Methods

The group began by identifying the types and causes of adverse events that are particularly harmful to patients. Major patient safety issues were identified using a multi-faceted, iterative approach: we first began with a literature search. We identified the major causes of harm and their underlying causes. We then consulted with experts on the committee as well as with external sources, such as the National Patient Safety Foundation, reviews by the Agency for Healthcare Research and Quality8 (which is the most comprehensive review of its kind) and epidemiologic studies from several nations including the US,9 Canada1 and Australia.10 We went back to the experts on the committee with the preliminary list of harms and their underlying causes for further feedback. After finalising the list, we shared it with external experts to get any final feedback. With the list of harms and their underlying causes formalised (table 1), we then sought out experts in each individual topic area to write a section of the report on their topic of expertise.

While there is some debate about the relationship between quality and safety, the Institute of Medicine in the US suggests that safety is one critical component of the delivery of high-quality care.11 Therefore, the committee chose to categorise the patient safety topics identified into structure, process and outcomes.12 The Agency for Healthcare Research and Quality (AHRQ) defines “structure” as the resources and organisational arrangements in place to deliver care, “process” as the activities of providers for delivering care and “outcomes” as the consequences of clinical activities by providers.13

Each identified topic was then reviewed in detail by lead experts who were asked to describe the basic epidemiology of the topic (eg, frequency, impact on care and preventability). These experts also identified major gaps in knowledge that need to be addressed through future research. Because healthcare systems, their focus and their impacts likely vary substantially based on the level of a nation's development, we asked experts to specify what is known about the individual topic in developing, transitional and developed nations. Nations were classified based on their category of national income (World Bank 2006 classification14), and nations listed as “high income” were considered developed, “medium income” as transitional and “low income” as developing. Each section was reviewed by at least three other members of the committee for completeness and balance. Finally, the entire report underwent external review by seven international experts (names and affiliations listed in the Acknowledgement section).

In this manuscript, we synthesise the key points from the individual sections and present the major findings.

Results

Overview

We identified 23 major patient safety topics for detailed examination (see table 1). Of these, eight were classified as structural, five as process and 10 as outcomes. While it was generally easier to separate the outcomes topics from others, the strong link between structural factors that underlie unsafe care and the unsafe processes they create necessarily lead to categorisation of some topics under one heading that could have arguably belonged in the other. The groupings were not meant to be definitive but created a practical framework for approaching the safety topics. Further, the topics covered activities and areas that could impact the health system at many different levels. For example, organisational accidents and failures obviously occur at the provider level and impact individual organisations such as hospitals or doctor practices. Counterfeit and substandard drugs, on the other hand, often represent failure to regulate and safeguard medications at the national level, impacting potentially all providers and patients in the country.

We found that unsafe patient care is common, that it is associated with significant morbidity and mortality throughout the world, and that much of the harm is likely amenable to intervention. Studies from the US suggest that approximately 3% to 4% of hospitalised patients suffer a serious adverse event,2 3 while studies from other developed nations, using less restrictive definitions, found that between 8% and 16% of hospitalised patients suffer an adverse event.1 4 A substantial proportion (between 30% and 50%) of adverse events are preventable, and most cause substantial harm, likely contributing to the deaths of tens of thousands of people in each of these nations. Although there have been few systematic prevalence studies in developing or transitional nations, existing evidence from these countries suggests that unsafe medical care is likely to be as—if not more—common15 16 than in developed nations. For example, one study of Mexican hospitals suggested that nearly one in four hospitalised patients developed a nosocomial infection.15

The data on structural and process factors that affect patient safety come almost exclusively from a small number of developed nations. This makes understanding the underlying causes of unsafe care or recommending solutions to improve safety in developing and transitional nations extremely difficult. Even in developed nations, there are still substantial gaps in knowledge about the structural and process factors that underlie unsafe care. We outline the key findings, beginning with these factors, and then the outcomes of such care.

Structural factors contributing to unsafe care

A major contributor to unsafe care is the breakdown of complex systems, which some have called “organisational accidents”. These breakdowns arise from combinations of factors originating at different levels of the system and can involve latent failures or poor oversight.17 A key structural issue that impacts safety is the inadequate number of qualified healthcare providers worldwide. Globally, 57 countries have an estimated deficit of 2.4 million doctors, nurses and midwives18 and thus face substantial challenges in meeting health-related Millennium Development Goals for improving the quality and safety of their healthcare systems. A closely related issue to inadequate staffing is production pressures, which refer to situations in which the optimal patient care capacity of a healthcare system or an individual healthcare provider has been exceeded. Providing for too many patients at one time distracts providers, forces greater reliability on memory to perform important actions and hinders effective communication among healthcare personnel. Each of these likely creates an environment for unsafe processes. For example, one analysis in 2005 identified communication problems as the single biggest cause of nearly 70% of sentinel events in the hospital setting in the US.19 The roles of organisational structure, capacity and communication breakdowns in producing unsafe care in developing and transitional countries have not been adequately studied.

A related threat hindering the delivery of safe care is provider fatigue. Doctors-in-training who work traditional 24-h shifts make 36% more serious medical errors in the care of their patients than comparable doctors not doing extended shift work. Fatigued doctors make up to five times as many serious diagnostic errors, report making four times as many fatigue-related errors that lead to a patient's death and suffer many more occupational injuries themselves.20–23 Although data on provider fatigue come primarily from developed nations, providers in developing and transitional nations are likely to be at least as susceptible to these threats.

We found that other key structural issues that may affect safety include the organisation's patient safety culture, which refers to shared attitudes, values and norms related to safety. A positive culture may result in improvements in safety practices through better communication, teamwork and knowledge sharing, although the evidence base underlying this assumption is weak. Further, there is scarce knowledge about how organisational factors combine with provider factors to affect patient safety culture. Another important topic includes the role that accreditation and regulations play in improving accountability and systems of care. While both are felt to likely impact safety, their actual level of influence on patient safety has not been empirically assessed and their roles are especially uncertain in developing and transitional countries.

Finally, human factors engineering (HFE) represents an important structural issue to understand the hazards of medical care and ways to minimise those risks. HFE techniques and heuristics can also assist in investigating adverse events24 when they do occur. Their primary value has been in improving troublesome design issues involving architecture, devices and clinical procedures25 (eg, anaesthesiology,26 surgery27 and nursing28). By understanding how individuals actually interface with technology, for example, HFE can reduce adverse event rates by maximising the human ability to use technology effectively.

Processes that underlie unsafe care

Misdiagnosis is understudied but represents a major type of error in healthcare with widely ranging rates of delayed and erroneous diagnosis. Even in the most highly developed countries with sophisticated technology, at least 10% to 15% of diagnoses are incorrect.29 The numbers from developing and transitional countries are surely higher and likely add substantial financial costs and create significant morbidity and mortality. For example, one review found high rates of overdiagnosis of malaria in developing nations with consequent underdiagnosis of pneumonia and other related disorders, leading to undertreatment and likely high rates of morbidity from the underlying condition.30 Another important failure of process is the lack of adequate follow-up of important tests. Data from developed countries suggest that only about half of critically important laboratory results indicating potentially life-threatening conditions were followed up by appropriate treatment in a timely manner.31 The rates of test follow-up in developing nations are also suboptimal and variable and cause serious lapses in patient care.32

Counterfeit and substandard drugs, defined as those that are mislabelled, missing active ingredients or include wrong active ingredients, pose a major risk to patient safety. It was classified by the group under unsafe processes (due to poor regulatory oversight of medication safety) but could be considered a structural failure of the healthcare system. Repeated use of counterfeit or substandard medicines can result in therapeutic failure, drug resistance or even death.33 34 Counterfeit drugs account for more than 10% of the global medicine market and up to 30% of medicines consumed in developing countries.33 35 It is likely that hundreds of thousands of people, if not more, die each year due to consumption of substandard medications,33 but the precise burden of the problem is unknown.

Unsafe injection practices also cause substantial morbidity and mortality in large parts of the world. In 2000, WHO estimated that some 16 billion injections were administered each year in transitional and developing nations and up to 40% of injections were given with syringes and needles reused without sterilisation; in some countries, this proportion was as high as 70%.36 These obviously contribute to high rates of infections with hepatitis viruses, human immunodeficiency virus (HIV) and other transmittable disease.

Outcomes of unsafe care

Adverse events can occur as a result of nearly any interaction with the healthcare system. Estimates from developed nations suggest that between 7.5% and 10.4% of hospitalised patients experience injuries due to medications alone.1 37 38 These adverse drug events (ADEs) cost tens of billions of dollars to healthcare systems around the world and have been estimated to contribute to 140 000 deaths each year in the US alone.39 40 Best estimates suggest that 28% to 56% of ADEs are preventable.41 The rates of ADEs from developing and transitional nations are largely unknown. Although the official rates of medication use are much lower in developing and transitional countries, the actual use of medications in these nations is hard to quantify.42 Therefore, whether ADEs are less common in these nations is unknown but they likely represent a major source of patient harm and economic costs.

The best evidence suggests that medical devices can also cause substantial harm. Errors that underlie device-related injuries are often categorised into three groups: manufacturer-related errors, user errors and use or design errors.43–50 In the US, more than 1 million adverse medical device events occur annually, at a rate of 6.3 events per 1000 patient days.51 Studies by WHO suggest that adverse medical device events might be particularly problematic in developing countries, where medical equipment is often improperly maintained or replaced, placing patients at great risk.52 53 One study from a transitional nation found that the rates of infection from medical devices alone were 34.2 per 1000 patient days in the hospital.53

Surgery and anaesthesia also present substantial safety risks. In the US, estimates suggest that surgical adverse events account for 48% of all adverse events and are preventable 54% to 74% of the time.2 54 The few available studies from the developing world have found surgical adverse event rates to be as much as fivefold to 10-fold higher.55–60 Improving the use of evidence-based practices could potentially reduce these rates dramatically.61

Nosocomial infections are reported to occur in approximately 5% to 10% of hospitalised patients in developed nations15 and between 25% and 40% in developing nations.15 16 One in four patients in intensive care may acquire an infection during a stay in hospital and one estimate suggests that these rates are twice as high in developing countries.16 Common types of nosocomial infections include nosocomial pneumonia, catheter-related infections and surgical infections. These events are not only common but also highly preventable. They represent a major source of morbidity and mortality, as well as substantial associated financial costs to health systems.

Another important source of infections from medical care is the use of unsafe blood products. Recent estimates indicate that 5% to 15% of HIV infections in developing countries result from unsafe blood transfusion.62 63 Unsafe blood poses a high risk for transmission of other blood-borne infections including hepatitis B, hepatitis C, syphilis, malaria, Chagas disease and West Nile fever. Studies demonstrate that nearly three in five countries lack an established quality system to screen collected blood for HIV and that 88.5% of blood units in sub-Saharan Africa are not screened for HIV in a reliable manner.64 65 It is not known what fraction of blood products in the developing world is tainted. Safe blood products are a particular concern for women of childbearing age, for whom severe haemorrhaging is a leading cause of maternal mortality.

Outcomes of unsafe care in vulnerable populations

With an estimated 7.6 million infant deaths during the perinatal period each year and approximately 600 000 deaths in women due to pregnancy or childbirth (99% of which occur in developing countries), maternal and child health remains a major concern worldwide.66–68 Although many such deaths result from lack of access to care, many are also due to unsafe care. No well-designed studies address what fraction of the morbidity and mortality of women and newborns is attributable to a lack of access to care or to the receipt of unsafe, poor-quality care. Given the importance of understanding the causes of high maternal mortality in developing nations, deciphering the role that unsafe medical care plays is critically important.

The older people are particularly vulnerable to adverse events. Falls, for example, represent the most common patient safety injury for the older patients in hospitals in developed nations.69 Hip fractures remain common and only 14% to 21% of patients recover the ability to perform daily activities.70 Decubitus ulcers are also widespread in this group, risk factors of which include immobility, friction, incontinence, cognitive impairment and poor nutritional status.71–73 In the US between 1990 and 2001, decubitus ulcers were reported to be the cause of death for 114 380 persons (age-adjusted mortality rate, 3.79 per 100 000 population).74 Additionally, rates of ADEs among older patients are much higher than in the general population.75 76

Discussion

We examined the available research on patient safety and focused specifically on 23 topics and have summarised the major consequences of unsafe care and its underlying causes. Several key findings emerged. First, the available data suggest that harm from medical care is widespread and likely imposes a substantial burden on the world's population. Second, most evidence about safety comes from developed nations, although there is growing epidemiological evidence of poor clinical outcomes due to unsafe medical care in developing and transitional countries. Finally, the data on structural and process factors that contribute to unsafe medical care are almost exclusively from a small number of developed nations. Although some solutions are readily apparent, large gaps in knowledge need to be filled before more comprehensive solutions can be developed, particularly for transitional and developing countries.

The nature and extent of unsafe care are still poorly understood in developing and transitional nations. For 5 of 10 “outcomes” of unsafe care (ADEs, adverse medical device events, surgical errors, falls and decubitus ulcer), there were few data points from developing or transitional nations. Data suggest that the burden of harm from unsafe care is sizeable in developed nations and likely to be comparable if not greater in transitional and developing nations. However, the lack of more reliable information underscores the need for high-quality epidemiological studies from these nations. Even among areas with a known level of harm, such as tainted blood products and nosocomial infections, research is needed to better understand the burden they pose to the population and national health systems and the efficacy of existing prevention and harm minimisation strategies.

In developed nations too, information is still lacking in many areas. We know relatively little about harm outside the hospital or who among the chronically ill is at particular risk for developing decubitus ulcers. We still need to develop many more effective intervention strategies. For all nations, the role of organisational structures and processes is poorly understood. Although inadequate numbers of high-quality staff affect the safety of care, optimal staffing levels and the appropriate mix of skills is not known. How organisational and provider factors combine to affect patient safety needs greater understanding.

To ensure comparability of data across the world, standardised tools, measures and definitions are needed. The WHO World Alliance for Patient Safety has focused on this area.77 Beyond definitions, as practice patterns change, we will need robust mechanisms for ongoing data collection from a range of nations, especially developing and transitional ones. The potential benefits of improving safety are enormous. For example, studies suggest that establishing safe injection practices around the globe could prevent as many as 1.3 million early deaths each year and eliminate billions of dollars (US$) annually in direct medical costs.78 The primary implications for funders of healthcare are that we still need both new evidence about the extent of harm that occurs from unsafe medical care in developing countries, the impact that harm has on patients' distrust of those systems, and we need solutions that are able to be implemented locally and are cost-effective.

This study has important limitations. First, although we attempted to account for the most important topics in patient safety, due to constraints, we addressed only those that account for the most harm and some topics that are also important could not be addressed. Although this report could not be wholly comprehensive and was not meant to be a series of systematic reviews of each topic, we do believe we captured most of the major issues in safety and identified the main literature in these areas. Another important limitation was a lack of reliable data in many of the areas we covered, especially for developing countries. Although the limited available information suggests that the burden of harm from unsafe care is very large, it will be essential to obtain more reliable data. Further, whether the key lessons learnt from the developed nations are applicable to the developing world is largely unknown. Next, as has been mentioned previously, many of the topics could have been categorised into multiple areas (ie, structure or process) and those categorisations were not meant to be definitive. Finally, in the interest of brevity, we necessarily omitted many details that are available in the larger report.

In conclusion, patients seek care to reduce their suffering. Based on research from the past two decades, we know that while the healthcare system cures disease and alleviates pain, it can also cause largely preventable harm and suffering. This evidence should not be interpreted as an acceptable cost of providing healthcare. Our review suggests that harm occurs too often and that much of it is avoidable. Reducing harm will require targeted, well-designed and appropriately managed research to gain greater understanding of its causes and contributing factors, especially in transitional and developing countries. The next generation of research should therefore focus both on better definitions of the problem and on effective solutions that reduce harm in medical care.