You are here

  • Home
  • Archive
  • Volume 31, Issue 1
  • Determining the frequency and preventability of adverse drug reaction-related admissions to an Irish University Hospital: a cross-sectional study

Article Text

Article menu

Download PDFPDF

Original article
Determining the frequency and preventability of adverse drug reaction-related admissions to an Irish University Hospital: a cross-sectional study
  1. Fiona Ahern1,
  2. Laura J Sahm2,3,
  3. Deirdre Lynch1,
  4. Suzanne McCarthy1,2
  1. 1Pharmacy Department, Cork University Hospital, Cork, Ireland
  2. 2School of Pharmacy, University College Cork, Cork, Ireland
  3. 3Pharmacy Department, Mercy University Hospital, Cork, Ireland
  1. Correspondence to Dr Suzanne McCarthy, School of Pharmacy, University College Cork, Cavanagh Pharmacy Building, College Road, Cork, Ireland; s.mccarthy{at}ucc.ie

Abstract

Background Adverse drug reactions (ADR) cause considerable morbidity and mortality.

Methods This 4-week study was undertaken in Cork University Hospital, Ireland, for all admissions from the emergency department (ED). A panel independently reviewed patients with suspected ADRs. Causality assessment was performed using the Naranjo ADR probability scale and the Hallas criteria was used to assess preventability of the ADRs.

Results During the study period, 1258 patients were admitted from the ED; of these, 856 patients were included in the study; 75 patients (8.8%) had an ADR-related admission. Over half were deemed to be ‘possibly’ or ‘definitely’ avoidable. The level of agreement between reviewers using the Naranjo and Hallas criteria was very low.

In the ADR group (n=75), 50.7% were men compared with 53.1% in the non-ADR group (n=781). The median age for patients in the ADR group was 73 years compared with 45 years in the non-ADR group. The average number of prescribed drugs per patient in the ADR group was 7.5 (SD±3.8) compared with 2.4 (SD±3.6) in the non-ADR group. Classified by drug type, 74.2% of the ADRs were attributed to cardiovascular and central nervous system drugs.

Conclusions This study estimated the incidence of ADR-related admissions to an Irish hospital at 8.8%, with 57.3% of these deemed to have been potentially avoidable. Older patients were more likely to have an ADR-related admission. Prescribers must be aware of this increased likelihood of an ADR when prescribing new drugs to this patient population, and regularly review treatment.

  • research, epidemiology

https://doi.org/10.1136/emermed-2012-201945

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    WHO has defined an adverse drug reaction (ADR) as ‘any response to a drug which is noxious, unintended and occurs at doses normally used for prophylaxis, diagnosis or therapy of disease, or for modification of physiological function’.1 This definition excludes non-adherence with prescribed medication, drug abuse, overdose (intentional or unintentional), treatment failure and errors in drug administration.2 ,3

    ADRs cause considerable morbidity and mortality, and so have implications for patients in that they can reduce patients’ quality of life, lead to prolonged stay in hospital or possibly even cause fatality.4 Pirmohamed et al4 in 2004, estimated that ADRs are responsible for fatalities in 0.15% of all patients admitted. It is estimated that serious ADRs are responsible for more than 100 000 deaths annually in the USA.2

    Estimates for the incidence of ADRs in the literature vary, depending on the definition of ADR used, study setting and study population.5 In addition, the incidence of ADR-related admissions is often underestimated due to the lack of documentation in patient medical notes.4 ,6

    In a 2010 systematic review on ADR-related admissions, 95 studies were included.7 These studies estimated a prevalence of ADR-related admission of 0.1% to 54%, and the authors identified that there was a higher prevalence reported in studies that examined all hospital admissions compared with those only in acute hospitals.7 There is also wide variation in the reported figures for preventability of ADRs in the literature ranging from 3.9% to 92%.8 ,9

    Although several studies have been performed in the UK, other European countries and the USA, there are limited data published on the incidence of ADR-related admissions in Ireland. Two unpublished studies conducted in similar-sized hospitals in Ireland (in Limerick and Dublin) determined varying prevalence of ADR-related admission; the prevalence in the Limerick hospital being almost twice that of the Dublin hospital (12.4% and 6.5%, respectively) (Breslin et al, personal communication; Desmond et al, personal communication).

    Aims and objectives

    • To determine the frequency of ADR-related admissions to Cork University Hospital (CUH).

    • To perform causality assessment of the ADRs.

    • To assess preventability of the ADRs identified.

    Methods

    Setting

    CUH and the Cork University Maternity Hospital is a combined 850-bed hospital site serving a population of over 620 000 people. CUH also acts as a tertiary referral centre for a population of over 1 million.10

    In 2009, 50 557 people presented to the emergency department (ED) in CUH resulting in 14 525 admissions.

    Data collection

    This study was undertaken in the ED in CUH for 4 weeks in June/July 2010. All admissions to the ED, between 00:01 on Sunday to 24:00 on Thursday, were identified using the ED admissions logbook; this data collection strategy was chosen to coincide with the working days (Monday–Friday) of the investigator (FA), and to ensure that all patients were reviewed by the investigator <24 h of admission. All admissions to the ED, including psychiatry and paediatric admissions, are entered into the admissions logbook.

    Data collected by the investigator included: date of admission, admitting physician, gender, date of birth, if an ADR was suspected, presenting complaint, patient's past medical history, initial diagnosis and medication on admission. For the purpose of this study, the chosen definition of an ADR was that defined by WHO.1

    For patients where a suspected/confirmed ADR had taken place (as determined by FA), follow-up (using information from medical notes) was completed at ward level. For patients where no suspected ADR had occurred, data collection forms were collected, and a 10% random sample was reviewed by a second clinical pharmacist to ensure validity of results. No additional cases were identified.

    Exclusion criteria

    All patients admitted to the ED on Friday and Saturday were excluded, as these patients would have been reviewed >24 h after admission. Patients who were transferred from the ED to another hospital were also excluded from the study due to difficulty in patient follow-up. Patients admitted with an overdose, whether intentional or non-intentional, were excluded, as this did not fall within WHO definition of an ADR.

    Review panel

    A review panel was established to review patients with suspected ADRs. This panel was made up of the investigator (FA), one academic/clinical pharmacist at the School of Pharmacy, University College Cork, a clinical pharmacist from CUH and a medical registrar.

    Each member of the review panel independently undertook a causality assessment using the Naranjo ADR probability scale.11 The Naranjo criteria classify the probability that an ADR is related to drug therapy based on a list of weighted questions that consider: prior reports of a reaction, whether there is a temporal relationship; the effects of dechallenge and rechallenge, whether a dose-response relationship exists; the possibility of alternative aetiologies, the presence of objective confirmation of an ADR, whether there is a past medical history of any similar reaction to the same or similar drugs. The Naranjo tool is one of the most widely used tools for assessing ADR causality.12 The Naranjo score assigned by the panel members to each patient classified the ADR as; ‘doubtful’, ‘possible’, ‘probable’ or ‘definite’. The final classification assigned to each patient was determined using the following algorithm:

    • All panel members had the same classification for the patient.

    • The majority of the panel (three members) had agreed on a particular classification.

    • If the panel was split between classifications, the classification assigned by the medical registrar was chosen for the final classification. The registrar was chosen as the ‘gold standard’ because although it was possible that a pharmacist would have a greater knowledge of the adverse effects of drugs, the registrar would have a greater knowledge on the medical conditions patients had, and so, could identify symptoms of a disease that may be similar to adverse effects of a drug. This was particularly relevant among the elderly patients who often presented with multiple comorbidities.

    Any patient where the final classification was deemed ‘doubtful’ was considered not to have had an ADR-related admission. This was also the criteria for classifying ADRs used in a similar 2008 prospective study by Zed et al.13

    Preventability of the ADRs was determined using the Hallas criteria for preventability.14 A recent systematic review on the preventability of ADRs identified that the Hallas criteria was used most frequently to assess preventability.15 Each member of the review panel assigned a preventability classification to each patient. Where agreement did not occur among all reviewers, the classification algorithm above was used to determine the final classification.

    Inter-rater reliability amongst the panel members was analysed using κ values, and categorised according to the classification scheme by McGinn et al,16 as presented in table 1.

    View this table:
    Table 1

    Qualitative classification of κ values as degree of agreement beyond chance16

    Study size

    A sample size of 667 patients was required to estimate the prevalence of ADR-related admissions, based on a prevalence of 12.4% from research conducted in an Irish hospital (Breslin et al, personal communication), a precision level of 2.5% and a 95% CI.

    Statistical methods

    Data manipulation and analyses were undertaken using Microsoft Excel Mac (2008) and Stata/MP V.11.0 (StataCorp, College Station, Texas, USA). For normally distributed data, t test and χ2 tests were used for comparison of continuous and categorical variables, respectively. Equivalent non-parametric tests were used where appropriate.

    Ethics approval

    Approval was sought and granted from the Clinical Research Ethics Committee of the Cork Teaching Hospitals, Ireland.

    Results

    Patient selection

    There were 1258 patients admitted to the ED during the study period. Figure 1 presents the selection process of patients for the study, including the reasons for patient exclusion.

    Figure 1
    Figure 1

    Flow diagram of patient selection for study.

    Frequency of ADR-related admissions to CUH

    Of the 856 patients who were included in the study, 75 (8.8%) were deemed to have had an ADR-related admission (95% CI 7.0 to 10.9%).

    Patient demographics

    Table 2 displays the patient demographics for the two cohorts of patient identified. No difference was observed in the percentage of men and women in the two groups (p=0.533). By contrast, a significant difference was observed between the two groups in respect of age, whereby those in the ADR group were older (p<0.05).

    View this table:
    Table 2

    Patient demographics for patients with and without an adverse drug reaction (ADR)-related admission

    Naranjo assessment tool for assessing causality

    Of the 75 ADR-related admission cases identified, 51 (68%) were classified ‘possible’, 21 (28%) classified ‘probable’ and three (4%) classified ‘definite’ by the review panel. For ten of the 75 patients, there was no clear majority among the panel, and so the classification by the registrar was chosen. There were no patients with an ADR that was classed as ‘doubtful’, so none of the original 75 patients with a suspected ADR were excluded. The κ score calculated was 0.16 indicating only slight agreement among the reviewers.

    Hallas criteria to determine preventability of ADRs

    Of the 75 ADR cases identified, four (5.3%) were classified as ‘definitely avoidable’, 39 (52%) ‘possibly avoidable’, 25 (33.3%) ‘unavoidable’ and seven (9.3%) ‘unclassifiable’ by the review panel. For 17 of the 75 patients, there was no clear majority, and so, the classification by the registrar was chosen. The Hallas κ score was less than 0.2, again indicating only slight agreement among members of the review panel.

    Drugs implicated in ADR-related admissions

    The 75 patients with an ADR-related admission were taking a significantly higher number of drugs prior to admission compared with the non-ADR group (p<0.05), with an average of 7.5 drugs (SD=3.8). The minimum number of drugs a patient was taking was one, and the maximum number was 16. In the non-ADR group, data on number of drugs taken were available for 662 patients. The average number of drugs that patients were taking was 2.4 (SD=3.6). The maximum number of drugs a patient was taking was 20. There were a total of 349 patients not taking any drugs prior to admission, highlighting the fact that many patients were admitted due to non-drug causes, for instance, alcohol-related injuries, road traffic accidents, sports injuries, deliberate self harm and assaults.

    A total of 157 drugs, corresponding to 82 individual drugs, were considered to have caused an ADR-related admission. The drug most frequently implicated was ramipril, (an ACE inhibitor), followed by bisoprolol (a β-blocker).

    The 82 individual drugs were further analysed according to the WHO Anatomical and Therapeutic Chemical (ATC) classification,17 as presented in table 3. Two drugs were classified as herbal medicines as they did not have an ATC code. A total of 74.2% of the ADRs were attributed to drugs used to treat conditions of the cardiovascular (CV) and central nervous system (CNS).

    View this table:
    Table 3

    Drugs identified as causing an adverse drug reaction-related admission, classified by anatomical and therapeutic chemical (ATC) class17

    ADRs experienced by patients

    Drugs affecting the CV system

    In total, 73 CV drugs were associated with an ADR in 41 of the 75 patients with a suspected ADR. The most common group was diuretics (n=22), which led to patients being admitted with dehydration, hyponatraemia and acute kidney injury (AKI). The ACE inhibitor ramipril caused ADRs due to orthostatic hypotension or AKI. Other antihypertensive drugs caused bradycardia, hypotension and collapse, which possibly led to patients having a fall and being admitted with fractures.

    Aspirin (n=5) and warfarin (n=4) caused gastrointestinal (GI) bleeding and gastritis. One of the cases considered to be a ‘definite’ ADR-related admission was due to warfarin that caused an upper GI bleed in a patient. The potassium channel activator, nicorandil, caused perianal and skin ulceration in two patients. Three patients had increased levels of transaminases, as determined by liver function tests (LFTs) while taking the statins rosuvastatin and atorvastatin.

    Drugs affecting the CNS

    Fifty different CNS drugs were identified as causing an ADR in 30 of the 75 patients; the most frequent group among these was analgesics causing nausea, constipation, confusion and upper GI bleed. Nine selective serotonin reuptake inhibitors (SSRIs) were responsible for ADRs. The most common of these was hyponatraemia due to escitalopram, which occurred in seven patients. Two other patients had a seizure while taking fluoxetine and sertraline, respectively. Seven antipsychotic drugs were associated with an ADR-related admission. One patient had a ‘definite’ ADR, due to being admitted with lithium toxicity. Other antipsychotic-associated ADRs included confusion, delirium and patients having tonic-clonic seizures due to the seizure-lowering threshold effect of antipsychotics. One patient admitted with a tonic-clonic seizure had been taking both clozapine and lithium. Two patients were admitted with ataxia-type symptoms while taking antiepileptic drugs. One patient taking a monoamine oxidase inhibitor developed migraine-type symptoms.

    Other suspected ADRs experienced by patients

    One patient experienced dizziness and hypoglycaemia as a result of being on insulin and one patient experienced AKI while taking oral antidiabetic tablets including metformin.

    Eight patients were admitted due to an ADR while undergoing chemotherapy. Ten different drugs were implicated. One patient who was taking cisplatin was considered to have had a ‘definite’ ADR, while six patients had a ‘probable’ ADR and one patient had a ‘possible’ ADR. The more common ADRs experienced in these patients were fatigue, vomiting, diarrhoea and neutropenia. One patient experienced neuropathy while taking bortezomib, and another patient also experienced neuropathy with thalidomide. One patient developed thrombocytopenia while taking hydroxycarbimide.

    Discussion

    Main findings

    This study sought to determine the prevalence of ADR-related admissions to CUH in Ireland. From a sample of 856 patients, 75 patients (8.8%, 95% CI 7.0 to 10.9%) were deemed to have had an ADR-related admission.

    Patient demographics

    No difference in gender distribution was observed in the ADR and non-ADR groups, suggesting that in this study sample, gender was not a factor associated with ADR-related admissions. This finding differs from previous studies, whereby ADR-related admissions were twice as likely to occur in women as in men.3 ,18

    The patients in the ADR group were significantly older than those in the non-ADR group. This finding is in line with data from the literature.19 It is suggested that the incidence of an ADR occurring increases with age, particularly in those over 70 years of age.18 A meta-analysis study reported that ADR-related hospitalisation among the elderly is four times higher than for the non-elderly people.5

    ADRs are also common in children.1 ,3 However, in this study, the youngest patient admitted to CUH with a suspected ADR was 20 years old. The majority of children admitted during the study period, were not taking any medications prior to admission. Commonly, children were admitted with fractures due to falls while playing, or due to viral illnesses. One child was admitted with an unintentional overdose of mefloquine tablets, however, overdose is not included in WHO definition1 of an ADR, and so this child was not classed as having an ADR.

    Frequency of ADR-related admissions to CUH

    The frequency of ADR-related admissions to CUH was 8.8% (95% CI 7.0 to 10.9%). This concurs with similar studies in the literature. Information on prevalence of ADR-related admissions specifically for Ireland is relatively limited. A 2002 systematic review of prospective and retrospective studies on ADRs in hospital patients estimated a weighted mean ADR prevalence of 7.5% (95% CI 7.2 to 7.8%) for seven UK and Ireland studies.

    Overall, there is wide variation in the figures reported in the literature ranging from 0.16% to 15.7%, with an overall median of 5.3%.20

    Drugs commonly associated with ADR-related admissions

    The current study observed that patients in the ADR group were prescribed a higher mean number of drugs compared with patients in the non-ADR group. It is highlighted in the literature that the prevalence of ADRs increases with the number of drugs a patient is taking.18 The CV and CNS drugs were most commonly associated with ADR-related admissions to CUH accounting for 74.2% of admissions. This is in line with current evidence in the literature. It has been estimated that six different classes of drugs; antibiotics, anticoagulants, digoxin, diuretics, hypoglycaemics and non-steroidal anti-inflammatory drugs (NSAIDS) account for 60–70% of all hospital ADR-related admissions.18 A systematic review of preventable drug-related hospital admissions reported that aspirin accounted for 16% of admissions, diuretics 16%, NSAIDs 11% and anticoagulants 8%.21

    Instruments used to review ADRs

    Use of the Naranjo assessment tool

    All the 75 suspected ADR-related admissions were classified as ‘possible’ or stronger. The Naranjo κ score was 0.16, which indicates only slight agreement among the review panel.

    Assessment of causality is inherently difficult as, often, discriminating information that would clearly rule in or out a case is missing. Studies in the literature have reported on the difficulties associated with assessing causality and how, often, the level of agreement between raters is poor.22 ,23 While the Naranjo assessment tool is one of the generally accepted and widely used methods for causality assessment, primarily as it offers a standardised and simple methodology, it does have limitations, and many criteria were unable to be assessed, such as the inclusion of rechallenge and the use of placebo, which would not be done routinely in clinical practice. The low level of agreement using the Naranjo assessment tool in this study concurs with the experience reported elsewhere.24

    Use of the Hallas score

    Of the 75 ADRs reviewed by the panel, four (5.3%) were classified as ‘definitely avoidable’ and 39 (52%) ‘possibly avoidable’. Wide variability is reported in the literature with figures for preventability of ADRs ranging from 3.9% to 92%.8 ,9

    The Hallas κ score was less than 0.2, indicating slight agreement on the preventability of the ADRs among the review panel. Of the four patients who had an ADR-related admission that was deemed to be ‘definitely avoidable’, one patient had suffered from lithium toxicity. If she had her lithium levels checked on a regular basis, this might have allowed her lithium dose to be adjusted to prevent her presenting with symptoms of lithium toxicity. Another patient was being treated with both clozapine and lithium, and presented with tonic-clonic seizures. Antipsychotics are known to lower seizure threshold.25

    Limitations of the study

    There were a number of limitations associated with this study. Time constraints meant that recruitment was conducted for five days a week based on the pharmacy working hours, instead of a full seven-day week. Patients not included for this reason were briefly reviewed in terms of their reasons for admission, and it is not thought that their exclusion would significantly affect the generalisability of the data. There was also difficulty accessing some medical notes within the allowed time frame, and so these patients had to be excluded from the study. For some of the patients, their medical notes were incomplete, and so it was not possible to obtain all the clinical information required to determine if an ADR had taken place. The ED notes, and the patient medical notes, were used as the source of information when determining whether a patient's admission was related to an ADR; this may have introduced some information bias to the data. Although all the ADRs identified resulted in an admission to hospital, the clinical significance of the ADRs was not evaluated further; this is important work which should be conducted in the future.

    Conclusions

    During the study period, 8.8% of admissions to CUH were due to suspected ADRs. Over half (57.3%) were deemed to be possibly or definitely avoidable. The Naranjo assessment tool to assess causality was deemed suboptimal due to inflexibilities with its use. Drugs affecting the CV and CNS accounted for 74.2% of ADR-related admissions. Gender was not associated with an increased prevalence of ADR-related admissions. Patients with an ADR were prescribed a higher mean number of drugs compared with patients without an ADR. Older patients were particularly susceptible to having an ADR-related admission. Prescribers must, therefore, ensure that they use the lowest effective dose in this patient population, and regularly review and monitor treatment.

    Acknowledgments

    Ms Gail Kerrisk and Ms Mari Hughes, Pharmacy Department, Cork University Hospital, for analysis and interpretation of data collected during the study. Dr Sarah O'Connell, Infectious Diseases Registrar, Cork University Hospital, for analysis and interpretation of data collected during the study. Ms Emma Cagney and the staff of the Pharmacy Department, Cork University Hospital, for assistance with data collection during the study. Professor Stephen Cusack and the staff of the Emergency Department, Cork University Hospital, for assistance during the data collection period of the study. Fiona Ahern used the results of this study, for her thesis in part fulfilment of the MSc in Clinical Pharmacy completed through the School of Pharmacy, University College Cork, Ireland.

    References

    1. ↵
      1. Stuart MC,
      2. Kouimtzi M,
      3. Hill SR
      . WHO model formulary. Geneva: World Health Organisation, 2008.
    2. ↵
      1. Lazarou J,
      2. Pomeranz BH,
      3. Corey PN
      . Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA 1998;279:1200–5.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Vervloet D,
      2. Durham S
      . Adverse reactions to drugs. BMJ 1998;316:1511–14.
      OpenUrlFREE Full Text
    4. ↵
      1. Pirmohamed M,
      2. James S,
      3. Meakin S,
      4. et al
      . Adverse drug reactions as cause of admission to hospital: prospective analysis of 18 820 patients. BMJ 2004;329:15–19.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Beijer HJ,
      2. de Blaey CJ
      . Hospitalisations caused by adverse drug reactions (ADR): a meta-analysis of observational studies. Pharm World Sci 2002;24:46–54.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Patel H,
      2. Bell D,
      3. Molokhia M,
      4. et al
      . Trends in hospital admissions for adverse drug reactions in England: analysis of national hospital episode statistics 1998–2005. BMC Clin Pharmacol 2007;7:9.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Leendertse AJ,
      2. Visser D,
      3. Egberts AC,
      4. et al
      . The relationship between study characteristics and the prevalence of medication-related hospitalizations: a literature review and novel analysis. Drug Saf 2010;33:233–44.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Rogers S,
      2. Wilson D,
      3. Wan S,
      4. et al
      . Medication-related admissions in older people: a cross-sectional, observational study. Drugs Aging 2009;26:951–61.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Zargarzadeh AH,
      2. Emami MH,
      3. Hosseini F
      . Drug-related hospital admissions in a generic pharmaceutical system. Clin Exp Pharmacol Physiol 2007;34:494–8.
      OpenUrlCrossRefPubMedWeb of Science
    10. ↵
      Health Service Executive. Profile of Cork University Hospital. 2010 (accessed: 1st Aug 2010).
    11. ↵
      1. Naranjo CA,
      2. Busto U,
      3. Sellers EM,
      4. et al
      . A method for estimating the probability of adverse drug reactions. Clin Pharmacol Ther 1981;30:239–45.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Hamilton HJ,
      2. Gallagher PF,
      3. O'Mahony D
      . Inappropriate prescribing and adverse drug events in older people. BMC Geriatr 2009;9:5.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Zed PJ,
      2. Abu-Laban RB,
      3. Balen RM,
      4. et al
      . Incidence, severity and preventability of medication-related visits to the emergency department: a prospective study. CMAJ 2008;178:1563–9.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Hallas J,
      2. Harvald B,
      3. Gram LF,
      4. et al
      . Drug related hospital admissions: the role of definitions and intensity of data collection, and the possibility of prevention. J Intern Med 1990;228:83–90.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Hakkarainen KM,
      2. Hedna K,
      3. Petzold M,
      4. et al
      . Percentage of patients with preventable adverse drug reactions and preventability of adverse drug reactions–a meta-analysis. PLoS One 2012;7:e33236.
      OpenUrlCrossRefPubMed
    16. ↵
      1. McGinn T,
      2. Wyer PC,
      3. Newman TB,
      4. et al
      . Tips for learners of evidence-based medicine: 3. Measures of observer variability (kappa statistic). CMAJ 2004;171:1369–73.
      OpenUrlFREE Full Text
    17. ↵
      1. Anatomical Therapeutical Chemical (ATC) classification index.
      http://www.whocc.no/atc_ddd_index/.
    18. ↵
      1. Wiffen P,
      2. Gill M,
      3. Edwards J,
      4. et al
      . Adverse drug reactions in hospital patients. A systematic review of the prospective and retrospective studies. Bandolier Extra 2002; June:1–15.
    19. ↵
      1. Onder G,
      2. Petrovic M,
      3. Tangiisuran B,
      4. et al
      . Development and validation of a score to assess risk of adverse drug reactions among in-hospital patients 65 years or older: the GerontoNet ADR risk score. Arch Intern Med 2010;170:1142–8.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Kongkaew C,
      2. Noyce PR,
      3. Ashcroft DM
      . Hospital admissions associated with adverse drug reactions: a systematic review of prospective observational studies. Ann Pharmacother 2008;42:1017–25.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Howard RL,
      2. Avery AJ,
      3. Slavenburg S,
      4. et al
      . Which drugs cause preventable admissions to hospital? A systematic review. Br J Clin Pharmacol 2007;63:136–47.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Arimone Y,
      2. Miremont-Salame G,
      3. Haramburu F,
      4. et al
      . Inter-expert agreement of seven criteria in causality assessment of adverse drug reactions. Br J Clin Pharmacol 2007;64:482–8.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Arimone Y,
      2. Begaud B,
      3. Miremont-Salame G,
      4. et al
      . Agreement of expert judgment in causality assessment of adverse drug reactions. Eur J Clin Pharmacol 2005;61:169–73.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Garcia-Cortes M,
      2. Lucena MI,
      3. Pachkoria K,
      4. et al
      . Evaluation of naranjo adverse drug reactions probability scale in causality assessment of drug-induced liver injury. Aliment Pharmacol Ther 2008;27:780–9.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      British Medical Association. Royal pharmaceutical society of Great Britain: British National Formulary. 59th edn. London: Pharmaceutical Press, 2010.

    Footnotes

    • Contributors All authors contributed to the design of the study. FA collected the data for the study. FA, LS and SM were involved in the analyses of the data. All authors contributed to the preparation of the manuscript.

    • Competing interests None.

    • Ethics approval Clinical Research Ethics Committee of the Cork Teaching Hospitals.

    • Provenance and peer review Not commissioned; externally peer reviewed..