The hospital environment is highly complex and continuously presents new challenges to the epidemiology team. There are currently 5,815 hospitals in the United States, and this number has been relatively stable over the past decade (1). Among hospitals, 86% are community based (including university hospitals), 8% are nonfederal psychiatric hospitals, and 4% are run by the federal government. US hospitals have 951,045 staffed beds and handle more than 37 million admissions each year. In 2006, there were 119 million visits to hospital emergency rooms and 102 million visits to hospital outpatient departments (2).

Patients move through the hospital environment rapidly. In 2006, the average length of stay per inpatient was only 4.8 days, compared to 5.4 days in 1995 and 6.5 days in 1985 (2). With this rapid turnover, it is easy for a patient to acquire an infection in the hospital and be discharged before the infection is diagnosed or becomes manifest. This creates a problem in case finding and thus affects the accuracy of attack rates.

Overall, 7.3% of the US population was hospitalized at least once in 2006, which is slightly less than the 7.7% figure for 1997. Although the major discharge diagnoses have remained relatively static over the past decade, the severity of illness has increased in many centers. In part, this is a reflection of advances in the outpatient environment, which can now handle mild or moderate illnesses without hospitalization. Third-party payers have also restricted or eliminated inpatient reimbursement for patients deemed to have milder illnesses, causing hospitals to try to limit admissions in this group. In addition, the inpatient population is influenced by population demographics. For example, as Americans have become more obese, type 2 diabetes has become more prevalent in the inpatient and outpatient settings (3). Diabetes may increase the severity of many comorbid illnesses and the susceptibility to healthcare-acquired infections or surgical complications. Thus, although fewer patients are hospitalized and length of stay has decreased, severity of illness has increased in the inpatient population. The result is an increased susceptibility to adverse events and increased difficulty in detecting and preventing such events.

Healthcare is a major part of the US economy, and hospitals contribute a significant proportion of this cost. The total expenses for US hospitals exceed $690 billion per year (1), accounting for 31% of all national health expenditures and rising steadily in the new millennium (2). To the extent that adverse events are preventable, they represent an opportunity to reduce cost and improve quality.

The hospital environment includes healthcare workers as well as patients, and both groups have a right to be protected from harm. Almost six million workers are employed by hospitals in the United States (2), and many of the nation’s 1.7 million physicians work at least partially in an inpatient setting. Hospitals also provide a training site for the country’s 71,000 medical students (4), 145,000 nursing students in entry-level baccalaureate programs (5), and 109,000 house staff (6).

The practice and study of healthcare epidemiology is founded on the principle of nonmaleficence, as expressed in the phrase primum non nocere: “first do no harm.” Unfortunately, the hospital environment has great potential to do harm. In the year 2000, the Institute of Medicine published
their seminal report To Err is Human (7), demonstrating that medical errors were a leading cause of death and injury in the United States. Based on data from previous studies, the report estimated that 44,000 to 98,000 Americans die annually in hospitals as a result of potentially preventable medical errors. As a result, medical errors have become at least the eighth most common cause of death in this country. In this context, medical errors are defined as potentially preventable adverse events, including medication errors (8), accidental injuries, misdiagnoses, healthcare-associated infections, and others. Importantly, the goal of healthcare epidemiology is to improve patient outcomes and not all errors result in adverse outcomes.

Although the overarching goal of healthcare epidemiology is to improve health outcomes by reducing adverse events, an effective program can also save money for the hospital. Preventable adverse events are extremely costly. For example, one recent review estimated that ventilatorassociated pneumonias cost approximately $23,000 per case, catheter-associated bloodstream infections cost over $18,000, and wound infections from coronary bypass surgery cost approximately $18,000 (9). Another study showed that wound infections from orthopedic infections cost approximately $18,000 per case (10). Because there are large numbers of healthcare-associated infections, these costs add up rapidly. It has been estimated that more than 1.7 million healthcare-associated infections occur each year: 290,000 surgical site infections, 250,000 pneumonias, 250,000 bloodstream infections, and 561,000 urinary tract infections (11). Overall, the direct cost of healthcare-associated infections ranges from $28 billion to $45 billion each year (12). This total does not include the indirect costs of lost productivity.

Among noninfectious adverse events, medication errors and misdiagnoses are among the most common and serious events, respectively (13). Medication errors include using the wrong dose, providing illegible prescription orders, using ambiguous abbreviations, giving the medication to the wrong patient, not paying attention to drug interactions or allergies, and using the wrong medication. It has been estimated that each hospital patient experiences approximately one medication error per day (8). Most of these errors do not result in apparent harm, but some are serious. It is estimated that medication errors cause 1.5 million adverse outcomes and 7,000 deaths each year in the United States (8).

Misdiagnoses are particularly problematic, because they necessarily involve human judgment that is necessarily fallible. Yet judgment can be augmented by systems that facilitate effective diagnoses. For example, the turnaround time on tests could be reduced so that clinicians have more data on which to base their diagnoses. Electronic medical records reduce errors by ensuring that a complete, legible record is available instantly to all members of the patient care team (14). Improved communication among physicians and between the outpatient and inpatient settings allows decisions to be based on a more complete understanding of the patient’s history and condition. Much remains to be done in these areas. The prevention of wrong diagnoses and delayed diagnoses represents an important area for future study (15).

Having established that healthcare-associated adverse events are common and problematic, the question remains: What proportion is truly preventable? The efficacy of infection control has been established by several studies. A 2003 review of these studies (16) found that infection control measures reduced rates of healthcare-associated infections by 10% to 70%. This broad range reflects the varying effectiveness of the interventions that were studied, the baseline infection rates, and the settings for the studies. Overall, the authors estimated that at least 20% of healthcare-associated infections could be prevented by effective healthcare epidemiology programs. Although this represents a cost savings of $6 to $7 billion per year (12), more recent studies suggest that the 20% figure underestimates the potential to reduce rates (17,18) even for the most problematic infections.

Given the large number of medication doses that are given each day in a busy hospital, eliminating errors may seem like a daunting task. However, studies have shown that most errors are due to illegible prescriptions, ambiguous abbreviations, overlooking known drug allergies, or writing for the wrong dose of drug (19,20,21). A simple but expensive solution is to use computer-based prescribing instead of a handwritten record. Studies have shown that computerized physician order entry of medication orders consistently reduces errors, although the effect on adverse outcomes varies among studies (14,22).

Thus, although healthcare in the hospital setting has the potential to cause harm, much of the harm can be prevented by an effective healthcare epidemiology program. Hospitals should be motivated to support such programs to improve outcomes, reduce costs, and comply with regulatory requirements (23), even—and some might argue especially—in an era of limited resources (24,25).