TY - JOUR
AU - Barber, Nick
AB - Abstract Objectives The aim was to adapt a US adverse drug event (ADE) trigger tool for UK use, and to establish its positive predictive value (PPV) and sensitivity in comparison to retrospective health record review for the identification of preventable ADEs, in a pilot study on one hospital ward. Methods An established US trigger tool was adapted for UK use. We applied it retrospectively to 207 patients' health records, following up positive triggers to identify any ADEs (both preventable and non-preventable). We compared the preventable ADEs to those identified using full health record review. Key findings We identified 168 positive triggers in 127 (61%) of 207 patients. Seven ADEs were identified, representing an ADE in 3.4% of patients or 0.7 ADEs per 100 patient days. Five were non-preventable adverse drug reactions and two were due to preventable errors. The prevalence of preventable ADEs was 1.0% of patients, or 0.2 per 100 patient days. The overall PPV was 0.04 for all ADEs, and 0.01 for preventable ADEs. PPVs for individual triggers varied widely. Five preventable ADEs were identified using health record review. The sensitivity of the trigger tool for identifying preventable ADEs was 0.40, when compared to health record review. Conclusions Although we identified some ADEs using the trigger tool, more work is needed to further refine the trigger tool to reduce the false positives and increase sensitivity. To comprehensively identify preventable ADEs, retrospective health record review remains the gold standard and we found no efficiency gain in using the trigger tool. adverse drug events, prescribing errors, trigger tool Introduction It is difficult and time-consuming to study the prevalence of medication-related harm. In general, patients' health records need to be read in detail to identify evidence of harm. This approach is not feasible for routine monitoring and is too expensive for many research studies. As a potential alternative, various ‘trigger tools’ have been developed and used, particularly in the USA. These are collections of triggers such as abnormal laboratory values and antidote prescriptions; a positive trigger then leads to more extensive investigation to identify whether or not harm has occurred.[1] Various trigger tools, for both prospective and retrospective application, and for both computerised and manual use, have been developed for medication-related harm (adverse drug events; ADEs) in particular[2–4] and for adverse events in general.[1,5,6] Others are intended for the identification of adverse drug reactions (ADRs) alone.[7–12] Such methods are now becoming widely advocated for use outside of the USA, for example by the English Patient Safety First campaign (www.patientsafetyfirst.nhs.uk). However, while there is significant experience with such tools in the USA, studies of their use in UK hospitals are published only in abstract form,[13–15] and it is not known how applicable or practical they are for use in the UK setting. None of the papers included in a recent systematic review[16] were from the UK and the only non-US studies examined ADRs alone and excluded other types of medication-related harm. There has also been little work to explore the validity of trigger tools. Several studies have reported the ADEs identified by trigger tools but not by other methods,[2,7] but few have studied the ADEs identified by other methods but not by the trigger tool. Finally, most studies conducted to date have focused on all ADEs or ADRs alone; few have specifically examined preventable ADEs, which are arguably more important. Our aims were to adapt an ADE trigger tool from the USA for use in the UK, and to test its sensitivity in comparison to case note review for the identification of preventable ADEs in a pilot study on one hospital ward. We also calculated the positive predictive value (PPV) of the tool overall, and for each of the individual triggers, for the identification both of all ADEs and of preventable ADEs alone. Methods Adapting trigger tool methodology for use in the UK We adapted a published US trigger tool[2] for UK use, as described elsewhere.[17] Briefly, reference ranges were changed to reflect UK units, and some drugs changed to reflect UK practice. The 23 original US triggers plus 23 proposed UK equivalents were then sent to six experts (three clinical pharmacologists, two clinical pharmacists and a medication safety expert at the UK National Patient Safety Agency) for comment and approval. Some minor comments were incorporated and the final UK version agreed. The tool was designed for manual use during the retrospective review of patients' health records.[2] Definitions An ADE was defined as any harm caused by medication use, where ‘harm’ was defined very broadly as any identifiable physiological or physical changes. We did not look for ADEs that led to admission; only those occurring in the inpatient setting were included. ADEs were considered to be preventable if they resulted from a medication error,[18] or non-preventable if they did not (and were thus attributed to an ADR). We defined two types of medication error: prescribing errors and medication administration errors. A prescribing error was defined, according to an established definition, as a prescribing decision or prescription-writing process that results in an unintentional, significant (i) reduction in the probability of treatment being timely and effective or (ii) increase in the risk of harm, when compared to generally accepted practice.[19] The definition is accompanied by lists of events that should and should not be included as prescribing errors,[19] and includes errors originating in both the prescribing decision and in prescription writing. Medication administration errors were defined as any difference between the medication ordered, including any pharmacists' endorsements, and that administered to the patient.[20] Medication administration errors could arise from dispensing errors as well as errors made by nursing staff during medication preparation and administration. An ADR was defined according to the World Health Organization as any response to a drug which is noxious and unintended and that occurs at doses used in humans for prophylaxis, diagnosis or therapy.[21] Setting and subjects Data were collected retrospectively on a complex surgical ward of a London teaching hospital during two 4-week periods. The first was about 3 months before the introduction of an electronic prescribing system (April 2003) and the second was 6 months afterwards (November/December 2003), as we were concurrently testing this, as well as other methods, for the evaluation of electronic prescribing.[17] However, data from both periods have been combined for the present paper as we would not expect the change to affect the performance of the trigger tool. Where patients were on the study ward prior to the beginning of the relevant period, or remained on the ward at the end of that period, only those medication orders written within the study period were included. The study hospital used paper-based multidisciplinary health records, which included paper-based drug charts for inpatient admissions, plus electronic results for all laboratory tests. For patients admitted to the study ward following the implementation of electronic prescribing, paper print-outs of the medication prescribed and administered were filed in the health records and could be examined retrospectively. Ethical approval was obtained from Riverside Research Ethics Committee. Data collection Retrieval of health records and all data collection took place between September 2004 and January 2005. As described previously,[17] inpatients who were on the study ward at any time during either 4-week data-collection period (April 2003 and November/December 2003) were identified retrospectively from the ward's admission book and their health records retrieved from the health records library. Where patients' health records were not initially available, repeated attempts were made to retrieve them. For each patient whose records were retrieved, the same research pharmacist completed the trigger tool using their paper health records plus electronic laboratory data. For each positive trigger, the health records were reviewed in more detail to identify any associated ADEs. The researcher also recorded the time taken to complete each trigger tool review and each full health record review. The ADEs identified were classified by the researchers into preventable ADEs (further subdivided into those resulting from prescribing errors and those resulting from medication administration errors) and non-preventable ADEs (those resulting from ADRs). This classification was done independently by two researchers and any disagreements resolved by discussion. Comparison with health record review Before completing the trigger tool, the research pharmacist also completed a full health record review for each patient whose health records were retrieved, studying the drug charts in particular detail for the medication orders written and any medication errors that occurred during the study period. A retrospective review form was used to guide this process.[17] Laboratory data were examined only if considered relevant in relation to the patient's medication or clinical condition, and only the relevant parameters checked. Details were recorded of any medication errors identified, together with whether or not they resulted in harm. The health record review focused on the identification of preventable ADEs only; we did not look for ADRs using this method. Analysis We calculated the overall sensitivity of the trigger tool for identifying preventable ADEs, when compared with case note review. Sensitivity was defined as: number of patients where the trigger tool identified a true positive preventable ADE/(number of patients where trigger tool identified a true positive preventable ADE plus the number for whom the trigger tool gave a false negative). We also calculated the PPV for the tool overall, and for each individual trigger, for identifying ADEs in general and for preventable ADEs in particular. PPV was defined as: the number of patients in whom a positive trigger indicated a true positive ADE/(number of patients in whom a positive trigger indicated a true positive ADE plus number of patients where a positive trigger did not indicate an ADE). Results The final version of the UK trigger tool consisted of 23 triggers (Table 1). There was a total of 276 patients listed in the ward's admission book for the two study periods. For 11 patients there was no record of admission to the study ward or to the hospital in their health records. For another six, while a record of admission existed, there was insufficient information to be able to conduct a review. Finally, for another 52 patients, the health records could not be retrieved as they were booked out to other clinical areas and could not readily be located. We therefore retrospectively retrieved and reviewed the health records for 207 (75%) of the 276 patients. In 36 of these cases (17%) no laboratory results were available relating to the study period. However, these patients were still included. In total, 939 patient days fell within the study period for the 207 patients studied. Table 1 Numbers of positive triggers, their positive predictive values (PPVs) and adverse drug events (ADEs) identified Code . UK trigger . Potential problem identified . Positive trigger (% of patients) . Total ADEs (PPV) . ADRs (PPV) . Preventable ADEs (PPV) . Data from medication chart T1 Chlorphenamine/loratadine/hydrocortisone Hypersensitivity reaction 9 (4%) 2 (0.22) 2 (0.22) 0 (0) T2 Vitamin K (phytomenadione) Over-anticoagulation with warfarin 2 (1%) 0 (0) 0 (0) 0 (0) T3 Flumazenil Over-sedation with benzodiazepines 0 0 0 0 T4 Anti-emeticsa Nausea/emesis related to drug use 101 (46%) 1 (0.01) 1 (0.01) 0 (0) T5 Naloxone Over-sedation with narcotic 0 0 0 0 T6 Anti-diarrhoealsb ADE 36 (16%) 0 (0) 0 (0) 0 (0) T7 Calcium resonium Hyperkalaemia due to renal impairment/drug 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T22 Unexpected medication stop ADE 2 (1%) 2 (1.00) 2 (1.00) 0 (0) Data from health records T20 Over-sedation, lethargy, falls, hypotension Related to overuse of medication 3 (1%) 0 (0) 0 (0) 0 (0) T21 Rash Drug-related/ADE 3 (1%) 2 (0.67) 2 (0.67) 0 (0) T23 Transfer to higher level of care, such as critical care area Adverse event 2 (1%) 0 (0) 0 (0) 0 (0) Biochemical/haematological/microbiological data T8 Activated partial thromboplastin time ratio >3 Over-anticoagulation with heparin 2 (1%) 2 (1.00) 2 (1.00) 0 T9 International normalized ratio >6 Over-anticoagulation with warfarin 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T10 White blood count <3×109/l Neutropenia related to drug or disease 1 (0.5%) 0 (0) 0 (0) 0 (0) T11 Serum glucose <2.8 mmol/l Related to insulin use or oral antidiabetics 0 0 0 0 T12 Rising serum creatinine Renal insufficiency related to drug use 4 (2%) 0 (0) 0 (0) 0 (0) T13 Clostridium difficile-positive stool Exposure to antibiotics 1 (0.5%) 1 (1.00) 1 (1.00) 0 T14 Digoxin level >2 µg/l Toxic digoxin level 0 0 0 0 T15 Lidocaine level >5 µg/ml Toxic lidocaine level 0 0 0 0 T16 Gentamicin/tobramycin levels peak >10 mg/l, trough >2 mg/l Toxic levels of antibiotics 0 0 0 0 T17 Amikacin levels peak >30 mg/l, trough >10 mg/l Toxic levels of antibiotics 0 0 0 0 T18 Vancomycin level >26 mg/l Toxic levels of antibiotics 0 0 0 0 T19 Theophylline level >20 mg/l Toxic levels of drugs 0 0 0 0 Code . UK trigger . Potential problem identified . Positive trigger (% of patients) . Total ADEs (PPV) . ADRs (PPV) . Preventable ADEs (PPV) . Data from medication chart T1 Chlorphenamine/loratadine/hydrocortisone Hypersensitivity reaction 9 (4%) 2 (0.22) 2 (0.22) 0 (0) T2 Vitamin K (phytomenadione) Over-anticoagulation with warfarin 2 (1%) 0 (0) 0 (0) 0 (0) T3 Flumazenil Over-sedation with benzodiazepines 0 0 0 0 T4 Anti-emeticsa Nausea/emesis related to drug use 101 (46%) 1 (0.01) 1 (0.01) 0 (0) T5 Naloxone Over-sedation with narcotic 0 0 0 0 T6 Anti-diarrhoealsb ADE 36 (16%) 0 (0) 0 (0) 0 (0) T7 Calcium resonium Hyperkalaemia due to renal impairment/drug 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T22 Unexpected medication stop ADE 2 (1%) 2 (1.00) 2 (1.00) 0 (0) Data from health records T20 Over-sedation, lethargy, falls, hypotension Related to overuse of medication 3 (1%) 0 (0) 0 (0) 0 (0) T21 Rash Drug-related/ADE 3 (1%) 2 (0.67) 2 (0.67) 0 (0) T23 Transfer to higher level of care, such as critical care area Adverse event 2 (1%) 0 (0) 0 (0) 0 (0) Biochemical/haematological/microbiological data T8 Activated partial thromboplastin time ratio >3 Over-anticoagulation with heparin 2 (1%) 2 (1.00) 2 (1.00) 0 T9 International normalized ratio >6 Over-anticoagulation with warfarin 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T10 White blood count <3×109/l Neutropenia related to drug or disease 1 (0.5%) 0 (0) 0 (0) 0 (0) T11 Serum glucose <2.8 mmol/l Related to insulin use or oral antidiabetics 0 0 0 0 T12 Rising serum creatinine Renal insufficiency related to drug use 4 (2%) 0 (0) 0 (0) 0 (0) T13 Clostridium difficile-positive stool Exposure to antibiotics 1 (0.5%) 1 (1.00) 1 (1.00) 0 T14 Digoxin level >2 µg/l Toxic digoxin level 0 0 0 0 T15 Lidocaine level >5 µg/ml Toxic lidocaine level 0 0 0 0 T16 Gentamicin/tobramycin levels peak >10 mg/l, trough >2 mg/l Toxic levels of antibiotics 0 0 0 0 T17 Amikacin levels peak >30 mg/l, trough >10 mg/l Toxic levels of antibiotics 0 0 0 0 T18 Vancomycin level >26 mg/l Toxic levels of antibiotics 0 0 0 0 T19 Theophylline level >20 mg/l Toxic levels of drugs 0 0 0 0 * ADE, adverse drug event; ADR, adverse drug reaction; PPV, positive predicitve value. a Anti-emetics: droperidol, ondansetron, promethazine, hydroxyzine, prochlorperazine, metoclopramide, cyclizine, granisetron or domperidone. b Anti-diarrhoeals: loperamide, diphenoxylate, codeine or co-phenotrope. Open in new tab Table 1 Numbers of positive triggers, their positive predictive values (PPVs) and adverse drug events (ADEs) identified Code . UK trigger . Potential problem identified . Positive trigger (% of patients) . Total ADEs (PPV) . ADRs (PPV) . Preventable ADEs (PPV) . Data from medication chart T1 Chlorphenamine/loratadine/hydrocortisone Hypersensitivity reaction 9 (4%) 2 (0.22) 2 (0.22) 0 (0) T2 Vitamin K (phytomenadione) Over-anticoagulation with warfarin 2 (1%) 0 (0) 0 (0) 0 (0) T3 Flumazenil Over-sedation with benzodiazepines 0 0 0 0 T4 Anti-emeticsa Nausea/emesis related to drug use 101 (46%) 1 (0.01) 1 (0.01) 0 (0) T5 Naloxone Over-sedation with narcotic 0 0 0 0 T6 Anti-diarrhoealsb ADE 36 (16%) 0 (0) 0 (0) 0 (0) T7 Calcium resonium Hyperkalaemia due to renal impairment/drug 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T22 Unexpected medication stop ADE 2 (1%) 2 (1.00) 2 (1.00) 0 (0) Data from health records T20 Over-sedation, lethargy, falls, hypotension Related to overuse of medication 3 (1%) 0 (0) 0 (0) 0 (0) T21 Rash Drug-related/ADE 3 (1%) 2 (0.67) 2 (0.67) 0 (0) T23 Transfer to higher level of care, such as critical care area Adverse event 2 (1%) 0 (0) 0 (0) 0 (0) Biochemical/haematological/microbiological data T8 Activated partial thromboplastin time ratio >3 Over-anticoagulation with heparin 2 (1%) 2 (1.00) 2 (1.00) 0 T9 International normalized ratio >6 Over-anticoagulation with warfarin 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T10 White blood count <3×109/l Neutropenia related to drug or disease 1 (0.5%) 0 (0) 0 (0) 0 (0) T11 Serum glucose <2.8 mmol/l Related to insulin use or oral antidiabetics 0 0 0 0 T12 Rising serum creatinine Renal insufficiency related to drug use 4 (2%) 0 (0) 0 (0) 0 (0) T13 Clostridium difficile-positive stool Exposure to antibiotics 1 (0.5%) 1 (1.00) 1 (1.00) 0 T14 Digoxin level >2 µg/l Toxic digoxin level 0 0 0 0 T15 Lidocaine level >5 µg/ml Toxic lidocaine level 0 0 0 0 T16 Gentamicin/tobramycin levels peak >10 mg/l, trough >2 mg/l Toxic levels of antibiotics 0 0 0 0 T17 Amikacin levels peak >30 mg/l, trough >10 mg/l Toxic levels of antibiotics 0 0 0 0 T18 Vancomycin level >26 mg/l Toxic levels of antibiotics 0 0 0 0 T19 Theophylline level >20 mg/l Toxic levels of drugs 0 0 0 0 Code . UK trigger . Potential problem identified . Positive trigger (% of patients) . Total ADEs (PPV) . ADRs (PPV) . Preventable ADEs (PPV) . Data from medication chart T1 Chlorphenamine/loratadine/hydrocortisone Hypersensitivity reaction 9 (4%) 2 (0.22) 2 (0.22) 0 (0) T2 Vitamin K (phytomenadione) Over-anticoagulation with warfarin 2 (1%) 0 (0) 0 (0) 0 (0) T3 Flumazenil Over-sedation with benzodiazepines 0 0 0 0 T4 Anti-emeticsa Nausea/emesis related to drug use 101 (46%) 1 (0.01) 1 (0.01) 0 (0) T5 Naloxone Over-sedation with narcotic 0 0 0 0 T6 Anti-diarrhoealsb ADE 36 (16%) 0 (0) 0 (0) 0 (0) T7 Calcium resonium Hyperkalaemia due to renal impairment/drug 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T22 Unexpected medication stop ADE 2 (1%) 2 (1.00) 2 (1.00) 0 (0) Data from health records T20 Over-sedation, lethargy, falls, hypotension Related to overuse of medication 3 (1%) 0 (0) 0 (0) 0 (0) T21 Rash Drug-related/ADE 3 (1%) 2 (0.67) 2 (0.67) 0 (0) T23 Transfer to higher level of care, such as critical care area Adverse event 2 (1%) 0 (0) 0 (0) 0 (0) Biochemical/haematological/microbiological data T8 Activated partial thromboplastin time ratio >3 Over-anticoagulation with heparin 2 (1%) 2 (1.00) 2 (1.00) 0 T9 International normalized ratio >6 Over-anticoagulation with warfarin 1 (0.5%) 1 (1.00) 0 (0) 1 (1.00) T10 White blood count <3×109/l Neutropenia related to drug or disease 1 (0.5%) 0 (0) 0 (0) 0 (0) T11 Serum glucose <2.8 mmol/l Related to insulin use or oral antidiabetics 0 0 0 0 T12 Rising serum creatinine Renal insufficiency related to drug use 4 (2%) 0 (0) 0 (0) 0 (0) T13 Clostridium difficile-positive stool Exposure to antibiotics 1 (0.5%) 1 (1.00) 1 (1.00) 0 T14 Digoxin level >2 µg/l Toxic digoxin level 0 0 0 0 T15 Lidocaine level >5 µg/ml Toxic lidocaine level 0 0 0 0 T16 Gentamicin/tobramycin levels peak >10 mg/l, trough >2 mg/l Toxic levels of antibiotics 0 0 0 0 T17 Amikacin levels peak >30 mg/l, trough >10 mg/l Toxic levels of antibiotics 0 0 0 0 T18 Vancomycin level >26 mg/l Toxic levels of antibiotics 0 0 0 0 T19 Theophylline level >20 mg/l Toxic levels of drugs 0 0 0 0 * ADE, adverse drug event; ADR, adverse drug reaction; PPV, positive predicitve value. a Anti-emetics: droperidol, ondansetron, promethazine, hydroxyzine, prochlorperazine, metoclopramide, cyclizine, granisetron or domperidone. b Anti-diarrhoeals: loperamide, diphenoxylate, codeine or co-phenotrope. Open in new tab ADEs identified using the trigger tool A total of 168 positive triggers was identified in 127 (61%) of the 207 patients. Table 1 summarises the positive responses for each of the 23 triggers, and indicates those that related to ADEs. A total of seven ADEs was identified (Table 2), representing an ADE in 3.4% of patients or 0.7 ADEs per 100 patient days. Five were ADRs, and two were prescribing errors. No medication administration errors resulting in harm were identified using the trigger tool. The prevalence of harm from preventable ADEs (those due to medication prescribing errors) was 1.0% of patients, or 0.2 per 100 patient days. Table 2 Summary of the adverse drug events identified . Type of adverse drug event . Identified using trigger tool . Positive triggers . Identified using health record review . Rash following prescription for norfloxacin, requiring chlorphenamine ADR Yes T1, T21 and T22 NA Rash following prescription for norfloxacin ADR Yes T21 and T22 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA Nausea and vomiting following morphine administration ADR Yes T4 NA Hypotension in patient prescribed regular doxazosin ADR Yes T20 NA Patient developed Clostridium difficile-positive stool after several days treatment with intravenous cefuroxime ADR Yes T13 NA Patient prescribed a total of 120 mmol potassium in intravenous fluids over a 3-day period, without checking a recent serum potassium level; this resulted in serum potassium of 4.6 mmol/l (desired range 3.5–5 mmol/l) on day 2, and 6.7 mmol/l by day 3, which required treatment with calcium resonium PE Yes T7 Yes A patient on warfarin with a target INR of 2–3 had an INR of 3.4, and then was prescribed ciprofloxacin 500 mg twice daily with no reduction in warfarin dose; the enhanced anticoagulant effect resulted in an INR of 6.1 the following day PE Yes T9 Yes A patient's usual ferrous sulphate tablets were not prescribed on admission, resulting in their haemoglobin dropping from 10.9 to 9.8 g/dl PE No NA Yes Patient usually took moxonidine 300 µg twice daily, which was not prescribed on admission; their blood pressure increased from 111/77 mmHg on admission to 215/120 mmHg the following day, when the error was identified and the drug prescribed PE No NA Yes Diltiazem XL 180 mg OD not given on two consecutive days; the patient complained of palpitations which resolved once a dose was given later on the second day MAE No NA Yes . Type of adverse drug event . Identified using trigger tool . Positive triggers . Identified using health record review . Rash following prescription for norfloxacin, requiring chlorphenamine ADR Yes T1, T21 and T22 NA Rash following prescription for norfloxacin ADR Yes T21 and T22 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA Nausea and vomiting following morphine administration ADR Yes T4 NA Hypotension in patient prescribed regular doxazosin ADR Yes T20 NA Patient developed Clostridium difficile-positive stool after several days treatment with intravenous cefuroxime ADR Yes T13 NA Patient prescribed a total of 120 mmol potassium in intravenous fluids over a 3-day period, without checking a recent serum potassium level; this resulted in serum potassium of 4.6 mmol/l (desired range 3.5–5 mmol/l) on day 2, and 6.7 mmol/l by day 3, which required treatment with calcium resonium PE Yes T7 Yes A patient on warfarin with a target INR of 2–3 had an INR of 3.4, and then was prescribed ciprofloxacin 500 mg twice daily with no reduction in warfarin dose; the enhanced anticoagulant effect resulted in an INR of 6.1 the following day PE Yes T9 Yes A patient's usual ferrous sulphate tablets were not prescribed on admission, resulting in their haemoglobin dropping from 10.9 to 9.8 g/dl PE No NA Yes Patient usually took moxonidine 300 µg twice daily, which was not prescribed on admission; their blood pressure increased from 111/77 mmHg on admission to 215/120 mmHg the following day, when the error was identified and the drug prescribed PE No NA Yes Diltiazem XL 180 mg OD not given on two consecutive days; the patient complained of palpitations which resolved once a dose was given later on the second day MAE No NA Yes ADR, adverse drug reaction; APTT, activated partial thromboplastin time; INR, international normalised ratio; MAE, medication administration error; NA, not applicable; OD, once daily; PE, prescribing error; T, trigger number (see Table 1). Open in new tab Table 2 Summary of the adverse drug events identified . Type of adverse drug event . Identified using trigger tool . Positive triggers . Identified using health record review . Rash following prescription for norfloxacin, requiring chlorphenamine ADR Yes T1, T21 and T22 NA Rash following prescription for norfloxacin ADR Yes T21 and T22 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA Nausea and vomiting following morphine administration ADR Yes T4 NA Hypotension in patient prescribed regular doxazosin ADR Yes T20 NA Patient developed Clostridium difficile-positive stool after several days treatment with intravenous cefuroxime ADR Yes T13 NA Patient prescribed a total of 120 mmol potassium in intravenous fluids over a 3-day period, without checking a recent serum potassium level; this resulted in serum potassium of 4.6 mmol/l (desired range 3.5–5 mmol/l) on day 2, and 6.7 mmol/l by day 3, which required treatment with calcium resonium PE Yes T7 Yes A patient on warfarin with a target INR of 2–3 had an INR of 3.4, and then was prescribed ciprofloxacin 500 mg twice daily with no reduction in warfarin dose; the enhanced anticoagulant effect resulted in an INR of 6.1 the following day PE Yes T9 Yes A patient's usual ferrous sulphate tablets were not prescribed on admission, resulting in their haemoglobin dropping from 10.9 to 9.8 g/dl PE No NA Yes Patient usually took moxonidine 300 µg twice daily, which was not prescribed on admission; their blood pressure increased from 111/77 mmHg on admission to 215/120 mmHg the following day, when the error was identified and the drug prescribed PE No NA Yes Diltiazem XL 180 mg OD not given on two consecutive days; the patient complained of palpitations which resolved once a dose was given later on the second day MAE No NA Yes . Type of adverse drug event . Identified using trigger tool . Positive triggers . Identified using health record review . Rash following prescription for norfloxacin, requiring chlorphenamine ADR Yes T1, T21 and T22 NA Rash following prescription for norfloxacin ADR Yes T21 and T22 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA High APTT ratio for patient on heparin infusion; infusion had been stopped and dose reduced by nursing staff as appropriate ADR Yes T8 NA Nausea and vomiting following morphine administration ADR Yes T4 NA Hypotension in patient prescribed regular doxazosin ADR Yes T20 NA Patient developed Clostridium difficile-positive stool after several days treatment with intravenous cefuroxime ADR Yes T13 NA Patient prescribed a total of 120 mmol potassium in intravenous fluids over a 3-day period, without checking a recent serum potassium level; this resulted in serum potassium of 4.6 mmol/l (desired range 3.5–5 mmol/l) on day 2, and 6.7 mmol/l by day 3, which required treatment with calcium resonium PE Yes T7 Yes A patient on warfarin with a target INR of 2–3 had an INR of 3.4, and then was prescribed ciprofloxacin 500 mg twice daily with no reduction in warfarin dose; the enhanced anticoagulant effect resulted in an INR of 6.1 the following day PE Yes T9 Yes A patient's usual ferrous sulphate tablets were not prescribed on admission, resulting in their haemoglobin dropping from 10.9 to 9.8 g/dl PE No NA Yes Patient usually took moxonidine 300 µg twice daily, which was not prescribed on admission; their blood pressure increased from 111/77 mmHg on admission to 215/120 mmHg the following day, when the error was identified and the drug prescribed PE No NA Yes Diltiazem XL 180 mg OD not given on two consecutive days; the patient complained of palpitations which resolved once a dose was given later on the second day MAE No NA Yes ADR, adverse drug reaction; APTT, activated partial thromboplastin time; INR, international normalised ratio; MAE, medication administration error; NA, not applicable; OD, once daily; PE, prescribing error; T, trigger number (see Table 1). Open in new tab The overall PPV was 0.04 for all ADEs and 0.01 for preventable ADEs. PPVs for the 23 individual triggers, for both ADEs and preventable ADEs, ranged from 0 to 1.0 (Table 1). Five triggers had PPVs of 1.0; these were calcium resonium, unexpected medication stop, activated partial thromboplastin time (APTT) ratio of more than 3.0, international normalised ratio (INR) of more than 6.0, and Clostridium difficile diarrhoea. However, since most triggers did not identify any ADEs, the PPV of the majority was zero. Nine triggers were not positive in any patient. Of the 168 positive triggers, 137 (82%) were the result of two triggers and led to the identification of only one (non-preventable) ADE. These were ‘prescription for anti-emetics’ and ‘prescription for anti-diarrhoeals’. It was noted that in most cases the anti-emetics were prescribed to be given ‘when required’, and no doses were actually given. Many of the positive ‘anti-diarrhoeals’ triggers were the result of codeine prescribed as an analgesic. Comparison with health record review Health record review identified four prescribing errors and one medication administration error (Table 2). The prevalence of preventable ADEs identified using this method was 2.4% of patients, or 0.5 per 100 patient days. The sensitivity of the trigger tool for identifying preventable ADEs was 0.4, when compared to health record review. Application of the trigger tool required an average of 4 min per patient and each full record review an average of 44 min. Discussion Main findings In a pilot study on one UK hospital ward, we identified 0.7 ADEs per 100 bed days, and 0.2 preventable ADEs per 100 patient days, using a UK adaptation of a US ADE trigger tool. The sensitivity of the trigger tool for identifying preventable ADEs was 0.40, when compared to health record review. There were many false positive triggers; overall PPVs were only 0.04 for all ADEs and 0.01 for preventable ADEs. PPVs for individual triggers varied widely; many had a PPV of zero; five had a PPV of 1.0 for all ADEs. Strengths and limitations While only a pilot study, this is the first UK study to test a trigger tool in detail in the secondary care setting, and the first to compare a trigger tool with another method of studying preventable ADEs. This is also one of the first studies to test in detail a trigger tool for identifying all types of ADE rather than just ADRs. We have thus identified the performance characteristics of a trigger tool for identifying preventable ADEs, and can suggest ways in which it could be improved. There are several limitations. We used a relatively small sample of patients, in one ward, and it is therefore not known how generalisable our results will be to other clinical specialties. Our sampling approach, based on a four-week period, meant that data collected for some of our patients were truncated, preventing calculation of ADE rates per patient admission. There are known limitations associated with retrospective health record review; the incidence of certain types of errors, such as medication administration errors, will be under-estimated. We did not explore inter-rater reliability for either the health record review or the trigger tool. Finally, the health record review was not conducted independently to the trigger tool. Our researcher completed the health records review first, so that completion of the trigger tool immediately prior to completing the less-structured health record review, leading to the ADEs identified using the trigger tool being foremost in the researcher's mind, would not bias data collection. Comparison with previous work The prevalence of ADEs and the PPVs are similar to the lower end of the range of others' findings. We discovered slightly fewer ADEs, including preventable ADEs, than other researchers in the field. Previous studies using similar trigger tools have identified 1.3–2.8 ADEs per 100 patient days,[10,22–24] and 0.3–0.4 preventable ADEs per 100 patient days.[22,24] Our figures, 0.7 and 0.2, respectively, are therefore slightly lower than those previously reported. Other studies present ADE rates per 1000 doses of medication,[2] or per 100 admissions[2,4,23] and cannot directly be compared to our figures. Overall PPVs for ADE trigger tools in the literature range from 0.04 to 0.23.[2,22–24] Our PPV, at 0.04, was again at the lower end of this range. There are many potential reasons for our ADE rates, and thus PPVs, being lower. First, the combinations of triggers used in the literature vary considerably, as highlighted recently,[16] as do definitions of ADEs and preventability. Second, we studied a single surgical ward in one hospital. There may be differences in underlying ADE rates in different settings; for example, a study in 86 US hospitals suggests variation in ADE rates for different groups of hospitals.[2] Finally, we did not have any laboratory results for 17% of patients, which may mean some ADEs were not detectable from our sample. For preventable ADEs, we report a sensitivity of 0.40 when compared to health record review. This is very similar to the figure of 0.42 reported in the US literature[24] for preventable ADEs, when compared to chart review plus stimulated voluntary report. A study in Korea using a more complex trigger tool linked to patient diagnoses reported a higher sensitivity of 0.79 for all ADEs (both preventable and non-preventable) when compared to retrospective health record review.[23] Implications It is important to be able to identify the avoidable harm caused by medicines, as this is something which can then be targeted by management strategies. A fundamental tenet of creating a safer system is that the unsafe aspects of the system must be measurable, so that the effectiveness of any change can be assessed. Preventable ADEs are relatively uncommon as a proportion of prescribing and administration acts, but the sheer volume of prescribing and administration of medicines means they are a significant cause of morbidity and mortality.[25] The question is: how can we best detect preventable ADEs in such a way that they can be measured in routine settings as part of a quality improvement cycle? Spontaneous reporting captures a negligible proportion of events; health record review captures most, if not all, but requires a competent reviewer and is very resource-intensive. Is a trigger tool the answer? It is well documented that trigger tools identify more events than routine spontaneous reporting systems.[22,26] However, in comparison to retrospective review of patients' health records, the sensitivity for identifying preventable ADEs was only 0.40 in our study. There are several reasons why this may be the case. First, there are many types of preventable ADE that will not be identified using the triggers used. These include non-prescription of medication that was clinically indicated, such as thromboprophylaxis, the omission of which can lead to serious patient harm. Second, triggers relating to drug levels will not identify harm if no levels have been taken. Third, advances in clinical practice mean that some triggers may now be less relevant. For example, the aminoglycoside-related triggers are unlikely to be relevant where once-daily dosing is used, as is current practice in most UK clinical settings. In our study, application of the trigger tool required only 4 min per patient; however, this was done following the health record review and so the researcher was already familiar with the information in each patient's set of health records. It would be expected that it would take considerably longer if it was done alone. It has been estimated that it takes about 20 min to complete a trigger tool review in isolation for one set of health records,[2] which equates to 69 h for 207 patients. In our study, it took an average of 44 min for each full health record review: 151.8 h for 207 patients. We therefore estimate that it takes 34.5 h to identify each preventable ADE using the trigger tool, and 30.4 h to identify each preventable ADE using retrospective health record review; there are thus no efficiency benefits in using the trigger tool. The overall PPV was low in our study, and 61% of patients had at least one positive trigger, requiring a more detailed review of their health records. This could be considerably improved if two triggers – anti-diarrhoeals and anti-emetics – were excluded, or if anti-emetics were considered only if actually administered. Excluding these triggers would have resulted in a PPV of 0.06 instead of 0.01 for preventable ADEs, and a PPV of 0.19 instead of 0.04 for all ADEs. At present the limited ability of the trigger tool to detect preventable ADEs, and its ability to miss whole classes of preventable ADEs, does not make it an attractive alternative to record review (which may also be used to detect other causes of error and avoidable harm). Honing the triggers, and using the advent of electronic prescribing and electronic administration records to permit automated prospective screening,[24] could lead to a trigger tool that was relatively quick and efficient. There would need to be a regular review of triggers to ensure a tool's usefulness, and it would need calibration against record review at least biannually to identify what it was missing, and identify potential new triggers. Future research should also explore inter-rater reliability. Conclusions This is the first UK study to test a trigger tool in detail in the secondary care setting, and the first to compare with another method of studying preventable ADEs. In a pilot study on one UK hospital ward, we identified 0.7 ADEs per 100 bed days, and 0.2 preventable ADEs per 100 bed days. The overall PPV was 0.04 for all ADEs and 0.01 for preventable ADEs. Removal of two triggers would dramatically increase the efficiency of the tool; more work is needed to elucidate the optimum combination of triggers for the UK secondary care setting. The sensitivity of the trigger tool for identifying preventable ADEs was only 0.40, when compared to health record review. To comprehensively identify all preventable ADEs, health record review remains the gold standard. Declarations Conflicts of interest The Author(s) declare(s) that they have no conflicts of interest to disclose. Funding The research was supported by the Department of Health's Patient Safety Research Programme. Acknowledgements The Centre for Medication Safety and Service Quality is affiliated with the Centre for Patient Safety and Service Quality at Imperial College Healthcare NHS Trust which is funded by the National Institute of Health Research. We acknowledge the help of Matthew Reynolds for assistance with data analysis. This paper has not been, and will not be, published in whole or in part in any other journal. The data relating specifically to the prescribing errors detected using both the trigger tool and retrospective review of medical notes were previously included in [17]. References 1 Resar RK et al. Methodology and rationale for the measurement of harm with trigger tools . Qual Saf Health Care 2003 ; 12 (suppl. 2 ): ii39 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 2 Rozich JD et al. Adverse drug event trigger tool: a practical methodology for measuring medication-related harm . Qual Saf Health Care 2003 ; 12 : 194 – 200 . Google Scholar Crossref Search ADS PubMed WorldCat 3 McMullin ST et al. Automated system for identifying potential dosage problems at a large university hospital . Am J Health Syst Pharm 1997 ; 54 : 545 – 549 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Kilbridge PM et al. Implementation of a system for computerized adverse drug event surveillance and intervention at an academic medical center . J Clin Outcomes Manag 2006 ; 13 : 94 – 100 . Google Scholar OpenURL Placeholder Text WorldCat 5 Szekendi MK et al. Active surveillance using electronic triggers to detect adverse events in hospitalized patients . Qual Saf Health Care 2006 ; 15 : 184 – 190 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Griffin FA , Classen DC. Detection of adverse events in surgical patients using the Trigger Tool approach . Qual Saf Health Care 2008 ; 17 : 253 – 258 . Google Scholar Crossref Search ADS PubMed WorldCat 7 Tegeder I et al. Retrospective analysis of the frequency and recognition of adverse drug reactions by means of automatically recorded laboratory signals . Br J Clin Pharmacol 1999 ; 47 : 557 – 564 . Google Scholar Crossref Search ADS PubMed WorldCat 8 Thuermann PA et al. Detection of adverse drug reactions in a neurological department: comparison between intensified surveillance and a computer-assisted approach . Drug Saf 2002 ; 25 : 713 – 724 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Azaz-Livshits T et al. Computerized surveillance of adverse drug reactions in hospital: pilot study . Br J Clin Pharmacol 1998 ; 45 : 309 – 314 . Google Scholar Crossref Search ADS PubMed WorldCat 10 Dormann H et al. Incidence and costs of adverse drug reactions during hospitalisation: computerised monitoring versus stimulated spontaneous reporting . Drug Saf 2000 ; 22 : 161 – 169 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Haffner S et al. Detecting adverse drug reactions on paediatric wards . Drug Saf 2005 ; 28 : 453 – 464 . Google Scholar Crossref Search ADS PubMed WorldCat 12 Dormann H et al. Implementation of a computer-assisted monitoring system for the detection of advserse drug reactions in gastroenterology . Aliment Pharmacol Ther 2004 ; 19 : 303 – 309 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Boon G et al. Specificity of trigger drugs as indicators of adverse drug events . Abstract presented at the UKCPA Symposium, Hinckley, November. Leicestershire: UKCPA, 2006 . 14 McCarthy S , Cavell G. Application of trigger tool methodology to the detection of preventable ADEs assciated with warfarin therapy – a pilot study . Pharm World Sci 2008 ; 30 : 302 – 303 . Google Scholar OpenURL Placeholder Text WorldCat 15 Cavell G , Thakrar R. Detecting adverse drug events using trigger tool methodology . Clin Pharm 2009 ; Suppl. 1 : S12 – S13 . Google Scholar OpenURL Placeholder Text WorldCat 16 Handler SM et al. A systematic review of the performance characteristics of clinical event monitor signals used to detect adverse drug events in the hospital setting . J Am Med Inform Assoc 2007 ; 14 : 451 – 458 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Franklin BD et al. Methodological variability in detecting prescribing errors and consequences for the evaluation of interventions . Pharmacoepidemiol Drug Saf 2009 ; 18 : 1 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat 18 Singh R et al. Experience with a trigger tool for identifying adverse drug events among older adults in ambulatory primary care . Qual Saf Health Care 2009 ; 18 : 199 – 204 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Dean B et al. What is a prescribing error? Qual Health Care 2000 ; 9 : 232 – 237 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Dean B , Barber N. Validity and reliability of observational methods for studying medication administration errors . Am J Health Syst Pharm 2001 ; 58 : 54 – 59 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Edwards IR , Aronson JK. Adverse drug reactions: definitions, diagnosis, and management . Lancet 2000 ; 356 : 1255 – 1259 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Takata GS et al. Development, testing and findings of a pediatric-focused trigger tool to identify medication-related harm in US children's hospitals . Pediatrics 2008 ; 121 : 927 – 935 . Google Scholar Crossref Search ADS WorldCat 23 Hwang S-H et al. Evaluation of a computer-based adverse-drug-event monitor . Am J Health Syst Pharm 2009 ; 65 : 2265 – 2272 . Google Scholar Crossref Search ADS WorldCat 24 Jha AK et al. Identifying adverse drug events: development of a computer-based monitor and comparison with chart review and stimulated voluntary report . J Am Med Inform Assoc 2002 ; 5 : 305 – 314 . Google Scholar Crossref Search ADS WorldCat 25 Vincent C et al. Adverse events in British Hospitals: preliminary retrospective record review . Br Med J 2001 ; 322 : 517 – 519 . Google Scholar Crossref Search ADS WorldCat 26 Sharek PJ et al. Adverse events in the neonatal intensive care unit: development, testing, and findings of an NICU-focused trigger tool to identify harm in North American NICUs . Pediatrics 2006 ; 118 : 1332 – 1340 . Google Scholar Crossref Search ADS PubMed WorldCat © 2010 The Authors. IJPP © 2010 Royal Pharmaceutical Society of Great Britain This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © 2010 The Authors. IJPP © 2010 Royal Pharmaceutical Society of Great Britain
TI - Testing a trigger tool as a method of detecting harm from medication errors in a UK hospital: a pilot study
JF - International Journal of Pharmacy Practice
DO - 10.1111/j.2042-7174.2010.00058.x
DA - 2010-09-14
UR - https://www.deepdyve.com/lp/oxford-university-press/testing-a-trigger-tool-as-a-method-of-detecting-harm-from-medication-AyXpkmlDQy
SP - 305
EP - 311
VL - 18
IS - 5
DP - DeepDyve
ER -