In Press). doi: 10.1111/ijn.12227. Härkänen M, Kervinen M, Ahonen J, Voutilainen A, Turunen H and VehviläinenJulkunen K. Patient-specific risk factors of adverse drug events in adult inpatients – evidence detected using the Global Trigger Tool method. Journal of Clinical Nursing, 2014 (In Press). doi: 10.1111/jocn.12714.

III

Härkänen M, Ahonen J, Kervinen M, Turunen H and Vehviläinen-Julkunen K. The factors associated with medication errors in adult medical and surgical inpatients: a direct observation approach with medication record reviews. Scandinavian Journal of Caring Sciences, 2014 (In Press). doi: 10.1111/scs.12163.

IV

Härkänen M, Turunen H and Vehviläinen-Julkunen K. Differences between medication administration errors detected using incident reports, Global Trigger Tool method and direct observations. Re-submitted, 2014.

The publications were adapted with the permission of the copyright owners. Summary also includes previously unpublished material.

XIV

XV

Contents 1 INTRODUCTION .............................................................................. 2 MEDICATION-RELATED ADVERSE OUTCOMES AND CONTRIBUTING FACTORS .............................................................. 2.1 Medication management .............................................................. 2.1.1 Medication process and its challenges in hospitals in Finland ......................................................................................... 2.1.2 Nurses' medication education and skills ........................... 2.2 Models for measuring safety ........................................................ 2.3 Definitions of medication-related adverse outcomes ............... 2.4 Methods for detecting medication-related adverse outcomes 2.4.1 Incident reports ...................................................................... 2.4.2 Global Trigger Tool method................................................. 2.4.3 Observation method .............................................................. 2.5 Contributing factors to medication-related adverse outcomes 2.6 Summary of the study background .............................................

1

3 3 3 4 6 8 9 11 13 15 17 19

3 PURPOSE, AIMS AND OBJECTIVES ............................................

20

4 MATERIAL AND METHODS ......................................................... 4.1 Study setting and design ............................................................... 4.2 Sample, data collection and analysis .......................................... 4.2.1 Medication-related incident reports ................................... 4.2.2 Patients' records and Global Trigger Tool ......................... 4.2.3 Observations of administered drugs .................................. 4.2.4 Secondary analysis of administration errors .................... 4.3 Validity and reliability of the study ............................................. 4.4 Ethical issues ...................................................................................

21

5 RESULTS .............................................................................................. 5.1 Medication-related adverse outcomes ........................................ 5.1.1 Types of medication-related adverse outcomes ................ 5.1.2 Severity of medication-related adverse outcomes ............ 5.2 Contributing factors of medication-related adverse outcomes 5.2.1 Work environment factors.................................................... 5.2.2 Team factors ........................................................................... 5.2.3 Emplyee-related factors ........................................................ 5.2.4 Patient-specific factors .......................................................... 5.2.5 Medication-related factors .................................................... 5.3 Summary of the study results.......................................................

21 22 22 22 24 25 26 27 28 28 28 29 31 32 32 33 34 34 36

XVI

6 DISCUSSION ...................................................................................... 6.1 Discussion of the study results .................................................... 6.1.1 Medication-related outcomes .............................................. 6.1.2 Contributing factors to medication-related outcomes ..... 6.1.3 Comparing medication-related outcomes detected using different methods ................................................................ 6.2 Discussion of the limitations and strengths of the study .........

37 37 37 39 43 44

7 CONCLUSIONS .................................................................................

46

8 RECOMMENDATIONS ................................................................... 8.1 Recommendations for further research ...................................... 8.2 Recommendations for clinical practice ....................................... 8.3 Recommendations for leadership and policy level ................... 8.4 Recommendations for education of nurses ................................

47

9 REFERENCES ......................................................................................

APPENDICES

47 48 48 49 50

XVII

Abbreviations ADE

Adverse drug event

AE

Adverse event

ADM

Automated dispensing machines

ADR

Adverse drug reaction

BCMA

Bar code medication administration

BPOC

Bar code point of care

CI

Confidence interval

CVI

Content validity index

CVR

Content validity ratio

ECTS

European credit transfer and accumulation system

ED

Emergency department

EU

European Union

GTT

Global Trigger Tool

HaiPro

Finnish incident reporting database

ICU

Intensive care unit

IV

Intravenous

MAE

Medication administration error

MAR

Medication administration records

ME

Medication error

NCCMERP

National Coordinating Council for Medication Error Reporting and Prevention

OR

Odds ratio

RCA

Root Cause Analysis

RN

Registered nurse

WHO

World Health Organization

XVIII

1 Introduction Medication management is a complex multi-stage and multi-disciplinary process, involving physicians, pharmacists, nurses and patients. Problems in the medication process can occur at any stage from prescribing, dispensing and administering, to recording and reporting. Medication management is an important part of registered nurses’ (RNs’) daily work worldwide. At the same time, it represents one of the most challenging tasks for nurses. Medication administration, in particular, is a typical nursing task. In that phase nurses should be able to recognize errors and prevent errors from affecting the patient, thus nurses are the final stage of defense in the medication process (Adhikari et al. 2014) and are working in the “sharp end” of the medication process where failures are visible and accidents are experienced (Morath & Turnbull 2005). Nurses have dual roles in the medication process; they both generate errors and prevent them (Barker et al. 2002). Health care is rapidly changing due to an aging population in need of more complicated and challenging care (Amalberti 2013). In Finland, the use of different medications is increasing every year (Finnish Medicines Agency Fimea & Social Insurance Institution 2013). New technology, multiple providers, competition, and the growing ubiquity of information increase the complexity of health care (Clancy et al. 2008). If this complexity is not managed, dangerous situations can occur (Unver et al. 2012). Worldwide health care systems are facing increasing difficulties in securing adequate resources and recruiting adequately trained personnel to maintain the quality of care. Reports of dangerous medication errors in the media and other signals have suggested that medication errors in health care may be increasing. (Kuitunen et al. 2008.) These carry serious implications for patients’ well-being, morbidity, mortality, and the costs for the health system and society (Pham et al. 2011, Ford et al. 2006). They also cause human suffering (Nilsson et al. 2012), for patients and family members as well as health-care professionals (Gerven et al. 2014). It is estimated that among the EU member states, 8% to 12% of all patients admitted to hospitals, suffer from adverse events whilst receiving care (European Commission 2010). In Finland, it is estimated that 700-1700 patient die every year because of adverse outcomes of care. Moreover, adverse events may cost as much as 1 billion euros annually. (National Institute for Health and Welfare 2014, 2011.) One of the aims of the European Commission is to improve the safety of care for patients (European Commission 2006). In Finland, the basis for patient safety is determined by the Health Care Act (1326/2010) which obligates to plan the patient safety work and by the Finnish patient safety strategy of 2009-2013 (Ministry of Social Affairs and Health 2009a). Since 2007, a web-based reporting system called HaiPro has been used for collecting information of incidents and adverse events in most hospitals and social organizations. Still, promoting of patient safety in Finland at the administration level has started much later than for example in other Nordic countries (National Institute for Health and Welfare 2011). The quantity of publications concerning patient and medication safety has increased rapidly. For the year 1997, there were only 80 studies in the PubMed database using the search term “patient safety”. By years 2012-2013 the same search term yielded 3000 and even more publications annually. The number of publications using search term “medication safety” also increased in the PubMed database, but those studies represent an obvious minority of all safety studies. (Figure 1.) A significant increase in the number of publications can be seen after year 2000, when the Institute of Medicine published the report “To err is human” (Kohn et al. 2000). This report revealed that almost 100,000 patient die annually in the United States of America due to medical errors made in hospitals. Still, the latest report demonstrates that the corresponding estimation may even be fourfold that figure, suggesting that developments in the field of patient safety have been largely

2

ineffective (James 2013). Thus, novel and much more effective strategies and tools are needed to address the complex quality challenges confronting health care (Chassing 2013).

Figure 1. The number of publications concerning patient safety and medication safety in the PubMed database, years 1997-2013

Continuing research that deepens our current understanding of the problem and improves the quality and safety of care is needed. According to the World Health Organization, worldwide research into patient safety is still in its infancy. New knowledge is required to measure and understand the risks and causes of harm and to develop solutions that prevent, or reduce, the effect of harm. (WHO 2012.) Medication safety represents one of the major concerns in patient safety globally (Kim & Bates 2013), thus greater attention should be devoted to studying, identifying and preventing problems in the medication processes. Medication errors and adverse events can be studied by different methods. Every method has its strengths and limitations, and they identify different problems (MeyerMassetti et al. 2011, Naessens et al. 2009). Although medication error measurement methods and prevention strategies are important areas of research, a fully adequate method does not exist (Ford et al. 2006), because none of the existing methods, if employed solely, can detect all patients with treatment-related damages (Matlow et al. 2011). Thus, it has been recommended that different methods be combined to reveal the problems in clinical practice (Naessens et al. 2009, Olsen et al. 2007). The purpose of this study is to present a more comprehensive and valid estimate of the problems that arise in the medication process in hospitals by gaining information of medication-related adverse outcomes and contributing factors. The study also outlines important methodological information concerning the differences between detection methods by comparing the information on medication-related adverse outcomes. Specifically the study aims to examine medication-related adverse outcomes and contributing factors of hospital patients detected using incident reports, Global Trigger Tool method and observation methods, to study the associations between adverse outcomes and contributing factors, and to compare differences between the detection methods. This type of research is relatively new in Finland and in nursing science. Even though single medication safety studies have been conducted, combining the incident reports, the Global Trigger Tool method and observation method for comparing differences of medication-related adverse outcomes of hospital patients provides an innovative strategy compared to the international field of safety research.

3

2 Medication-related Adverse Outcomes and Contributing Factors 2.1 MEDICATION MANAGEMENT 2.1.1 Medication process and its challenges in hospitals in Finland In Finland, pharmaceutical services are guided, supervised and developed under the Finnish Medicines Agency (Fimea), National Supervisory Authority for Welfare and Health (Valvira), and National Institute for Health and Welfare all of whom work under the Ministry of Social Affairs and Health. Medication management is a health care activity that is carried out by educated and trained health care professionals in multi-professional collaboration. Every professional has his/her own role to ensure effective and safe medication management. (Ojala 2012, Ministry of Social Affairs and Health 2009b.) Medications are usually dispensed and administered based on a physician’s instructions and orders. There are also some specially educated nurses who enjoy a limited right to prespcribe medications in Finland (Valvira 2014, Act 1088/2010, 1089/2010). The physician or specially educated nurse will decide whether a treatment should be commenced, adjusted or discontinued in collaboration with the patient. The patient should understand why the medication is prescribed, how it should be used, and what are the possible adverse effects of the drugs. Safe medication requires that the drug choice is correct and suits to the patient's condition. Risk areas in the prescribing consist of: delays or errors in the patients’ diagnoses; failure to condsider the patients’ allergies, contra-indications or interactions; problems identifying the correct drugs; difficulties monitoring the effect of care; and shortcomings in patient counseling. (Ministry of Social Affairs and Health 2009b.) It is the nurses’ responsibility to understand the medication order correctly, prepare the medication dose correctly, and administer the medication to ensure that the patient receives the right medication and dose, at the appropriate time, using approved administration techniques and routes. The risk areas for implementing the medications are deviations during receiving the medication orders including problems in documenting and communication, deviations in dispensing, preparing or administering medications, or problems in monitoring and informing the effect of care, and problems in patient counseling. (Ministry of Social Affairs and Health 2009b.) The medication processes in hospitals differ from primary care settings. For example intravenous medications are quite common in hospital use, and that requires that nurses are able to manage different administration techniques, and that they possess knowledge of the specific characteristics of many different drugs, some which may be very unusual (Ojala 2012). Usually registered nurses are responsible for medication management in hospitals differently than in primary care where also practical nurses have medication management responsibilities. Pharmacists work in many hospital wards. In 2011, there were 157 clinical pharmacists in hospital wards in Finland (Ahonen et al. 2013), but the number is increasing continuously. Work content varies a lot between wards, but most pharmacists are responsible for the logistic of medications, preparing and dispensing of drugs, and negotiation of difficulties related to medications. They also educate personnel involved in medication and counsel patients. (Ahonen et al. 2013, Ojala 2012.) Based on previous experiences (Toppinen et al. 2009), clinical pharmacists are important for increasing safety in medication processes, but they should be used even more to increase medication knowledge among health professionals and patients than has been the case thus far. From the perspective of the hospital pharmacy, the risk areas of medication process are problems in the flow of

4

information during the processing of orders, problems in preparing and delivering drugs, and deviations in guidance and advice (Ministry of Social Affairs and Health 2009b). In Finland, the use of different medications is increasingly on the rise (Finnish Medicines Agency Fimea & Social Insurance Institution 2013). Development of the medicine brings new medicines and technologies, which are even more impressive, but controlled use of them sets higher requirements (National Institute for Health and Welfare 2011). Health care is changing rapidly due to the aging population that demands more complicated and challenging care (Amalberti 2013). The need for health care services is increasing as the population is aging, but the available resources have not kept pace in proportion. There is a need to change practices and structures in the future to better reflect the current operating environments. (Ojala 2012.) Technology can provide crucial assistance. The unit dose dispensing systems and electronic documenting are already widely used in many health care organisations in Finland. (Ojala 2012.) The unit dose dispensing systems still confront various practical and systemic level challenges including costs of the service (Mäntylä et al. 2013). Other medication technology includes automated dispensing machines (ADMs), medication administration records (MARs), bar coding technology including bar code medication administration (BCMA) and bar code point of care (BPOC), and intravenous (IV) medication safety systems or smart IV pumps. Still, the significance of these technologies and the relationship between technology and adverse drug events, patient outcomes, overall costs and quality of care has not been studied in depth. (Wulff et al. 2011.) The flow of patients’ medication information between organizations and health professionals still presents many problems in Finland (Ojala 2012, National Institute for Health and Welfare 2011) as medication management requires a multidisciplinary approach and effective interdisciplinary communication (Choo et al. 2010). One solution is to implement electronic prescribing and electronic medical records. The Act on Electronic Prescriptions (61/2007) mandates the introduction of electronic prescriptions for pharmacies, healthcare units, and self-employed persons with practices in healthcare units' premises. Under the Act on the Electronic Processing of Client Data in Social and Health Care Services (159/2007), public healthcare organizations are obliged to enter patient records in a nationally centralized archive. Deployment of the centralized archive is mandatory for private healthcare organizations, if they have an electronic system for longterm storage of patient records. The aim of the Act is to enhance the data security of patient information processing, patients' access to information, and provision of healthcare services with better patient safety and efficiency. Adverse events caused by medications are a major cause of hospitalization. Many of these admissions could be prevented by different verifications and audits. Preexisting databases can already be used for identification of interactions between drugs. Still, interactions are only part of the problem in the medication management. The complexity of patients’ medications is in many cases a challenge. Thus, the multi-professional cooperation, evaluation of medications, patient-specific medication plan, use of evidencebased treatments, counseling and additional education of medication management skills plays a potentially important role. (Ojala 2012, Ministry of Social Affairs and Health 2011). 2.1.2 Nurses’ medication education and skills Nurses play a vital role in different phases of a patient's medication process in hospitals and thus need adequate competence to fulfill their role (Sulosaari et al. 2011). Implementing medications safely and with high quality requires appropriate knowledge and skills. In Finland, the education of registered nurses (Bachelor of Health Care) consists of 210 European Credit Transfer and Accumulation System (ECTS) credits and it takes about 3.5 years to study in the polytechnics (AMK) (Ministry of Education 2006). An EU-directive (2013/55/EU) concerning professional qualifications guides the content of education. In 2014, there is a nationwide project in Finland which aims to standardize the curriculum of

5

nurse education among the various polytechnics, and to develop a more homogenous national standard for nursing skills (Finnish Nurses Association 2014). The Act concerning health care professionals (559/1994) stipulates that registered nurses must have a valid qualification and must be registered by the National Supervisory Authority for Welfare and Health (Valvira) which mantains the register of health care personnel. During education, registered nurses should complete at least 9 ECTS studies of pharmacological treatments. According to the Ministry of Education (2006), registered nurses should implement prescribed medications safely and follow the impact and effectiveness of the treatment. In addition, registered nurses are held responsible managing drug calculation and clinical pharmacology, intravenous drug and fluid therapy and blood transfusions, patient counseling, and identifying the risk of medication process and using the information of medication incidents for developing medication process. The health professionals that are performing medications should understand the whole medication process as a part of patient care process. This includes understanding why, how and what kind of medications are administered. But technical know-how of medications is not sufficient to guarantee safe medication management. Implementing medications also requires jurisprudence, ethical, pharmacological, physiological, pathophysiological, and mathematical skills and knowledge. (Ministry of Social Affairs and Health 2009b.) Registered nurses' medication competence consists of theoretical, practical, and decisionmaking competence. These broad areas include the following subject areas: anatomy and physiology, pharmacology, communication, interdisciplinary collaboration, information seeking, mathematical and medication calculation, medication administration, medication education, assessment and evaluation, documentation and promotion of medication safety as part of patient safety. (Saano & Taam-Ukkonen 2013, Sulosaari et al. 2012.) Nurses are expected to possess a comprehensive medication competence to be able to conduct their duties safely and effectively. Once nursing students graduate, they are immediately expected to be able to administer medications safely (Sulosaari et al. 2011). Still, Health Care Surveillance Authorities have noted problems in the knowledge and skills of health professionals, and especially nurses’ know-how of pharmacotherapy is observed to be incomplete (Ministry of Social Affairs and Health 2009b). Nurses’ own assessments have confirmed that they consider their own pharmacology skills to be insufficient and have found pharmacology to be a difficult subject (Grandell-Niemi et al. 2005). Problems of nursing students’ and graduated nurses’ mathematical and calculation skills have also been revealed. This is an important finding, because one mistake in calculation can cause a significant medication error which might result in a life-threatning situation to patients. (McMullan et al. 2010, Grandell-Niemi et al. 2006.) These insufficient skills increase the risk of adverse outcomes in the medication process. It is important to ensure adequate medication competence to guarantee safe practice upon graduation to nursing profession. However, previous studies revealed that the amount of medication education varied between the polytechnics in Finland. Moreover, the content of teaching is quite comprehensive and consequentely the theoretical basis of medication care remain underemphasized even though the understanding of such principles is essential for safe delivery of medication care. (Sulosaari et al. 2013.) Skills and competence of medication management is ensured by a skill test that should be completed every two to five years according to guidelines of each organization (Saano & Taam-Ukkonen 2013). Additional education of the medication management is required to fulfill the need of medication competence. One example of the additional education is the Finnish eLearning material directed at all members of the staff carrying out pharmacotherapy (LOVE 2014). The eLearning material is based on Safe Pharmacotherapy, National Guide for Pharmacotherapy in Social and Health Care (Ministry of Social Affairs and Health 2009b). The learning material, which is widely used in most hospitals in Finland, provides an opportunity to maintain the knowledge and skills in pharmacotherapy and medication management.

6

2.2 MODELS FOR MEASURING SAFETY Quality of care is challenging to measure and assess because it is highly dependent on how broadly health and health responsibility are defined. In Donabedian’s conceptual model, quality of care draws from three categories: “structure,” “process,” and “outcome". Structure denotes the context in which care is delivered, process means the transactions between patients and providers during care, and outcome signifies the effects of care. (Donabedian 1997.) Safety of care, or in other words, patient safety can be seen as part of the outcome of care, and thus part of the quality of care. For example, the Oxford Dictionary (2014) defines safety as “the condition of being protected from or unlikely to cause danger, risk, or injury” and provides a synonym for safety as “harmlessness, lack of side effects”. The Institute of Medicine (IOM) defines safety as “freedom from accidental injury”, and according to the World Health Organization (WHO 2014a), the simplest way to define patient safety is “the prevention of errors and adverse effects to patients associated with health care”. When safety is defined using its opposite: lack of safety, then the ability to prevent adverse outcomes requires finding the causes of what goes wrong and safety is improved by eliminating those causes and improving barriers to prevent them. To prevent adverse outcomes and eliminate or reduce risks, it is necessary to understand why they occur. To ensure specific outcomes, prepare appropriate responses, and heighten readiness, it is important to understand what can occur and when. (Hollnagel 2011.) After measuring harm, and understanding their causes and contributing factors, it is possible to find solutions to increase the safety of care (WHO 2014b). (Figure 2.)

Figure 2. Research cycle: strengthening capacity for patient safety research (WHO 2014b)

For studying adverse outcomes of care and finding their causes, different methods and models can be used. The oldest model for describing the cause-effects relationships among accidents is the Domino model developed by Heinrich in 1929 (published in 1931). This model assumes that accidents are the culmination of a series of events or circumstances, which occur in a specific order. Therefore, accidents can be prevented by finding and eliminating their possible causes. Heinrich posed his model in terms of a single domino leading to an accident. The premise here is that human errors cause accidents. This model describes situations in a simple environment where relations are linear, meaning that one cause can describe the effect of another one. (Sabet et al. 2013.)

7

Root cause analysis (RCA) is another example of trying to understand cause-effect relationships through linear thinking. It is a method of problem solving that tries to identify the root causes of faults or problems. The type of analysis has been used since the 1950s typically as a reactive method. Root cause analysis is not a single, sharply defined methodology; there are many different tools, processes, and philosophies for performing it. Some examples of RCA techniques include such things as “Asking why 5 times”, causal tree, and decision table. A RCA should be performed as soon as possible after the error occurs. If the identified root cause can be removed that prevents the undesirable event from recurring. (Rooney & Heuvel 2004, Williams 2001.) The health care system is extremely complex. A system becomes complex when there are many combinations of events at the point of time (combinatorial complexity) or when a simple event with feedback is repeated over time (dynamic complexity). For example, the medication process is highly complex and prone to errors (Clancy et al. 2008), and work processes in the health care environment are so complicated that series of minor failures that by themselves woud do no harm inevitably combine with latent or hidden flaws (Morath & Turnbull 2005). Characteristics of a complex system include unpredictability and nonlinearity, which means that stimulus and response are unequal. (Pitkäaho 2011, Clancy et al. 2008.) In a complex system containing a large number of interacting elements, with non-linear relationships, system is constantly changing and dynamic meaning that the whole is greater than the sum of its parts. In a complex system, situations repeat accidently, and cause effect relations can be perceived only retrospectively. (Snowden & Boone 2007.) The Cynefin Framework (Figure 3) developed by Dave Snowden in 1999 provides a typology of prevailing operative contexts that are divided into five domains which explains the differences between diverse systems. When the complexity of a system increases, it is more difficult to understand it (Amagoh 2008). When a system is complex, a simple, linear causeeffect model cannot adequaly describe it. ORDER Simple

Complicated

- Cause-effect is obvious

- Repeat cause-effect

- Repeatable - Predictable - Good practice

- Best practice

DISORDER Complex

Chaotic

- Cause-effect perceived only retrospectively

- No relationship between cause-effects

- Repeat accidentally - Unpredictable

- Novel practice

- Emergent practice

UNORDER

Figure 3. A description of the Cynefin Framework (Adapted from Snowden & Boone 2007)

8

The most famous of the complex, linear cause-effect models is the Human error model developed by James Reasons (Reason 1990), which is also called the “Swiss cheese model”. Instead of blaming individuals for forgetfulness, inattention, or moral weakness (the person approach), the system approach takes into account the conditions under which individuals work and it tries to build defenses to avert errors or mitigate those effects. It is assumed that accidents are the result of a combination of active failures committed by persons, and latent conditions in the system, and that the accidents can be prevented by strengthening the barriers and defenses. Thus, in the system approach, striving for a comprehensive management of errors includes several targets for improvement: the person, the team, the task, the workplace, and the institution as a whole. (Reason 2000.) This study focuses on examining medication-related adverse outcomes and on understanding their contributing factors in the complex hospital environment. It employes the classification of error producing conditions presented by Dean et al. (2002) to describe the contributing factors. The classification is slightly modified and the contributing factors are delivered under the following categories: work environment, team, person = employee, medication-related, and patient-specific.

2.3 DEFINITIONS OF MEDICATION-RELATED ADVERSE OUTCOMES In this study, a medication-related adverse outcome refers to problems in the medication process caused by health professionals when providing care to patients. This concept includes both medication incidents (medication errors and near misses that are prevented before affecting patient) and medication-related adverse events: adverse drug events caused by medication errors or adverse drug reactions caused by the properties of the drugs administered. Inconsistency in defining medication errors has been confirmed as in the study that analysed definitions of medication errors from 45 studies and found that 26 different forms of wording were employed. Besides the word “error”, also “failure”, “deviation”, or “discrepancy” are typically employed. (Lisby et al. 2010). This study employes the commonly used definition provided by The National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP 2014) which defines a medication error as “any preventable event that may cause or lead to inappropriate medication use or patient harm while the medication is in the control of the health care professional, patient, or consumer”. Operationally, medication errors can be defined as a wrong dose, drug, delivery route, documentation, preparation, time, administration technique, administration of defunct drug, or omission of a prescribed drug. The term adverse drug event (ADE) (used as a synonym for harm) means an unintended physical injury resulting from or contributed to by medical care that requires additional monitoring, treatment or hospitalization, or that results in death (Griffin & Resar 2009). In this study, harmful situations that required patient monitoring are not included as ADEs, but will be studied separately. Monitoring is defined as need of observing or recording relevant physiological or psychological signs. Need of intervention is defined as a situation that may include change in therapy or active medical/surgical treatment. Intervention necessary to sustain life included cardiovascular and respiratory support (e.g., cardiopulmonary resuscitation, defibrillation, intubation). (NCCMERP 2001.) In this study ADEs are judged as preventable if they are the result of medication errors. This definition is commonly used in many previous studies (e.g. James 2013, Jylhä et al. 2011, Franklin et al. 2010, Singh et al. 2009, de Wet & Bowie 2009, Szekendi et al. 2006). If an ADE occurs without an error, it is classified as an adverse drug reaction (ADR) caused by the properties of the drug. (Figure 4).

9

MEDICATION-RELATED ADVERSE OUTCOMES

Medication incidents

- Near misses prevented before affecting the patient - Medication errors (ME), that cause or cause not harm

Medicationrelated adverse events

- Preventable adverse drug events (ADEs) - Caused by an medication error

- Adverse drug reactions (ADRs) - Caused by the properties of the drug

Figure 4. The relationship between medication incidents and medication-related adverse events

2.4 METHODS FOR DETECTING MEDICATION-RELATED ADVERSE OUTCOMES Medication errors and adverse events can be detected by various methods, such as analysing incident reports recorded by health professionals, using retrospective patient chart reviews such as the Global Trigger Tool (GTT) method, or using observations of clinical practice. Every method has its strengths and limitations, and they identify different problems (Meyer-Massetti et al. 2011, Franklin et al. 2009, Naessens et al. 2009). Medication error research does not have a fully adequate method (Barker et al. 2002, Ford et al. 2006), because no single method can detect all patients with treatment-related damages (Matlow et al. 2011). Still, non-voluntary detection methods have been found to be more effective in detecting events than self-report methods (Snyder & Fields 2010). Because of the limitation and strength of every research methods, previous studies suggest that a combination of different methods be used to reveal the problems in clinical practice and to understand the true effectiveness of different interventions (Franklin et al. 2009, Naessens et al. 2009, Olsen et al. 2007). In studies that have been used for studying medication errors, different variations and combinations of the methods can be found. This makes the findings partly contradictory and difficult to compare. Methodological variations between studies present a potential barrier to understand the best protocols to increase the patient’s safety (McLeod et al. 2013). Although detection and analysis of errors is important in understanding failure-prone aspects of health care and designing strategies to prevent and mitigate these failures, there is special value in quantifying actual adverse events (Griffin & Resar 2009).

10

Literature pertaining to every method used in this study for studying medication-related problems (incident reports, the Global Trigger Tool method, and observation method) was reviewed. Searches were conducted via the following databases: Cinahl, PubMed, Scopus, Cochrane, and Medic (Finnish language). Only peer-reviewed research or review articles published in English or Finnish language were accepted. Previously published literature concerning medication-related incident reports was reviewed in 2014 by using the following keywords: medication AND (error* OR incident*) AND reporting in Cinahl, PubMed, Scopus, and Cochrane databases, and (lääkeh* OR lääkity*) AND (poikkeama OR virhe) keywords in Medic database. In PubMed and Scopus databases, timeframe 2010─2014 was used for limiting the search results, while in Cochrane, Cinahl, and Medic databases timeframe 2004─2014 was used. Studies concerning only single medicines, or diseases were excluded, as well as, studies with outpatient setting. Only medication-related incident reports were included. On the bases of full text and after removing duplicates, 5 research articles were included. In addition, other relevant articles (2), found by using reference lists were included (Appendix 1.) In 2013, previous studies published between 2003 and ─2013 on adverse events detected by using the Global Trigger Tool method were reviewed using the following keywords: “trigger tool” in Cinahl, PubMed, Scopus, and Cochrane databases, and (lääkeh* OR lääkity*) AND haitta* keywords in Medic database. For this literature review, only trigger tool studies made in adult inpatients were included. Studies concerning only single medicines, or diseases were excluded, as well as, studies with outpatient setting, or concerning falls, or pressure ulcers. Studies related to development of the method, or studying the validity and reliability of the method without study results were also excluded. 21 articles were accepted on the basis of the full text. After removing 11 duplicates, and adding three other relevant articles, found by using reference lists, there was a total of 13 research and review articles. During the summer of 2014, articles published after primary literature search were reviewed again, and on that basis two more articles were added (published 2013─2014). (Appendix 2.) Literature published 2003─2013 on studies using observation method for studying medication errors in adult inpatients were reviewed using the following keywords and search terms: observation* AND (“medication errors” OR “medication incidents”) in Cinahl, PubMed, Scopus, and Cochrane databases, and (havainno* OR huomioin*) AND lääk* keywords in Medic database. For limiting search result, in Scopus database subject area of nursing was used, and in PubMed database the age limit under 18 was used. Studies using mixed-methods (where observation is one of the used methods), or using some intervention (medication card filling, computerized provider order entry, medication trolleys, e-prescribing, automated prescribing/dispensing, bar-code assisted administration) were excluded. Studies concerning interruptions in medication process were also excluded, as well as studies that only focused on specific diseases, intravenous administration, pharmacist work’s observation, enteral feeding tube, single drugs, medication communication, hospital admission/transmission, or prescribing errors. Review articles, and studies made in neonatal, pediatric, or outpatient settings were also exluded from the review. On the basis of full texts, 10 articles were accepted after removing seven duplicates. In summer 2014, a second review was conducted, but there were no new articles published to include. (Appendix 3.)

11

2.4.1 Incident reports An important factor in improving patient safety in medication processes is to collect and analyse information about errors. Most hospitals worldwide gather information concerning incidents that are either voluntarily or mandatorily reported to hospitals’ databases by health professionals. The primary purpose of reporting these errors is to learn from each experience for avoiding those in the future, and to gather data on errors to identify opportunities for improvement and to monitor progress in the prevention of errors at the organizational level (Savage et al. 2005, Leape 2002). The incident reporting has become a widely used method for studying medication errors, first and foremost, because it is quite commonly used in health care organizations for assessment of medication errors or adverse events. In addition, the data collection through incident reporting system is relatively easy and low cost (Meyer-Massetti et al. 2011), because data is already collected and available to use for analysing. As a limitation, incident reports do not usually provide an adequate assessment of clinical adverse events (Olsen et al. 2007). Reported incidents do neither represent the reality of prevalence of medication errors, because only part of medication errors are noticed, and only 10 to 20 percent of them are ever reported (Griffin & Resar 2009). The under-reporting of medication errors can compromise patient safety (Alshaikh et al. 2013, Hartnell et al. 2012), but under-reporting may be either intentional or unintentional. A health professional may fail to recognize the error, or forger to report it. (Ford et al. 2006.) Major barriers to reporting are lack of time or fear of consequences (Holmström et al. 2012, Meyer-Massetti et al. 2011, Ulanimo et al. 2007), blame culture, training and coordination of reporting (Holmström et al. 2012). Reporting would be made more frequently if reporting is made easier, and if they emloyees are adequately educated about reporting, and if they received timely feedback (Hartnell et al. 2012), and by bringing administrative support (Alshaikh et al. 2013). The frequency of reported incidents is dependent on the thoroughness with which errors are sought, the methodology that is used, patient population, and the definition of errors (Ford et al. 2006). Because only parts of errors are reported, the incident reports can not be used for studying prevalence of errors. Instead of studying the frequency or prevalence of errors, those should be used for learning of errors for avoiding them in the future. Table 1 present the results of several studies that have used medication-related incident reports for describing such things as error types. Different study settings, different definitions and classification systems, make it challenging to compare the results between studies. Still, most studies typically report wrong doses, wrong drugs and omissions, as well as medication administration errors.

12 Table 1. Examples of studies (n=5) describing stages and types of medication errors reported to hospitals incident reporting systems Study

Setting

Alshaikh et al. 2013 Saudi Arabia

Medication error reports (n=949) over a 1-year period from November 2009 to November 2010 in university teaching hospital. A total of (n=406) medication errors, 40 near misses and no adverse drug reactions were reported between 1 March 2008 and 28 February 2010 at a large, specialist UK psychiatric hospital. A total of (n=839,553) medication errors reported to MEDMARX system from 537 hospitals between 1999 and 2005. ICU** and non-ICU setting were compared.

Haw & Cahill 2013 UK

Latif et al. 2013 USA

Song et al. 2008 China

Stage of medication errors* 89.4% Prescribing 3.9% Dispensing 2.1% Transcribing 1.6% Administering

Types of medication errors* 44.4% Prescribing 31.3% Improper dose 2.6% Wrong patient 2.4% Wrong route 1.8% Omission

89% Administration 7% Prescribing 4% Dispensing

27.8% 18.2% 12.1% 10.5% dose 6.7%

ICU 44% Administering 23% Prescribing 22% Transcribing/ documenting 9% Dispensing Non-ICU 33% Administering 23% Transcribing/ documenting 22% Dispensing 21% Prescribing

A total of (n=1278) medication incident reports between January 2004December 2006 in university –affliated acute general hospital in Hong Kong. Thomas & A total of (n=12,084) 61% Administration Panchagnula incident reports from 151 26% Prescribing 2008 organisations reported to UK the UK National patient Safety Agency between 1st August 2006-28th February 2007 * only the most common types are presented in the table ** ICU = Intensive care unit

Missing signature Omission Wrong dose Unauthorized extra Wrong drug

ICU 26% Omission 24% Improper dose 9.7% Unauthorized drug 8.2% Wrong time 5.5% Extra dose Non-ICU 28% Omission 20% Improper dose 12% Unauthorized drug 7.4% Wrong time 6.1% Extra dose 36.5% Wrong strength/ dosage 16.7% Wrong drug 7.7% Wrong frequency 7.0% Wrong formulation 6.9%Wrong patient Administration incidents: Incorrect checking of drug Rate of infusion / administration Omitted drug Wrong dose Delay in starting / giving

13

2.4.2 Global Trigger Tool –method In 2003, the Institute for Healthcare Improvement (IHI) developed the Global Trigger Tool (GTT) method. The GTT method is a retrospective review of a random sample of inpatient hospital records using “triggers” to identify possible adverse events (Griffin & Resar 2009). Triggers are clues of adverse events that may have occurred during care. The triggers of adverse drug events are, for example, abnormal laboratory test, patient symptoms, or antidotes of drugs. The original tool was developed for inpatient adult patients, including different parts of hospital care. In the later development of the method, also specific medication-related trigger tool was developed for identifying ADEs (IHI 2004). Development of the method has been expanded, and specific trigger tools can be found in mental health, surgical, pediatric, perinatal, intensive care, neonatal intensive care, nursing home, and outpatient settings. Also translations for different languages and modifications for responding country-specific differences have been made. (IHI 2014.) Identifying adverse events from patients’ records in traditional ways can be timeconsuming, difficult, and expensive. The GTT method is a potential and effective alternative (Franklin et al. 2010), and it is considered to be the most accurate and efficient method to identify adverse events (Health Quality & Safety Commission 2013). GTT method has revealed adverse events even in every third or fourth of adult inpatients. Even 70% of detected adverse events have been judged to have been preventable. The vast majority of adverse events are mild and cause only temporary harm to patients, but part of them are severe and cause permanent harm, life- threatening consequences, or even patient’s death. Most of the previous studies have evaluated the severity of adverse events using widely used NCCMERP index (2001) ranging from No harm (A) to Patients death (I). (Table 2.) In addition to showing the capacity of an effective measurement of adverse events, the GTT method has proven to be a valid, reliable, well-documented, easily adopted, and promising measurement, at least if the training process has been adequate (Classen et al. 2011, Naessens et al. 2010, Adler et al. 2008, Classen et al. 2008). The previous studies have proved GTT method’s high specificity, moderate sensitivity, and favorable interrater and intrarater reliability (Sharek et al. 2011), and its’ to be easy-to-use and practical (Griffin & Classen 2008). When compared to other methods, trigger tools can examine medicationrelated problems longitudinally. The GTT method is also more efficient, less laborintensive, and reproducible than ordinary chart review method (Meyer-Massetti et al. 2011). In addition, GTT is able to identify more harmful adverse events than traditional incident reporting systems (Nilsson et al. 2012, Classen et al. 2011, Naessens et al. 2010) or even detect events which would have gone unnoticed by other methods (Doupi 2012). Still, there are discrepancies between studies and some concerns have been raised concerning the usefulness of the resulted data using the GTT method, including concerns about its reliability (Health Quality & Safety Commission 2013), and non-specificity (Good et al. 2011) and the sensitivity of the results, as well as efficiency of the method (Franklin et al. 2010). Some have warned to avoid comparisons between hospitals based on the GTT method. At the very least, such comparisons should not be done without substantial training to achieve better agreement between reviewer teams. (von Plessen et al. 2012, Schilmeijer et al. 2012). Some studies have raised concerns of its time-consuming and labour intensive nature. Data warehouse, automated processes, or computer based screening can be more effective when compared to traditional paper-based screening. (O’Leary et al. 2013, Naessens et al. 2010.) Still, previous studies have found relatively poor agreement between traditional trigger tool method and data warehouse based on screening, and a combination of these has been recommended (O’Leary et al. 2013).

14 Table 2. Prevalence, preventability and severity of adverse drug events (ADEs) and adverse events (AEs) detected by trigger tool method Study

Prevalence of ADEs /AEs

de Boer et al. 2013 The Netherlands

91 ADEs in 76 patients, 262 patients reviewed, 29% of randomly selected inpatients had at least one ADE 62 ADEs in 240 patients, 26% (26 ADEs per 100 admissions)

Carnevali et al. 2013 Belgium Franklin et al. 2010 UK

Preventability of ADEs /AEs 7.6% were preventable

Severity of ADEs / AEs 93.4% mild/moderate 6.6% severe or lifethreatening

35% were preventable

68% level E (*) 27% level F 3% level G 2% level H 0% level I

7 ADEs in 207 patients, 3.4% of randomly selected inpatients had at least one ADE 138 AEs in 125 patient, 854 patients reviewed, 14.6% of randomly selected inpatients had at least one AE 2772 AEs in 16 172 patients, 17.1% of randomly selected inpatients had at least one AE 118 ADEs were detected which occurred in 62 patients.

28.6% were preventable

Landigran et al. 2010 USA

588 AEs were found among 2341 patients, 25.1% of randomly selected inpatients had at least one AE

63.1% were preventable

Nilsson et al. 2012 Sweden

41 AEs were found in 25 patients, 128 record reviewed, 19.5% of randomly selected inpatients had at least one AE 54 AEs in 250 patients, 21.6% of randomly selected inpatients had at least one AE 42 AEs in 250 patients, 20% of randomly selected inpatients had at least one AE

54% were preventable

Griffin & Classen 2008 USA Kennerly et al. 2013 USA Klopotowska et al. 2013 The Netherlands

O’Leary et al. 2013 USA Schildmeijer et al. 2012 Sweden Seynaeve et al. 2011 Belgium Szekendi et al. 2006 USA

230 ADEs occurred in 79 patients

47% level E (*) 44% level F 8.7% level G,H,I

70.3% were preventable

26.5% mild 41.0% moderate 41.0% severe 18.1% life-threatening 2.4% fatal 41.7% level E (*) 42.7% level F 2.9% level G 8.5% level H 2.4% level I

11.1% were preventable

51.9% of AEs were serious

58% were preventable

45% level E (*) 48% level F 1% level G 5% level I 96% level E (*) 4% level F

At least one AE or medical 64% were preventable 85% level E or F (*) error was identified in 243 4% level G, (74%) of reviewed patient 10% level H records 1% level I von Plessen et al. In total, 687 AEs were 96% level E or F (*) 2012 identified in 11487 patient Denmark days. The percentage of harmed patients was 25%. (*) NCCMERP level of harm: E Temporary harm to the patient and required intervention F Temporary harm to the patient and required initial or prolonged hospitalization G Permanent patient harm H Intervention necessary to sustain life I Patient death

15

2.4.3 Observation method Only a portion of the errors in clinical practice are noticed and reported (Griffin & Resar 2009). Thus, observation method, which affords a view of the reality of clinical practice, can reveal additional and undetected information concerning problems in the medication process. Observation as a method of detecting medication errors had been used since the 1960s, and had been found to be an effective method (Barker et al. 2002). When direct observations, chart reviews and incident reports for studying medication errors were compared, the observation method was the most effective for detecting errors (Haw et al. 2007). In previous studies using the direct observational approach in adult inpatients, methods including classifications of incidence, types and severity of detected errors have varied. The hospital settings used have also been varied, and studies have been conducted in different countries. All the aforementioned facts make it even more challenging to compare the results of previous studies. By comparing a variety of study results (Table 3), usually observed errors do not cause harm to patients. Still, employing the worst case scenario, and assuming the error potential the cause harm, the possible severity rate of detected errors can be seen to be even higher. There is evidence that medication errors do occur in half of all opportunities of errors when including wrong time errors and calculating the lack of identity control of patients as errors (Lisby et al. 2005). Still, a great variability of the incidence of errors can be found. In most cases, the wrong time of medication administration or wrong administration technique have been the most common error types (e.g. Berdot et al. 2012, Vazin & Delfani 2012, Kelly & Wright 2012, Maricle et al. 2007), and also inappropriate aseptic technique has been common (Gokhman et al. 2012). These are typically cases that are easy to observe but maybe because of low clinical impact, are rarely reported through reporting systems in hospitals. This can be seen as potential strength of the method. Other strengths of the observation method are that knowledge of errors by subjects, willingness to report, remembering, or the ability to communicate are not required. The ways to increase validity of the method is to use experienced observers and training them to unobstrusive, objective, and nonjudgemental. (Barker et al. 2002.) The weaknesses of the observation method are its time-consuming and resource intensive nature (Meyer-Massetti et al. 2011). Thus, the size of the data and generalization of the findings can be limited. Furthermore, structured observation method offers only information of the observed moment, and many correlating factors to errors may remain understudied without understanding the examinee’s views of the situation and the factors that may affect to errors (Meyer-Massetti et al. 2011). Possible threats to the validity of observational data are the Hawthorne effect and observer interference (Keers et al. 2013). Still, previous studies concerning validity and reliability of the observational method for detecting medication errors proved that observation of nurses during drug administrations did not affect the error rate. Thus concerns over the reliability and validity of the observation method may be unfounded. (Dean & Barber 2001.)

16 Table 3. Incidence, types, and severity of medication errors detected by using observation method in adult inpatients Study, hospital setting Berdot et al. 2012, France Immunology-cardiology, nephrology, vascular medical, and cardiovascular surgical wards / teaching hospital Chua et al. 2009, Malaysia Hematology ward / teaching hospital

Incidence of errors 27.6% of administrations

Types of errors 72.6% wrong time errors, 14% omissions, 3.7% unauthorized drug errors

Severity of errors 94% of errors had no clinical impact, 6% had a serious or significant impact on patients

11.4% of all opportunities for errors

25.2% incorrect time 16.3% incorrect adm. technique 14.1% unauthor. drug

Gokhman et al. 2012, USA Medical emergency team care / tertiary care, academic medical center Haw et al. 2007, UK Two elderly long-stay wards / psychiatric hospital

296 errors in 186 administered doses

66% inappropriate aseptic technique

10.4% of the administration errors were considered as potentially lifethreatening 14% of errors the potential for harm existed

25.9% of all opportunities for errors

Most errors were of doubtful or minor severity

Kelly & Wright 2012, UK One stroke and one care of older people ward in four hospitals Kopp et al. 2006, USA medical/surgical ICU / tertiary care academic medical center

38.4% of all opportunities for errors

30.1% unauthorized tablet crushing or capsule opening, 27.1% omission, 23.6% failure to record administration 72.1% wrong time 8.0% wrong preparation 6.5% other 4.9% omission

One error for every five doses of medication administered.

23% omissions 20% wrong dose 16% wrong drug

Lisby et al. 2005, Denmark medical and surgical department / University Hospital

43% of all opportunities for errors

Maricle et al. 2007, USA Medical ICU, medical cardiac telemetry, acute psychiatric unit / tertiary care hospital Patanwala et al. 2010, USA Academic, tertiary ED

5% of administrations

Administration errors: 150/166 lack of identity control 18/166 wrong time 12/166 wrong delivery 34% wrong technique 32% wrong time 19.5% omission

Of the potential ADEs, 61.8% were significant, 29.1% serious, and rest life-threatening/fatal. Of the actual ADEs, 42.9% were significant, 57.1% serious, and none lifethreatening or fatal In worst case scenario 20– 30% of medication errors were assessed as potential adverse drug events.

1 medication error for every 5 medication orders written and 4 doses administered.

Not classified

Vazin & Delfani 2012, Iran Internal ICU / university hospital

7.6% of all opportunities for errors

Administration errors: 33.3% wrong technique 16.3% wrong time 16.3% wrong preparation

The overall mean harm score of the 35 incidents analysed was 4.1 on a scale of 0–10.

Not judged

No errors resulted in permanent harm to the patient or contributed to initial or pro-longed hospitalization. 94% of errors caused no harm or required only patient monitoring

17

2.5 CONTRIBUTING OUTCOMES

FACTORS

TO

MEDICATION-RELATED

ADVERSE

Studies published from 2003 to ─2013 of contributing factors to medication errors in adult inpatients were reviewed via Cinahl, PubMed, Scopus, Cochrane, and Medic databases in 2013 using the following keywords and search terms: (“medication error” OR “medication incident”) AND (risk* OR “contributing factor” OR antecedent* OR cause* OR determinant*) NOT (child* OR pediatric* OR paediatric*) in Cinahl, PubMed, Scopus, and Cochrane databases, and keywords/search terms (lääkeh* OR lääkity*) AND (virhe* OR poikkea*) in Medic database. Non-hospital settings, case studies, and studies concerning enteral feeding tubes, epidural vein cannulation, single medicines, or single diseases were excluded, as well as studies focusing only patient, physician, pharmacist, care-giver, or student perspectives of medication errors. In Scopus database, search was limited only to nursing because of large amount of search result. On the basis of full text, 25 articles were accepted after removing five duplications, and adding 6 articles found by manual search. In summer 2014, articles published after primary literature review searches were reviewed again, but not any new articles was included. (Appendix 4.) In studies (n=25) that were included to this literature review, mostly incident reports (n=8) were used, but also observation method (n=4), questionnaires (n=4), interviews (n=2), mixed methods (n=2), record reviews (n=2), and literature reviews (n=3). The factors contributing to medication errors are multisided and in this study those include the following factors (modified using the classification of error producing conditions presented by Dean et al. 2002): work environment, team, person (employee), patient-specific, and medication-related. (Table 4.) Work environmental factors can contribute to medication errors. Those factors include distractions (e.g. Unver et al. 2012, Pham et al. 2011, Roughead & Semple 2009, Mrayyan et al. 2007), inadequacy of staff or resources or performance deficit (e.g. Alshaikh et al. 2013, Latif et al. 2013, Lawton et al. 2012). In addition, problems in job security (Wilkins & Shields 2008), in ward climate (Lawton et al. 2012), or lack of existing protocols or guidelines (Ruuhilehto et al. 2011, Nichols et al. 2008) have found contributing to medication errors. Of the team factors, low support (Wilkins & Shields 2008), lack of guidance (Nichols et al. 2008), problems in communication (e.g. Pham et al. 2011, Ruuhilehto et al. 2011, Roughead & Semple 2009, Nichols et al. 2008) contributed to errors. Problems in flow of information because of handwriting, used abbreviations, and problems in transcription (e.g. Alshaikh et al. 2013, Latif et al. 2013, Pham et al. 2011, Hewitt 2010, Hicks et al. 2004, Mayo & Duncan 2004) have contributed to medication-related problems as well. According to previous studies, person (employee) factors contributing to medication errors are lack of experience (e.g. Pham et al. 2011, Westbrook et al. 2011, Tang et al. 2007), lack of knowledge or education (e.g. Alshaikh et al. 2013, Kane-Gill et al. 2010, Chang & Mark 2009, Kopp et al. 2006), slips in attention or memory lapses (e.g. Nichols et al. 2008, Kopp et al. 2006), or rule violations, failure to follow protocol (e.g. Latif et al. 2013, Pham et al. 2011, Hewitt 2010). Additional factors contributing to errors include: fatigue or working overtime, role overload, stress, heavy workload (e.g. Lawton et al. 2012, Unver et al. 2012, Hewitt 2010, Wilkins & Shields 2008), number of patient under the nurse’s care (Berdot et al. 2012), or personal neglect (Tang et al. 2007). Some patient-specific factors have been associated with medication errors, including amount of medicines patients used (Ben-Yehuda et al. 2011, Patanwala et al. 2010), complex issues of patients/multi-morbidity (e.g. Ruuhilehto et al. 2011, Seynaeve et al. 2011, Nichols et al. 2008), as well as length of hospital stay (Ben-Yehuda et al. 2011). In those cases where patient was unfamiliar, or the identification of patient was not done (Unver et al. 2012, Westbrook et al. 2011, Nichols et al. 2008), or patient condition was not paid attention (Winterstein et al. 2004) have found to be connected to increased risk as well. Of the medication-related factors, intravenous administration (Westbrook et al. 2011), injections (Berdot et al. 2012), and unfamiliar medication (Nichols et al. 2008) also

18

contribute to medication errors. In addition, problems in drug identification, confusion between 2 drugs with similar names or packaging, miscalculation, and problems related to medication labels/packaging contribute to medication errors (Unver et al. 2012, Hewitt 2010, Armitage et al. 2007, Mrayyan et al. 2007, Kopp et al. 2006, Mayo & Duncan 2004). Table 4. Contributing factors of medication errors in previous studies (n=25) (classification of error producing conditions presented by Dean et al. 2002) Contributing factors of medication errors Work environment factors: - Distractions during work (by other co-workers, patients, or events in the unit) - Perceived staffing or resource inadequacy, performance deficit - Low job security - Ward climate - Lack of protocols / guidelines - Work shift (night) - Extra-long shifts - Drug distribution system Team factors: - Low co-workers support - Lack of guidance - Poor communication - Handwriting difficult to read, abbreviations, transcription inaccurate / omitted Person / employee related factors: - Lack of experience, new staff - Lack of (drug) knowledge, education, knowledge deficit - Slips (in attention) and memory lapses - Rule violations, failure to follow protocol - Fatigue, working overtime - Role overload, stress, heavy workload, number of patient under the nurse’s care. - Personal neglect Patient-specific factors: - Amount of medicines - Complex issues, amount of diseases - Unfamiliar patient, Lack of patient identification - Patient conditions not considered - Length of hospital stay Medication-related factors: - Intravenous administration, - Injections - Error in drug identification, confusion between 2 drugs with similar names or packaging, miscalculation - Unfamiliar medication - Medication labels / packaging are of poor quality / damaged

Supporting literature

- Alshaikh et al. 2013, Unver et al. 2012, Pham et al. 2011, Hewitt 2010, Brady et al. 2009, Roughead & Semple 2009, Nichols et al. 2008, Mrayyan et al. 2007, Mayo & Duncan 2004 - Alshaikh et al. 2013, Latif et al. 2013, Lawton et al. 2012, Ruuhilehto et al. 2011, Kane-Gill et al. 2010, Wilkins & Shields 2008, Hicks et al. 2004 - Wilkins & Shields 2008 - Lawton et al. 2012 - Ruuhilehto et al. 2011, Nichols et al. 2008 - Nichols et al. 2008 - Hewitt 2010, Nichols et al. 2008 - Brady et al. 2009 - Wilkins & Shields 2008 - Nichols et al. 2008 - Pham et al. 2011, Ruuhilehto et al. 2011, Roughead & Semple 2009, Nichols et al. 2008, Armitage et al. 2007 - Alshaikh et al. 2013, Latif et al. 2013, Pham et al. 2011, Hewitt 2010, Hicks et al. 2004, Mayo & Duncan 2004 - Alshaikh et al. 2013, Pham et al. 2011, Westbrook et al. 2011, Tang et al. 2007 - Alshaikh et al. 2013, Kane-Gill et al. 2010, Brady et al. 2009, Chang & Mark 2009, Tang et al. 2007, Kopp et al. 2006, Winterstein et al. 2004 - Nichols et al. 2008, Kopp et al. 2006, Winterstein et al. 2004 - Latif et al. 2013, Pham et al. 2011, Hewitt 2010, Kane-Gill et al. 2010, Nichols et al. 2008, Armitage et al. 2007, Kopp et al. 2006, Hicks et al. 2004 - Unver et al. 2012, Hewitt 2010, Roughead & Semple 2009, Nichols et al. 2008, Wilkins & Shields 2008, Mayo & Duncan 2004 - Alshaikh et al. 2013, Lawton et al. 2012, Pham et al. 2011, Seynaeve et al. 2011, Brady et al. 2009, Roughead & Semple 2009, Nichols et al. 2008, Wilkins & Shields 2008, Tang et al. 2007, Berdot et al. 2012 - Tang et al. 2007 - Ben-Yehuda et al. 2011, Patanwala et al. 2010 - Ben-Yehuda et al. 2011, Ruuhilehto et al. 2011, Seynaeve et al. 2011, Nichols et al. 2008 - Unver et al. 2012, Westbrook et al. 2011, Nichols et al. 2008 - Winterstein et al. 2004 - Ben-Yehuda et al. 2011 - Westbrook et al. 2011 - Berdot et al. 2012 - Unver et al. 2012, Hewitt 2010, Kopp et al. 2006, Mayo & Duncan 2004 - Nichols et al. 2008 - Armitage et al. 2007, Mrayyan et al. 2007

19

2.6 SUMMARY OF THE STUDY BACKGROUND Medication-related problems including medication incidents and adverse events can be studied using different research methods, and there exists a comprehensive literature devoted to the general topic of medication safety problems. Previous studies have revealed that problems in the medication process are common, but only a portion of them cause serious consequences to patients. The factors contributing to medication-related adverse outcomes are multisided in the complex medication process and include the following factors (using the classification of error producing conditions presented by Dean et al. 2002): work environment, team, person (employee), patient-specific, and medicationrelated. (Figure 5.) Because of limitations of each method, previous studies using single method have provided limited information concerning problems in the medication process and their contributing factors. In addition, comparing the results between studies is challenging because of differing definitions, study designs and settings. Furthermore, there are a limited number of studies of medication safety focusing on nursing or conducted in Finland, and none using the combination of incident reports, GTT and observation methods. Thus, more detailed information on medication-related problems is required to achieve a deeper understanding and to identify areas of development for increasing safety in the medication processes in hospitals.

Work environment - Distractions - Resource inadequacy - Low job security - Lack of guidelines - Ward climate - Poor design of the work

Team - Lack of guidance, communication, support

Employee - Lack of skills, education - Slips and lapses -Workload, stress, fatigue - Neglect, rule violations

Patient - Care intensity, complexity and length - Unidentified / unmonitored patient

Medication - Intravenous administration, injections - Confusion with the medications

CONTRIBUTING TO

Medicationrelated

Medication incidents

adverse events

- Near misses prevented before affecting the patient - Medication errors (ME), that cause or cause not harm

- Preventable adverse drug events (ADEs) - Caused by an medication error

Figure 5. Summary of the study framework

- Adverse drug reactions (ADRs) - Caused by the properties of the drug

20

3 Purpose, Aims and Objectives The purpose of this study is to present a more comprehensive and valid estimate of the problems that arise in the medication process in hospitals by gaining information of medication-related adverse outcomes and contributing factors. The study also outlines important methodological information concerning the differences between detection methods by comparing the information on medication-related adverse outcomes. Specifically the study aims to examine medication-related adverse outcomes and contributing factors of hospital patients detected using incident reports, Global Trigger Tool method and observation methods, to study the associations between adverse outcomes and contributing factors, and to compare differences between the detection methods. The specific research objectives are: 1) To examine the incidence, types, severity and contributing factors of medicationrelated adverse outcomes among hospital patients (Article I, Article II, Article III, Article IV, summary) 2) To compare the types, severity and medication-related factors of medication errors detected using three detection methods (Article IV, summary) 3) To study the associations between medication-related adverse outcomes and contributing factors (Article II, Article III, summary) The findings can be utilizes in clinical practice, at the leadership and policy levels, and for educational purposes to better understand medication-related adverse outcomes and their contributing factors, as well as for purposes of continuing research which aims to improve medication safety. The experiences included in the study describe the limitations and strengths of the current detection methods which can be used for developing research methodology both in the field of safety research and health research including nursing science.

21

4 Materials and Methods This study consist of four sub-studies (Article I-IV), including different samples, timeframe, setting, designs, data collection and analysis methods. (Table 5.) Table 5. Study sample, time, setting, design, data collection, and analysis in four sub-studies (Articles I-IV) Sub-

Sample, time & setting

Design

study/

Data

Data analysis

collection

Article I

Medication-related incident

Retrospective

Register study

reports (n=671) during year

Descriptive statistics, cross-tabulation

2010, one university hospital, organisation level, Finland II

III

Randomly selected patient

Global Trigger

Descriptive statistics,

records (n=463) during year

Retrospective

Tool (GTT)

logistic regression

2011, one university hospital,

method,

analysis, Mann-Whitney

organisation level excluding

register study

U-test, Student’s t-test,

psychiatric, rehabilisation and

Pearson’s chi-squared

pediactric patients, Finland

test

Medication administration

Structured,

Descriptive statistics,

observations (n=1058) with

Prospective

direct

logistic regression

record reviews (n=122) during

observation

analysis, Cohen’s kappa

April to May 2012, in one

method and

university hopitals’ four surgical

record review

Secondary

coefficient of Agreement, MannWhitney U-test, Pearson chi-squared test, Fisher’s exact test Descriptive statistics,

analysis

Pearson’s chi-squared

and medical wards, Finland

IV

Medication administration

Retrospective

errors (n=451) detected using incident reports, GTT and

test

observation methods in one university hospital, Finland

4.1 STUDY SETTING AND DESIGN The study was conducted in one university hospital in Finland, which has 800 beds and provides specialised medical care to 860,000 inhabitants. About 90,000 inpatients are treated annually in the hospital. Incident reports from the hospital during 1st of January to 31th of December in 2010 were analysed retrospectively. Randomly selected patient records of hospitals wards during 1st of January to 31th of December in 2011 were analysed retrospectively using Global Trigger Tool method. Direct observations were conducted in hospital’s four medical and surgical wards (cardiology, internal medicine and nephrology, gastroenterology, traumatology) during April to May 2012. Finally, all medication administration errors that were detected during analysing incident reports, using Global Trigger Tool method, or observations were re-analysed retrospectively.

22

4.2 SAMPLE, DATA COLLECTION AND ANALYSIS 4.2.1 Medication-related incident reports The data was collected from a hospital’s web-based error-reporting database (HaiPro). All health care professionals who were involved in patient care were able to report incidents on this database. The reporting process is easy and fast, and it can be done voluntarily and anonymously on any workstation computer in hospital wards. The database collects information on incidents with web-based forms (Appendix 5). There was a total of 1617 incident reports in the database in 2010, of which (n=671) were medication-related and collected for this study. Incident reporting forms included recorded descriptions of the incidents describing what happened, the cause of the incidents, the consequences of the incidents for the patient, and the circumstances and other possible factors contributing to the incidents. In addition, reporters suggested the ways in which the incidents could be prevented in the future. The data was collected in collaboration with the researcher (MH) and the hospital’s head administrator of the HaiPro programme. During data collection, duplicates of the reports were removed. Those incident reports retained for the study were printed for analysis. After reviewing the incident reports (n=671) the classification of incidents according to their nature including severity and type were classified. The qualitative material was quantified and classified into different categories. The incidences that could not be classified reliably were classified as “not known”. Two experts analysed the material in collaboration (the researcher=MH and hospital’s pharmacist/head administrator of the HaiPro programme). The classifying of errors was jointly agreed. When differences arose as to how best to classify the errors, a consensus was reached. The data was processed statistically using SPSS statistics 19.0 for Windows. Cross-tabulation (Metsämuuronen 2009) was used when analysing the connection between how errors were detected and the phase of the medication process (stage). (Article I.) 4.2.2 Patients’ records and the Global Trigger Tool A random sample of 20 patients’ records was collected retrospectively every two weeks over the course of a year in 2011 to review a total of 480 patients’ records. The treatment period analysed included any field of specialization such as medical, surgical, cardiology, neurology, gynecology, obstetric, oncology, dermatology, pulmonary, ophthalmology, and otorhinolaryngology. In any given analysed treatment period, a patient could receive care in many fields of specialisation in addtition to being treated in the emergency department, intensive care unit, or operating room. The inclusion criteria were: patient’s length of stay (at least 24 hours) and patient’s age (18 years or older). Patients that were treated in psychiatric and rehabilitation wards were excluded. Specific electronic records used in the intensive care unit, emergency department, and operating room were excluded along with non-electronic paper-based lists (e.g. blood pressure or blood glucose). The university hospital’s information technology (IT) service provided randomised data sampling. Despite precautions taken in the selection of the randomised data sample using inclusion and exclusion criteria, 17 patients’ records that included psychiatric treatment periods entered the selection and were subsequently excluded because they did not meet the criteria. The total number of patients’ records collected for analysis was 463. Data was collected using two primary record reviewers with nursing backgrounds (researcher=MH and research assistant) who reviewed all patient records (n=463) independently to identify positive triggers. The trigger list used in this study (Appendix 6) was modified using GTT medication module triggers (Griffin & Resar 2009) and the trigger tool for measuring adverse drug events (IHI 2004) lists. A multi-professional research group, including a clinical physician, a clinical pharmacist, a professor of nursing science,

23

and a researcher (MH) modified the list. Some laboratory tests’ reference values were modified for the Finnish hospital population. Some triggers remained as contained in the original trigger tool lists, others were added, and some others (use of anti-diarrheal drugs, use of sodium polystyrene for hyperkalemia or rash, or transfer to a higher level of care) were removed from the original trigger lists. The final trigger list for data collection consisted of 22 -items (Table 6). Table 6. The modified trigger list containing 22 -items and information of modified and original reference values / triggers Triggers M1 Clostridium difficile stool M2 Partial Thromboplastin Time (P-APTT) M3 AntiFxa M4 International Normalised Ratio (INR) M5 Blood glucose M6 Digoxin M7 Leucocytes / white blood cell M8 Thrombocytes / platelet M9 Creatinine or Blood urea nitrogen (BUN) M10 Potassium

Modified reference values/ triggers Positive stool >100 seconds

Original GTT* reference values / triggers Positive stool >100 seconds

>10 U/ml

Does not exist in the original tools* >6

>5 1.5 nmol/l 15 mmol/l 4.7 mmol/l

2 ng/ml Health systems > Patient safety. Retrieved from: http://www.euro.who.int/en/health-topics/Health-systems/patient-safety WHO. World Health Organization. (2014b). Patient safety > Patient Safety Research > Strengthening capacity for patient safety research. Retrieved from: http://www.who.int/patientsafety/research/strengthening_capacity/measuring_harm/en/ Wilkins K & Shields M. (2008) Correlates of medication error in hospitals. Health reports 19 (2), 1-12.

63

Williams P. (2001). Techniques for root cause analysis. Baylor University Medical Center Proceedings 14(2), 154–157. Winterstein AG, Johns TE, Rosenberg EI, Hatton RC, Gonzalez-Rothi R & Kanjanarat P. (2004). Nature and causes of clinically significant medication errors in a tertiary care hospital. American Journal Health-System Pharmacy 61(18), 1908-1916. WMA. World Medical Association. (2013). Declaration of Helsinki. Ethical Principles for Medical Research Involving Human Subjects. JAMA 310(20), 2191-2194. Wulff K, Cummings GG, Marck P & Yurtseven O. (2011). Medication administration technologies and patient safety: a mixed-method systematic review. Journal of Advanced Nursing 67(10), 2080-95.

Appendix 1. Articles using medication-related incident reports (n=7) searched in 2014

Search terms: medication AND (error* OR incident*) AND reporting

In Finnish: (lääkeh* OR lääkity*) AND (poikkeama OR virhe)

Limitations: years 2010-2014, Cinahl and Medic 2004-2014, English or Finnish language CINAHL n=110

PUBMED n=106

SCOPUS n=78

COCHRANE n=125

MEDIC

n=30

Accepted on the basis of the tittle or abstract CINAHL n=18

PUBMED n=5

SCOPUS n=15

COCHRANE n=0

MEDIC

n=7

Accepted on the basis of the full text

CINAHL

n=3

PUBMED n=0

SCOPUS

n=4

COCHRANE n=0

MEDIC

n=0

Duplications (n=2) removed, other relevant articles (n=2) added

A total of n=7 research articles

Appendix 2. Articles using Trigger tool method (n=13) searched in 2013, and additional search in 2014, (n=2) articles

In Finnish: (lääkeh* OR lääkity*) AND haitta*

Search terms: “trigger tool”

Limitations: years 2003-2013, English or Finnish language

CINAHL n=38

PUBMED n=64

SCOPUS n=84

COCHRANE n=2

MEDIC n=69

Accepted on the basis of the tittle or abstract CINAHL

n=21

PUBMED n=35

SCOPUS n=28

COCHRANE n=1

MEDIC

n=0

Accepted on the basis of the full text

CINAHL

n=3

PUBMED n=8

SCOPUS n=10

COCHRANE n=0

MEDIC

n=0

Duplications (n=11) removed, other relevant articles (n=3) added

A total of n=13 research articles 2003-2013 Additional search in 2014, n=2 articles

Appendix 3. Articles using observation method for studying medication errors in adult inpatients (n=10) searched in 2013, and additional search in 2014 (n=0) articles

Search terms: Observation* AND (“medication errors” OR “medication incidents”)

In Finnish: (havainno* OR huomioin*) AND lääk*

Limitations: years 2003-2013, English or Finnish language, in Scopus database subject area only nursing, in PubMed database only adults CINAHL n=173

PUBMED n=124

SCOPUS n=100

COCHRANE n=14

MEDIC

n=18

Accepted on the basis of the tittle or abstract CINAHL n=10

PUBMED n=18

SCOPUS n=15

COCHRANE n=0

MEDIC

n=0

Accepted on the basis of the full text

CINAHL

n=5

PUBMED

SCOPUS

n=6

n=5

COCHRANE n=0

MEDIC

Duplications (n=6) removed, other relevant articles (n=0) added

A total of n=10 research articles 2003-2013

n=0

Appendix 4. Articles studying contributing factors to medication errors in adult inpatients (n=25) searched in 2013, and additional search in 2014 (n=0) articles

Search terms: (“medication error” OR “medication incident”) AND (risk* OR “contributing factor” OR antecedent* OR cause* OR determinant*) NOT (child* OR pediatric* OR paediatric*)

In Finnish: (lääkeh* OR lääkity*) AND (virhe* OR poikkea*)

Limitations: years 2003-2013, English or Finnish language, in Scopus database subject area only nursing CINAHL n=104

PUBMED n=111

SCOPUS n=118

COCHRANE n=29

MEDIC n=75

Accepted on the basis of the tittle or abstract CINAHL n=24

PUBMED n=37

SCOPUS n=19

COCHRANE n=3

MEDIC n=6

Accepted on the basis of the full text

CINAHL n=6

PUBMED n=11

SCOPUS n=6

COCHRANE n=0

MEDIC n=1

Duplications (n=5) removed, other relevant articles (n=6) added

A total of n=25 research and review articles 2003-2013

Appendix 5. HaiPro incident reporting form

Appendix 6. Data collection form for Global Trigger Tool –study

GTT data collection form Form (nro): Date of discharge: Patient (code): The number of drugs used: Diagnoses: Reviewers Triggers 1.

2.

(Translated 8/2014) Ward: Length of hospital stay (days): Age: Gender: Type of living: Adverse event: (a short description)

Severity of ADE: (NCCMERP E-I)

Stage of an error: (1-5)

Type of an error: (1- 12)

MI – Clostridium difficile stool M2 – P-APTT >100 s M3 - AntiFxa >10 U/ml M4- INR > 5 M5 – Blood glucose < 3,5 mmol/l M6 – Digoxin > 1,5 nmol/l M7- Leucocytes < 3 E9/l M8- Trombocytes < 80 E9/l M9- (BUN) urea > 15 mmol/l creatinine > 120 µmol/l M10- Potassium < 3,4 tai > 4,7 mmol/l M11 - K-vitamin M12 – Antihistamines M13 - Flumazenil M14 - Naloxone M15 – Bridion (Sugammadex) M16 – Antiemetic M17 - Oversedation/sleepiness M18 – Hypotension, dizziness, fall M19- Bradycardia < 45/min M20- Prolonged QT time M21 - Abrupt medication stop M22 - Other

A total number of detected triggers:

A total number of detected adverse events:

A detailed description of the detected events:

Occurred during the hospital stay:

Occurred before admission:

NCCMERP classification of the severity: A: Circumstances or events that have the capacity to cause error B: An error did not reach the patient C:An error reached the patient, but did not cause patient harm D: An error reached the patient and required monitoring to confirm that it resulted in no harm to the patient E: An error may have contributed / resulted in temporary harm and required intervention F: An error may have contributed / resulted in temporary harm and required initial or prolonged hospitalisation G: An error may have contributed to or resulted in permanent patient harm H: An error that required intervention necessary to sustain life I: An error that may have contributed to or resulted in the patient’s death

CLASSIFICATION OF THE ERRORS Stage of medication process: 1. Prescription error 3. Documentation error 5. Preparation error

2. Administration error 4. Dispensing error

Type of error:: 1. Wrong dosege 2. Wrong drug 5. Wrong technique 6. Wrong drug from administration route 9. Discontinued drug 11. Error in preparing the drug

3. Omission of drug 7. Wrong patient 10. Error in documenting 12. Other

4. Wrong timing 8. Wrong

Appendix 7. Structured observation form for direct observations

Medication observation form

(Translated 8/2014)

Observation number:

Ward (code):

Date:

Time of day:

PATIENT (the information to be acquired by observing the patient) Patient (code):

Did patient ask about the drug dosage? Y / N

Did patient's condition allow asking about it? Y / N

Did the patient notice an error? Y / N

DRUG / INFUSION (based on observation and a later verification from patient records) Drug:

Dose given:

Drug form:

Route of administration:

Solvent used and its volume (intravenous drugs): NURSE (based on observations and information to be acquired from the nurse) Nurse (code):

Has received orientation to administer medication at the ward: Y / N

Work experience at the ward (years):

In the nursing profession (years):

Number of days before the previous day-off from work:

Nurse seems tired: Y / N

Nurse seems sick: Y / N

Nurse seems nervous: Y / N

WORK ENVIRONMENTAL FACTORS (information to be acquired at the ward) Number of patients under observed nurse’s care:

Number of bed-ridden patients:

Medicine cabinet clean: Y / N

Separate medicine room: Y / N

Limited access to medicine room: Y / N

Spacious: Y / N

Well-lit: Y / N

Distractions and interruptions during observations: Large number of people in the medication room: Y / N Busy atmosphere: Y / N

Noise: Y / N

Answering the phone: Y / N

Answering patient calls: Y / N

Patient interrupts: Y / N

Personnel interrupts: Y / N

Other distractions or interruptions: Y / N

, what kind?

TEAM FACTORS (based on observations) Did the nurse ask for help from other nurses/pharmacists: Y / N Was the medication dose double-checked? Y / N

, how?

Patient record review after observations: Was there a prescription for this drug? Y / N Was the prescription same as the prepared / administered drug? Y / N Was the administration route correct? Y / N Was the drug form correct? Y / N Was the drug administration timed correctly? Y / N Were the drug and dose the same in the medication card as in the prescription? Y / N Were the administered/prepared drug and dose the same as in the medication card? Y / N

PATIENT (the information to be collected from patient records) Age:

Care intensity classification:

Gender:

Current living arrangement:

Regularly taken drugs/day:

Drugs taken when necessary/day:

The number of times oral drugs are administered/day:

The number of times other drugs are administered/day:

Intravenous drugs/day: The number of times intravenous drugs are administered/day:

PREPARING THE DRUG (if observing preparation process) Did the nurse use gloves? Y / N

Did the nurse clean the table? Y / N

Was the waste properly disposed of? Y / N

Was the label properly entered? Y / N

Did the nurse use medication guidelines to prepare the product? Y / N (if necessary, check the medication guidelines after observation): Were tablets crushed? Y / N

License to crush tablets? Y / N

Were tablets dissolved? Y / N

License to dissolve tablets? Y / N

Were capsules opened? Y / N

License to open capsules? Y / N

Were solvent products and their amounts correct (i.v)? Y / N

ADMINISTERING THE DRUG The rate / duration of administering intravenous drug: Was the patient identified before drug administration? Y / N

, how?

Was it confirmed that the patient received / took the drug? Y / N Was the patient monitored after drug administration? Y / N Was any specific drug-related guideline followed during the administration? Y / N Did the nurse use gloves during administration? Y / N

(if necessary, check the medication guidelines after observation): Was rate of intravenous drug administration correct? Y / N Would the drug administration have required monitoring the patient’s condition? Y / N Should any special guideline have been followed? Y / N

, which one?

Has an error been detected? Observer: Y / N (percentage of the consensus of reviewers is to be calculated)

Additional information:

Verifier of the observation: Y / N

This sheet is to be filled if an error has been detected:

CLASSIFICATION OF THE ERRORS

Stage of medication process: Prescription error Y / N

Administration error Y / N

Dispensing error Y / N

Preparation error Y / N

Documentation error Y / N

Type of error: Wrong dosege Y / N

Wrong drug Y / N

Omission of drug Y / N

Wrong timing Y / N

Wrong administration technique Y / N

Wrong drug from Y / N

Wrong patient Y / N

Wrong administration route Y / N

Discontinued drug Y / N

Error in documenting Y / N

Error in preparing the drug Y / N

Other Y / N

,what?

THE SEVERITY OF THE ERROR: What kind of consequences could the error could cause to the patient? (NCCMERP classification A-I) Observer:

Verifier:

Physician:

NCCMERP classification of the severity of the error A: Circumstances or events that have the capacity to cause error B: An error did not reach the patient C:An error reached the patient, but did not cause patient harm D: An error reached the patient and required monitoring to confirm that it resulted in no harm to the patient E: An error may have contributed / resulted in temporary harm and required intervention F: An error may have contributed / resulted in temporary harm and required initial or prolonged hospitalisation G: An error may have contributed to or resulted in permanent patient harm H: An error that required intervention necessary to sustain life I: An error that may have contributed to or resulted in the patient’s death

Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients Patients’ medication process is a complex multi-stage and multiprofessional process and an important part of nurses’ daily work. In this study, medicationrelated adverse outcomes were analysed using medicationrelated incident reports, patients’ records using the Global Trigger

dissertations | 260 | MARJA HÄRKÄNEN

MARJA HÄRKÄNEN

MARJA HÄRKÄNEN

Medication-related Adverse Outcomes and Contributing Factors among Hospital Patients

Tool method, and observations of medication administrations. The study demonstrated that medication-related adverse outcomes are common and three methods produced different information about these outcomes.

Publications of the University of Eastern Finland Dissertations in Health Sciences

Publications of the University of Eastern Finland Dissertations in Health Sciences isbn 15461+61654