Received: 27 March 2020 Revised: 15 January 2021 Accepted: 17 February 2021 DOI: 10.1111/bcp.14799 SYSTEMATIC REVIEW AND META-ANALYSIS Epidemiology of drug-related deaths in European hospitals: A systematic review and meta-analysis of observational studies Eva Montané1,2 | Xavier Castells3 1 Department of Clinical Pharmacology, Hospital Universitari Germans Trias i Pujol, Aims: To perform a systematic review of observational studies on the epidemiology Badalona, Barcelona, Spain of drug-related death (DRD) in patients requiring hospitalisation or while hospitalised 2 Department of Pharmacology, Therapeutics and Toxicology, Universitat Autònoma de (hospital-acquired DRD). Barcelona, Barcelona, Spain Methods: We conducted a systematic review of observational studies investigating 3 Department of Medical Sciences, University the occurrence rate of DRD episodes among deceased inpatients. Two independent of Girona, Girona, Spain researchers assessed eligibility criteria, extracted data and evaluated the risk of bias. Correspondence Both quality assessment and meta-analysis were performed. Dr Eva Montané, Carretera de Canyet s/n, 08916 Badalona, Barcelona, Spain. Results: From 1351 identified potential studies, 6 retrospective studies were Email: emontane.germanstrias@gencat.cat. included. DRD occurrences rates were 7.3% (95% confidence interval [CI] 4.1–12.5) among deceased inpatients and 0.13% (95% CI 0.04–0.40) among hospitalised patients. During hospitalisation, acquired-DRD represented 2.7% (95% CI 1.0–6.9) of inpatient deaths and occurred in 0.05% (95% CI 0.01–0.23) of hospitalised patients. However, these estimates have to be viewed with caution because there was signifi- cant heterogeneity (I2 > 97%). None of the studies were considered to be at high risk of bias according to the criteria of the NIH Quality Assessment Tool. The most com- mon adverse drug reactions related to death were haemorrhages due to anti- thrombotic drugs (39%, 95% CI 26.5–53.2) and infections in drug-immunosuppressed patients (27.5%, 95% CI 16.7–41.7). Conclusion: We found that the DRD occurrence rate of deceased hospital inpatients has been infrequently studied in Europe. Our findings suggest that drugs are an important cause of death in hospitals. The limited number of studies in European countries stresses the need for more research in this area. KEYWORDS adverse drug reaction, epidemiology, hospital mortality, incidence/prevalence, meta-analysis, observational studies 1 | I N T RO DU CT I O N Several studies and some meta-analyses have been published in different settings or populations, mainly focusing on assessing the Adverse drug reactions (ADRs) are a known problem in terms of incidence of ADRs leading to hospitalisation.5–10 Regarding the epide- 1,2 morbidity, costs and mortality. In Europe, an estimated 5% of all miology of fatal ADRs among hospitalised patients, a meta-analysis hospital admissions are due to ADRs, and ADRs are the fifth most conducted by Lazarou et al. reported an incidence of 0.32% in the common cause of hospital death.3 Extrapolated data from a meta- USA4 and a review conducted by Bouvy et al. reported an incidence analysis revealed that in Europe about 197 000 deaths occur annually of 0.5% in Europe.11 Recently, Patel et al. published a meta-analysis 4 due to ADRs. including 48 international studies showing an incidence of fatal ADRs Brit Jnl Clinical Pharma. 2021;87:3659–3671. wileyonlinelibrary.com/journal/bcp © 2021 British Pharmacological Society 3659 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3660 MONTANÉ AND CASTELLS of 0.2%.12 Differences in the prevalence of diseases that impact drug objective, providing epidemiological data (such as the number of consumption could result in large differences in drug-related admitted patients and the number of deceased patients in the hospi- death (DRD) incidences between low- and high-income countries (0.3 tal) or data allowing their calculation, and performed in European vs. 3%).4,13 countries. The occurrence rate of DRD in hospitalised patients is an impor- The exclusion criteria were clinical trials, review articles, system- tant epidemiological indicator of mortality caused by drugs. Recently atic reviews, meta-analyses, commentaries and editorials, as well as Montané et al.14 found that DRDs among deceased inpatients was studies with other designs (e.g. spontaneous reporting or case 7%. This is less than half of Pardo Cabello's finding,15 despite both reports). No studies assessing mortality rate among patients treated in studies being from Spain. Meta-analysis in the context of a systematic specific hospital services or with specific drugs or drug classes were review is the most reliable method for understanding the findings included. Studies performed in Asian, American, Oceanian or African from studies with different and sometimes conflicting results. We per- countries were also excluded from the study. formed a systematic review and meta-analysis aiming to assess the Titles and/or abstracts of studies retrieved using the search strat- incidences and occurrence rates of DRD in hospital settings from egy and those from additional sources were screened by the primary studies performed in European countries and studying the sources of author to identify studies that potentially met the inclusion criteria. between-study variability of results. We limited these to this specific The full text of these potentially eligible studies was retrieved and geographic area in order to obtain more homogenous data taking into independently assessed for eligibility by the same researcher. Doubts account the fact that European countries have quite similar health about any particular study were settled after a thorough discussion systems, with comparable disease prevalence and patterns of drug with the second author. 16,17 consumption. To our knowledge, no meta-analysis has studied the incidence of DRD among deceased inpatients so far. The approach of this meta- 2.3 | Data extraction analysis differs from others that included studies evaluating ADRs that led to hospital admission or occurred during hospitalisation, regardless Both authors retrieved the study data independently. A of whether the outcome was fatal or not. standardised, prepiloted form was used to extract data from the included studies for assessment of study quality and evidence syn- thesis. Extracted information included: year of publication, country 2 | MATERIALS AND METHODS where the study was conducted, study setting, study period, study design, study methodology, ADR definition, causality assessment This systematic review and meta-analysis was conducted in tool, study population with demographics of participants and base- accordance with the Preferred Reporting Items for Systematic line characteristics, incidences of DRD and hospital-acquired DRD 18 Reviews and Meta-Analyses statement [Prisma statement]. The among inpatients and overall deceased patients, characteristics of study was registered in PROSPERO International prospective register suspected drugs (using the Anatomical Therapeutic Chemical classi- of systematic reviews in December 2018. Registration number: fication20), drug–drug interactions, characteristics of ADR, median CRD42019105618.19 hospital stay, and risk factors for and preventability of ADR. Dis- crepancies were resolved through discussion between the researchers. Missing relevant and necessary data were requested 2.1 | Data sources and search strategy from authors of the study. Three databases (PubMed, Web of Science and SCOPUS) were searched from inception to July 2019. Search strings were con- 2.4 | Methodological quality structed using Boolean operators (AND, OR) and combining key- words. The following syntax was used in each database: (([adverse To assess the risk of bias in individual studies, 2 raters (E.M. and X.C.) drug effect] OR “adverse drug reaction”) AND “Hospital Mortality” independently assessed the methodological quality using the NIH AND (incidence OR prevalence)), without language restriction Quality Assessment Tool for Observational Cohort and Cross- (Electronic Supplementary Material [ESM] 1). Furthermore, the cita- Sectional Studies.21 This quality tool comprises 14 items or criteria tions and references of the included studies and review articles were and guidance for rating. Six of the 14 criteria (43%) were not applica- screened for relevant articles. ble due to the design of the studies (retrospective cohorts), these were items 4, 5, 7, 10, 12 and 13 (Table 1). The quality score was cal- culated by dividing the number of criteria that were present (yes) by 2.2 | Study selection criteria the total number of applicable criteria. A quality of 100% was obtained if all 8 applicable criteria were present in the study and stud- The inclusion criteria were observational studies that assessed the ies were considered to be at high risk of bias if they achieved a DRD occurrence rate among all deceased inpatients as a main study score <50%. 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MONTANÉ AND CASTELLS 3661 TABLE 1 Quality rating of included studies Juntti- Pardo Lapatto- Pardo Criteria/studies Patinen30 Cabello32 Reiniluoto31 Cabello15 Grévy29 Montané14 1. Was the research question or objective in this paper clearly Yes Yes Yes Yes Yes Yes stated? 2. Was the study population clearly specified and defined? Yes Yes Yes Yes Yes Yes 3. Was the participation rate of eligible persons at least 50%? Yes No Yes Yes Yes Yes 4. Were all the subjects selected or recruited from the same or NA NA NA NA NA NA similar populations (including the same time period)? Were inclusion and exclusion criteria for being in the study prespecified and applied uniformly to all participants? 5. Was a sample size justification, power description, or variance NA NA NA NA NA NA and effect estimates provided? 6. For the analyses in this paper, were the exposure(s) of interest Yes Yes Yes Yes Yes Yes measured prior to the outcome(s) being measured? 7. Was the timeframe sufficient so that 1 could reasonably NA NA NA NA NA NA expect to see an association between exposure and outcome if it existed? 8. For exposures that can vary in amount or level, did the study Yes No No No Yes Yes examine different levels of the exposure as related to the outcome (e.g. categories of exposure, or exposure measured as continuous variable)? 9. Were the exposure measures (independent variables) clearly Yes Yes Yes Yes Yes Yes defined, valid, reliable and implemented consistently across all study participants? 10. Was the exposure(s) assessed more than once over time? NA NA NA NA NA NA 11. Were the outcome measures (dependent variables) clearly Yes Yes Yes Yes Yes Yes defined, valid, reliable and implemented consistently across all study participants? 12. Were the outcome assessors blinded to the exposure status NA NA NA NA NA NA of participants? 13. Was loss to follow-up after baseline 20% or less? NA NA NA NA NA NA 14. Were key potential confounding variables measured and No Yes No Yes No Yes adjusted statistically for their impact on the relationship between exposure(s) and outcome(s)? Quality score (n, %) 7/8 6/8 6/8 7/8 7/8 8/8 87.5% 75% 75% 87.5% 87.5% 100% NA: not applicable. 2.5 | Data analysis As the most common DRDs have been found to be haemorrhages and infections in drug-immunosuppressed patients,14,22,23 we The occurrence rate of DRD among deceased inpatients was assessed the occurrence rate of these specific types of DRD, which calculated using the number of DRDs as the numerator and the total were calculated by dividing the number of patients dying from each number of deceased inpatients during the study period as the denomi- specific type of DRD by the total number of patients dying from nator; for the incidence of DRD among hospitalised patients, the num- DRDs. Both internal and external haemorrhages were considered. ber of DRDs was used as the numerator and the total number of Infections were considered both when disseminated and also when hospital admissions during the study period as the denominator. The affecting a single organ. rate of hospital-acquired DRD was calculated using the number of hospital-acquired DRDs as the numerator and the total number of DRDs as the denominator. For the incidence of hospital-acquired 2.6 | Statistical analysis DRD among hospitalised patients, the number of hospital-acquired DRDs was used as the numerator and the total number of hospital Individual results were pooled using a random effects model. admissions during the study period as the denominator. Statistical heterogeneity was determined by calculating the I2 statistic, 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3662 MONTANÉ AND CASTELLS which measures the percentage of variance across studies due to het- 3.2 | Study characteristics and measures erogeneity rather than chance.24 We initially planned to study the sources of statistical heteroge- All the included studies were observational retrospective cohorts neity by means of meta-regression using patient characteristics and performed in the last 20 years (from 2000 to 2015) in University the type of causality scale administered as covariates. However, the Hospitals of 3 European countries: France (1 study), Finland (2 final number of studies identified was too low for this to be studies) and Spain (3 studies). The study period was 12 months in all undertaken. studies except for 1 that lasted 22 months.15 All the studies Publication bias was analysed by means of funnel plot, and Begg reported the number of DRDs and the number of deceased inpa- and Mazumdar rank correlation test and Egger's test of the tients. All studies reported the number of hospital admissions except intercept.25,26 Statistical analysis was achieved with comprehensive for 1 that reported an approximate value.31 The number of hospital- 27 meta-analysis v3. acquired DRDs was reported in all studies but 2 (in 1 of which it could be calculated).15,30 Table 2 summarises the characteristics of the included studies. 3 | RESULTS 3.1 | Study selection 3.3 | Assessment of methodological quality The database searches identified 1351 potentially relevant citations. On the basis of the NIH Quality Assessment tool for observational After duplicate removal, a total of 1029 studies were reviewed, 1016 studies, no study scored under 50%, which was our prespecified cri- (98.7%) were excluded after screening of the titles and abstracts. terion for low quality. The quality score ranged from 75 to 100%. After full text review of the remaining 13 studies, 5 publications met Three studies did not report details of exposures (item 8), such as the inclusion criteria. One article identified was an abstract published doses, routes of administration and time of starting treatment. 28 at the society's journal, for which detailed data were available in a Descriptions of potential confounding variables measured and doctoral thesis published online.29 Thus, a total of 6 studies were adjusted statistically (item 14) were also lacking in 3 studies included in the review (Figure 1: PRISMA Diagram).14,15,29–32 (Table 1). FIGURE 1 Study flow diagram MONTANÉ AND CASTELLS TABLE 2 Characteristics of included studies Main suspected Drug–drug Study author Study design ADR definition Causality assessment drug class (ATC) interactions (%) Main adverse drug reactions Juntti-Patinen30 Retrospective WHO - WHO-UMC L and B NR - Haemorrhages (35%, 26/75) - Infections in immunosuppressed patients (29%, 22/75) Pardo Cabello32 Retrospective WHO - WHO-UMC M and B 65% (11/17) - Haemorrhages (71%, 12/17) - Naranjo algorithm modified Lapatto-Reiniluoto31 Retrospective WHO - WHO-UMC L and B NR - Haemorrhages (33%, 17/52) - Infections in immunosuppressed patients (25%, 13/52) Pardo Cabello15 Retrospective WHO - Wulff B and N 30% (77*/256) - Haemorrhages (19%, 48/256) - Cardiac arrhythmia (14%, 35/256) Grévy29 Retrospective WHO - French method L and B 30.4% (56/184)) - Haemorrhages (46%, 85/184) - Naranjo algorithm - Infections in immunosuppressed patients (16%, 30/184) Montané14 Retrospective European directive - WHO-UMC L and B 44% (32/73) - Haemorrhages (47%, 34/73) - Naranjo algorithm - Infections in immunosuppressed patients (44%, 32/73) ADR: adverse drug reaction; ATC: Anatomical Therapeutic Chemical classification where B is Blood and blood forming organs, L is Antineoplastic and immunomodulating agents, M is Musculoskeletal System, and N is Nervous system; DRD: drug-related death; NR: not reported; WHO: World Health Organization; WHO-UMC: World Health Organization–Uppsala Monitoring Centre; *: calculated. 3663 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3664 TABLE 2 Continued DRD hospitalization DRD potentially Study author DRD patients’ age (mean, range) DRD patients’ sex (% women) stay mean, range (d) preventable % (tool) Risk factors 30 Juntti-Patinen 65 (3–93) 45.3(34/75) 8.2 (1–81) NR NR Pardo Cabello32 73 (39–91) 29.4 (5/17) 7.9 (2–39) NR - number of drugs - Charlson comorbidity index - length stay Lapatto-Reiniluoto31 74 (25–99) 36.5 (19/52) NR NR NR 15 Pardo Cabello 77 (69–86) 50.4 (129/256) NR NR - number of drugs - comorbidity Grévy29 76 (31–100) 43.5 (80/184) 15.5 (1–158) 34% (63/184; Olivier) NR Montané14 70 (19–94) 27.4 (20/73) 8.8 (0–57) 47% (34/73; Schumok-Thornton) - age - sex - number of drugs - Charlson comorbidity index ADR: adverse drug reaction; ATC: Anatomical Therapeutic Chemical classification where B is Blood and blood forming organs, L is Antineoplastic and immunomodulating agents, M is Musculoskeletal System, and N is Nervous system; DRD: drug-related death; NR: not reported; WHO: World Health Organization; WHO-UMC: World Health Organization–Uppsala Monitoring Centre; *: calculated. MONTANÉ AND CASTELLS 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MONTANÉ AND CASTELLS 3665 3.4 | ADR definition 3.6 | Characteristics of patients with DRD All studies but 1 used the World Health Organization (WHO) defini- Patients with DRDs were predominantly men (mean proportion of tion for an ADR as “a response to a drug that is noxious and 61.2%, ranging from 50 to 73%), with a mean age of 74.2 years (rang- unintended and that occurs at doses normally used in humans for pro- ing from 65 to 77 y). For studies that provided data (all except 2), the phylaxis, diagnosis or therapy of disease”.33 The remaining study used mean hospitalisation stays ranged from 8 to 16 days (Table 2). the definition proposed in the latest European pharmacovigilance leg- islation (Directive of the European Parliament and of the Council 2010/84/EU) as “a response to a medicinal product which is noxious 3.7 | Adverse drug reactions and unintended”.34 The European definition widened the WHO defi- nition of ADR (including medication errors, abuse, misuse as well as Haemorrhages were the most common ADR with proportion of doses and indications not authorized for the regulatory agencies or 39.0% (95% CI 26.5–53.2), ranging from 19 to 71% of DRDs (ESM 6). off-label uses). Infections in drug-immunosuppressed patients were the second lead- ing cause of DRD, with a proportion of 27.5% (95% CI 16.7–41.7), ranging from 16 to 44% (ESM 7). Another commonly reported cause 3.4.1.Pooled DRD incidences and occurrence rates of DRD in 1 study was cardiac arrhythmia (14%; Table 2). A total of 676 548 patients were hospitalised over 2000–2015. Of the 7578 patients who died, 657 died as a result of an ADR. The 3.8 | Suspected drugs number of DRD per year ranged from 17 to 256,15,32 the number of in-hospital deceased per year ranged from 820 to 1708,29,32 Cytostatic agents, classified as group L (Antineoplastic and and the number of hospital admitted patients per year ranged from immunomodulating agents), and antithrombotic agents, classified as 14,31 21 483 to 400 000. The overall incidence of DRD among group B (Blood and blood forming organs), were the most commonly hospitalised patients was 0.13% (95% confidence interval involved drug type. Drug–drug interactions were reported in 4 studies. [CI] 0.04–0.40), ranging from 0.01 to 0.74% (ESM 2). The occur- The proportion of drug–drug interactions was 37.4% of DRDs (95% rence rate of DRD among deceased inpatients was 7.3% (95% CI CI 28.3–47.6); ranging from 30 to 64% (ESM 8; Table 1). All except 4.1–12.5), ranging from 3 to 18% (ESM 3). The incidence of 1 of the DRDs due to drug–drug interactions, were due to pharmaco- hospital-acquired DRD among hospitalised patients was 0.05% dynamic and synergistic interactions. (95% CI 0.01–0.23, I2 = 98.9%) ranging from 0.03 to 0.4% (ESM 4). The occurrence rate of hospital-acquired DRD among deceased inpatients was 2.7%, (95% CI 1.0–6.9, I2 = 97.6%) ranging from 0.6 3.9 | Avoidability assessment to 11.5% (ESM 5; Table 3). As the final number of studies included in the statistical analysis Only 2 studies assessed potential avoidability of DRDs, using 2 differ- was low, we did not determine the sources of between-study variabil- ent tools: Olivier score39 and Schumock–Thornton criteria.40 One ity by means of meta-regression. study29 found that the main reasons for causing preventable DRDs were that drugs were not appropriate for the patient's condition and that the duration of treatment was too long. The other study14 found 3.5 | Causality assessment that the drug–drug interactions involved in the DRDs were the main cause of avoidability. Proportions of DRD avoidability were 34 and All the studies reported the tool used for causality assessment; 3 stud- 47%, respectively (Table 2). ies used 2 different causality tools.14,29,32 The WHO–Uppsala Moni- toring Centre criteria and the Naranjo algorithm were the most frequently used tools.35,36 The French algorithm and the Wulff criteria 3.10 | Risk factors assessment were the other tools used for causality assessment in 1 study each37,38 (Table 2). In 2 studies using the WHO–Uppsala Monitoring Three (50%) studies reported the assessment of risk factors for DRDs. Centre criteria, cases were included only if DRD attribution causality The number of administered drugs and the effect of comorbidity were 30,31 was certain or probable, excluding all possible DRD cases. In the the most consistent findings of these studies, with positive associa- other 2 studies using the Naranjo algorithm, DRD were probable or tions between these risk factors and the rate of DRD. The number of possible related to drugs in 100 and 98,7% of patients, respec- administered drugs was assessed using mean, median and ≥10 drugs, tively.14,29 Finally, the study using the Wulff criteria, DRDs were clas- depending on the study. The comorbidity was assessed using the sified as suspected of causing death or suspected of contributing to Charlson score in 2 studies and having ≥4 diseases in 1 study. Mixed deaths. results were found for sex and age, with 2 studies finding no 3666 TABLE 3 Study characteristics and incidences of drug-related death (DRD) and hospital-acquired DRD Number of admitted Number of deceased Number of deceased Number of Number of hospital-acquired Study author Country Study period year patients (A) inpatients (B) inpatients assessed (C) DRDs (D) DRDs (E) Juntti-Patinen30 Finland 2000 141 484 1547 1511 75 NR 32 Pardo Cabello Spain 2004 21 541 820 289 17 17 Lapatto-Reiniluoto31 Finland 2012 Near 400 000 1708 1708 52 10 15 Pardo Cabello Spain 2009 01–10/2010* 34 590 1400 1388 256 161** Grévy29 France 2014 57 450 1646 1646 184 69 14 Montané Spain 2015 21 483 1135 1036 73 6 Meta-analysis % (CI 95%) Europe 2000–2015 676 548 8256 7578 657 263 *: 22 months; **: calculated DRD: drug-related death; CI: confidence interval. MONTANÉ AND CASTELLS 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MONTANÉ AND CASTELLS TABLE 3 Continued DRD incidence of DRD occurrence rate of Hospital-acquired DRD incidence Hospital-acquired DRD occurrence Hospital-acquired DRD Study author hospitalized patients (D/A) deceased inpatients (D/C) of inpatients (E/A) rate of deceased inpatients (E/C) incidence of DRD (E/D) Juntti-Patinen30 0.053% (75/141484) 4.96% (75/1511) Not calculable Not calculable Not calculable 32 Pardo Cabello 0.08% (17/21541) 5.88% (17/289) 0.08% (17/21541) 5.88% (17/289) 100% (17/17) Lapatto-Reiniluoto31 0.013% (52/400000) 3.04% (52/1708) 0.0025% (10/400000) 0.58% (10/1708) 19.2% (10/52) Pardo Cabello15 0.74% (256/34590) 18.44% (256/1388) 0.5% (161/34590) 11.6% (161/1388) 63% (161/256) Grévy29 0.32% (184/57450) 11.18% (184/1646) 0.12% (69/57450) 4.19% (69/1646) 37.5% (69/184) Montané14 0.34% (73/21483) 7.05% (73/1036) 0.028% (6/21483) 0.58% (6/1036) 8.2% (6/73) Meta-analysis % (CI 95%) 0.1% (0.0–0.4) 7.3% (4.1–12.5) 0.1% (0.0–0.2) 2.7% (1.0–6.9) 37.4% (17.7–62.4) *: 22 months; **: calculated DRD: drug-related death; CI: confidence interval. 3667 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3668 MONTANÉ AND CASTELLS association with DRD and another finding that men and younger having other comorbidities and pharmacokinetic changes that facili- patients have a higher risk of DRD. tate the appearance of ADRs and fatal outcomes.43–45 Regarding the characteristics of ADR, there was also a high vari- ability in the reporting of the types of ADRs identified, making a sum- 3.11 | Publication bias marized report problematic. The 2 most common ADRs related to death were haemorrhages and infections in drug-immunosuppressed All funnel plots were reasonably symmetrical (ESM 9–12) and neither patients, which was similar to other meta-analyses or population- Egger nor Begg tests were statistically significant for any outcome. based studies.12,23 Both these ADRs are type A, that is dose-depen- dent, augmented pharmacological effect, predictable and potentially avoidable reactions.46 For cerebral haemorrhages in patients with 4 | DISCUSSION head trauma or high blood pressure, antithrombotic drugs could be a contributing factor. Accordingly, 2 pharmacological groups were In this meta-analysis, we investigated the epidemiology of DRDs in responsible for most ADR, chemotherapy (group L) and anti- the hospital setting. We found that the DRD occurrence rate of thrombotic agents (group B), which was similar to the findings of other deceased hospital inpatients has been infrequently studied. All the meta-analyses.12,41,42 Chemotherapy-induced neutropenia and subse- included studies were performed in the last 2 decades. Overall, DRD quent infections were an important cause of mortality, where medulla occurrence rate of deceased hospital inpatients was 7.3% and >1/3 aplasia is a pharmacologic effect often necessary for chemotherapy were hospital-acquired. Both rates are high and denote that is a rele- effectiveness, but which can lead to patient death. Antithrombotic vant healthcare problem in European countries. Another important agents were associated with a significant risk of haemorrhage. Con- finding of this meta-analysis is the DRD incidence among hospitalised trary to other studies, nonsteroidal anti-inflammatory agents and car- patients, which was 0.1%. Other studies with different methodology, diovascular drugs (such as digoxin, diuretics or renin-angiotensin such as a European review and a meta-analysis including worldwide system inhibitors) were not frequently involved in DRD in this studies reported higher incidences, which were 0.5 and 0.2% respec- study.12,41 Moreover, nervous system agents and anti-infectives were 11,12 tively. The incidence of hospital-acquired DRD among hos- frequently involved in DRDs in studies assessing data from voluntary pitalised patients in our study was almost 0.1%. This result is similar reporting registers of ADR but not in our study.47,48 This difference to that of the recently published meta-analysis by Patel et al., despite could be explained by a bias in voluntary registers, where type A and differences in the study design, which included cross-sectional well-known ADRs are underreported, making voluntary registers not studies.41 useful for calculating ADR prevalence. Although all the included studies had a similar study design and Drug–drug interactions were present in >1/3 of deceased length, there was high heterogeneity in the incidence results due to patients, mainly due to pharmacodynamic interactions. In chemother- variability in the methodological characteristics, as follows: selection apy, combining drugs that have similar pharmacological effect but dif- and assessment of deceased patients were not well reported in all ferent mechanisms of action are common and necessary to achieve studies. When they were, in 1 study a preselection was carried out the desired immunosuppressant effect. A common synergic drug–drug (from predefined diagnoses of death suggesting ADR) leading to an interaction is a cytotoxic agent with a corticosteroid.49 Regarding anti- 14 underestimation of DRD incidence. Even though most studies used thrombotic agents, they are sometimes prescribed by medical special- WHO's definition for ADR, there was significant variability in causality ists for different diseases. A typical example of synergic drug–drug assessment of DRD cases, using at least 4 different causality methods, interaction is when patients receive AAS for ischaemic cardiopathy, which could lead to differences in ADR attribution. Moreover, in cilostazol for peripheral arteriopathy and anticoagulation for atrial 30,31 2 studies, the DRD cases were included only if attribution causal- fibrillation. ity was certain or probable, excluding possible DRD cases, thus possi- Avoidability of DRD was not widely assessed in the included bly leading to an underestimation of the number of DRD cases. A studies, but when it was, rates were higher than in other studies clear example of this is a cerebral haemorrhage in an anticoagulated which also assessed fatal drug poisoning.50 This could be explained as patient with a preceding trauma, which could be classified as a possi- the tools used differed, as well as the population studied. ble DRD or as a probable DRD with a contributing role, depending on The risk factors related to DRD were not systematically assessed. the study. There was also great variability in the sample size of stud- The number of drugs and comorbidities were the most commonly ies, with the number of patients admitted and the number of patients reported risk factors related to DRD in this and other meta-ana- assessed, varying from 300 to almost 2000. lyses.51 Conversely, older age and female sex, despite having been Demographic characteristics of patients with DRD showed a pre- associated with an increased risk for ADR-related hospitalisation,52 dominance of elderly men, which surprisingly diverged from other were not associated with DRD in 2 studies included in this meta- studies where the proportion of women is higher23; although elderly analysis and were associated with a reduced risk of DRD in another. 41,42 people are also the group with higher DRD incidences. This could Differences in how the effect of confounding covariates were be due to elderly patients often being polymedicated, as well as analysed may explain the discrepancies between these findings. 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MONTANÉ AND CASTELLS 3669 We limited our meta-analysis to studies conducted in European hospital, and in at least 1 in 14 deceased patients, denoting that is a countries in order to have greater homogeneity of data. Surprisingly, relevant healthcare issue. Despite the similarity of the design of the all the included studies were single-centre and were performed in only studies included, high heterogeneity was found. The number of drugs 3 European countries and 4 university hospitals, leading to low repre- and comorbidities were the most common risk factors related to sentativeness of the obtained results. In Spain and Finland, DRD. The limited number of studies in European countries highlights researchers compared the studies' results at 2 different periods of the need for more research in this area. time. This review clearly shows that few studies have been conducted using this epidemiological methodology that assesses DRD cases AC KNOW LEDG EME NT S among deceased inpatients. Therefore, multicentre studies conducted We would like to thank the authors of the included studies: Professor in several European countries assessing the epidemiology of DRD are Pertti Neuvonen and Dr Outi Lapatto-Reiniluoto from Finland, Dr needed to gain better knowledge of DRD incidences among deceased Puche from Spain, and Dr Grévy from France for their help providing inpatients in this geographic area. In the same way, in low or middle- the lacking study data. The authors sincerely appreciate the help of income countries, epidemiological DRD data are also scarce. The Alethea Charlton for reviewing the language of this manuscript. results of a study conducted in South Africa with a methodology simi- This research did not receive any specific grant from funding lar to the studies included in this review, showed that epidemiological agencies in the public, commercial, or not-for-profit sectors. incidence data for DRD of inpatients were extremely high; and anti- infectives for the management of HIV and tuberculosis were the drugs COMPETING INTER ESTS most commonly implicated in DRD.13 There are no competing interests to declare. CONT RIB UTOR S 4.1 | Strengths and limitations Dr Montané conceived the idea, contributed to the study design, con- ducted the literature search, reviewed the selected cases, undertook The main limitation of this meta-analysis is that it was limited to stud- data acquisition of included cases, performed analysis and interpreta- ies assessing incidence of DRD among deceased inpatients. Therefore, tion of the data, and wrote the first draft of manuscript. studies that assessed fatal ADR as part of a broader study investigat- Dr Castells contributed to study design, review of the selected ing all ADRs in hospitalised patients were excluded. Other study limi- cases, data acquisition of included cases, analysis and interpretation tations are related to the quality of the studies included, which, in of the data, and writing the manuscript. general, were deemed to have low risk of bias. Large statistical hetero- geneity was found, which reduces confidence in the incidences and OR CID rates calculated. Reporting bias can threaten the validity of any meta- Eva Montané https://orcid.org/0000-0002-1691-7652 analysis. However, no evidence of small study effects was suggested Xavier Castells https://orcid.org/0000-0002-2619-7273 by the funnel plots, nor by the Egger and Begg tests, indicating that publication bias is unlikely. Nevertheless, it must be stressed that the RE FE RE NCE S validity of these plots and tests is low when the number of studies 1. Sultana J, Cutroneo P, Trifirò G. Clinical and economic burden of included in the meta-analysis is, like in our review, also low.53 The adverse drug reactions. J Pharmacol Pharmacother. 2013;4(Suppl 1): restriction of including studies conducted in Europe limits S73-S77. https://doi.org/10.4103/0976-500X.120957 generalisability. 2. Watanabe JH, McInnis T, Hirsch JD. Cost of Prescription Drug- Related Morbidity and Mortality. Ann Pharmacother. 2018;52(9):829- A strength of this study is that we performed a systematic review 837. https://doi.org/10.1177/1060028018765159 using the most important data bases, searching for a long period of 3. Strengthening pharmacovigilance to reduce adverse effects of time and with no language restriction, which allowed us to include as medicines. MEMO/08/782. Brussels, December – November 2008. many studies as possible. Another relevant strength of the present http://europa.eu/rapid/press-release_MEMO-08-782_en.htm. Accessed September 23, 2020. study was the fact that we contacted the authors of the studies and 4. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reac- hence were able to include 1 more study and calculate the maximum tions in hospitalized patients. A meta-analysis of prospective studies. of the DRD and hospital-acquired DRD occurrence rates, as well as JAMA. 1998;279(15):1200-1205. DRD and hospital-acquired DRD incidences for meta-analysis. No 5. Taché SV, Sönnichsen A, Ashcroft DM. Prevalence of adverse drug events in ambulatory care: a systematic review. Ann Pharmacother. studies were excluded for not providing enough data to calculate inci- 2011;45(7–8):977-989. https://doi.org/10.1345/aph.1P627 dence or occurrence rates. 6. Lisha J, Annalakshmi V, Maria J, Padmini D. Adverse Drug Reactions in Critical Care Settings: A Systematic Review. Curr Drug Saf. 2017;12(3):147-161. https://doi.org/10.2174/157488631266 5 | C O N CL U S I O N 6170710192409 7. Oscanoa TJ, Lizaraso F, Carvajal A. Hospital admissions due to adverse drug reactions in the elderly. A Meta-Anal Eur J Clin Our findings suggest that in European hospitals drugs are an impor- Pharmacol. 2017;73(6):759-770. https://doi.org/10.1007/s00228- tant cause of death, occurring in 1 out of 1000 patients admitted to 017-2225-3 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 3670 MONTANÉ AND CASTELLS 8. Smyth RM, Gargon E, Kirkham J, et al. Adverse drug reactions in 27. Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive meta- children--a systematic review. PLoS One. 2012;7(3):e24061. https:// analysis version 3.3.070. Englewood, NJ: Biostat; 2014. doi.org/10.1371/journal.pone.0024061 28. Grévy A, Lepelley M, Khouri C, Mallaret M, Roustit M. Drug-related 9. Impicciatore P, Choonara I, Clarkson A, Provasi D, Pandolfini C, deaths: retrospective analysis in French Universitary Hospital. Clin Bonati M. Incidence of adverse drug reactions in paediatric in/out- Ther. 2017;39(8):e93-e94. https://doi.org/10.1016/j.clinthera.2017. patients: a systematic review and meta-analysis of prospective stud- 05.293, https://doi.org/10.1016/j.clinthera.2017.05.293 ies. Br J Clin Pharmacol. 2001;52(1):77-83. https://doi.org/10.1046/j. 29. Grévy A. Décès imputables aux médicaments: analyse rétrospective 0306-5251.2001.01407.x au CHU de Grenoble-Alpes en 2014. Sciences pharmaceutiques. 10. Davies EC, Green CF, Taylor S, Williamson PR, Mottram DR, 2017. hdumas-01719243i. https://pdfs.semanticscholar.org/8339/ Pirmohamed M. Adverse drug reactions in hospital in-patients: a pro- ae2e3f19875bef471805976e10e82eb73efd.pdf?_ga=2.50913321. spective analysis of 3695 patient-episodes. PLoS One. 2009;4(2): 1629777483.1564047146-1795757843.1562769786. Accessed e4439. https://doi.org/10.1371/journal.pone.0004439 September 23, 2020. 11. Bouvy JC, De Bruin ML, Koopmanschap MA. Epidemiology of 30. Juntti-Patinen L, Neuvonen PJ. Drug-related deaths in a university Adverse Drug Reactions in Europe: A Review of Recent Observa- central hospital. Eur J Clin Pharmacol. 2002;58(7):479-482. tional Studies. Drug Saf. 2015;38(5):437-453. 31. Lapatto-Reinilouto O, Patinen L, Niemi M, Backman JT, Neuvonen PJ. 12. Patel TK, Patel PB. Mortality among patients due to adverse drug Drug-Related Inadvertent Deaths in a University Hospital--A Declin- reactions that lead to hospitalisation: a meta-analysis. Eur J Clin ing Trend. Basic Clin Pharmacol Toxicol. 2015;117(6):421-426. Pharmacol. 2018;74(6):819-832. https://doi.org/10.1007/s00228- 32. Pardo Cabello AJ, González Contreras LG, Manzano Gamero MV, 018-2441-5 Gómez Jiménez FJ, Puche CE. Prevalence of fatal adverse drug reac- 13. Mouton JP, Mehta U, Parrish AG, et al. Mortality from adverse drug tions in hospitalized patients. Int J Clin Pharmacol Ther. 2009;47(10): reactions in adult medical inpatients at four hospitals in South Africa: 596-602. a cross-sectional survey. Br J Clin Pharmacol. 2015;80(4):818-826. 33. World health organization technical report series (1972, no 498). 14. Montané E, Arellano AL, Sanz Y, Roca J, Farré M. Drug-related deaths International Drug Monitoring. The Role of the National Centres. in hospital inpatients: A retrospective cohort study. Br J Clin Report of a WHO Meeting. Geneva, Switzerland. http://apps.who. Pharmacol. 2018;84(3):542-552. int/iris/bitstream/10665/40968/1/WHO_TRS_498.pdf. Accessed 15. Pardo Cabello AJ, Del Pozo GE, Gómez Jiménez FJ, Mota September 23, 2020. Rodríguez C, Luna Del Castillo J, Cañas E. Drug-related mortality 34. Commission Directive 2010/84/EU of the European Parliament and among inpatients: a retrospective observational study. Eur J Clin the Council of 15 December 2010 amending as regards Pharmacol. 2016;72(6):731-736. pharmacovigilance, Directive 2001/83/EC on the Community code 16. Gaeta M, Campanella F, Capasso L, et al. An overview of different relating to medicinal products for human use. Official Journal of the health indicators used in the European Health Systems. J Prev Med European Union 31.12.2010: L348/74-L348/99. http://ec.europa. Hyg. 2017;58(2):E114-E120. eu/health/files/eudralex/vol-1/dir_2010_84/dir_2010_84_en.pdf. 17. Simó MJ. Utilización de medicamentos en España y en Europa [Use of Accessed September 23, 2020. prescription drugs in Spain and Europe]. Aten Primaria. 2012;44(6): 35. World Health Organization. The use of the WHO-UMC system for stan- 335-347. https://doi.org/10.1016/j.aprim.2011.06.009 dardized case causality assessment. Uppsala: The Uppsala Monitoring 18. Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred Centre. https://www.who-umc.org/media/164200/who-umc- reporting items for systematic reviews and meta-analyses: the PRI- causality-assessment_new-logo.pdf. Accessed September 23, 2020. SMA statement. PLoS Med. 2009;6(7):e1000097. https://doi.org/10. 36. Naranjo CA, Busto U, Sellers EM, et al. A method for estimating the 1371/journal.pmed.1000097 probability of adverse drug reactions. Clin Pharmacol Ther. 1981;30 19. PROSPERO: International prospective register of systematic reviews. (2):239-245. http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID= 37. Bégaud B, Evreux JC, Jouglard J, Lagier G. Imputation of the unex- CRD42019105618 . Accessed September 23, 2020. pected or toxic effects of drugs. Actualization of the method used in 20. WHO. Collaborating Centre for Drug Statistics Methodology, France. Therapie. 1985;40(2):111-118. Guidelines for ATC classification and DDD assignment 2019. 38. Wulff HR. Rational diagnosis and treatment. Oxford, England: Oslo, Norway, 2018. http://www.whocc.no. Accessed September 23, Blackwell; 1981. 2020. 39. Olivier P, Caron J, Haramburu F, et al. Validation d'une échelle de 21. National Heart Lung and Blood Institute. Quality Assessment Tool for mesure: exemple de l'échelle française d'évitabilité des effets Observational Cohort and Cross-Sectional Studies - NHLBI, NIH. indésirables médicamenteux. Therapie. 2005;60(1):39-45. National Institutes of Health. https://www.nhlbi.nih.gov/health- 40. Schumock GT, Thornton JP. Focusing on the preventability of adverse topics/study-quality-assessment-tools. Accessed September 23, drug reactions. Hosp Pharm. 1992;27(6):538. 2020. 41. Patel PB, Patel TK. Mortality among patients due to adverse drug 22. Buajordet I, Ebbesen J, Erikssen J, Brørs O, Hilberg T. Fatal adverse reactions that occur following hospitalisation: a meta-analysis. Eur J drug events: the paradox of drug treatment. J Intern Med. 2001;250 Clin Pharmacol. 2019;75(9):1293-1307. https://doi.org/10.1007/ (4):327-341. s00228-019-02702-4 23. Wester K, Jönsson AK, Spigset O, Druid H, Hägg S. Incidence of fatal 42. Oscanoa TJ, Lizaraso F, Carvajal A. Hospital admissions due to adverse drug reactions: a population based study. Br J Clin Pharmacol. adverse drug reactions in the elderly. A meta-analysis. Eur J Clin 2008;65(4):573-579. https://doi.org/10.1111/j.1365-2125.2007. Pharmacol. 2017;73(6):759-770. https://doi.org/10.1007/s00228- 03064.x 017-2225-3 24. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsis- 43. Mangoni AA, Jackson SH. Age-related changes in pharmacokinetics tency in meta-analyses. BMJ. 2003;327(7414):557-560. and pharmacodynamics: basic principles and practical applications. Br 25. Begg CB, Mazumdar M. Operating characteristics of a rank correla- J Clin Pharmacol. 2004;57(1):6-14. https://doi.org/10.1046/j.1365- tion test for publication bias. Biometrics. 1994;50(4):1088-1101. 2125.2003.02007.x 26. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta- 44. Routledge PA, O'Mahony MS, Woodhouse KW. Adverse drug reac- analysis detected by a simple, graphical test. BMJ. 1997;315(7109): tions in elderly patients. Br J Clin Pharmacol. 2004;57(2):121-126. 629-634. https://doi.org/10.1136/bmj.315.7109.629 https://doi.org/10.1046/j.1365-2125.2003.01875.x 13652125, 2021, 10, Downloaded from https://bpspubs.onlinelibrary.wiley.com/doi/10.1111/bcp.14799 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [29/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License MONTANÉ AND CASTELLS 3671 45. Shi S, Mörike K, Klotz U. The clinical implications of ageing for ratio- 52. Angamo MT, Chalmers L, Curtain CM, Bereznicki LR. Adverse-Drug- nal drug therapy. Eur J Clin Pharmacol. 2008;64(2):183-199. https:// Reaction-Related Hospitalisations in Developed and Developing doi.org/10.1007/s00228-007-0422-1 Countries: A Review of Prevalence and Contributing Factors. Drug 46. Rawlins MD, Thompson JW. Mechanisms of adverse drug reactions. Saf. 2016;39(9):847-857. https://doi.org/10.1007/s40264-016- In: Davies DM, ed. Textbook of adverse drug reactions. Oxford: Oxford 0444-7 University Press; 1991:18-45. 53. Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, 47. Mittmann N, Liu BA, Iskedjian M, et al. Drug-related mortality in Welch VA (editors). Chapter 13: Assessing risk of bias due to missing Canada (1984-1994). Pharmacoepidemiol Drug Saf. 1997;6(3):157- results in a synthesis. In: Cochrane Handbook for Systematic Reviews of 168. https://doi.org/10.1002/(SICI)1099-1557(199705)6:3<157:: Interventions version 6.1 (updated September 2020). Cochrane, 2020. AID-PDS260>3.0.CO;2-J www.training.cochrane.org/handbook. Accessed December 4, 2020. 48. Leone R, Sottosanti L, Luisa Iorio M, et al. Drug-related deaths: an analysis of the Italian spontaneous reporting database. Drug Saf. 2008;31(8):703-713. https://doi.org/10.2165/00002018-200831 SUPPORTING INF ORMATION 080-00007 Additional supporting information may be found online in the 49. van Leeuwen RW, Brundel DH, Neef C, et al. Prevalence of potential Supporting Information section at the end of this article. drug-drug interactions in cancer patients treated with oral anticancer drugs. Br J Cancer. 2013;108(5):1071-1078. 50. Jönsson AK, Hakkarainen KM, Spigset O, Druid H, Hiselius A, Hägg S. Preventable drug related mortality in a Swedish population. How to cite this article: Montané E, Castells X. Epidemiology Pharmacoepidemiol Drug Saf. 2010;19(2):211-215. https://doi.org/10. of drug-related deaths in European hospitals: A systematic 1002/pds.1890 review and meta-analysis of observational studies. Brit Jnl 51. Leelakanok N, Holcombe AL, Lund BC, Gu X, Schweizer ML. Associa- Clinical Pharma. 2021;87(10):3659–3671. https://doi.org/10. tion between polypharmacy and death: A systematic review and meta-analysis. J Am Pharm Assoc2003. 2017;57(6):729-738e10. 1111/bcp.14799 https://doi.org/10.1016/j.japh.2017.06.002