Ir J Med Sci DOI 10.1007/s11845-014-1140-1
ORIGINAL ARTICLE
Adverse drug reactions as a cause of admission to a Dublin-based university teaching hospital D. Walsh • A. Lavan • A.-M. Cushen D. Williams
•
Received: 28 February 2014 / Accepted: 5 May 2014 Ó Royal Academy of Medicine in Ireland 2014
Abstract Background A meta-analysis of 25 international studies suggests that 4.2–6.0 % of medical admissions are the result of an adverse drug reaction (ADR). One Irish study has found that 8.8 % of admissions to a university teaching hospital were attributable to ADRs. Aim To develop and evaluate a process to detect ADRrelated medical admissions to a university teaching hospital in North Dublin. Methods A screening process was developed to detect ADR-related admissions based on a previous Scottish study. Having evaluated the accuracy of the screening process in a large Dublin-based university teaching hospital, the same methodology was then applied to medical admissions occurring over a 9-day period. Results The sensitivity and specificity of the screening process were 100 and 97 %, respectively. The incidence of ADR-related hospitalization from 137 admissions was 5.1 % (95 % CI 1.4–8.8 %). Of the ADRs, six were type A (predictable and preventable) and one was a type B (uncommon ADRs) reaction. Of the seven ADRs, two were considered to be unavoidable while five were potentially avoidable. Highrisk medications namely anticoagulants, antiplatelets and antihypertensives were identified as causative medications. Conclusions This study outlines the feasibility of screening for ADR-related admission in the hospital setting. ADRs constitute an important and avoidable cause of hospital admission.
D. Walsh (&) D. Williams Royal College of Surgeons, Dublin, Ireland e-mail: [email protected] A. Lavan A.-M. Cushen Beaumont hospital, Dublin, Ireland
Keywords Adverse drug reactions Hospital admissions Incidence Medications
Introduction Adverse drug reactions (ADRs) have been defined as ‘‘an appreciably harmful or unpleasant reaction resulting from an intervention relating to the use of a medicinal product’’ [1]. ADRs are a significant cause of excess morbidity and, in some cases, mortality; ADRs also lead to patient dissatisfaction with their care and have been shown to be a risk factor for future non-compliance with medications [2]. Additionally, ADRs can mimic disease leading to potentially inappropriate investigations and treatment [3]. A meta-analysis of 25 international studies suggested that 4.2–6 % (median 5.8 %) of all admissions to medical departments were attributable to ADRs [4]. In Britain, a prospective observational study with a cohort of 18,820 admissions found an incidence of 6.5 % (95 % CI 6.2–6.9 %) ADR-related admissions, and in 80 % of cases, the admission was directly attributable to the ADR [5]. A similar study based in a Scottish hospital found an incidence of 2.7 % (95 % CI 1.8–3.7 %) ADR-related hospital admissions [6]. One recent Irish study, based in a university teaching hospital in the Cork, reported an incidence of ADR-related admissions of 8.8 % [7]. An estimated 250,000 admissions per annum in the UK are related to ADRs, with a projected cost burden of £466 million/€706 million [8]. Evidence has shown a definite association between ADRs and increased length of stay across several studies [9]. One prospective study of inpatient admissions comparing similar groups found that patients admitted with an ADR stayed on average 20 days, while patients with no ADR had a shorter stay of 8 days [10]. There is also
123
Ir J Med Sci
considerable evidence that the majority of ADR-related admissions are avoidable. Pirmohamed et al. [5], using a validated tool to assess preventability, found that 72 % of ADR-related admissions were classed as definitely or possibly avoidable. A Scottish paper, based on the same methodology, reported that only 13.3 % of admissions were unavoidable [6]. Several other studies have reported similar findings with 60–70 % of ADRs being classes as preventable [11–14]. Many different approaches have been used to detect ADRs in the hospital setting including spontaneous reporting, chart review, systematic screening and computer-based detection programmes. The spontaneous reporting of ADRs has been shown to significantly underestimate the incidence of ADRs [15]. Our study involved a prospective systematic screening of all admissions using a team of a physician and a pharmacist based on the process devised by Hopf et al. [6]. The paper on which our methodology is based found pharmacist-led screening of admissions during ward rounds in a Scottish hospital to be a highly accurate process for detecting ADR-related admissions; this approach yielded a sensitivity and specificity of 88.2 and 90.6 %, respectively [6]. However, physician screening for ADRs has been criticized as inaccurate by some authors. One German study, by Dormann et al. [16], found that there was a low rate of detection of community-acquired ADRs upon hospital admission; only 56.9 % of ADRs detected on admission using intensive computerized surveillance were reported by physicians. Our prospective systematic screening approach incorporates a multi-faceted review of each suspected ADR involving clinical judgement, chart review and a validated gold-standard algorithm to assess probability, type and causative medication for each case. Evidence has shown that studies that employ multiple approaches are more likely to yield a higher incidence of ADRs [9]. There is a need to develop a screening method to identify hospital admissions potentially attributable to an ADR, which is both applicable and practical in an Irish hospital setting. The burden of ADRs to the Irish health system in terms of preventable morbidity, mortality and cost is an area which requires further research. This current study was conducted in a large Dublinbased tertiary referral hospital and focused on all medical admissions occurring consecutively during a 9-day period. The aims of the study were 1.
2.
To evaluate the sensitivity and specificity of a clinician-led screening process to identify ADR-related admissions. To determine from a sample of 137 patients admitted over 9 consecutive days the proportion of admissions
123
directly attributable to ADRs, and to categorize these ADRs in terms of type and causative medication.
Method Ethical approval was granted to prospectively screen admissions to at large university teaching hospital in Dublin using a similar methodology as used in a previous study [6]. A 2-day pilot study was conducted to calculate the sensitivity and specificity of the screening process before data collection began. The definition of an ADR was that of Edwards and Aronson and the criteria employed by Naranjo et al. [1, 18] were used to assess ADR probability. The Naranjo scale (Table 1) was used as the gold standard for assessing ADRs; this is a probability scale used to classify the likelihood of an ADR as doubtful, possible, probable or definite [18]. The ward pharmacist and physician conducting the screening were informed about the aims and methods of the study during a meeting which took place before the study, where the definitions and criteria used were explained. A screening form (see ‘‘Appendix’’) was used by a ward pharmacist and a medical registrar to document relevant patient data when an ADR-related admission was suspected. A suspected ADR was considered a true positive if BOTH the team responsible for that patient achieved consensus that an ADR was the likely
Table 1 Adapted Naranjo scale: an algorithm devised by Naranjo et al. used to determine the probability of ADR Adapted Naranjo probability scale
Yes
No
Not known
Previous report of same reaction
1
0 0
Adverse event after drug was administered
2
-1 0
Reaction improved when drug discontinued/ antagonist administered
1
0 0
Reaction reappeared when drug readministered
2
-1 0
Reaction could have been caused by something other than drug
-1
2 0
Reaction reappeared when placebo given
-1
1 0
Drug detected in blood in toxic concentration
1
0 0
Reaction more severe when dose increased/less severe when reduced Patient had similar reaction to drug/similar drug in a previous exposure
1
0 0
1
0 0
Definite: score of 9 or greater Probable: score of 5–8 Possible: score of 1–4 Doubtful: score of 0
Ir J Med Sci
cause of admission and the Naranjo ADR scale rated the likelihood of an ADR as possible, probable or definite (Table 1). A screen negative case was verified by analysing the patient notes and drug kardex. An overview of the process is described in Fig. 1. The 2-day pilot study recruited a cohort of 43 consecutive admissions all of which were screened by the pharmacist or physician. Each admission was then analysed to confirm whether or not it was accurately assessed by the team (as illustrated in Table 1); this allowed for the calculation of sensitivity and specificity. Having ascertained that the screening methodology was highly accurate with sensitivity and specificity figures of 97 and 100 %, respectively, we applied the same clinician-led screening process to all medical admissions to the hospital over a 9-day period. All medical patients admitted to the acute medical assessment unit, accident and emergency or directly to a ward from 03/08/2012 to 11/08/12 were screened for ADRs by a pharmacist or a medical registrar. Admissions were excluded if they were due to deliberate or unintentional overdose, relapse due to non-compliance or if the patient was under 16-year old as in previous studies [5, 6] Surgical admissions were not included. After the review process, all confirmed ADRs were classified into type A and type B reactions according to the definition of Rawlins et al. [19]. A type A ADR is defined
as a common, dose-related and predictable reaction. A type B ADR is an uncommon idiosyncratic reaction which is not related to the dose of the drug. ADRs were also classified into those which were unavoidable, possibly avoidable and definitely avoidable in accordance with definitions of Hallas et al. [20]. All data were coded and analysed using PASW version 18 for windows. Patient data including sex, age, and whether or not an ADR was suspected were coded. When an ADR was suspected the type of reaction and drug implicated was anonymized and entered into PASW. The results were presented as descriptive statistics using interquartile ranges and percentage frequencies with 95 % CI.
Results A total of 137 medical patients were admitted over a 9-day period from the 03/08/12 to the 11/08/12. Initially, 11 suspected ADRs were identified with the remaining 126 admissions screened as negative, i.e. non-cases. After the review process 7 of the 11 ADRs were deemed to be true positives, equating to an incidence rate of 5.1 % (95 % CI 1.4–8.8 %). Of the seven ADRs, six were type A reactions (dose-related and predictable) with 1 type B reaction. Highrisk groups of medications were represented in the ADRs
Fig. 1 Overview of the screening process undertaken in this study
Medical admissions
Screened by pharmacist or medical registrar
Screen negative (No ADR suspected)
Screen positive (ADR suspected)
Confirm by reviewing patient notes and drug kardex
Review using Naranjo scale and
True/False Negatives
consult with admitting team
True/False Positives
123
Ir J Med Sci Table 2 List of each ADR occurring during our study including nature of event, probability of ADR (determined by Naranjo scale), type of ADR, drug(s) implicated and preventability Confirmed ADRs causing admission Nature of ADR
Suspected drug(s)
ADR type
ADR score
Causality
Preventability
Gastric ulcer
Diclofenac
A
2
Possible
Possible
Hyperkalaemia
Captopril
A
3
Possible
Possible
Palpitations/long QT interval Postural hypotension
Clarithromycin Propranolol
B A
3 2
Possible Possible
Not avoidable Possible
Haematemesis
Aspirin and Warfarin
A
5
Probable
Not avoidable
Hyponatraemia
Bendroflumethiazide
A
8
Probable
Possible
Postural hypotension
Bisoprolol
A
3
Possible
Possible
Type A reaction: dose-related and related to the pharmacological effect of the medication. Typically preventable Type B reaction: Bizarre in nature, unrelated to mechanism of action of the medication. Typically not predictable ADR score: numerical score on Naranjo ADR probability scale (see Table 1) Causality: likelihood of a ADR explaining the reaction as per the Naranjo scale [18] Preventability: potential for preventing the adverse event as per criteria as defined by Rawlins et al. [19]
detected; beta-blockers (propranolol and Bisoprolol), a non-steroidal anti-inflammatory medication (Diclofenac), a diuretic (bendroflumethiazide), and antiplatelet and anticoagulation medications (aspirin and warfarin) were responsible for admission (Table 2). There were five ADRs which were classed as possibly avoidable, i.e. if there was greater effort above and beyond the obligatory demands of the given situation, the ADR could be avoided; two ADRs were classed as unavoidable (unpredictable event which occurred despite concordance with good medical practice). No ADR was identified as definitely preventable, i.e. not consistent with good medical practice (Table 2). All confirmed ADR-attributable reactions including clinical indication for admission, type of ADR, Naranjo probability scale result and preventability are described in Table 2. While the Naranjo scale assessed five ADRs which had been confirmed by the medical teams as ‘‘possible ADRs,’’ it must be taken into account that some of the Naranjo scale questions could not be answered based on clinical data available, e.g. drug serum levels. The median age of patients who had an ADR was 86 years. Polypharmacy (the concomitant use of greater than four medications) seen in five of the seven confirmed ADRs.
Discussion The incidence of ADRs found in this study of 137 admissions was 5.1 % (95 % CI 1.1–8.8 %). This figure is greater than that of Hopf et al. [4, 6] in Scotland but is within the range of a meta-analysis of 25 international studies of ADRs. The only other Irish study of ADRrelated admissions screened 856 patients admitted to a university teaching hospital in Cork and found an even
123
higher incidence of ADR-related hospitalization of 8.8 % [7]. This study used a panel to review cases using two validated criteria, one of which was the Naranjo scale (as used in our study); however, it must be noted that there was a high level of disagreement among the researchers using the two scales. Our methodology for detecting ADRs on admission is cost-effective and requires a relatively small number of clinicians. This approach also allows for the prospective screening of admissions, which has been shown to generate more accurate incidence data [17], and incorporates a multi-faceted review of each suspected ADR involving clinical judgement, chart review and a validated algorithm to assess probability, type and causative medication for each case. There has been considerable variation worldwide in the methodologies employed to detect ADRs, and consequently the incidence of ADRs reported. A systematic review of 95 prospective studies demonstrated widely varying incidence figures from 0.1 to 54 % [9]. Many factors influence the incidence of ADRs reported in a given study including type of hospital, geographical setting and method of detection [14, 23]. The incidence of ADRs has also been shown to vary with patient population and methods of detection [17]. In spite of this, the literature on ADRs is consistent in asserting that the majority of ADRs are preventable, it has been suggested that at least 70 % of all ADRs are preventable []. In our study, five of the seven ADRs were classed as definitely or possibly avoidable and six of the seven ADRs detected were classed as type A reactions (predictable reactions). A large body of evidence suggests that the majority of type A reactions are preventable [21, 22]. The high-risk groups of medications implicated in this study are consistent with the international literature. It is estimated that six groups of medications are
Ir J Med Sci
responsible for 60–70 % of all ADRs: antibiotics, anticoagulants, digoxin, diuretics, hypoglycaemic agents and non-steroidal anti-inflammatory drugs [9]. In our study four of the seven ADRs detected were represented in these highrisk groups. Similarly, our finding of polypharmacy in five of seven ADR patients is in keeping with current evidence [9]. This present study applies an established screening method to a tertiary hospital based in North Dublin but there is a need to generate further data on the burden of ADRs on the Irish healthcare system. Further information on the incidence and characteristics of ADRs can be used to inform interventions to reduce the risk of such events occurring. A meta-analysis of 95 studies showed that incidence figures for ADRs can vary substantially at national level depending on location, population studied and methodology [23]. At present, there is insufficient national data regarding the variation in ADR rates across different geographical regions and types of hospital. Similarly, more data are needed on the clinical significance of ADRs which have caused admission to gauge the associated morbidity and mortality.
incidence figure of 5.1 % (95 % CI 1.4–8.8 %). This study also made limited use of the Naranjo scale as an algorithm to predict the probability of ADRs. It was not possible to answer all relevant questions in the Naranjo scale to score ADRs, e.g. ‘‘drug detected in blood in toxic concentration,’’ ‘‘reaction more severe when dose increased,’’ etc.
Conclusion We have identified a significant number of ADR-related admissions (1 in every 20 admissions) and developed a sensitive and reliable screening tool to detect ADRs which resulted in admission to hospital. Establishing a true rate of ADRs in the Irish hospital setting will allow one to develop initiatives aimed at preventing this important cause of patient harm. Acknowledgments The authors would like to thank the Health Research Board of Ireland for generously funding this study as well as the staff of the acute medical assessment unit and accident and emergency department in Beaumont Hospital for their help during the study. Conflict of interest
None.
Limitations This study took place over a relatively short time-frame and thus could only generate a cohort of 137 patients leading to a relatively wide confidence interval around the
Appendix See Table 3.
123
Table 3 ADR screening form
Ir J Med Sci
123
Ir J Med Sci
References 1. Edwards IR, Aronson JK (2000) Adverse drug reactions: definitions, diagnosis, and management. Lancet 356(9237):1255–1259 (Epub 2000/11/10) 2. Green CF, Mottram DR, Rowe PH, Pirmohamed M (2000) Adverse drug reactions as a cause of admission to an acute medical assessment unit: a pilot study. J Clin Pharm Ther 25(5):355–361 (Epub 2000/12/21) 3. Farcas A, Sinpetrean A, Mogosan C, Palage M, Vostinaru O, Bojita M et al (2010) Adverse drug reactions detected by stimulated spontaneous reporting in an internal medicine department in Romania. Eur J Intern Med 21(5):453–457 4. Muehlberger N, Schneeweiss S, Hasford J (1997) Adverse drug reaction monitoring—cost and benefit considerations. Part I: frequency of adverse drug reactions causing hospital admissions. Pharmacoepidemiol Drug Saf 6(Suppl 3):S71–S77 (Epub 2004/04/10) 5. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ et al (2004) Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ 329(7456):15–19 (Epub 2004/07/03) 6. Hopf Y, Watson M, Williams D (2008) Adverse-drug-reaction related admissions to a hospital in Scotland. Pharm World Sci 30(6):854–862 (Epub 2008/07/26) 7. Ahern F, Sahm LJ, Lynch D, McCarthy S (2013) Determining the frequency and preventability of adverse drug reaction-related admissions to an Irish University Hospital: a cross-sectional study. Emerg Med J. doi:10.1136/emermed-2012-201945 8. Hitchen L (2006) Adverse drug reactions result in 250,000 UK admissions a year. BMJ 332(7550):1109 (Epub 2006/05/13) 9. Wiffen P, Gill M, Edwards J, Moore A (2002) Adverse drug reactions in hospital patients: a systematic review of the prospective and retrospective studies. Bandolier Extra 1–16 10. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, Pirmohamed M (2009) Adverse drug reactions in hospital inpatients: a prospective analysis of 3695 patient-episodes. PloS one 4(2):e4439 (Epub 2009/02/12) 11. Howard RL, Avery AJ, Howard PD, Partridge M (2003) Investigation into the reasons for preventable drug related admissions to a medical admissions unit: observational study. Qual Saf Health Care 12(4):280–285 (Epub 2003/08/05)
12. Gholami K, Shalviri G (1999) Factors associated with preventability, predictability, and severity of adverse drug reactions. Ann Pharmacother 33(2):236–240 (Epub 1999/03/20) 13. Dartnell JG, Anderson RP, Chohan V, Galbraith KJ, Lyon ME, Nestor PJ et al (1996) Hospitalisation for adverse events related to drug therapy: incidence, avoidability and costs. Med J Aust 164(11):659–662 (Epub 1996/06/03) 14. Beijer HJ, de Blaey CJ (2002) Hospitalisations caused by adverse drug reactions (ADR): a meta-analysis of observational studies. Pharm World Sci 24(2):46–54 (Epub 2002/06/14) 15. Williams D, Feely J (1999) Underreporting of adverse drug reactions: attitudes of Irish doctors. Ir J Med Sci 168:257–261 16. Dormann H, Criegee-Rieck M, Neubert A, Egger T, Geise A, Krebs S et al (2003) Lack of awareness of community-acquired adverse drug reactions upon hospital admission: dimensions and consequences of a dilemma. Drug Saf 26(5):353–362 (Epub 2003/03/26) 17. Kongkaew C, Noyce PR, Ashcroft DM (2008) Hospital admissions associated with adverse drug reactions: a systematic review of prospective observational studies. Ann Pharmacother 42(7):1017–1025 (Epub 2008/07/03) 18. Naranjo CA et al (1981) A method for estimating the probability of adverse drug reactions. Clin Pharmacol Ther 30(2):239–245 19. Rawlins MDTS (1998) Mechanisms of adverse drug reactions. In: Davies DMFR, de Galnville H (eds) Davies’s textbook of adverse drug reactions. Lippincott-Raven Publishers, Philadelphia 20. Hallas J et al (1990) Drug related hospital admissions: the role of definitions and intensity of data collection, and the possibility of prevention. J Intern Med 228(2):83–90 21. Davies EC, Green CF, Mottram DR, Pirmohamed M (2007) Adverse drug reactions in hospitals: a narrative review. Curr Drug Saf 2(1):79–87 (Epub 2008/08/12) 22. Mjorndal T, Boman MD, Hagg S, Backstrom M, Wiholm BE, Wahlin A et al (2002) Adverse drug reactions as a cause for admissions to a department of internal medicine. Pharmacoepidemiol Drug Saf 11(1):65–72 (Epub 2002/05/10) 23. Leendertse AJ, Visser D, Egberts AC et al (2010) The relationship between study characteristics and the prevalence of medication-related hospitalizations: a literature review and novel analysis. Drug Saf 33:233–244
123
ORIGINAL ARTICLE
Adverse drug reactions as a cause of admission to a Dublin-based university teaching hospital D. Walsh • A. Lavan • A.-M. Cushen D. Williams
•
Received: 28 February 2014 / Accepted: 5 May 2014 Ó Royal Academy of Medicine in Ireland 2014
Abstract Background A meta-analysis of 25 international studies suggests that 4.2–6.0 % of medical admissions are the result of an adverse drug reaction (ADR). One Irish study has found that 8.8 % of admissions to a university teaching hospital were attributable to ADRs. Aim To develop and evaluate a process to detect ADRrelated medical admissions to a university teaching hospital in North Dublin. Methods A screening process was developed to detect ADR-related admissions based on a previous Scottish study. Having evaluated the accuracy of the screening process in a large Dublin-based university teaching hospital, the same methodology was then applied to medical admissions occurring over a 9-day period. Results The sensitivity and specificity of the screening process were 100 and 97 %, respectively. The incidence of ADR-related hospitalization from 137 admissions was 5.1 % (95 % CI 1.4–8.8 %). Of the ADRs, six were type A (predictable and preventable) and one was a type B (uncommon ADRs) reaction. Of the seven ADRs, two were considered to be unavoidable while five were potentially avoidable. Highrisk medications namely anticoagulants, antiplatelets and antihypertensives were identified as causative medications. Conclusions This study outlines the feasibility of screening for ADR-related admission in the hospital setting. ADRs constitute an important and avoidable cause of hospital admission.
D. Walsh (&) D. Williams Royal College of Surgeons, Dublin, Ireland e-mail: [email protected] A. Lavan A.-M. Cushen Beaumont hospital, Dublin, Ireland
Keywords Adverse drug reactions Hospital admissions Incidence Medications
Introduction Adverse drug reactions (ADRs) have been defined as ‘‘an appreciably harmful or unpleasant reaction resulting from an intervention relating to the use of a medicinal product’’ [1]. ADRs are a significant cause of excess morbidity and, in some cases, mortality; ADRs also lead to patient dissatisfaction with their care and have been shown to be a risk factor for future non-compliance with medications [2]. Additionally, ADRs can mimic disease leading to potentially inappropriate investigations and treatment [3]. A meta-analysis of 25 international studies suggested that 4.2–6 % (median 5.8 %) of all admissions to medical departments were attributable to ADRs [4]. In Britain, a prospective observational study with a cohort of 18,820 admissions found an incidence of 6.5 % (95 % CI 6.2–6.9 %) ADR-related admissions, and in 80 % of cases, the admission was directly attributable to the ADR [5]. A similar study based in a Scottish hospital found an incidence of 2.7 % (95 % CI 1.8–3.7 %) ADR-related hospital admissions [6]. One recent Irish study, based in a university teaching hospital in the Cork, reported an incidence of ADR-related admissions of 8.8 % [7]. An estimated 250,000 admissions per annum in the UK are related to ADRs, with a projected cost burden of £466 million/€706 million [8]. Evidence has shown a definite association between ADRs and increased length of stay across several studies [9]. One prospective study of inpatient admissions comparing similar groups found that patients admitted with an ADR stayed on average 20 days, while patients with no ADR had a shorter stay of 8 days [10]. There is also
123
Ir J Med Sci
considerable evidence that the majority of ADR-related admissions are avoidable. Pirmohamed et al. [5], using a validated tool to assess preventability, found that 72 % of ADR-related admissions were classed as definitely or possibly avoidable. A Scottish paper, based on the same methodology, reported that only 13.3 % of admissions were unavoidable [6]. Several other studies have reported similar findings with 60–70 % of ADRs being classes as preventable [11–14]. Many different approaches have been used to detect ADRs in the hospital setting including spontaneous reporting, chart review, systematic screening and computer-based detection programmes. The spontaneous reporting of ADRs has been shown to significantly underestimate the incidence of ADRs [15]. Our study involved a prospective systematic screening of all admissions using a team of a physician and a pharmacist based on the process devised by Hopf et al. [6]. The paper on which our methodology is based found pharmacist-led screening of admissions during ward rounds in a Scottish hospital to be a highly accurate process for detecting ADR-related admissions; this approach yielded a sensitivity and specificity of 88.2 and 90.6 %, respectively [6]. However, physician screening for ADRs has been criticized as inaccurate by some authors. One German study, by Dormann et al. [16], found that there was a low rate of detection of community-acquired ADRs upon hospital admission; only 56.9 % of ADRs detected on admission using intensive computerized surveillance were reported by physicians. Our prospective systematic screening approach incorporates a multi-faceted review of each suspected ADR involving clinical judgement, chart review and a validated gold-standard algorithm to assess probability, type and causative medication for each case. Evidence has shown that studies that employ multiple approaches are more likely to yield a higher incidence of ADRs [9]. There is a need to develop a screening method to identify hospital admissions potentially attributable to an ADR, which is both applicable and practical in an Irish hospital setting. The burden of ADRs to the Irish health system in terms of preventable morbidity, mortality and cost is an area which requires further research. This current study was conducted in a large Dublinbased tertiary referral hospital and focused on all medical admissions occurring consecutively during a 9-day period. The aims of the study were 1.
2.
To evaluate the sensitivity and specificity of a clinician-led screening process to identify ADR-related admissions. To determine from a sample of 137 patients admitted over 9 consecutive days the proportion of admissions
123
directly attributable to ADRs, and to categorize these ADRs in terms of type and causative medication.
Method Ethical approval was granted to prospectively screen admissions to at large university teaching hospital in Dublin using a similar methodology as used in a previous study [6]. A 2-day pilot study was conducted to calculate the sensitivity and specificity of the screening process before data collection began. The definition of an ADR was that of Edwards and Aronson and the criteria employed by Naranjo et al. [1, 18] were used to assess ADR probability. The Naranjo scale (Table 1) was used as the gold standard for assessing ADRs; this is a probability scale used to classify the likelihood of an ADR as doubtful, possible, probable or definite [18]. The ward pharmacist and physician conducting the screening were informed about the aims and methods of the study during a meeting which took place before the study, where the definitions and criteria used were explained. A screening form (see ‘‘Appendix’’) was used by a ward pharmacist and a medical registrar to document relevant patient data when an ADR-related admission was suspected. A suspected ADR was considered a true positive if BOTH the team responsible for that patient achieved consensus that an ADR was the likely
Table 1 Adapted Naranjo scale: an algorithm devised by Naranjo et al. used to determine the probability of ADR Adapted Naranjo probability scale
Yes
No
Not known
Previous report of same reaction
1
0 0
Adverse event after drug was administered
2
-1 0
Reaction improved when drug discontinued/ antagonist administered
1
0 0
Reaction reappeared when drug readministered
2
-1 0
Reaction could have been caused by something other than drug
-1
2 0
Reaction reappeared when placebo given
-1
1 0
Drug detected in blood in toxic concentration
1
0 0
Reaction more severe when dose increased/less severe when reduced Patient had similar reaction to drug/similar drug in a previous exposure
1
0 0
1
0 0
Definite: score of 9 or greater Probable: score of 5–8 Possible: score of 1–4 Doubtful: score of 0
Ir J Med Sci
cause of admission and the Naranjo ADR scale rated the likelihood of an ADR as possible, probable or definite (Table 1). A screen negative case was verified by analysing the patient notes and drug kardex. An overview of the process is described in Fig. 1. The 2-day pilot study recruited a cohort of 43 consecutive admissions all of which were screened by the pharmacist or physician. Each admission was then analysed to confirm whether or not it was accurately assessed by the team (as illustrated in Table 1); this allowed for the calculation of sensitivity and specificity. Having ascertained that the screening methodology was highly accurate with sensitivity and specificity figures of 97 and 100 %, respectively, we applied the same clinician-led screening process to all medical admissions to the hospital over a 9-day period. All medical patients admitted to the acute medical assessment unit, accident and emergency or directly to a ward from 03/08/2012 to 11/08/12 were screened for ADRs by a pharmacist or a medical registrar. Admissions were excluded if they were due to deliberate or unintentional overdose, relapse due to non-compliance or if the patient was under 16-year old as in previous studies [5, 6] Surgical admissions were not included. After the review process, all confirmed ADRs were classified into type A and type B reactions according to the definition of Rawlins et al. [19]. A type A ADR is defined
as a common, dose-related and predictable reaction. A type B ADR is an uncommon idiosyncratic reaction which is not related to the dose of the drug. ADRs were also classified into those which were unavoidable, possibly avoidable and definitely avoidable in accordance with definitions of Hallas et al. [20]. All data were coded and analysed using PASW version 18 for windows. Patient data including sex, age, and whether or not an ADR was suspected were coded. When an ADR was suspected the type of reaction and drug implicated was anonymized and entered into PASW. The results were presented as descriptive statistics using interquartile ranges and percentage frequencies with 95 % CI.
Results A total of 137 medical patients were admitted over a 9-day period from the 03/08/12 to the 11/08/12. Initially, 11 suspected ADRs were identified with the remaining 126 admissions screened as negative, i.e. non-cases. After the review process 7 of the 11 ADRs were deemed to be true positives, equating to an incidence rate of 5.1 % (95 % CI 1.4–8.8 %). Of the seven ADRs, six were type A reactions (dose-related and predictable) with 1 type B reaction. Highrisk groups of medications were represented in the ADRs
Fig. 1 Overview of the screening process undertaken in this study
Medical admissions
Screened by pharmacist or medical registrar
Screen negative (No ADR suspected)
Screen positive (ADR suspected)
Confirm by reviewing patient notes and drug kardex
Review using Naranjo scale and
True/False Negatives
consult with admitting team
True/False Positives
123
Ir J Med Sci Table 2 List of each ADR occurring during our study including nature of event, probability of ADR (determined by Naranjo scale), type of ADR, drug(s) implicated and preventability Confirmed ADRs causing admission Nature of ADR
Suspected drug(s)
ADR type
ADR score
Causality
Preventability
Gastric ulcer
Diclofenac
A
2
Possible
Possible
Hyperkalaemia
Captopril
A
3
Possible
Possible
Palpitations/long QT interval Postural hypotension
Clarithromycin Propranolol
B A
3 2
Possible Possible
Not avoidable Possible
Haematemesis
Aspirin and Warfarin
A
5
Probable
Not avoidable
Hyponatraemia
Bendroflumethiazide
A
8
Probable
Possible
Postural hypotension
Bisoprolol
A
3
Possible
Possible
Type A reaction: dose-related and related to the pharmacological effect of the medication. Typically preventable Type B reaction: Bizarre in nature, unrelated to mechanism of action of the medication. Typically not predictable ADR score: numerical score on Naranjo ADR probability scale (see Table 1) Causality: likelihood of a ADR explaining the reaction as per the Naranjo scale [18] Preventability: potential for preventing the adverse event as per criteria as defined by Rawlins et al. [19]
detected; beta-blockers (propranolol and Bisoprolol), a non-steroidal anti-inflammatory medication (Diclofenac), a diuretic (bendroflumethiazide), and antiplatelet and anticoagulation medications (aspirin and warfarin) were responsible for admission (Table 2). There were five ADRs which were classed as possibly avoidable, i.e. if there was greater effort above and beyond the obligatory demands of the given situation, the ADR could be avoided; two ADRs were classed as unavoidable (unpredictable event which occurred despite concordance with good medical practice). No ADR was identified as definitely preventable, i.e. not consistent with good medical practice (Table 2). All confirmed ADR-attributable reactions including clinical indication for admission, type of ADR, Naranjo probability scale result and preventability are described in Table 2. While the Naranjo scale assessed five ADRs which had been confirmed by the medical teams as ‘‘possible ADRs,’’ it must be taken into account that some of the Naranjo scale questions could not be answered based on clinical data available, e.g. drug serum levels. The median age of patients who had an ADR was 86 years. Polypharmacy (the concomitant use of greater than four medications) seen in five of the seven confirmed ADRs.
Discussion The incidence of ADRs found in this study of 137 admissions was 5.1 % (95 % CI 1.1–8.8 %). This figure is greater than that of Hopf et al. [4, 6] in Scotland but is within the range of a meta-analysis of 25 international studies of ADRs. The only other Irish study of ADRrelated admissions screened 856 patients admitted to a university teaching hospital in Cork and found an even
123
higher incidence of ADR-related hospitalization of 8.8 % [7]. This study used a panel to review cases using two validated criteria, one of which was the Naranjo scale (as used in our study); however, it must be noted that there was a high level of disagreement among the researchers using the two scales. Our methodology for detecting ADRs on admission is cost-effective and requires a relatively small number of clinicians. This approach also allows for the prospective screening of admissions, which has been shown to generate more accurate incidence data [17], and incorporates a multi-faceted review of each suspected ADR involving clinical judgement, chart review and a validated algorithm to assess probability, type and causative medication for each case. There has been considerable variation worldwide in the methodologies employed to detect ADRs, and consequently the incidence of ADRs reported. A systematic review of 95 prospective studies demonstrated widely varying incidence figures from 0.1 to 54 % [9]. Many factors influence the incidence of ADRs reported in a given study including type of hospital, geographical setting and method of detection [14, 23]. The incidence of ADRs has also been shown to vary with patient population and methods of detection [17]. In spite of this, the literature on ADRs is consistent in asserting that the majority of ADRs are preventable, it has been suggested that at least 70 % of all ADRs are preventable []. In our study, five of the seven ADRs were classed as definitely or possibly avoidable and six of the seven ADRs detected were classed as type A reactions (predictable reactions). A large body of evidence suggests that the majority of type A reactions are preventable [21, 22]. The high-risk groups of medications implicated in this study are consistent with the international literature. It is estimated that six groups of medications are
Ir J Med Sci
responsible for 60–70 % of all ADRs: antibiotics, anticoagulants, digoxin, diuretics, hypoglycaemic agents and non-steroidal anti-inflammatory drugs [9]. In our study four of the seven ADRs detected were represented in these highrisk groups. Similarly, our finding of polypharmacy in five of seven ADR patients is in keeping with current evidence [9]. This present study applies an established screening method to a tertiary hospital based in North Dublin but there is a need to generate further data on the burden of ADRs on the Irish healthcare system. Further information on the incidence and characteristics of ADRs can be used to inform interventions to reduce the risk of such events occurring. A meta-analysis of 95 studies showed that incidence figures for ADRs can vary substantially at national level depending on location, population studied and methodology [23]. At present, there is insufficient national data regarding the variation in ADR rates across different geographical regions and types of hospital. Similarly, more data are needed on the clinical significance of ADRs which have caused admission to gauge the associated morbidity and mortality.
incidence figure of 5.1 % (95 % CI 1.4–8.8 %). This study also made limited use of the Naranjo scale as an algorithm to predict the probability of ADRs. It was not possible to answer all relevant questions in the Naranjo scale to score ADRs, e.g. ‘‘drug detected in blood in toxic concentration,’’ ‘‘reaction more severe when dose increased,’’ etc.
Conclusion We have identified a significant number of ADR-related admissions (1 in every 20 admissions) and developed a sensitive and reliable screening tool to detect ADRs which resulted in admission to hospital. Establishing a true rate of ADRs in the Irish hospital setting will allow one to develop initiatives aimed at preventing this important cause of patient harm. Acknowledgments The authors would like to thank the Health Research Board of Ireland for generously funding this study as well as the staff of the acute medical assessment unit and accident and emergency department in Beaumont Hospital for their help during the study. Conflict of interest
None.
Limitations This study took place over a relatively short time-frame and thus could only generate a cohort of 137 patients leading to a relatively wide confidence interval around the
Appendix See Table 3.
123
Table 3 ADR screening form
Ir J Med Sci
123
Ir J Med Sci
References 1. Edwards IR, Aronson JK (2000) Adverse drug reactions: definitions, diagnosis, and management. Lancet 356(9237):1255–1259 (Epub 2000/11/10) 2. Green CF, Mottram DR, Rowe PH, Pirmohamed M (2000) Adverse drug reactions as a cause of admission to an acute medical assessment unit: a pilot study. J Clin Pharm Ther 25(5):355–361 (Epub 2000/12/21) 3. Farcas A, Sinpetrean A, Mogosan C, Palage M, Vostinaru O, Bojita M et al (2010) Adverse drug reactions detected by stimulated spontaneous reporting in an internal medicine department in Romania. Eur J Intern Med 21(5):453–457 4. Muehlberger N, Schneeweiss S, Hasford J (1997) Adverse drug reaction monitoring—cost and benefit considerations. Part I: frequency of adverse drug reactions causing hospital admissions. Pharmacoepidemiol Drug Saf 6(Suppl 3):S71–S77 (Epub 2004/04/10) 5. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ et al (2004) Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ 329(7456):15–19 (Epub 2004/07/03) 6. Hopf Y, Watson M, Williams D (2008) Adverse-drug-reaction related admissions to a hospital in Scotland. Pharm World Sci 30(6):854–862 (Epub 2008/07/26) 7. Ahern F, Sahm LJ, Lynch D, McCarthy S (2013) Determining the frequency and preventability of adverse drug reaction-related admissions to an Irish University Hospital: a cross-sectional study. Emerg Med J. doi:10.1136/emermed-2012-201945 8. Hitchen L (2006) Adverse drug reactions result in 250,000 UK admissions a year. BMJ 332(7550):1109 (Epub 2006/05/13) 9. Wiffen P, Gill M, Edwards J, Moore A (2002) Adverse drug reactions in hospital patients: a systematic review of the prospective and retrospective studies. Bandolier Extra 1–16 10. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, Pirmohamed M (2009) Adverse drug reactions in hospital inpatients: a prospective analysis of 3695 patient-episodes. PloS one 4(2):e4439 (Epub 2009/02/12) 11. Howard RL, Avery AJ, Howard PD, Partridge M (2003) Investigation into the reasons for preventable drug related admissions to a medical admissions unit: observational study. Qual Saf Health Care 12(4):280–285 (Epub 2003/08/05)
12. Gholami K, Shalviri G (1999) Factors associated with preventability, predictability, and severity of adverse drug reactions. Ann Pharmacother 33(2):236–240 (Epub 1999/03/20) 13. Dartnell JG, Anderson RP, Chohan V, Galbraith KJ, Lyon ME, Nestor PJ et al (1996) Hospitalisation for adverse events related to drug therapy: incidence, avoidability and costs. Med J Aust 164(11):659–662 (Epub 1996/06/03) 14. Beijer HJ, de Blaey CJ (2002) Hospitalisations caused by adverse drug reactions (ADR): a meta-analysis of observational studies. Pharm World Sci 24(2):46–54 (Epub 2002/06/14) 15. Williams D, Feely J (1999) Underreporting of adverse drug reactions: attitudes of Irish doctors. Ir J Med Sci 168:257–261 16. Dormann H, Criegee-Rieck M, Neubert A, Egger T, Geise A, Krebs S et al (2003) Lack of awareness of community-acquired adverse drug reactions upon hospital admission: dimensions and consequences of a dilemma. Drug Saf 26(5):353–362 (Epub 2003/03/26) 17. Kongkaew C, Noyce PR, Ashcroft DM (2008) Hospital admissions associated with adverse drug reactions: a systematic review of prospective observational studies. Ann Pharmacother 42(7):1017–1025 (Epub 2008/07/03) 18. Naranjo CA et al (1981) A method for estimating the probability of adverse drug reactions. Clin Pharmacol Ther 30(2):239–245 19. Rawlins MDTS (1998) Mechanisms of adverse drug reactions. In: Davies DMFR, de Galnville H (eds) Davies’s textbook of adverse drug reactions. Lippincott-Raven Publishers, Philadelphia 20. Hallas J et al (1990) Drug related hospital admissions: the role of definitions and intensity of data collection, and the possibility of prevention. J Intern Med 228(2):83–90 21. Davies EC, Green CF, Mottram DR, Pirmohamed M (2007) Adverse drug reactions in hospitals: a narrative review. Curr Drug Saf 2(1):79–87 (Epub 2008/08/12) 22. Mjorndal T, Boman MD, Hagg S, Backstrom M, Wiholm BE, Wahlin A et al (2002) Adverse drug reactions as a cause for admissions to a department of internal medicine. Pharmacoepidemiol Drug Saf 11(1):65–72 (Epub 2002/05/10) 23. Leendertse AJ, Visser D, Egberts AC et al (2010) The relationship between study characteristics and the prevalence of medication-related hospitalizations: a literature review and novel analysis. Drug Saf 33:233–244
123
