Observational descriptive studies establish the case definition of an infectious disease event by obtaining data for analysis from available primary (e.g., medical records) or secondary (e.g., infection control surveillance) sources. These data enable the characteristics of the population that has acquired the infection to be delineated according to (a) “person” (age, sex, race, marital status, personal habits, occupation, socioeconomic status, medical or surgical procedure or therapy, device use, underlying disease, or other exposures or events); (b) “place” (geographic occurrence of the health event or outbreak, medical or surgical service, place of acquisition of infection, or travel); and (c) “time” (preepidemic and postepidemic periods, seasonal variation, secular trends, or duration of stay in hospital). The information from descriptive studies might provide important clues regarding the risk factors associated with infection, and in each case it is hoped that an analysis of the collected data might be used to generate hypotheses regarding the occurrence and distribution of disease or infection in the population(s) being studied.

Observational analytic studies are designed to test hypotheses raised by the findings in descriptive investigations. The objectives of these studies are (a) to establish the cause and effects of infection in a population and (b) determine why a population acquired a particular infection in the first place. The three most common types of observational analytic studies are cohort studies, case-control studies, and prevalence or cross-sectional studies.

Cohort Studies In cohort studies, hypotheses that have been generated from previous (descriptive) studies are tested in a new population. A population of individuals (a cohort) that is free of the infection or disease of interest is recruited for study. The presence or absence of the suspected (hypothesized) risk factors for the disease is recorded at the beginning of the study and throughout the observation period. All members of the cohort population (e.g., all premature infants admitted to a neonatal intensive care unit during a defined time period) are followed over time for evidence or appearance of the infection or disease and classified accordingly as exposed or unexposed to specific risk factors. If the observation period begins at the present time and continues into the future or until the appearance of disease, the study is called a prospective cohort study. If the population studied is one that in the past was apparently free of the markers of disease on examination of records or banked laboratory specimens, it may be chosen for study if data on exposure to the suspected risk factors for disease also are available. The population may be followed to the present or until the appearance of disease. This type of study, common in occupational epidemiology, is called a historical or retrospective cohort study.

A key requirement of a cohort study is that participants be reliably categorized into exposed and unexposed groups. Relative risk, that is, the ratio of the incidence of the outcome in the exposed group to the incidence in the unexposed group, is used to measure the strength of an association between exposures or risk factors and disease. Cohort studies have the advantage of enabling identification and direct measurement of risk factors associated with disease, determination of the incidence of infection and disease, and ascertainment of the temporal relationship between exposure and disease. In cohort studies, observational bias may be less of a limitation on the validity or results, since the information on the presence of risk factors is recorded before the outcome of disease is established. To ensure sufficient numbers for analysis, cohort studies require continual follow-up of large populations for long periods unless the disease under investigation is one of high incidence. Cohort studies are, in general, more expensive and time-consuming to conduct and are not suitable for the investigation of uncommon infections or conditions. However, they render the most convincing nonexperimental approach for establishing causation.

Case-Control Studies In a case-control study, individuals (cases) who are already infected, ill, or meet a given case definition are compared with a group of individuals (controls) who do not have the infection, disease, or other outcome of medical interest. In contrast to cohort studies, participants in a case-control study are selected by manifestation of symptoms and signs, laboratory parameters, or a specific condition, disease, or outcome. Thus, the search for exposure of case and control subjects to potential risk factors remains a retrospective one. For case-control studies, the measure of association between exposures or risk factors and health outcome is expressed as an odds ratio, that is, the ratio of the odds of an exposure, event, or outcome occurring in a population to the odds in a control group, where the odds of an event is the ratio of the probability of it occurring to the probability of it not occurring.

The presence of significant differences in the exposure to risk factors among case versus control subjects suggests an etiologic (causal) association between those factors and the infection or disease defined by cases. Casecontrol methods are useful for studying infections, events, or outcomes likely associated with multiple risk factors or
low incidence rates; for investigating situations in which there is a long lag-time between exposure and outcome of interest; and for establishing etiologic associations or causation of a disease, infection, or other outcome when there is no existing information about the cause or source. In an attempt to reduce bias, control subjects might be selected from individuals matched with cases for selected characteristics, such as age, gender, socioeconomic status, or other variables not suspected or under investigation as risk factors. Compared with cohort studies, case-control studies may be conducted in relatively shorter time, are relatively less expensive, or may require a smaller sample size to execute. Limitations of case-control studies include selection bias in choosing case and control subjects; recall bias in which study subjects might have difficulty in remembering possible exposures; incomplete information on specific exposures; or risk factor data may be difficult to find (or remember). Case-control studies are not used to measure incidence or prevalence rates and, generally, are not capable of establishing temporal relationships between an exposure and outcome.

Prevalence or Cross-Sectional Studies In prevalence studies, the presence of putative risk factors and the disease under investigation is recorded in a survey of a study population at a specific point in time or within a (short) time period. The rates of disease among those with and without the suspected risk factors are compared. Thus, cross-sectional studies can establish association but not causation for suspected risk factors. Prevalence studies are relatively inexpensive and can be carried out rapidly if well-planned. However, they do not allow the ascertainment of risk factors at the beginning of disease nor do they enable one to establish a temporal sequence of risk factors preceding the infection or other outcome of interest. Point prevalence, period prevalence, and seroprevalence surveys are examples of cross-sectional studies.

In experimental studies, the investigator controls an exposure of individuals in a population to a suspected causal factor, a prevention measure, a therapeutic regimen, or some other specific intervention. These exposure modalities are randomly allocated to comparable groups, thereby minimizing confounding factors. Both the exposed and unexposed groups are monitored thereafter for specific outcomes (e.g., appearance of infection or disease, evidence of effective prevention or control of the disease, or cure). Experimental studies often are used to evaluate antimicrobial or vaccine treatment regimens and are generally expensive to conduct. Within healthcare settings, studies that examine restriction of certain antimicrobials or promotion of use of alternative antimicrobials for the control of antimicrobial resistance could be considered under the category of experimental. For ethical reasons, it is rarely possible to expose human populations to potential pathogens or to withhold a preventive measure that could potentially be beneficial to the patient. Unfortunately, animal hosts are not naturally susceptible to many agents of human disease. Thus, one has to be careful when extrapolating epidemiologic findings in animal experimental studies to the control of infections in human subjects.

Quasi-experimental studies: more recently, there has been an increase in the number of published papers describing results from these studies. This type of study shares the design characteristics of experimental studies but lacks random assignments of study subjects. Quasi-experimental studies are useful where randomization is impossible, impractical, or unethical. The main drawbacks of quasiexperimental studies are their inability to eliminate confounding bias or establish causal relationships.

The agents causing healthcare-associated infectious diseases are microorganisms ranging in size and complexity from viruses and bacteria to protozoa and helminths. Bacteria, fungi, and certain viruses have been the agents most recognized and studied as causes of healthcare-associated infections (11). For transmission to take place, the microorganism must remain viable in the environment until contact with the host has been sufficient to allow infection. Reservoirs that allow the agent to survive or multiply may be animate, as exemplified by healthcare worker carriage of staphylococci in the anterior nares or throat (12,13, 14, 15), or the inanimate environment, as demonstrated by Pseudomonas spp. colonization of sink areas, Legionella in hot or cold water supply systems (16, 17, 18, 19), Clostridium difficile spores on computer keyboards, or Serratia marcescens growing in contaminated soap or hand lotion preparations (20, 21, 22).

Certain intrinsic and genetically determined properties of a microorganism are important for it to survive in the environment. These include the ability to resist the effects of heat, drying, ultraviolet light, or chemical agents, including antimicrobials; the ability to compete with other microorganisms; and the ability to independently multiply in the environment or to develop and multiply within another host or vector. Intrinsic agent factors important to the production of disease include infectivity, pathogenicity, virulence, the infecting dose, the agent’s ability to produce toxins, its immunogenicity and ability to resist or overcome the human immune defense system, its ability to replicate only in certain types of cells, tissues, or hosts (vectors), its
ability to persist or cause chronic infection, and its interaction with other host mechanisms, including the ability to cause immunosuppression (e.g., HIV).

Once transferred to a host surface, the agent may multiply and colonize without invading or evoking a measurable host immune response (23, 24, 25). The presence of an agent at surface sites in the host does not define the presence of an infection. Nonetheless, patients so colonized may act as the reservoir source of transmission to other patients (26).

If infection takes place, a measurable immune response will develop in most hosts even if the infection is subclinical. The success of this process for the agent is increased in the nonimmune host and is most successful in the nonimmune, immunocompromised host. A microorganism’s ability to infect another host vector (e.g., yellow fever virus in mosquitoes) or another nonhuman reservoir (e.g., yellow fever virus in the monkey) is important in the epidemiology of certain infectious diseases in world populations at large but plays little role in healthcare infection epidemiology.

Infection depends on exposure of a susceptible host to an infecting agent. Exposure of the susceptible host to such agents is influenced by age, behavior, family associations, occupation, socioeconomic level, travel, avocation, access to preventive healthcare, vaccination status, or hospitalization. Whether or not disease takes place in the infected host and the severity of disease when it appears depend not only on the intrinsic virulence factors of the agent but more importantly on the pathogenicity of the interactions between the agent and the host. The host immune defenses attempt to prevent infection. Thus, any reduction in host defenses may allow infection to take place with a smaller dose of microorganisms or at a body site that is not usually susceptible to infection. A combination of reductions in host defense characteristics and the requirements for an agent to cause infection are typical of the interactions that allow acquisition of opportunistic infections in immunocompromised patients. A commonly cited model indicating the potential interactions between agent and host and the relationships among colonization, infection, and clinical and subclinical disease is shown in Figure 1-2 (27).

Host factors important to the development and severity of infection or disease may be categorized as intrinsic or extrinsic. Intrinsic factors include the age at infection; birth weight; sex; race; nutritional status (28); comorbid conditions (including anatomic anomalies) and diseases; genetically determined immune status; immunosuppression associated with other infections, diseases, or therapy; vaccination or immunization status; previous experience with this or similar agents; and the psychological state of the host (29). Colonization of the upper and lower respiratory tracts is more likely when the severity of illness increases in critically ill patients. This, along with other host impairments (e.g., reduced mucociliary clearance or changes in systemic pH), allows colonization to progress to invasive infection. Moreover, other clinical conditions may lead to an alteration in epithelial cell surface susceptibility to binding with bacteria, leading to enhanced colonization (23, 24, 25). Extrinsic factors include invasive medical or surgical procedures; medical devices, such as intravenous catheters or mechanical ventilators; sexual practices and contraception; duration of antimicrobial therapy and hospitalization; and exposure to hospital personnel.

The environment provides the mutual background on which agent-host interactions take place and contains the factors that influence the spread of infection. Environmental factors include (a) physical factors such as climatic conditions of heat, cold, humidity, seasons, and surroundings (e.g., intensive care units, outpatient clinics, long-term care facilities, or water reservoirs); (b) biologic factors (e.g., intermediary hosts such as insect or snail vectors); and (c) social factors (e.g., socioeconomic status, sexual behavior, types of food and methods of preparation, and availability of adequate housing, potable water, adequate waste disposal and healthcare amenities). These environmental factors influence both the survival and the multiplication of infectious disease agents in their reservoirs and the behavior of the host in housing, occupation, and recreation that relate to exposure to pathogens. Food- and water-borne diseases flourish in warmer months because of better incubation temperatures for the multiplication of the agent and recreational exposures of the host, whereas respiratory agents appear to benefit from increased opportunities for airborne and droplet transmission in the closed and closer living environments of the winter. In US hospitals, the frequency of hospital-acquired Acinetobacter spp. infections is increasing in critical care units and has been shown to be seasonal in nature (30). The seasonal variation in the incidence of this pathogen is thought to be due to changes in climate—summer weather increases the number of Acinetobacter spp. in the natural environment and transmission of this microorganism in the hospital environment during this season (30).

Within healthcare settings, the components of the agent, host, and environment triad interact in a variety of ways to produce healthcare-associated infections. For example, the
intensive care unit is now considered the area of highest risk for the transmission of healthcare-associated pathogens in US hospitals (31). Moreover, methicillin-resistant S. aureus (MRSA), VRE, and ceftazidime-resistant Pseudomonas aeruginosa are endemic in many intensive care units in these hospitals (31). The emergence of vancomycin-resistant S. aureus in US institutions highlighted the unwelcome but inevitable reality that this pathogen may become endemic in acute care settings (32). A complex interaction of contributory factors, such as inadequate hand washing and infection control practices among healthcare workers, fluctuating staffing levels, an unexpected increase in patient census relative to staffing levels in the intensive care unit, or an unprecedented increase in the number of severely ill patients with multiple invasive devices, could all contribute to the acquisition of hospital infections caused by one of these endemic microorganisms (33,34). Adding to the complexity of the process would be the unquantifiable mechanism of transmission of the agent from host to healthcare worker, healthcare worker to healthcare worker, and host to environment. Thus, acceptable measures for the prevention and control of healthcare-associated infection dictate that the healthcare epidemiologist looks at and analyzes the interrelationships among all components of the triad of agent, host, and environment (31).

It is well-known that the social environment is extremely significant in determining personal behavior that affects the direct transmission of agents, such as HIV via breast milk in regions of high HIV endemicity, gram-negative microorganisms via artificial nails worn by healthcare workers in US intensive care units (35), and pathogens that cause sexually transmitted diseases. What must be understood to be equally relevant is the impact of other factors in the social environment, such as the distribution of and access to medical resources; the use of preventive services (36, 37, 38); the enforcement of codes in food preparation, infection control practices, or occupational health practices; the extent of acceptance of breast-feeding for children (39, 40, 41); and the acceptance of advice on the appropriate use of antimicrobials (42, 43, 44,45,46). Also, there must be an appreciation by patients, relatives, and healthcare workers alike that at-risk patients (e.g., those born very prematurely have severe congenital abnormalities, the very elderly, or those with premorbid end-stage cardiac or pulmonary disease), who have numerous indwelling medical invasive devices, or who have undergone multiple invasive procedures or surgical procedures would be particularly susceptible to healthcare-associated infections that are likely nonpreventable. There must be an informed and ethically sound willingness to reject the extraordinary application of medical technology, including the inappropriate or repeated use of resistance-inducing antimicrobials when clinical evidence and experience suggest that the condition of the sick patient is untreatable or irreversible.

Microenvironments, including military barracks, dormitories, day-care centers, chronic disease institutions, ambulatory surgery and dialysis centers, and acute care hospitals, provide special venues for agent-host interactions. Historically, epidemics in these institutional environments provided the experience that drove the development and acceptance of control measures, guidelines, and infection control programs. Acute care hospitals, especially those offering regional secondary and tertiary care, remain the dominant examples of these environments. Changing patterns of outpatient practice, home healthcare, and technical advances in medicine have resulted in increasingly severely diseased and injured populations being managed in acute care facilities. Data from CDC demonstrate that the changing healthcare environments in the United States are resulting in larger intensive care unit populations while there has been a general decrease in the number of general medical beds (31).

Special units for intensive medical or surgical care for extensive burns, trauma, transplantation, and cancer chemotherapy frequently house patients with increased susceptibility to infection (47). In these patients, reduced inocula of pathogens or commensals are required to cause infection, infection may take place at unusual sites, and usually nonpathogenic agents may cause serious disease and death. Frequent opportunistic infections in these patients require repeated, broad, and extended therapy with multiple antimicrobials, leading to increasingly resistant resident microbial populations (31,46).