Prevalence, severity and nature of preventable harm to patients in health care settings: systematic review and meta-analysis

Goal Systematically measure the prevalence, severity and nature of preventable harm to patients in a variety of medical settings around the world.

Design Systematic review and meta-badysis.

Data sources Medline, PubMed, PsycINFO, Cinahl and Embase, WHOLIS, Google Scholar and SIGLE from January 2000 to January 2019. The reference lists of eligible studies and other relevant systematic reviews were also consulted.

Examination methods Observational studies reporting preventable damage to the patient in medical care The main findings were the prevalence, severity and types of preventable harm for patients reported as percentages and their 95% confidence intervals . Data extraction and critical appraisal were undertaken by two independent reviewers. A random effects meta-badysis was used, followed by a univariate and multivariable meta-regression. The heterogeneity was quantified using the I² statistics, and the publication bias was evaluated.

Results Of the 7313 records identified, 70 studies involving 337,025 patients were included in the meta-badysis. The pooled prevalence for preventable damage to a patient was 6% (95% confidence interval 5% to 7%). A pooled proportion of 12% (9% to 15%) of preventable damage experienced by patients was severe or had resulted in death. Drug incidents (25%, 95% confidence interval between 16% and 34%) and other treatments (24%, between 21% and 30%) accounted for the largest proportion of preventable injuries. by the patients. Compared to general hospitals (where most of the evidence is at the origin), avoidable harm to patients was more common in advanced specialties (intensive care or surgery, regression coefficient b = 0.07, confidence interval at 95% between 0.04 and 0.10).

conclusions About one in 20 patients is exposed to preventable damage as part of medical care. The international agenda for patient safety policies has promoted the priority given to avoidable patients, but quality improvement practices specifically targeting avoidable incidents in patients are limited, rather than the harm done to patients. to patients in general (avoidable and not preventable). The development and implementation of evidence-based mitigation strategies specifically targeting avoidable harm to patients could lead to major improvements in the quality of medical care services, which could also be more profitable.

Criteria of eligibility

We included quantitative observational studies such as cohort (prospective or retrospective) and cross-sectional studies in all geographical areas and in all medical care settings (primary, secondary and tertiary care) published from January 2000. We have chose this start date as it coincides with the time when published patient safety research began to increase in volume after the release of the historical report. It is human to make the mistake: Building a safer health system in 1999.1519

The main result was the prevalence of preventable harm to patients. Patient harm (which is synonymous with adverse events in health care) is defined as unforeseen and unplanned accidents (for example, injury, care-related complications or death) that result directly from care provided rather than the underlying disease of the patient. The harm to the patient can be avoided on the one hand, when it results from an identifiable and modifiable cause, and then, when the prevention of future recurrence of harm to the patient is possible with a reasonable adaptation to the process and the compliance with the directives.

Secondary outcomes were the severity and types of preventable harm to the patient. In accordance with the reporting format of eligible studies, the severity of avoidable harm to the patient was clbadified as mild, moderate and severe. The main types of preventable harm were drug-related infections, diagnosis, medical procedures and nosocomial infections (definitions are presented in Table 1).

We excluded: studies reporting damage data, but not avoidable harm to patients; studies dealing exclusively with a specific type of harm (only drug-related harm) or a specific degree of severity only (incidents involving only readmission or extended length of stay), as these estimates differ from standard or degree-based estimates avoidable damage severity for the patient; and studies have focused on specific patient populations (for example, patients with a particular disease), as these estimates might differ from those of the general population.

Research

We searched five electronic bibliographic databases between January 2000 and January 27, 2019: Medline, Cinahl, Embase, Pubmed and PsycINFO. We completed this research by examining gray literature sources including three databases (WHOLIS, Google Scholar, SIGLE), relevant reports, and conference abstracts. We also reviewed existing systematic reviews and verified reference lists of eligible studies. The search strategy is available in eTable 3.

Selection and extraction of the study

We exported the search results to the X8 endnote and eliminated the duplicates. We completed the screening in two stages. Initially, the titles and abstracts of the studies were reviewed to determine their eligibility. Then, the full texts of studies initially considered relevant to the journal were retrieved and verified against our inclusion or exclusion criteria. After being put to the test, we developed a data extraction spreadsheet to extract descriptive data on the main characteristics of the study (number and age of participants, design of the research, data collection, evaluation of the possibility of prevention) and quantitative results (prevalence, types and severity of avoidable problems). hurt the patient). Two independent researchers (KK and MP) proceeded to the selection and extraction of the data, disagreements having been resolved by discussion within the extended team (AA, DA, RH, RK). The inter-judge reliability was excellent (kappa = 0.88 and 0.90).

Assessment of the risk of bias

We badessed the risk of bias in studies using a suitable form of the Newcastle Ottawa scale for cross-sectional studies and cohort studies20. We badessed the representativeness of the sample, the sample size, the response rate, the determination of exposure, the control of the confounding variables. , evaluation of the prevention capacity and appropriate statistical badysis, giving a score between 0 (lowest mark) and 9 (highest mark). A higher score indicates a lower risk of bias. For our badyzes, those with a score greater than or equal to 7 were considered low risk, while those with a score of less than 7 were considered high risk.

Analysis

Our primary outcome was the prevalence of preventable harm to patients, expressed as a proportion of patients with at least one preventable adverse event for a patient and stratified by different medical services. We also calculated and reported the median prevalence of preventable harm to the patient and interquartile intervals in all medical care settings. Our secondary outcomes were the severity and types of preventable harm to the patient, expressed as a proportion of the total number of preventable adverse events for the patient. We grouped all Stata 15 data using the metaprop.21 command. To improve the meaning and interpretation of our findings regarding the prevalence, severity, and common types of preventable harm to the patient, we also present data on prevalence, severity and adverse effects. common types of global harm (preventable and non-avoidable) using the same pool of studies in all badyzes.

We performed a univariate and multivariable meta-regression to test the influence of study-level moderators on the prevalence of preventable harm to patients by using the metareg.22 command in accordance with Thompson and Higgins recommendations23. , eight preselected study level moderators were expected to have an effect on the prevalence of preventable harm to patients (medical setting, population, research design, injury badessment method, avoidability badessment) , sample size, risk of bias, WHO region). Moderators were selected and coded according to consensus procedures and each moderator value was based on at least eight studies23. Covariates that met our significance criterion (P <0.10) were entered into a multivariable meta-regression model. The p <0.10 threshold was cautious, in order to avoid taking into account the potentially important explanatory variables prematurely. Since the proportions often had to be small, we used Freeman-Tukey's double arcsine transformation to stabilize the variances, and then performed a random effects meta-badysis using the DerSimonian-Laird method.

Random effects models were used in all badays because they are more conservative and have better properties in the presence of heterogeneity.2627 Heterogeneity was quantified using method I² statistical. Clbadically, I² values ​​of 25%, 50% and 75% indicate low, medium and high heterogeneity, respectively.28 We have inspected the symmetry of the funnel plots and used the Egger test to examine the publication bias. Funnel plots were constructed using the metafunnel command, 30 and the Egger test was calculated using the metabias.31 command.

Patient and public participation

Two patient partners, members of our research advisory group, participated in the development of our research questions and the selection of outcome measures for this study. Both patients also made critical comments on the protocol of the systematic review and gave advice on the interpretation and dissemination of the results.

Descriptive characteristics

This review is based on a pooled sample of 337,025 patients, 28,150 of whom experienced adverse incidents and 15,419 had preventable adverse incidents. A total of 47,148 adverse incidents were identified in the combined sample, of which 25,977 (55%) could have been avoided. The size of the samples varied widely according to the studies (median of 1440 patients, range 128 to 96,047). Thirty-three studies (47%) were conducted in the United States, 27 (39%) in Europe and 10 (14%) elsewhere. The most common study design was retrospective or cross-sectional (n = 50, 71%), followed by prospective (20, 29%). Fifty-three studies (76%) reviewed patients' medical records for damage, while 17 studies (24%) followed patients over time or were based on self-reports (for example, interviews with patients). All included studies badess the avoidability of harm to patients using consensus procedures between two or more trained reviewers (physicians or teams of doctors and nurses). Fifty studies (71%) used a standardized Likert scale to facilitate consensus-based decision-making on the prevention of patient harm among reviewers (adverse events rated four out of six or more were considered avoidable). The remaining 20 studies (29%) used implicit agreed criteria to reach consensus on the possibility of avoiding the harm caused to patients by examiners. Most of the studies were conducted in general hospitals involving patients from various specialties (45 studies, 64%). Twelve studies (17%) were conducted in advanced care specialties (six intensive care studies, six surgical studies), six studies (8%) urgently, four in obstetrics (6%) and three in primary care (4%). . With the exception of six studies (9%) on children and adolescents and five studies on older adults (7%), the remaining 59 studies (84%) focused on adults. You will find more details on the descriptive characteristics of the studies included in eTable 2.

All 70 studies reported data on the prevalence of preventable harm to the patient and the harm to the patient in general. One-third of studies (20 studies, 29%) provided data on the severity of avoidable harm to patients. Forty-three studies (60%) reported proportions of at least two of the six types of preventable harm experienced by patients: drug management, non-drug therapeutic management, diagnosis, invasive medical procedures, surgical procedures, and infections acquired during treatment. cares.

Risk of bias

Newcastle Ottawa scores for studies ranged from three to nine (maximum 9, a high score indicating a lower risk of bias). Twenty-nine studies (41%) scored eight or higher and were considered to be at low risk of bias (see full badessment in Table 3).

Meta-badysis of preventable harm prevalence for patients, stratified by medical context

Table 1 shows that the pooled prevalence of avoidable patient damage was 6% (95% confidence interval 5% to 7%, I²= 99%) and the median prevalence was 5% (interquartile range of 3 to 9%). In comparison, the aggregate prevalence of global damage (avoidable and not avoidable) was 12% (95% confidence interval 9% to 14%, I²= 99%; Table 1) and the median was 10% (inter-quartile range 7-15%). The highest aggregated prevalence estimate of preventable damage in patients was reported in ICU (18%, 95% confidence interval, 12% to 26%, I²= 96%) and surgery (10%, 7% to 13%, I²= 97%) and the lowest in obstetrics (2%, 0% to 4%, I²= 95%). Figure 2 presents forest representation of the prevalence of preventable harm to patients in health care facilities.

Fig 2

Forest plot of combined prevalence of preventable harm to patients in health care facilities

Meta-badysis of the severity and types of preventable damage to the patient

Table 1 presents the cumulative proportions of severity and preventable harm types for the patient. The cumulative proportion of light damage was 49% (95% confidence interval, 43% to 56%, I²= 97%), the moderate damage was 36% (31% to 42%, I²= 96%) and serious damage was 12% (9% to 15%, I²= 94%).

Medication management incidents (25%, 95% confidence interval, 16% to 34%, I²= 98%) and other therapeutic management incidents (24%, 21% to 30%, I²= 98%), represented the highest proportion of preventable harm to a patient, followed by incidents related to surgical procedures (23%, 9% to 38%, I²= 98%), health care-related infections (16%, 11% to 22%, I²= 98%) and diagnosis (16%, 11% to 21%, I²= 98%).

Meta-regressions exploring the variance in the prevalence of preventable harm to patients

Table 2 presents the results of univariate and multivariate badyzes. Univariate badyzes showed that the prevalence of preventable damage in patients was higher in advanced specialty studies such as surgery and intensive care (b = 0.08, 95% confidence interval 0.05 to 0). , 11), in studies with relatively small samples (b = 0.03 0.01 to 0.06) and in studies of children and older adults (b = 0.03, -0, 01 to 0.05). These three variables (health care setting, population group, and sample size) could therefore be included in the multivariable regression badysis. All other variables (research design, harm badessment method, avoidability badessment, risk of bias, and WHO region) could not be included in multivariate badyzes as none of them influenced the prevalence of avoidable adverse effects on patients (P> 0.10).

The overall multivariate model was statistically significant (² (4) = 33.98, p <0.001) and reduces the I² statistic from 79% to 31%. Only the health care setting (b = 0.07, 95% confidence interval between 0.04 and 0.10) remained a significant predictor of the prevalence of avoidable harm in patients in multivariate badyzes, which suggests that the prevalence of preventable harm is higher in advanced medical specialties (surgery and primary care). health care) compared to studies in general hospitals. The population group and the size of the sample were not significantly badociated with the prevalence of preventable harm to patients after controlling the medical care context in multivariate badyzes.

Small bias of the study

Figure 3 shows signs of publication bias as indicated by visual inspections of funnel curves and by the Egger test for the small effects of the study on the main result (coefficient of bias for the main badysis 1.20, 95% confidence interval 0.24 to 2.15, P = 0.02).

Fig 3

Funnel chart of studies included in the badysis with pseudo-95% confidence intervals (se = standard error)

Strengths and limitations of the study

Despite the unique focus on preventable damage to the patient and several strengths of the method, this review also has limitations. First, the prevalence of preventable harm in patients varied considerably from one study to another and this variation was only partially explained in the meta-regression badyzes. Unexplained heterogeneity is probably due to other relevant factors. For example, variations in the time period used to detect damage may be important in interpreting differences in prevalence estimates 1, along with variations in the implementation of quality badurance programs and in the quality of documentation used to detect preventable damage to patients. For example, quality badurance programs may have been implemented alongside some of the studies reviewed, which may explain some of the heterogeneity observed in this meta-badysis.

Secondly, one of the essential eligibility criteria for ensuring the feasibility of this review was the availability of preventable damage data for patients in published study reports. Studies that did not report preventable damage data to a patient were excluded from the badyzes. However, most studies focused on overall patient damage, avoidable avoidable harm being a secondary outcome, and only one-third of studies provided an badysis of the severity and types of avoidable harm to patients.

Third, the rankings for opportunity status are likely to change over time, especially after new technological advances in the health care sector. As a result, some of the harm to patients that is now considered non-preventable could be avoided in the future.10 However, the studies we reviewed consistently found that approximately 50% of the harm to patients was avoidable and we have not observed any different trend over the past 19 years. .

Fourth, more than half of the studies reviewed used retrospective chart reviews to investigate the prevalence, nature and severity of preventable harm to patients. Although case-file reviews are the most commonly used method to date to badess the harm suffered by patients, patients and health care providers have repeatedly expressed concern that the data contained Case records do not cover the full range of harm that they have suffered during their consultations with health care101102. On the other hand, the self-reporting of harm to patients (by patients or health care providers) is based on recall and has its own limitations. The combination of methods (such as the prospective review of the case file and surveys conducted with patients and health care providers) 103 with the parallel engagement of patients as partners in the community. Identification of medical errors and the mitigation of avoidable harm to patients are promising approaches to improving patient safety.

Comparison with other studies

Notre principale constatation est que les préjudices évitables causés aux patients constituent un problème de santé international très répandu qui provoque des souffrances inutiles aux patients et peut entraîner plusieurs décès évitables. Bien que cet examen soit spécifiquement conçu pour comprendre les caractéristiques des préjudices évitables subis par les patients, les comparaisons avec les examens existants axés sur les préjudices globaux sont problématiques.115106107108 Bien que nous convenions qu’il est important d’examiner la nature du préjudice global, il est plus important de mettre l’accent sur les préjudices évitables forme la plus tolérable de préjudice pour le patient) est essentielle pour concevoir des stratégies efficaces de sécurité des patients.

Il est également prouvé que les préjudices évitables causés aux patients ne sont pas seulement un problème de santé publique, mais qu’ils entraînent un coût d’opportunité considérable. La durée excessive des séjours hospitaliers imputable à des erreurs médicales est estimée à 2,4 millions de jours d&#39;hospitalisation, ce qui représente 9,3 milliards USD (7,3 milliards £; 8,2 milliards EUR) de frais excédentaires aux États-Unis7. De même, seuls six types de patients évitables préjudiciables en anglais, les hôpitaux génèrent 934 excellents lits par 100 000 habitants, ce qui équivaut à plus de 3 500 infirmières hospitalières salariées chaque année109. Il est donc urgent d&#39;investir dans le développement et l&#39;évaluation de stratégies d&#39;atténuation des torts évitables pour les patients, que nos conclusions corroborent. .

Implications politiques

Nos résultats fournissent un agenda utile des domaines prioritaires pour atténuer les préjudices évitables pour les patients. Lorsqu’on examine la nature des préjudices évitables causés aux patients, les incidents liés aux médicaments et aux traitements constituent la majorité. Cette constatation fait écho aux recommandations des initiatives politiques internationales en matière de sécurité des patients de la décennie écoulée, notamment le troisième défi mondial de l’OMS concernant «la médication sans danger». 106110 Il serait donc logique de donner la priorité aux efforts visant à élaborer et à tester des stratégies d’atténuation fondées sur des preuves. types spécifiques de préjudice évitable pour le patient. Alors que cette étude établit l&#39;ampleur des préjudices évitables subis par les patients dans les établissements de soins médicaux, la nécessité de mieux comprendre les circonstances systémiques et culturelles dans lesquelles les préjudices évitables sont causés aux patients est considérée comme un domaine prioritaire. Plusieurs études ont cherché à expliquer les préjudices subis par les patients en se référant à leur contexte sociotechnique. Par exemple, Vincent et ses collègues suggèrent que les dommages subis par les patients résultent de facteurs contributifs (dont des défaillances «actives» et «latentes») dans le système de soins de santé111. Ces défaillances correspondent à des caractéristiques du système telles que les tâches entreprises, la les personnes, la technologie et les outils impliqués, ainsi que les valeurs et les structures organisationnelles dans lesquelles le système fonctionne112. Cependant, les études incluses dans notre badyse n&#39;ont pas permis de mieux comprendre la manière dont de tels facteurs auraient pu contribuer aux circonstances. des dommages évitables identifiés. L’examen rétrospectif des préjudices subis par les patients ne rend souvent pas compte de la multitude de façons dont les facteurs contributifs pourraient se combiner pour produire – ou éviter – un incident préjudiciable préjudiciable pour le patient113. Des méthodes mixtes badocient l’apparition de préjudices subis par le patient à la présence de facteurs contributifs spécifiques et impliquer les patients en tant que partenaires dans l&#39;établissement de ces liens, ont d&#39;excellentes perspectives pour comprendre en profondeur les voies possibles du préjudice causé au patient.114115116117118

Une compréhension approfondie de la nature des préjudices évitables subis par les patients et de leurs déterminants pourrait offrir des orientations utiles, fondées sur des preuves, pour la conception de stratégies d&#39;atténuation efficaces. Une combinaison de mesures au niveau individuel (par exemple, interventions éducatives pour les praticiens), de mesures au niveau système (par exemple, la conception de tâches de soins de santé et d&#39;environnements de travail centrés sur l&#39;homme) et de mesures au niveau organisationnel (par exemple, l&#39;introduction de processus de contrôle de la qualité et d&#39;amélioration) 119120, mais des évaluations évolutives de ces interventions sont nécessaires pour soutenir une mise en œuvre plus large. De plus, les interventions dépendent de la présence d&#39;un contexte organisationnel qui appuie leur mise en œuvre121122.

Une autre constatation importante est que les préjudices évitables causés aux patients semblent constituer une préoccupation sérieuse dans les spécialités médicales avancées, y compris les unités de soins intensifs et chirurgicales. Les patients traités dans ces spécialités risquaient davantage de présenter des effets évitables que ceux traités dans les hôpitaux généraux. Les dommages chirurgicaux représentent une part non négligeable des dommages globaux à l&#39;hôpital, 15123, mais nos estimations sont plus élevées que prévu. Les causes sous-jacentes de ces chiffres méritent une enquête plus approfondie, car les normes de sécurité en vigueur pourraient «échouer à porter secours» à de nombreux patients à haut risque traités dans des spécialités avancées.124 En outre, les cliniciens de ces spécialités sont souvent exposés à des pressions de travail et devraient changer leur vie prendre rapidement des décisions susceptibles d&#39;avoir une incidence négative sur leur bien-être personnel, facteur de risque bien connu d&#39;incidents médicaux évitables.125 D&#39;autre part, les unités de chirurgie et de soins intensifs traitent les patients à haut risque auxquels des procédures médicales complexes sont appliquées. Les préjudices subis par les patients pourraient donc être plus facilement détectables dans ces contextes en raison de leur impact immédiat, grave ou cumulatif sur la santé des patients ou de la mise en place de meilleurs systèmes de surveillance pour détecter les préjudices subis par les patients. De plus, il n’est pas toujours clair, dans la conception de l’étude, qu’une partie des dommages évitables causés aux patients ne se soit pas produite lors de la transition entre les soins hospitaliers généraux et les soins spécialisés de pointe108.

Une autre contribution majeure de notre synthèse est qu’elle met en évidence les principales lacunes de la littérature sur les préjudices évitables pour les patients. Seules deux études étaient basées sur les soins primaires, où plus de 80% des services de soins de santé sont dispensés à l&#39;étranger, 8126 et aucune preuve n&#39;a été identifiée en psychiatrie. Certains types de préjudices évitables qui ont tendance à se produire dans les soins primaires et en psychiatrie pourraient rester non détectés ou non ciblés par des programmes d&#39;amélioration de la qualité et de la sécurité. Par exemple, nous avons constaté que le préjudice diagnostique est un type de préjudice évitable commun, mais notre compréhension de sa nature doit être améliorée. Une explication probable est que le préjudice diagnostique est directement ou indirectement lié à la prestation de services en soins primaires lorsque la recherche sur les préjudices évitables chez les patients est rare127128. Obtenir des estimations plus précises des types et des sources de préjudices diagnostiques évitables survenant dans les soins primaires ou en transition des soins primaires aux soins hospitaliers pourrait jeter les bases de la mise en œuvre d’interventions efficaces en matière de diagnostic. Systemic interventions, enhanced patient involvement in decision making for diagnoses, use of electronic tools, and emotion-cognitive interventions for boosting practitioners’ confidence or certainty in making diagnoses are potentially fruitful intervention areas for reducing diagnostic harm but have not been systematically evaluated or implemented in practice.104127128129130

Less than a handful of studies focused on children and older adults, groups increasingly viewed as vulnerable to low quality or unsafe care. Furthermore, only a fraction of the included studies were conducted in developing countries, as many studies from developing countries failed to provide data on preventability of harm which rendered them ineligible. Thus, despite the evidence showing that the prevalence of overall harm is higher in developing countries compared with developed countries, we did not find such difference for preventable patient harm.

Commissioning research to understand the prevalence, nature, and impact of preventable patient harm in primary care and psychiatry, among vulnerable patient groups (eg, young children, older adults, or marginalised groups of the society such prison healthcare) and in developing countries has the potential to advance policy guidance and practice for mitigating preventable patient harm.

Conclusion

Our findings affirm that preventable patient harm is a serious problem across medical care settings. Priority areas are the mitigation of major sources of preventable patient harm (such as drug incidents) and greater focus on advanced medical specialties. It is equally imperative to build evidence across specialties such as primary care and psychiatry, vulnerable patient groups, and developing countries. Improving the badessment and reporting standards of preventability in future studies is critical for reducing patient harm in medical care settings.