pharmacoepidemiology and drug safety 2015; 24: 176–186 Published online 17 June 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/pds.3663
ORIGINAL REPORT
Adverse drug events in patients admitted to an emergency department: an analysis of direct costs Florian Meier1†, Renke Maas2†, Anja Sonst3, Andrius Patapovas4, Fabian Müller2, Bettina Plank-Kiegele2, Barbara Pfistermeister2, Oliver Schöffski1, Thomas Bürkle4 and Harald Dormann3* 1
Department of Health Management, Friedrich-Alexander-Universität Erlangen-Nürnberg, Nuremberg, Germany Institute of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany 3 Department of Emergency Medicine Hospital Fürth, Fürth, Germany 4 Department of Medical Informatics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany 2
ABSTRACT Purpose Several economic evaluations of adverse drug events (ADEs) exist, but the underlying methodology has not been standardized so far. The aim of the study was to combine prospective, intensive pharmacovigilance methods, and standardized accounting data to calculate direct costs of community-acquired ADEs (caADEs) contributing to emergency department (ED) admission and subsequent hospitalization. Methods A prospective observational study with three phases extending over 2 years was implemented in a 749 bed tertiary care hospital with an annual ED census of approximately 45 000 patients. The patient records of all adult non-trauma ED admissions were systematically analyzed by a team of emergency physicians, clinical pharmacologists, and pharmacists for potential ADE. Associated diagnosis related group costs were extracted from standardized accounting data. Results Of 2262 patients attending the ED during the study periods, the hospitalization of 366 patients (16.2%) was related to one or more caADEs of which 97.5% were considered predictable and 62.0% were classified as preventable. The mean caADE-related diagnosis related group costs were €2743 (95% bias-corrected and accelerated CI: €2498 to €3018). Extrapolated to a national scale, this corresponds to caADE-related costs of €2.245bn for the German health insurance funds, annually. Costs of €1.310bn could be attributed to events classified as predictable and preventable. Conclusions In an ED, caADEs are frequent, and a significant proportion of these events and their related costs appear to be predictable and preventable. The ED as a first-line provider for ADE cases appears to be an appropriate environment to implement strategic and operative improvements for enhanced patient safety. Copyright © 2014 John Wiley & Sons, Ltd. key words—adverse drug event; hospital admission; emergency department; pharmacovigilance; health economics; pharmacoepidemiology Received 15 October 2012; Revised 12 May 2014; Accepted 14 May 2014
INTRODUCTION Adverse drug events (ADEs) have long been recognized as a leading cause of morbidity and mortality and a relevant cost factor for health systems of industrialized countries.1–3 International studies suggest that the incidence of ADEs ranges from 0.17% to 65% depending on setting and patient group.4–6 The hospital admission rate due to ADEs may reach up to 6.5%, and the economic harm in the USA may amount from $2.58bn up to $47.445bn for hospitalization and a total of *Correspondence to: H. Dormann, Hospital Fürth, Department of Emergency Medicine, Jakob Henle Str. 1, 90766 Fürth, Germany. E-mail: harald.dormann@ klinikum-fuerth.de † Meier F and Maas R contributed equally to the manuscript.
Copyright © 2014 John Wiley & Sons, Ltd.
$76.558bn for the health care system, annually.7–10 Previous studies in Germany estimated that 5.3% of all hospital admissions are at least potentially ADE-related with estimated direct inpatient costs of some €400–500m per year.11–13 The primary aim of the present investigation was to estimate the direct costs of community-acquired ADEs (caADEs) requiring emergency department (ED) admission and subsequent hospitalization from a health insurance perspective in Germany. These cost estimations were to be based on the combination of prospective intensive pharmacovigilance methods to identify and characterize caADEs and hospital accounting data according to the German diagnosis related groups (G-DRG) system. To assess potential
direct hospital costs of community-acquired ades
savings, the predictability and preventability of caADEs were considered. Furthermore, medical issues such as critical medication contributing to hospitalization, the treated diseases and diagnostic and therapeutic interventions were also analyzed. METHODS
177
drug-related or not. For cases with one or more potential ADEs, the chart review was repeated by two boardcertified specialists in internal/critical care medicine and clinical pharmacology. Any disagreements in the screening process were discussed until consensus was reached. The rate of false negatives was below 3%.16
Study setting This observational study was part of a research project supported by The German Federal Ministry of Health within the “German Coalition for Patient Safety.” The study aim was an analysis of ADEs in the ED of a 749 bed tertiary care hospital with 35 415 inpatient cases in 2011 and an annual ED census of approximately 45 000 patients coming from more than 130 general practitioners (GPs), nursing homes, or outpatient care services or directly from home. In 2011, 14 313 non-trauma ED patients were hospitalized. The hospital is located in a city with 97 337 adult inhabitants (≥18 years) and responsible for the emergency services of a region with 194 194 adult residents (31 December 2011).14 The patient structure is diversified including urban as well as rural parts in the catchment area. Data collection
Definition of adverse drug event and medication error Adverse drug events were classified according to the following definitions in adverse drug reactions (ADRs) and events in which MEs were involved.18 In line with the definition of the World Health Organization, an ADR is a “noxious and/or unintended response to medication which occurs despite appropriate drug dosage for prophylaxis, diagnosis or therapy of the indicating medical condition” (please note: this definition is distinct from the ADR definition recently introduced by European regulatory authorities, which includes error-related events as well).19 An ME was defined as wrong and inadequate use of medication comprising, for example, ignored ontraindications, missing indications, and wrong dosage. Finally, all ADEs in this study were scored regarding probability, predictability, and preventability using the following internationally validated scores:
The data presented here were collected in a prospective cohort study of 2262 consecutive ED patients visiting the ED between 1 to 30 September 2010, 1 to 30 April 2011, and 6 November to 6 December 2011. All adult non-trauma patients admitted to the ED were eligible. In this analysis, we concentrated our efforts on the GP’s medication orders and the self-medication. All patient data collected in the ED were pseudonymized and transferred to the study database, which has been established for this research project.15
The evaluation of the causality/probability that a drug caused an ADR usually involves subjective, clinical judgment. To minimize bias, the ADR probability scale by the World Health Organization Uppsala Monitoring Centre was used to estimate the causality/probability of an ADE.20 Only ADRs scored “possible,” “probable,” or “certain” were considered for this study.
Intensive pharmacovigilance method
Predictability: Rieder score
The collection of the medication and clinical data as well as the definition of ADEs, medication errors (MEs), and the corresponding classification of events as causally related, predictable, and preventable has been described previously in detail.15–17 In brief, all patient records including anamnestic data, ED documentation, laboratory findings, and external sources, such as medical or nursing home charts, when available, were systematically analyzed by a team of emergency physicians, clinical pharmacologists, and pharmacists for potential ADEs. For this study, we focused on caADEs caused by drugs taken prior to ED admission. The patient’s clinical symptoms and diagnoses were evaluated if potentially
Adverse drug events were classified as predictable or unpredictable as described by Rieder.21 Predictable drug reactions are those ADEs that can be anticipated when planning therapy including, for example, side effects, toxicity, and drug-drug interactions.
Copyright © 2014 John Wiley & Sons, Ltd.
Probability: World Health Organization Uppsala Monitoring Centre
Preventability: Schumock questionnaire and medication errors To assess potential preventability of ADEs, the criteria proposed by Schumock and Thornton were applied, resulting in a seven question catalog given in Table 1.22 If one or more questions were answered with “yes,” Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
178
f. meier et al.
the ADE was classified as preventable. In addition, all MEs were judged to be preventable. Standardized classification systems of International Classification of Diseases, Operation and Procedure Codes, Diagnosis Related Group, and Anatomical Therapeutic Chemical Classification System Several standardized classification systems were used to describe diseases, procedures in hospital, and drugs. For primary (i.e., main diseases) and secondary diagnoses (i.e., concomitant diseases), the codes of the International Classification of Diseases (ICD) German Modification were used.23 In German hospitals, all major diagnostic and therapeutic interventions for inpatients are mandatorily coded using the German procedure classification codes (Operationen und Prozedurenschlüssel, OPS), which has been derived from the international classification of procedures in medicine.24,25 OPS and ICD are essential for DRG grouping (G-DRG Version 2010 and 2011).26 We used the Anatomical Therapeutic Chemical (ATC) Classification System for categorizing drugs.27 This classification system was enhanced with additional information. The defined daily dose (DDD) for all ATC groups of this study was extracted from the German Arzneiverordnungs-Report 2012, which lists the 3000 most frequently prescribed drugs for patients in the German statutory health insurance.28 Economic data collection and perspective Germany has a national data standard of documentation and coding rules for collecting hospital inpatient data in the G-DRG system, which is defined in Section 21 of the Hospital Remuneration Act (KHEntgG). Any German hospital must deliver this data set annually to the Institute for the Hospital Remuneration System. This standardized data set can be enhanced with accounting information by applying a certified DRG grouper software. Thereby, the relative cost weight of the assigned DRG has been multiplied with the Table 1. Schumock et al. questionnaire for characterizing the preventability of adverse drug events (ADE) Items of the Schumock et al. questionnaire: 1) Was the drug involved in the ADE considered inappropriate for the patient’s clinical condition? 2) Was the dose, route, and frequency of administration inappropriate for the patient’s age, weight, and disease state? 3) Have required therapeutic drug monitoring or other necessary laboratory tests not performed? 4) Was there a history of allergy or previous reactions to the drug? 5) Was a drug interaction involved in the reaction? 6) Was a toxic serum drug level documented? 7) Was poor compliance involved in the reaction?
Copyright © 2014 John Wiley & Sons, Ltd.
DRG base rate of €2982 (equal for 2010 and 2011) to calculate the actual reimbursement rates.29 Medical data including ICD, OPS, and DRG codes as well as reimbursement rates for all inpatients of the study cohort have been extracted with a query interface from the hospital information system. Case consolidations, which were patients with one single DRG covering several hospital stays such as preadmission, inpatient stay, or post-discharge, were excluded from the study because the reimbursement could not be divided into the different hospital stays. DRG costs reported in this study are calculated on the same basis used by the German statutory health insurance companies. A profit margin, which would be relevant for a business view, was not calculated. Patient visits to a German hospital for outpatient care are flat fee reimbursed according to the Section 120 of Social Code Book V. Such outpatient health care costs were not addressed in this analysis. Extrapolating the direct hospital costs for Germany In order to estimate the direct costs for Germany, we first calculated the expected yearly total ADE cases for the hospital using the ratio of ADE-related hospitalization of the three study phases for all non-trauma ED hospitalized patients aged 18 years or more. The ED is the sole provider for emergency care in the catchment area. Thus, the extrapolation covers the adult population. Because the age structure of the catchment area was similar to Germany’s age distribution, age standardization was not necessary. In the next step, we divided the German adult population by the catchment area population and created a multiplication factor (68 624 000/194 194 = 353.4).30 Multiplying the factor with the estimated yearly total of the hospital that represents the catchment area, we obtained the number of patients with a potential ADE-related hospital stay in Germany per year. Finally, we multiplied these extrapolated numbers with the mean DRG costs per case from our study. Software and statistical analysis Pseudonymized patient and ADE data were stored in secuTrial (secuTrial, Version 4.5.0.13, interActive Systems GmbH, Berlin, BLN, Germany). For economic data storage and data preparation, we used SQL Server 2008 (Microsoft SQL Server 2008 R2, Microsoft Corporation, Redmond, WA, USA). The explorative analysis was done with DeltaMaster (DeltaMaster, version 5.5.2, Bissantz & Company GmbH, Nuremberg, BAY, Germany), and statistical analysis was conducted using SPSS 20 (IBM SPSS Statistics for Windows, version 20.0, IBM Corporation, Armonk, NY, USA). Results are Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
direct hospital costs of community-acquired ades
presented with mean, standard deviation (±SD), median with interquartile range (IQR), and range for continuous data, and in absolute numbers and percentages for discrete data. Comparison between groups was performed using the two-sided chi-square test. Statistical significance was established as p < 0.05. Cost data often show a non-normal distribution with right-skewness, thus we additionally used non-parametric bootstrap techniques (10 000 replications) to handle uncertainties and calculated the bias-corrected and accelerated 95% confidence interval (BCa 95% CI).31–33 RESULTS Study population and characterization of community-acquired adverse drug events The mean age of the 2262 evaluated non-trauma patients (study phase I: 752, II: 754, III: 756) consisting of 1090 men and 1172 women was 62.1 years (±20.2, median 67.0, IQR 47–79, range 17–102). Out of these, 366 (16.2%) cases with a mean age of 72.8 years (±15.7, median 76.0, IQR 67–84, range 18–102) had at least one caADE that contributed to ED admission and subsequent hospitalization. In 334 patients (91.3%) one caADE, in 30 patients (8.2%) the combination of two caADEs, and in two patients three caADEs (0.5%) contributed to hospitalization. Further patient characteristics are given in Table 2. The following numbers refer to the 400 caADEs caused by 318 ADRs and 82 ME-related events, which were responsible for hospitalization. On the basis of the World Health Organization Uppsala Monitoring Centre probability scale, 35 (8.8%) events were Table 2. Characteristics of hospitalized patients Number of patients
Total patients Age (years) 18–39 40–49 50–59 60–69 70–79 80–89 90+ Sex Male Female Number of drugs per patient 1 2–5 6–10 11+
n
%
366
100.0
19 17 24 46 107 131 22
5.2 4.6 6.6 12.6 29.2 35.8 6.0
150 216
41.0 59.0
14 113 147 92
3.8 30.9 40.2 25.1
Copyright © 2014 John Wiley & Sons, Ltd.
179
considered as “certainly” associated with a specific drug. In 88 (22.0%) events, the association was classified probable/likely, and in 277 (69.3%) events, it was considered possible. Most caADEs were classified as predictable (97.5%) or preventable (62.0%). The most frequent predictable caADEs were side effects (60.0%) and drug-drug interactions (35.5%) as shown in Table 3. Drugs associated with community-acquired adverse drug events Patients hospitalized because of caADEs took on average 7.6 (±4.1, median 7.0, IQR 5–11, range 1–35) drugs. The ATC drug groups most frequently involved in caADEs were antithrombotic agents (9.9%), beta blockers (9.4%), and antipsychotics (6.9%). Overall, the drugs from the 10 most common ATC groups (394 drugs) comprised 56.8% of all drugs. The proportion of caADEs judged to be preventable varied between the ATC groups. It was considered the highest in the group of selective calcium channel blockers (100.0%), dopaminergic agents (94.4%), and ACEinhibitors (87.2%). Considering the prescribed DDD in Germany in 2011, antipsychotics (300.0 million DDD) and dopaminergic agents (120.7 million DDD) had the fewest DDD among the 10 most common ATC groups. Though, the proportions of caADE-related hospitalizations per annually prescribed DDD of antipsychotics (0.160 = 48/300.0 million DDD), dopaminergic substances (0.149), and antithrombotic agents (0.049) are bigger than those of frequently prescribed agents like ACE-inhibitors (0.009) or calcium channel blockers (0.010), as shown in Table 4. Thus, the likelihood of suffering a caADE-related hospital admission per prescribed DDD is higher in these ATC groups. International classification of diseases, diagnosis related groups, operation and procedure code, and principal departments The mean number of primary and secondary diagnoses was 6.9 (±3.6, median 6.0, IQR 4–9, range 1–20) per case. The most frequent primary diagnoses in patients presenting with caADEs grouped by ICD chapters were diseases of the circulatory system (n = 71) with a mean length of stay (LOS) of 8.6 days (±7.3, median 7.0, IQR 4–12, range 0–32), followed by diseases of the digestive system (n = 61) with a mean LOS of 5.7 days (±4.9, median 5.0, IQR 3–7, range 0–27) and symptoms, signs, and abnormal clinical and laboratory findings, not elsewhere classified (n = 48) with a mean LOS of 4.2 days (±2.3, median 4.0, IQR 2.3–6, range 0–9) and comprises 53.7% of all cases. For more details, the 10 Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
180
f. meier et al.
Table 3. Classification, predictability, and preventability of community-acquired adverse drug events (caADEs) Number of caADEs
Preventability—yes
Preventability—no
n (%)
n (%)
n (%)
400 (100.0) 318 (79.5) 82 (20.5) 400 (100.0) 142 (35.5) 2 (0.5) 240 (60.0) 6 (1.5) 1 (0.3) 7 (1.8) 1 (0.3) 1 (0.3)
248 (100.0) 166 (66.9) 82 (33.1) 248 (100.0) 135 (54.4) 2 (0.8) 105 (42.3) 3 (1.2) 0 1 (0.4) 1 (0.4) 1 (0.4)
152 (100.0) 152 (100.0) 0 152 (100.0) 7 (4.6) 0 135 (88.8) 3 (2.0) 1 (0.7) 6 (3.9) 0 0
caADE classification Adverse drug reactions caADEs involving medication errors Predictability Predictable—drug-drug interaction Predictable—secondary effect Predictable—side effect Predictable—toxicity Unpredictable—idiosyncratic Unpredictable—immuno-allergy Unpredictable—intolerance Not specified
p
ORIGINAL REPORT
Adverse drug events in patients admitted to an emergency department: an analysis of direct costs Florian Meier1†, Renke Maas2†, Anja Sonst3, Andrius Patapovas4, Fabian Müller2, Bettina Plank-Kiegele2, Barbara Pfistermeister2, Oliver Schöffski1, Thomas Bürkle4 and Harald Dormann3* 1
Department of Health Management, Friedrich-Alexander-Universität Erlangen-Nürnberg, Nuremberg, Germany Institute of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany 3 Department of Emergency Medicine Hospital Fürth, Fürth, Germany 4 Department of Medical Informatics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany 2
ABSTRACT Purpose Several economic evaluations of adverse drug events (ADEs) exist, but the underlying methodology has not been standardized so far. The aim of the study was to combine prospective, intensive pharmacovigilance methods, and standardized accounting data to calculate direct costs of community-acquired ADEs (caADEs) contributing to emergency department (ED) admission and subsequent hospitalization. Methods A prospective observational study with three phases extending over 2 years was implemented in a 749 bed tertiary care hospital with an annual ED census of approximately 45 000 patients. The patient records of all adult non-trauma ED admissions were systematically analyzed by a team of emergency physicians, clinical pharmacologists, and pharmacists for potential ADE. Associated diagnosis related group costs were extracted from standardized accounting data. Results Of 2262 patients attending the ED during the study periods, the hospitalization of 366 patients (16.2%) was related to one or more caADEs of which 97.5% were considered predictable and 62.0% were classified as preventable. The mean caADE-related diagnosis related group costs were €2743 (95% bias-corrected and accelerated CI: €2498 to €3018). Extrapolated to a national scale, this corresponds to caADE-related costs of €2.245bn for the German health insurance funds, annually. Costs of €1.310bn could be attributed to events classified as predictable and preventable. Conclusions In an ED, caADEs are frequent, and a significant proportion of these events and their related costs appear to be predictable and preventable. The ED as a first-line provider for ADE cases appears to be an appropriate environment to implement strategic and operative improvements for enhanced patient safety. Copyright © 2014 John Wiley & Sons, Ltd. key words—adverse drug event; hospital admission; emergency department; pharmacovigilance; health economics; pharmacoepidemiology Received 15 October 2012; Revised 12 May 2014; Accepted 14 May 2014
INTRODUCTION Adverse drug events (ADEs) have long been recognized as a leading cause of morbidity and mortality and a relevant cost factor for health systems of industrialized countries.1–3 International studies suggest that the incidence of ADEs ranges from 0.17% to 65% depending on setting and patient group.4–6 The hospital admission rate due to ADEs may reach up to 6.5%, and the economic harm in the USA may amount from $2.58bn up to $47.445bn for hospitalization and a total of *Correspondence to: H. Dormann, Hospital Fürth, Department of Emergency Medicine, Jakob Henle Str. 1, 90766 Fürth, Germany. E-mail: harald.dormann@ klinikum-fuerth.de † Meier F and Maas R contributed equally to the manuscript.
Copyright © 2014 John Wiley & Sons, Ltd.
$76.558bn for the health care system, annually.7–10 Previous studies in Germany estimated that 5.3% of all hospital admissions are at least potentially ADE-related with estimated direct inpatient costs of some €400–500m per year.11–13 The primary aim of the present investigation was to estimate the direct costs of community-acquired ADEs (caADEs) requiring emergency department (ED) admission and subsequent hospitalization from a health insurance perspective in Germany. These cost estimations were to be based on the combination of prospective intensive pharmacovigilance methods to identify and characterize caADEs and hospital accounting data according to the German diagnosis related groups (G-DRG) system. To assess potential
direct hospital costs of community-acquired ades
savings, the predictability and preventability of caADEs were considered. Furthermore, medical issues such as critical medication contributing to hospitalization, the treated diseases and diagnostic and therapeutic interventions were also analyzed. METHODS
177
drug-related or not. For cases with one or more potential ADEs, the chart review was repeated by two boardcertified specialists in internal/critical care medicine and clinical pharmacology. Any disagreements in the screening process were discussed until consensus was reached. The rate of false negatives was below 3%.16
Study setting This observational study was part of a research project supported by The German Federal Ministry of Health within the “German Coalition for Patient Safety.” The study aim was an analysis of ADEs in the ED of a 749 bed tertiary care hospital with 35 415 inpatient cases in 2011 and an annual ED census of approximately 45 000 patients coming from more than 130 general practitioners (GPs), nursing homes, or outpatient care services or directly from home. In 2011, 14 313 non-trauma ED patients were hospitalized. The hospital is located in a city with 97 337 adult inhabitants (≥18 years) and responsible for the emergency services of a region with 194 194 adult residents (31 December 2011).14 The patient structure is diversified including urban as well as rural parts in the catchment area. Data collection
Definition of adverse drug event and medication error Adverse drug events were classified according to the following definitions in adverse drug reactions (ADRs) and events in which MEs were involved.18 In line with the definition of the World Health Organization, an ADR is a “noxious and/or unintended response to medication which occurs despite appropriate drug dosage for prophylaxis, diagnosis or therapy of the indicating medical condition” (please note: this definition is distinct from the ADR definition recently introduced by European regulatory authorities, which includes error-related events as well).19 An ME was defined as wrong and inadequate use of medication comprising, for example, ignored ontraindications, missing indications, and wrong dosage. Finally, all ADEs in this study were scored regarding probability, predictability, and preventability using the following internationally validated scores:
The data presented here were collected in a prospective cohort study of 2262 consecutive ED patients visiting the ED between 1 to 30 September 2010, 1 to 30 April 2011, and 6 November to 6 December 2011. All adult non-trauma patients admitted to the ED were eligible. In this analysis, we concentrated our efforts on the GP’s medication orders and the self-medication. All patient data collected in the ED were pseudonymized and transferred to the study database, which has been established for this research project.15
The evaluation of the causality/probability that a drug caused an ADR usually involves subjective, clinical judgment. To minimize bias, the ADR probability scale by the World Health Organization Uppsala Monitoring Centre was used to estimate the causality/probability of an ADE.20 Only ADRs scored “possible,” “probable,” or “certain” were considered for this study.
Intensive pharmacovigilance method
Predictability: Rieder score
The collection of the medication and clinical data as well as the definition of ADEs, medication errors (MEs), and the corresponding classification of events as causally related, predictable, and preventable has been described previously in detail.15–17 In brief, all patient records including anamnestic data, ED documentation, laboratory findings, and external sources, such as medical or nursing home charts, when available, were systematically analyzed by a team of emergency physicians, clinical pharmacologists, and pharmacists for potential ADEs. For this study, we focused on caADEs caused by drugs taken prior to ED admission. The patient’s clinical symptoms and diagnoses were evaluated if potentially
Adverse drug events were classified as predictable or unpredictable as described by Rieder.21 Predictable drug reactions are those ADEs that can be anticipated when planning therapy including, for example, side effects, toxicity, and drug-drug interactions.
Copyright © 2014 John Wiley & Sons, Ltd.
Probability: World Health Organization Uppsala Monitoring Centre
Preventability: Schumock questionnaire and medication errors To assess potential preventability of ADEs, the criteria proposed by Schumock and Thornton were applied, resulting in a seven question catalog given in Table 1.22 If one or more questions were answered with “yes,” Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
178
f. meier et al.
the ADE was classified as preventable. In addition, all MEs were judged to be preventable. Standardized classification systems of International Classification of Diseases, Operation and Procedure Codes, Diagnosis Related Group, and Anatomical Therapeutic Chemical Classification System Several standardized classification systems were used to describe diseases, procedures in hospital, and drugs. For primary (i.e., main diseases) and secondary diagnoses (i.e., concomitant diseases), the codes of the International Classification of Diseases (ICD) German Modification were used.23 In German hospitals, all major diagnostic and therapeutic interventions for inpatients are mandatorily coded using the German procedure classification codes (Operationen und Prozedurenschlüssel, OPS), which has been derived from the international classification of procedures in medicine.24,25 OPS and ICD are essential for DRG grouping (G-DRG Version 2010 and 2011).26 We used the Anatomical Therapeutic Chemical (ATC) Classification System for categorizing drugs.27 This classification system was enhanced with additional information. The defined daily dose (DDD) for all ATC groups of this study was extracted from the German Arzneiverordnungs-Report 2012, which lists the 3000 most frequently prescribed drugs for patients in the German statutory health insurance.28 Economic data collection and perspective Germany has a national data standard of documentation and coding rules for collecting hospital inpatient data in the G-DRG system, which is defined in Section 21 of the Hospital Remuneration Act (KHEntgG). Any German hospital must deliver this data set annually to the Institute for the Hospital Remuneration System. This standardized data set can be enhanced with accounting information by applying a certified DRG grouper software. Thereby, the relative cost weight of the assigned DRG has been multiplied with the Table 1. Schumock et al. questionnaire for characterizing the preventability of adverse drug events (ADE) Items of the Schumock et al. questionnaire: 1) Was the drug involved in the ADE considered inappropriate for the patient’s clinical condition? 2) Was the dose, route, and frequency of administration inappropriate for the patient’s age, weight, and disease state? 3) Have required therapeutic drug monitoring or other necessary laboratory tests not performed? 4) Was there a history of allergy or previous reactions to the drug? 5) Was a drug interaction involved in the reaction? 6) Was a toxic serum drug level documented? 7) Was poor compliance involved in the reaction?
Copyright © 2014 John Wiley & Sons, Ltd.
DRG base rate of €2982 (equal for 2010 and 2011) to calculate the actual reimbursement rates.29 Medical data including ICD, OPS, and DRG codes as well as reimbursement rates for all inpatients of the study cohort have been extracted with a query interface from the hospital information system. Case consolidations, which were patients with one single DRG covering several hospital stays such as preadmission, inpatient stay, or post-discharge, were excluded from the study because the reimbursement could not be divided into the different hospital stays. DRG costs reported in this study are calculated on the same basis used by the German statutory health insurance companies. A profit margin, which would be relevant for a business view, was not calculated. Patient visits to a German hospital for outpatient care are flat fee reimbursed according to the Section 120 of Social Code Book V. Such outpatient health care costs were not addressed in this analysis. Extrapolating the direct hospital costs for Germany In order to estimate the direct costs for Germany, we first calculated the expected yearly total ADE cases for the hospital using the ratio of ADE-related hospitalization of the three study phases for all non-trauma ED hospitalized patients aged 18 years or more. The ED is the sole provider for emergency care in the catchment area. Thus, the extrapolation covers the adult population. Because the age structure of the catchment area was similar to Germany’s age distribution, age standardization was not necessary. In the next step, we divided the German adult population by the catchment area population and created a multiplication factor (68 624 000/194 194 = 353.4).30 Multiplying the factor with the estimated yearly total of the hospital that represents the catchment area, we obtained the number of patients with a potential ADE-related hospital stay in Germany per year. Finally, we multiplied these extrapolated numbers with the mean DRG costs per case from our study. Software and statistical analysis Pseudonymized patient and ADE data were stored in secuTrial (secuTrial, Version 4.5.0.13, interActive Systems GmbH, Berlin, BLN, Germany). For economic data storage and data preparation, we used SQL Server 2008 (Microsoft SQL Server 2008 R2, Microsoft Corporation, Redmond, WA, USA). The explorative analysis was done with DeltaMaster (DeltaMaster, version 5.5.2, Bissantz & Company GmbH, Nuremberg, BAY, Germany), and statistical analysis was conducted using SPSS 20 (IBM SPSS Statistics for Windows, version 20.0, IBM Corporation, Armonk, NY, USA). Results are Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
direct hospital costs of community-acquired ades
presented with mean, standard deviation (±SD), median with interquartile range (IQR), and range for continuous data, and in absolute numbers and percentages for discrete data. Comparison between groups was performed using the two-sided chi-square test. Statistical significance was established as p < 0.05. Cost data often show a non-normal distribution with right-skewness, thus we additionally used non-parametric bootstrap techniques (10 000 replications) to handle uncertainties and calculated the bias-corrected and accelerated 95% confidence interval (BCa 95% CI).31–33 RESULTS Study population and characterization of community-acquired adverse drug events The mean age of the 2262 evaluated non-trauma patients (study phase I: 752, II: 754, III: 756) consisting of 1090 men and 1172 women was 62.1 years (±20.2, median 67.0, IQR 47–79, range 17–102). Out of these, 366 (16.2%) cases with a mean age of 72.8 years (±15.7, median 76.0, IQR 67–84, range 18–102) had at least one caADE that contributed to ED admission and subsequent hospitalization. In 334 patients (91.3%) one caADE, in 30 patients (8.2%) the combination of two caADEs, and in two patients three caADEs (0.5%) contributed to hospitalization. Further patient characteristics are given in Table 2. The following numbers refer to the 400 caADEs caused by 318 ADRs and 82 ME-related events, which were responsible for hospitalization. On the basis of the World Health Organization Uppsala Monitoring Centre probability scale, 35 (8.8%) events were Table 2. Characteristics of hospitalized patients Number of patients
Total patients Age (years) 18–39 40–49 50–59 60–69 70–79 80–89 90+ Sex Male Female Number of drugs per patient 1 2–5 6–10 11+
n
%
366
100.0
19 17 24 46 107 131 22
5.2 4.6 6.6 12.6 29.2 35.8 6.0
150 216
41.0 59.0
14 113 147 92
3.8 30.9 40.2 25.1
Copyright © 2014 John Wiley & Sons, Ltd.
179
considered as “certainly” associated with a specific drug. In 88 (22.0%) events, the association was classified probable/likely, and in 277 (69.3%) events, it was considered possible. Most caADEs were classified as predictable (97.5%) or preventable (62.0%). The most frequent predictable caADEs were side effects (60.0%) and drug-drug interactions (35.5%) as shown in Table 3. Drugs associated with community-acquired adverse drug events Patients hospitalized because of caADEs took on average 7.6 (±4.1, median 7.0, IQR 5–11, range 1–35) drugs. The ATC drug groups most frequently involved in caADEs were antithrombotic agents (9.9%), beta blockers (9.4%), and antipsychotics (6.9%). Overall, the drugs from the 10 most common ATC groups (394 drugs) comprised 56.8% of all drugs. The proportion of caADEs judged to be preventable varied between the ATC groups. It was considered the highest in the group of selective calcium channel blockers (100.0%), dopaminergic agents (94.4%), and ACEinhibitors (87.2%). Considering the prescribed DDD in Germany in 2011, antipsychotics (300.0 million DDD) and dopaminergic agents (120.7 million DDD) had the fewest DDD among the 10 most common ATC groups. Though, the proportions of caADE-related hospitalizations per annually prescribed DDD of antipsychotics (0.160 = 48/300.0 million DDD), dopaminergic substances (0.149), and antithrombotic agents (0.049) are bigger than those of frequently prescribed agents like ACE-inhibitors (0.009) or calcium channel blockers (0.010), as shown in Table 4. Thus, the likelihood of suffering a caADE-related hospital admission per prescribed DDD is higher in these ATC groups. International classification of diseases, diagnosis related groups, operation and procedure code, and principal departments The mean number of primary and secondary diagnoses was 6.9 (±3.6, median 6.0, IQR 4–9, range 1–20) per case. The most frequent primary diagnoses in patients presenting with caADEs grouped by ICD chapters were diseases of the circulatory system (n = 71) with a mean length of stay (LOS) of 8.6 days (±7.3, median 7.0, IQR 4–12, range 0–32), followed by diseases of the digestive system (n = 61) with a mean LOS of 5.7 days (±4.9, median 5.0, IQR 3–7, range 0–27) and symptoms, signs, and abnormal clinical and laboratory findings, not elsewhere classified (n = 48) with a mean LOS of 4.2 days (±2.3, median 4.0, IQR 2.3–6, range 0–9) and comprises 53.7% of all cases. For more details, the 10 Pharmacoepidemiology and Drug Safety, 2015; 24: 176–186 DOI: 10.1002/pds
180
f. meier et al.
Table 3. Classification, predictability, and preventability of community-acquired adverse drug events (caADEs) Number of caADEs
Preventability—yes
Preventability—no
n (%)
n (%)
n (%)
400 (100.0) 318 (79.5) 82 (20.5) 400 (100.0) 142 (35.5) 2 (0.5) 240 (60.0) 6 (1.5) 1 (0.3) 7 (1.8) 1 (0.3) 1 (0.3)
248 (100.0) 166 (66.9) 82 (33.1) 248 (100.0) 135 (54.4) 2 (0.8) 105 (42.3) 3 (1.2) 0 1 (0.4) 1 (0.4) 1 (0.4)
152 (100.0) 152 (100.0) 0 152 (100.0) 7 (4.6) 0 135 (88.8) 3 (2.0) 1 (0.7) 6 (3.9) 0 0
caADE classification Adverse drug reactions caADEs involving medication errors Predictability Predictable—drug-drug interaction Predictable—secondary effect Predictable—side effect Predictable—toxicity Unpredictable—idiosyncratic Unpredictable—immuno-allergy Unpredictable—intolerance Not specified
p
