Progress in science and technology has led to dramatic, worldwide improvements in health and longevity. However, this progress is associated with a level of complexity, distractions, and system opacity, which hampers our ability to reliably produce optimal and safe outcomes [22].

Healthcare can be viewed as a complex adaptive system and concepts from complexity science and engineering will undoubtedly play an increasing role in the design of new care delivery systems and models [3]. Numerous studies document the worldwide unacceptable rates of patient harm and the negative consequences of variations in care [23–29]. In addition, poor quality, i.e., the difference between optimal outcomes and what actually exists, is characterized by overuse, underuse, and misuse of healthcare resources [30–35]. Although progress to date has been slow, continued efforts to understand the root causes of suboptimal levels of quality will ultimately lead to a more reliable, high-value healthcare system [36, 37].

Poor quality and errors stem from a fragmented, multilayered, and “siloed” system of care with diffuse accountability, staggering amount of information, and pressures to function at the margins of the system’s capacity [38]. When combined with human fallibility, complexity leads to process variability and poorly coordinated medical care as well as inconsistent standards and inadequate care transitions (Table 1.2) [7, 39]. Other factors such as strong production pressures, time constraints, and a rigidly hierarchical culture also contribute to a system of unreliable, inconsistent, and too often dangerous care.

Many patients are injured during the course of their treatment and some die from these injuries. In New York hospitals, for example, 3.7 % of patients out of 30,121 randomly selected records suffered adverse events during their hospitalization and approximately 70 % of these resulted in disability lasting less than 6 months, 2.6 % caused permanently disabling injuries, and 13.6 % led to death [25]. In a study of hospitals in Colorado and Utah, surgical adverse events accounted for two-thirds of all events [40].

Serious, entirely preventable surgical events, known as “never events,” continue to occur despite extensive efforts to thwart them. Perioperative mistakes such as retained surgical equipment, burns and positioning injuries, as well as wrong-site, wrong-patient, and wrong-procedure events should never occur in any patients [9, 41]. When combined with other events such as medication errors, accidental punctures and lacerations, and other mistakes, these events constitute considerable aggregate risk for the surgical patient.

The US Centers for Disease Control and Prevention (CDC) estimates that each year 1.7 million HAIs occur in US hospitals each year, resulting in 99,000 deaths and an estimated $20 billion in healthcare costs [42]. Healthcare-acquired conditions such as infections are a costly plague to patients and the healthcare system. When patients are admitted to a hospital, they should not suffer a preventable healthcare-associated infection (HAI) . Unfortunately, surgical team members still have low hand washing compliance rates upon entering the operating room ranging from 2.9 to 10 %, thus contributing to surgical infections [43, 44]. Unfortunately, HAIs affect 5–10 % of all hospitalized patients in the USA annually [42]. HAIs such as surgical site infections, pneumonia, and infections of implanted devices can lead to death or serious chronic disability and are largely if not entirely preventable.

In New York City, hospital-acquired staphylococcus infections alone cost $400 million. In 2014, a survey by the CDC which described the burden of HAIs in US hospitals reported that about 75,000 patients with HAIs died during their hospitalizations [42]. More than half of these occurred outside of the intensive care unit. Most alarming is that many hospital-acquired bacterial infections have developed resistance to, at least, one of the antibiotics traditionally used to treat them [45]. Antibiotic stewardship and infection-reduction programs include discriminate antibiotic therapy as well as reliable use of appropriate infection prevention measures (hand hygiene, skin preparation and depilation techniques, gloves, gowns, air handling, cleaning, etc.) [46].

Research indicates that unnecessary variation harms patients, leads to poor quality, and results in high levels of waste [2, 49–51]. Furthermore, it appears that much of the current variation in surgical care reflects inconsistent application of evidence-based practice standards as applied to clinical decision making and the use of technology or methods for which there is no evidence or wide acceptance. Much practice variation and many clinical decisions seem to be influenced by non-patient-related factors such as geographic, age-related, racial, socioeconomic, and ethnic disparities that have been demonstrated to exist for a variety of conditions [52–56].

The rates of many surgical procedures including vascular surgery, coronary artery bypass operations, lung surgery, and other types of procedures vary as much as tenfold across geographic regions [1, 2, 49, 52, 57, 58]. Substantial practice variation has also been shown to exist between surgeons, even within the same medical center [59]. For example, when selecting patients with prostate cancer for radical prostatectomy, a study demonstrated considerable variability among surgeons at a high-volume academic center [60]. The study suggested that publicly reporting individual practice patterns at the surgeon level could potentially decrease the overtreatment of low-risk prostate cancer [61]. These phenomena are not due solely to insurance coverage variations and they are well found in countries with universal health coverage such as Great Britain and Canada [62].

In another example, poor adherence to well-accepted national guidelines for preoperative testing has been shown to lead to overuse. Feng et al. found that women undergoing mid-urethral sling surgery were subjected to unnecessary testing during preparation for surgery [63]. In this study, approximately two-thirds of complete blood counts and coagulation profiles were not indicated. Additionally, 22 % of chest radiographs and 6 % of electrocardiograms were not obtained despite being indicated. One study demonstrated that 31 % of patients undergoing total knee arthroplasty did not have an indication for the procedure and an additional 21 % had inconclusive indications [30].

The appropriateness criteria have not been developed for most common surgical procedures and many of the existing ones are outdated [32, 34, 64–69]. It is anticipated that investments in comparative effectiveness research will yield meaningful contributions towards the development of appropriateness criteria and reduce practice variation in the future. A broad, coordinated effort will be required to ensure adherence to practice guidelines and other tools which promote the application of evidence-based practice standards to address variation in the use of surgical procedures. Ultimately, an approach which incorporates a shared decision-making paradigm involving patients and physicians should ensure that proper diagnostic evaluation has been done and appropriate treatments are offered [67].

Studies have found that only between 10 and 20 % of routine medical practice has a basis in scientific research [70–72]. Much of what is done in clinical practice is based on tradition or opinion in the absence of valid clinical knowledge or with inadequate evidence for what is best for a given patient. Quite often, these treatments are effective, but the lack of concrete data underscores the need for healthcare organizations and individual practitioners to follow their outcomes and compare them to other centers. Risk-adjusted surgical registries such as the American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP) allow opportunities for improvement to continuously improve the efficiency and effectiveness of surgical care [73].

Effective care occurs when the benefits of an intervention outweigh the risks. Overuse occurs when patients receive treatments, tests, or medications when there is no evidence that such treatment will improve a patient’s outcome and may expose the patient to unnecessary risks. The associated cost from overuse is staggering, particularly for certain conditions and procedures. It has been estimated, for example, that the number of unnecessary hysterectomies in the USA impacts approximately 80,000 women and adds a cost of $320 million annually [76].

Underuse occurs when healthcare providers neglect to give patients medically indicated care or to fail to follow accepted practices. Care for vulnerable individuals such as the elderly and children falls short of acceptable standards for a wide variety of conditions. Patients do not receive the appropriate and timely care necessary which often leads to additional and more severe complications resulting in poor outcomes and adding to healthcare costs needlessly. An in-depth study of lower extremity vascular procedures for critical limb ischemia, for example, showed a significant variability of amputation rates when comparing areas with different intensity of vascular care suggesting that patients in some areas are far less likely to receive limb salvage procedures [77].

The care provided to different segments of the population does not appear to be evenly distributed and many studies have documented racial and socioeconomic disparities in both treatments and outcomes [54–56]. For example, in patients with early-stage non-small-cell lung cancer (NSCLC) , receiving of curative-intent surgery was significantly less for black patients than for whites in every state in the USA [52]. Such unequal care has been documented for a number of different surgical treatments such as obesity surgery, cancer care, and cardiovascular procedures [2, 62, 78].

Disparities also occur in populations with special vulnerability to adverse events such as the very old, mentally ill, trauma patients and the very young often due to their inability to participate actively in their own care mainly due to communication barriers [53, 79]. Older people, for example, may suffer varying degrees of impairment in vision and hearing as well as cognitive deficits and may not be able to understand or communicate with their caregivers. These problems are compounded when serious illness or trauma occurs contributing to these difficulties and potentially leading to errors [80]. Infants and children are also at greater risk of serious errors particularly related to medications with devastating effects [71, 81]. In culturally and ethnically diverse populations, individuals with limited language skills or literacy are also vulnerable to disparities and communication failures often occur which potentially lead to misunderstandings and errors [82–84].

Quality can be assessed both explicitly and implicitly. Explicit quality measures are developed prospectively and are well defined. Explicit measures are evidence based and their construct validity and reliability have been verified through independent observations. Unfortunately, the majority of surgical care currently can only be evaluated implicitly. Implicit measurements of quality are generally based on subjective evaluation [67, 85–87]. While clinical databases and disease registries such as NSQIP and the Society of Thoracic Surgery’s National Databases have developed well-defined process and outcome measures, they are only applicable to a limited range of surgical procedures and participation is voluntary [73, 88]. Much of what constitutes “surgical care” currently falls outside the range of our ability to objectively compare and mostly relies on subjective interpretation. Furthermore, implicit quality measures are based on expert judgment by peers or by proxies of quality including processes of care but do not measure true quality. For example, using the perspective of the three domains of quality proposed by Donabedian (structure, process, and outcomes), structural measures such as hospital or surgeon volume are relatively easy to obtain [85]. However, the relationship between volume and quality is not always clear. In general, hospitals or surgeons performing large numbers of a particular surgical procedure may have lower mortality; however, other factors including severity case-mix and other unmodifiable, often, intangible factors also contribute to poor results. Adjusting for risk requires the use of sophisticated analytic methods with inherent limitations and not all risk factors can be captured. Furthermore this approach is currently limited to a narrow range of procedures.