Bennett & Brachman's Hospital Infections, 5th Edition

46

Public Reporting of Healthcare-Associated Infection Rates

Michael Edmond

The movement to mandate that hospitals publicly disclose information on rates of healthcare-associated infections (HAIs) has gained momentum rapidly. In 2003, Illinois and Pennsylvania became the first states to enact legislation that mandated reporting. Since then, eight additional states have enacted legislation, and the vast majority of remaining states are evaluating the issue in their legislative bodies.

The incidence and impact of HAIs are primarily driven by progress in medical care. The development of invasive diagnostic and therapeutic modalities, which have revolutionized medical care, bypass anatomic and physiologic barriers and markedly increase the risk of HAIs. This is compounded by the increase in severity of illness in inpatients along with an increased proportion of these patients who are immunosuppressed via their underlying diseases or transplantation, or cytotoxic therapies. Problems in the healthcare delivery system have contributed to the problem as well. The decreasing profitability of hospitals has caused some to decrease funding for infection control programs, and the nursing shortage has had an impact on the quality of care delivered. Thus, for a myriad of reasons, the incidence of HAIs remains problematic.

Over the past few years, the once hidden magnitude of HAIs has been exposed by the popular press. In response to this and a grassroots campaign by Consumers Union, the organization that publishes Consumer Reports [1], states are enacting legislation mandating the reporting of HAIs and disclosure of infection rates to the public.

The concept of mandatory reporting of HAIs and other healthcare quality issues converges well with the emergence of consumer-driven health care, the newest paradigm to attempt to control healthcare costs. Unlike managed care that controls costs by limiting the supply of health care, consumer-driven health care attempts to control costs by limiting the demand for health care. This is accomplished through the use of health savings accounts and the provision of incentives for healthcare consumers to seek involvement in decisions about their health and their health care [2]. Thus, the well-informed consumer is an integral feature of consumer-driven health care.

Whether public reporting of healthcare quality data is effective remains unknown. In a systematic review of the literature performed by the Centers for Disease Control and Prevention (CDC), the authors found that few rigorous studies adequately addressed the issue of effectiveness and no conclusions could be drawn [3]. They also noted that no studies have addressed public reporting of HAIs.

Healthcare-Associated Infections: The Scope of the Problem

With any public policy issue, it is important to estimate the impact of the problem for which legislation is intended to address. It is estimated that 2 million persons [4],

P.802

or 5–10% of hospitalized patients in the United States, develop HAIs each year [5,6]. These infections account for an estimated 90,000 deaths and have an attributable cost of $4.5 billion [7].

From the standpoint of public policy, it is important to point out that the focus should be on HAIs that are preventable because interventions and policies will have no impact on infections that cannot be prevented—but what fraction of HAIs can be prevented? In the only recent study designed to answer this question, Harbarth et al. performed a meta-analysis of 24 multimodal intervention studies to decrease HAIs reported in the medical literature from 1990–2002. They determined that the preventable proportion of HAIs ranges from 10–70%, with the best overall estimate at 20–30%[8]. Unfortunately, when HAIs are discussed in the popular press, it is seldom pointed out that many of these infections cannot be prevented, leading to unrealistic expectations on the part of healthcare consumers.

Table 46-1 illustrates the potential impact of mandatory reporting of HAIs. Each year, 7% of Americans are hospitalized [9]. If 5–10% of inpatients develop HAIs and 10–70% of these HAIs are preventable, the annual incidence of preventable HAIs is 0.0035% to 0.245%. Applied to 2006 census data (U.S. population 298 million) [10], we can estimate that the number of persons affected by preventable HAIs in the United States ranges from approximately 100,000 to 1.5 million yearly. Finally, by oxestimating the effect of a mandatory reporting and disclosure program at a range of 10–50% reduction in HAIs, we can determine that the estimated number of persons who will be affected ranges from approximately 10,000 to 700,000 annually.

TABLE 46-1 POTENTIAL ANNUAL, NATIONAL IMPACT OF MANDATORY REPORTING AND DISCLOSURE OF HEALTHCARE-ASSOCIATED INFECTIONS (HAIS)

Estimate Number of Persons Affected U.S. population 298,000,000 Proportion of population hospitalized annually 7% 20,860,000 Proportion of inpatients developing an HAI 5–10% 1,043,000–2,086,000 Proportion of HAIs that are preventable 10–70% 104,300–1,460,200 Effectiveness of mandatory reporting (% reduction in preventable HAIs) 10–50% 10,430–730,100

Assumptions Underlying the Policy of Mandatory Reporting and Disclosure

The mandatory reporting movement is predicated on 10 assumptions (Table 46-2), all of which must be true for a completely successful outcome. However, at the present time there is little reason to believe that all of these assumptions are true, as discussed, and for some, data are not currently available to either confirm or refute.

TABLE 46-2 ASSUMPTIONS UNDERLYING MANDATORY REPORTING AND PUBLIC DISCLOSURE OF HEALTHCARE-ASSOCIATED INFECTION (HAI) RATES

1. Transparency, open exchange of information, and accountability are important societal values. 2. HAIs are preventable. 3. Valid data on HAI rates will be produced. 4. Consumers make rational decisions about choices in health care. 5. Consumers will understand and use data on HAI rates. 6. Consumers are able to choose their site of medical care and are willing to change their site of care. 7. Consumers who use HAI rate data will make decisions that will improve the quality of their care. 8. Market forces will provide incentive for hospitals to lower HAI rates. 9. Positive outcomes will outweigh negative unintended consequences. 10. Health care is a commodity.

1. Transparency, open exchange of information, and accountability are important societal values. At the heart of the consumer movement is the desire to diminish information asymmetry so that consumers are able to select providers of high-quality care. Consumer advocates argue that consumers currently do not have the data necessary to make decisions regarding their health care. Thus, release of HAI rates by all hospitals should empower consumers. When hospitals attempt to block disclosure, they risk the loss of trust by their patients because of the assumption made by many that the hospital must have something to hide. Conversely, when hospitals disclose their quality data, they pursue a transparent approach that demonstrates

P.803

accountability to the public and honors the public's right to know.

2. HAIs are preventable. While the medical literature is replete with examples of how to prevent HAIs through best practices and technological advances, the proportion of HAIs that can be prevented remains unknown. As noted, Harbarth et al. estimated the proportion of HAIs that are preventable at 10–70%. However, some newer studies demonstrating the ability to dramatically reduce nosocomial bloodstream infections [11,12] may be pushing the preventable proportion of HAIs to the higher end of that range.
3. Valid HAI data will be produced. Given the complexities of surveillance and the difficulties in risk adjustment, delivering valid data to consumers will not be easy. Careful attention must be directed to surveillance methodology. This will require standardization of HAI case definitions, surveillance strategies, and data sources. Moreover, the data must be risk adjusted to account for the severity of illness and the complexity of care offered at each hospital. Without risk adjustment, hospitals with the sickest patients will appear to be providing lower quality care simply on the basis of higher crude HAI rates. Standardization and risk adjustment are imperative to produce meaningful interhospital comparisons. While this can be addressed in mandatory reporting legislation, arriving at valid risk adjustment remains extraordinarily difficult.
4. Consumers make rational decisions about choices in health care. In other words, do consumers make decisions regarding their health care that maximize their welfare? There has been little research focused on how patients reach such decisions; however, it seems likely that the more urgent the required treatment, the less likely that the patient will proceed with a rational, well-planned investigation of the options with regard to where to seek treatment. In the setting of a major health crisis, patients rely on the recommendations of their physicians, family members, and friends, and often need to reach decisions relatively quickly. A well-publicized, illustrative anecdote is the decision by former President Bill Clinton to have coronary artery bypass graft (CABG) surgery at the hospital in the State of New York with the highest mortality rate for that procedure [13].
5. Consumers will understand and use reported data on HAI rates. It is important to realize that reports on healthcare quality are designed by experts and policy makers whose understanding of the healthcare system informs their decision on the specific indicators that should be used to measure quality. However, the end user of the data, the consumer, may not be able to work backward from the indicator to the bigger picture of quality [14]. Overall, consumers have a poor understanding of quality of care indicators, and this is worse in patients with low socioeconomic status. A significant proportion of the population does not have the reading proficiency to understand quality report cards [15]. Moreover, consumers do not use indicators which they do not understand [16].

Given that HAI rates are among the newest metrics to be released to consumers, it is unknown at this point how frequently these specific data are used by consumers. However, two recent reviews concluded that consumers rarely seek out this information and that it has a modest impact on medical decision making [17,18].

A national survey of 2,102 adults in 2004 by the Kaiser Family Foundation assessed how consumers access and use healthcare quality information [19]. Only 19% of those surveyed stated that they had used quality information in the last year to make a healthcare decision. When asked which sources of information on hospital quality they are very likely to use, 65% of respondents reported friends, family, and co-workers, and 65% also reported their doctor or nurse. However, fewer said they would consult a Web site (37%), order a booklet (20%), contact a state agency (18%), or use a newspaper or magazine (16%). When asked which sources of information would have “a lot of influence” in selecting a new doctor, 61% reported their physician and 52% friends or family, but only 37% would use patient surveys; 17% would contact a government agency, 14% would use insurance company data, and only 10% would visit a Web site.

When asked if they would prefer a hospital that is familiar or one that is more highly rated, 61% of respondents preferred a hospital that is familiar. Similarly, when asked if they would prefer a surgeon who has treated family or friends versus a surgeon who is more highly rated, respondents were nearly evenly split.

In a survey of 474 patients who had undergone CABG surgery in Pennsylvania, a state that publishes a report card on CABG mortality, only 12% of the patients were aware of the quality report before surgery and only 2% stated that it had impacted their choice of surgeon [20]. A recent survey of 510 Medicare patients who underwent surgery found that only 11% looked for information to compare hospitals and 48% stated they would not use hospital performance data if they required surgery in the future [21]. Last, a survey of 1,500 persons in New York state found that 18.5% used available data on healthcare quality in medical decision making [22].

Although the public's desire to access healthcare quality data may change as consumers become more educated, more data become available, and more individuals have access to and are more comfortable with online information sources, at the present time it appears that a minority of individuals are interested in these data and prefer the recommendations of their healthcare providers, family, and friends.

6. Consumers are able to choose their site of care and are willing to change their site of care. Many patients are unable to choose their site of care due to their health insurance plan. Twenty-four percent of Americans are enrolled in health maintenance organizations (HMOs), and 95% of covered workers are enrolled in a managed care plan (HMO, preferred provider organization, or point-of-service plan) [23].

P.804

Thus, a significant proportion of the population has little choice in healthcare venue or may have some choice that comes with financial penalty.

A recent analysis of New York CABG quality data from 1989–2002 showed that public reporting of hospital performance had no impact on changes in market share for hospitals [24]. However, it could be argued that even if patients are unwilling to use healthcare quality reports or change their site of care, third-party payers will use the data to direct their members to hospitals that demonstrate higher quality. There is also little evidence to date, however, to support that argument [24,25,26,27]. It also could be argued that if public reporting is effective in improving the overall quality of care in a given state, even patients who are unwilling to change their site of care may experience a benefit [28].

7. Consumers who use data reported on HAIs will make decisions that will improve their care. This assumption depends on two other assumptions: that comparative data on hospital HAI rates are valid and that healthcare consumers can and will change their site of care in response to the reported data.
8. Market forces will provide incentive for hospitals to improve quality. Public reporting can theoretically improve quality in four ways: (1) remediation (hospitals make a concerted effort to improve quality), (2) restriction (licensing and accreditation organizations use the data to restrict provision of care by poor performers), (3) removal (poor performers discontinue providing services), and (4) stimulation of competition between providers on the basis of improving quality to improve market share.

Most of the data that support these assumptions come from observational studies. Some have cited the 21% reduction in CABG mortality in New York following public reporting of mortality rates as evidence of the positive effect of market forces [29] However, others believe that the decline in mortality was due to other factors, such as the avoidance of surgery on high-risk patients. Nonetheless, some low-volume surgeons with high mortality rates stopped performing CABGs after mortality rates were published [30]. One experimental study in Wisconsin compared the number of quality improvement activities in hospitals that had publicly reported quality data to those who received private reports on their performance without public reporting to those who had neither private nor public reporting. Hospitals with publicly reported quality data performed significantly more quality improvement activities than did the other two groups of hospitals [31]. The difference was even greater when the subset of hospitals that had received poor quality ratings were compared.

9. Positive outcomes will outweigh negative unintended consequences. The impacts of public reporting of HAIs are myriad. If the data are collected via nonstandardized methodologies, comparing the resulting HAI rates from hospitals will not be meaningful. This may lead consumers to make choices that are not congruent with their wishes. Even if data were collected appropriately, without adequate risk adjustment, those hospitals that care for the sickest patients will appear to have higher HAI rates. In addition, Marshall et al. have described seven other unintended consequences of public reporting on the quality of medical care [32]:

• Tunnel vision. This occurs when quality improvement efforts are concentrated on areas being measured to the detriment of other important areas. For example, a hospital might focus efforts on the area of bloodstream infection (BSI) because this infection rate is publicly reported, while ignoring rising rates of ventilator-associated pneumonias that are not required to be publicly reported.
• Suboptimization. This is defined as pursuing narrow organizational objectives at the expense of strategic coordination. In a hospital placing great emphasis on decreasing nosocomial BSIs, multiple groups (e.g., infection control, performance improvement, unit nurses) may each develop competing interventions to reduce infections that duplicate work and data collection. This type of problem could be avoided by developing a multidisciplinary team that involves members of all the stakeholder groups.
• Myopia. Hospitals may concentrate on short-term issues and lose sight of the long-term outcomes. This unintended consequence is particularly worrisome. Given the high stakes associated with mandatory disclosure of HAI rates, hospitals with high rates should be motivated to improve infection control activities, an obvious desirable outcome. However, some hospitals, especially those with particularly high rates, may seek solutions that are not appropriate in a desperate attempt to rapidly lower HAI rates. Of greatest concern is that hospitals will resort to antimicrobial prophylaxis for patients with commonly used medical devices (urinary catheters, central venous catheters, and mechanical ventilators). This will almost surely yield a short-term result in lower HAI rates, but the long-term consequences of rapidly accelerating antimicrobial resistance associated with such practices could be very problematic.
• Convergence. Convergence is defined as placing more emphasis on being exposed as an outlier than on efforts to perform in an outstanding fashion. Hospitals could aim for average or median HAI rates instead of pushing to drive their rates to the absolute minimum, which should be the goal.
• Ossification. This occurs when organizations avoid experimentation with new approaches out of fear of poor performance. This may be particularly problematic in academic medical centers where innovative approaches to decrease HAIs may be discouraged due to concerns that HAIs may increase.
• Gaming. Hospitals game the system when they alter behavior to gain strategic advantage. For example, quality indicators that are measured via administrative data can be affected by coached changes in the coding of diagnoses because risk adjustment depends on the coding of co-morbid conditions.

P.805

• Misrepresentation. Unfortunately, with public reporting comes an inherent incentive for hospitals to structure their surveillance activities to produce a system with suboptimal sensitivity. In the absence of mandatory reporting, it is in the best interest of hospitals to maximally detect HAIs for quality improvement purposes and to decrease unreimbursed costs of care associated with infections. However, in the setting of mandatory reporting and disclosure, it may be overall more economically advantageous for hospitals to detect fewer HAIs because disclosure of high infection rates may lead patients to seek care elsewhere. It is important to note that in the hypothetical case of two hospitals that have identical HAI rates, the hospital with the best surveillance system will appear to have a higher infection rate, a phenomenon known as surveillance bias.

In addition, it is important to consider the opportunity cost of public reporting. Given that public reporting is an unfunded mandate in almost all states that have enacted laws and that hospital budgets are a zero-sum game, diversion of resources from other programs and problems is of great concern.

Some striking examples of unintended consequences have been associated with public reporting of health quality data. A recent large, multicenter study compared patients in Michigan (where there is no public reporting for percutaneous coronary intervention [PCI] outcomes) to those in New York (where outcome reporting is mandated) [33]. The Michigan patients had a higher incidence of co-morbidities, a higher incidence of high-risk indications for PCI, and an observed mortality rate that was nearly 2-fold higher. When multivariate analysis was used to control for co-morbidities, there was no significant difference in mortality rates between the two states. The authors of this study postulated that the difference in mortality rates may have been due to physicians in New York avoiding PCI on high-risk patients due to fear of increasing their publicly reported mortality rates.

A separate study surveyed all interventional cardiologists in New York, and 65% responded [34]. Of those responding, 79% disclosed that the public reporting of mortality statistics influenced their decision on whether to perform PCI on individual patients, and 83% believed that some patients may not receive PCI because of public reporting of mortality rates. Two surveys of cardiothoracic surgeons in states reporting mortality rates following CABG revealed similar findings. In a survey of Pennsylvania surgeons, 63% reported they were less willing to operate on severely ill patients following public reporting [35]. When surgeons in New York were surveyed, 62% reported refusing to operate on high-risk patients due to public reporting of mortality rates [36].

Another phenomenon noted after public reporting started was that CABGs were performed on 19% fewer African Americans and Hispanics in New York [37]. These racial disparities lasted for 9 years. It is thought that the surgeons assumed that racial minorities were at higher risk for poor outcomes and thus avoided performing surgery on them.

In summary, with regard to CABG and PCI, several unintended consequences emerged. These included denial of aggressive therapy to high-risk patients, punishing physicians willing to treat high-risk patients, and racial profiling. As Hughes and Mackay note, the real question regarding the impact of public reporting is whether it results in better delivery of health care or in better risk avoidance via shifting of high-risk patients to other practitioners or hospitals [38].

10. Health care is a commodity. In the United States, unlike many other countries, health care is treated as a commodity rather than as a basic human right. The unfortunate downside of this is that a large segment of the U.S. population is unable to purchase it health care because they lack the resources. Public reporting of quality indicators stems from and reinforces the commodity concept.

Minority and low-income persons are overrepresented among those without health insurance. Some unique issues with regard to public reporting apply to these groups. Metrics addressing quality of health care typically emphasize diseases, procedures, and health status rather than more relevant issues to the underserved (e.g., proximity of services, availability of appointments, financial barriers, and navigability of bureaucracy) [39]. These patients are least likely to be able to choose their healthcare venue. In addition, vulnerable populations may be affected by unintended consequences. These include the avoidance of minority and low-income patients because they may decrease quality scores, the opportunity costs that affect these patients and increase their marginalization, and damage to institutions that care for a disproportionate number of vulnerable patients.

An inherent flaw in the U.S. healthcare system is that some nonemergency services for the uninsured are a discretionary component of hospitals' budgets at a time when fiscal stress is increasing in hospitals due to decreasing revenues and unfunded mandated programs. In 2004, one-third of U.S. hospitals posted an operating budget deficit [40]. Because ultimately a hospital's finances are a zero-sum game, we must examine the opportunity cost of mandatory reporting and disclosure. This raises important questions: From where will these resources be diverted? How can we ensure that an improvement in healthcare for those able to afford it does not result in denying it to those who cannot?

Options for Reporting

The ideal mandatory reporting and disclosure program is characterized by four important characteristics:

1. The accuracy of data collection is maximized.
2. The methodology for data collection and analysis is standardized in all hospitals to be compared.
3. The costs to hospitals and state agencies are minimized.

P.806

4. The end product delivered is valid, easy to access, useful to consumers, and fair to hospitals.

To meet surveillance needs of individual hospitals, how HAI data are collected is not especially important provided that it is consistent. However, interhospital comparisons are impossible when hospitals use different methodologies [41]. Thus, requiring hospitals to report their data without a mandated methodology would do little to inform consumers and likely mislead them. In addition, there must be mechanisms to adjust the rates for risk factors and severity of illness; otherwise, hospitals that care for the sickest patients will appear to be poor performers when compared to hospitals that offer lower levels of care.

At the present time, there is no standard methodology used by U.S. hospitals to define or detect HAIs. Each hospital can choose which type of infection (e.g., urinary tract infection [UTI], BSI, pneumonia, surgical site infection [SSI]) to track, how to define the infection, and the methods to use to determine which patients have the infection.

The CDC's National Healthcare Safety Network (NHSN, formerly the National Nosocomial Infections Surveillance [NNIS] system) offers an excellent approach with standardized infection definitions utilizing multiple sources of data that are applied in a specified fashion [42]. However, only 5% of U.S. hospitals participate in the surveillance system [43], and the remaining 95% are free to track HAIs in any manner they choose.

Despite being the oldest, most highly developed, and the only U.S. national HAI surveillance system, an important problem with NHSN is that the sensitivity for detection of HAIs is suboptimal. For the major infections, the sensitivity is 59% for catheter-associated UTI, 67% for SSIs, 68% for nosocomial pneumonia, and 85% for nosocomial BSIs [44]. Moreover, with the added pressure of public disclosure, underreporting of HAIs may become even more problematic.

As states evaluate mandatory reporting legislation, several questions must be addressed: Which indicators will be included? What data sources will be used? What populations will be surveyed? How will the data be risk adjusted? How will the results be validated?

Indicators

To date, consumer advocates have focused on public reporting of outcome indicators. Although SSIs, ventilator-associated pneumonia (VAP), and central line-associated BSIs are relatively infrequent events, their impact in terms of morbidity, mortality, and cost are great. While nosocomial UTIs are associated with low mortality risk and are relatively inexpensive, they occur at higher frequency than the other infections. Validated definitions for each of these infections have been developed by CDC.

Because surveillance for BSIs is triggered by a positive blood culture and the case definition is relatively simple to apply, surveillance for these infections is straightforward. On the other hand, case definitions for VAP are complicated. The major difficulty with surgical site infection surveillance relates to case ascertainment. Because at least 50% of these infections occur after hospital discharge, capturing all SSIs is difficult, particularly in hospitals that do not have a centralized medical record or an electronic medical record. Thus, the validity of SSI rates may be questionable. Acknowledging the pitfalls associated with SSI surveillance, CDC's Healthcare Infection Control Practices Advisory Committee (HICPAC) recommends central line-related BSIs and SSIs as the best outcome indicators for public reporting [45].

It is important to note that as the interest in avoidance of HAIs has increased, in part due to an increased level of interest by healthcare consumers, analysis of surveillance data is becoming more intense. In an effort to prevent infections, each case identified via surveillance may undergo scrutiny by clinicians who may question the diagnosis from a clinical perspective. However, surveillance definitions are typically not developed for use by clinicians in making decisions regarding therapy. This was demonstrated in a study of VAP in trauma patients in which Miller et al. found that 31% of patients without pneumonia as clinically defined by bronchoalveolar lavage criteria were classified as having pneumonia using CDC definitions [46].

More recently, process indicators have gained attention. In general, process measures are practices proven to decrease HAI incidence. Thus, they provide direct measures of performance that have been linked to outcomes. Berenholtz et al. demonstrated spectacular decreases in catheter-related BSIs in a surgical ICU by focusing on the process of central venous catheter insertion. Other examples include head-of-bed elevation to prevent nosocomial pneumonia, avoidance of vascular catheters in the femoral area to prevent BSI, hand hygiene, and influenza vaccination of healthcare workers. As compared to outcome measures, process indicators are easier to define and measure and do not require risk adjustment. Monitoring these indicators can be accomplished by personnel with less training than outcome measures, which can require complex definitions and the review of multiple data sources. Feedback of process indicators has been shown to more forcefully drive compliance with best practices than does feedback of outcomes indicators [47]. The major disadvantage of process indicators is that they are less meaningful to consumers. The CDC's HICPAC recommends that states include both outcomes and process measures in their mandatory reporting programs.

Sources of Data

Clinical data from multiple sources (e.g., physician notes, laboratory data, radiology reports) collected by trained infection control practitioners in real time via active surveillance remains the gold standard data source. Through

P.807

NNIS, there is more than three decades of U.S. experience with this methodology. Unfortunately, such data acquisition is labor intensive and costly. For this reason, most hospitals have limited surveillance for HAIs to their intensive care units and to selected surgical procedures.

Use of administrative claims data as a surrogate for clinical data collected via concurrent surveillance has been advocated by some and forms the basis for public reporting in Pennsylvania. The primary advantage of identifying HAIs via these data are low cost and high efficiency because the data already exist for every hospital discharge, and electronic reports searching for ICD-9 codes of interest can be generated rapidly. However, the use of coding data, which were designed for billing, not clinical purposes, is very problematic. Use of administrative data for HAI detection shifts case ascertainment from trained infection control practitioners (ICPs) to medical records abstractors with little medical knowledge. Thus, surveillance using these data is highly prone to misclassification bias.

Another problem is that this coding does not distinguish between conditions present on admission and those that develop after admission, which makes coding for complications of care, including HAIs, problematic. Moreover, the codes for complications are vaguely defined, often depend on physician documentation, and require interpretation by the coder, all of which produce variability in rates of complications among hospitals [48]. In a review of nearly 500 inpatient records from two different states, McCarthy et al. found that for surgical patients, 31% of the coded complications had no objective clinical evidence in the medical record to support the diagnosis; for medical patients, 44% of the complications lacked evidence [49]. An Italian study evaluated the use of ICD-9 codes for SSI surveillance and found that depending on the codes used, sensitivity for detection of infection ranged from 10–21%[50].

A recent study from a children's hospital in Pennsylvania compared active surveillance by ICPs to identification of HAIs via administrative data [51]. Active surveillance by ICPs had a sensitivity of 76%, positive predictive value of 100%, and a negative predictive value of 99%, whereas administrative data had a sensitivity of 61%, positive predictive value of 20%, and a negative predictive value of 99%. When cases were identified by administrative data only, further review revealed that 90% were misclassified because the infection originated before hospital admission, no infection was present, or no device predisposing the patient to infection had been in situ. One potential improvement in the use of ICD-9 codes would be to note whether each condition coded was present at admission (“date stamping”) [52], which is currently required only in California and New York [53].

Legislative Activity

Fourteen states have now enacted laws regarding mandatory reporting of HAIs (Table 46-3) with many differences in data sources, metrics to be reported, and mechanisms for reporting and release to the public. Four additional states (Alaska, Indiana, Texas, and Utah) have passed proposals to study the issue and recommend legislative actions. Twenty-one other states considered bills in their 2006 legislative sessions.

Many states evaluating legislation have adopted the Consumer's Union Model Hospital Infections Disclosure Act. This requires acute care hospitals to report quarterly to their state health departments infection rates for SSIs, VAP, nosocomial central line-related BSI, nosocomial UTIs, and other nosocomial infections at the discretion of the state's health department. It is recommended that patient race, ethnicity, and primary language be reported to assess racial and language disparities. The act requires state health departments to create an advisory committee with representatives from public and private hospitals, direct care nurses, physicians, epidemiologists with HAI expertise, academic researchers, consumer organizations, health insurers, health maintenance organizations, and purchasers of health insurance (e.g., employers). In addition, the act requires the state health department and advisory committee to validate the HAI rates and publish an annual report with comparative risk-adjusted HAI rates available to the public via Web site.

The advantages of the Consumers Union bill are that it focuses on the major HAIs and mandates risk adjustment and validation. The disadvantages are that it requires hospitalwide surveillance, which is resource intensive, and mandates the formation of a large advisory committee, which dilutes the expertise of healthcare epidemiologists and ICPs.

To achieve the goals of the Model Hospital Infections Disclosure Act as proposed by Consumers Union, most hospitals would need to significantly increase the resources provided to their infection control programs. A recent survey of acute care hospitals in Virginia revealed that 64% of hospitals had only one ICP full-time equivalent (FTE) and at 86% of hospitals, the ICPs had other major responsibilities [54]. Moreover, had Virginia mandated reporting of all HAIs, an estimated 160 ICPs would have been required statewide at an estimated cost of $11.5 million yearly.

Virginia, Colorado, and Tennessee chose to use an existing surveillance network, NHSN, to avoid the costs of establishing a surveillance system de novo and to be able to benchmark its hospitals against other hospitals nationally. Hospitals will submit raw HAI data to CDC and then report their risk-adjusted infection rates to the state department of health. Colorado's law also requires hospitals to submit their infection data to NHSN. Most of the states have developed an advisory committee to assist in the analysis of data and the methodology for public disclosure and most will post comparative data on HAI rates on a public Web site. Few of the states have made provisions for validation of the data or funding to offset costs. Some states have built in a delay (up to three years)

P.808

P.809

before data submission is required to allow hospitals to gain experience with the surveillance methodology.

TABLE 46-3 STATE LAWS ENACTED REGARDING REPORTING OF NOSOCOMIAL INFECTIONS

State Year Data Source/Venues Targeted Metrics Reported Reporting & Release Mechanisms SSI, surgical site infections; VAP, ventilator-associated pneumonia; CL-BSI, central line associated bloodstream infection; HAI, healthcare-associated infections. Illinois 2003 Administrative claims and clinical data; applies to hospitals and ambulatory surgery centers Class I SSI, VAP, CL-BSI occurring during hospitalization Mandatory quarterly reports to the Department of Public Health which then submits to the General Assembly a summary report to be published on its Web site. Pennsylvania 2003 Administrative claims data; applies to hospitals All HAI Data are reported to the PA Health Care Cost Containment Council, which releases data to the public. Florida 2004 Data source not specified, although current report available online utilizes administrative claims data; applies to health care facilities Not specified A Web site maintained by the Agency for Health Care Administration (www.FloridaCompareCare.gov) reports facility rates for infections due to medical care and postoperative sepsis, along with process indicators from the Surgical Infection Prevention Project. Missouri 2004 Data source not specified; applies to hospitals and ambulatory surgery centers Class I SSI, VAP, CL-BSI Data collection, analysis and reporting rules to be recommended by an advisory committee. Department of Health to publish a quarterly report on its Web site. Nevada 2005 Data source not specified; applies to medical facilities SSI, VAP, CL-BSI, nosocomial UTI Hospitals report to the Health Division of the Department of Human Resources. No provision for public disclosure. New York 2005 Clinical data; applies to general hospitals CL-BSI and SSI occurring in critical care units Hospitals are required to report no more frequently than every 6 months; the commissioner of health shall establish a state wide database of all reported hospital-acquired infection information organized so that consumers, hospitals, healthcare professionals, purchasers, and payers may compare hospitals to each other, to regional and statewide averages and, when available, to national data. Virginia 2005 Clinical data using CDC definitions for nosocomial infections; applies to acute care hospitals Infections and target populations to be determined by the State Board of Health Acute care hospitals required to report selected indicators to the National Healthcare Safety Network and forward adjusted infection rates to the State Health Department; data may be released to the public on request. Colorado 2006 Applies to hospitals, ambulatory surgery centers, dialysis centers Cardiac SSI, orthopedic SSI, CL-BSI; physicians are required to report cardiac and orthopedic SSIs identified postdischarge Acute care hospitals required to report selected indicators to the National Healthcare Safety Network. The Department of Public Health will include risk-adjusted infection rates in an annual report to be distributed widely, published on its Web site and released to the public on request. Connecticut 2006 Data sources to be determined by an advisory committee; applies to hospitals Metrics to be determined by an advisory committee Annual report on infection rates produced by the Department of Public Health will be posted on the department's Web site. Maryland 2006 Data sources to be determined by the Maryland Health Care Commission; applies to hospitals and ambulatory surgery centers Metrics to be determined by the Maryland Health Care Commission Reporting and release mechanisms to be determined by the Maryland Health Care Commission. New Hampshire 2006 Data source not specified; applies to hospitals CL-BSI, VAP, SSI, adherence rates of central line insertion practices, surgical antimicrobial prophylaxis, coverage rates of influenza vaccination for health care personnel and patients Hospitals report to the State Department of Health and Human Resources no more frequently than quarterly. The department's Web site will report infection rates both exclusive and inclusive of adjustments for potential differences in risk factors for each reporting hospital; an analysis of trends in the prevention and control of infection rates in hospitals across the state; regional and, if available, national comparisons for the purpose of comparing individual hospital performance; and a narrative describing lessons for safety and quality improvement. South Carolina 2006 Clinical data; applies to hospitals SSI, VAP, CL-BSI, other infections decided by the Department of Health and Environmental Control in consultation with an advisory committee Hospitals submit reports at least every 6 months to the Department of Health and Environmental Control, which then publishes an annual report on its Web site. Tennessee 2006 Clinical data using CDC definitions for HAIs; applies to hospitals with an average daily census ≥25 inpatients and outpatient facilities performing on average ≥25 procedures daily CL-BSI in ICUs, excluding burn units and level 1 trauma units; SSI for CABG Acute care hospitals required to report selected indicators to the National Healthcare Safety Network. Hospital-specific CL-BSI rates to be reported on the Department of Health's Web site for hospitals with >30 central line insertions per year, updated every 6 months with the most recent 4 quarters of data. For CABG SSI, only an aggregate statewide rate will be released. Vermont 2006 Data source not specified; applies to hospitals Metrics that are “valid, reliable, and useful, including comparisons to appropriate industry benchmarks” to be determined by the health commissioner in consultation with representatives of specified groups Data on HAI rates to be included along with other quality metrics in hospital community reports.

Nevada's law is unique in that it mandates reporting of HAIs, but the data are not disclosed to the public. While this approach may be effective in improving the quality of care, it does not meet the needs of the consumer in acquiring the data needed to choose their site of care. Also, without public disclosure, hospitals may not be viewed as being held accountable to the public.

There has been some interest in the establishment of federal legislation on mandatory reporting and disclosure to establish a single national standard rather than have states develop varying standards. The National Quality Forum is currently in the process of developing national standards for reporting of HAIs, and a Congressional hearing explored the issue recently.

Conclusion

In the end, the benefit to the healthcare consumer due to the transparency provided by mandatory reporting and disclosure of HAIs should trump the risk to hospitals' reputations and finances, and states are increasingly enacting laws to ensure this. This is an excellent incentive for hospitals to commit resources to prevent HAIs. However,

P.810

standardization of methodology for detecting infections, accurate risk adjustment, and validation of reported infection rates is paramount in providing consumers with reliable data. Last, the impact of such legislation, including intended and unintended consequences, should be evaluated after implementation.

References

1. Consumers Union. Stop Hospital Infections.org (www.consumersunion.org/campaigns/stophospitalinfections/about.html) accessed 10 april 2006.
2. Bachman RF. Consumer-driven health care: the future is now. Benefits Q2004;20:15–22.
3. McKibben L, Fowler G, Horan T, Brennan PJ. Ensuring rational public reporting systems for health care-associated infections: systematic literature review and evaluation recommendations. Am J Infect Control2006;34:142–149.
4. Center for Disease Control and Prevention. Public health focus: surveillance, prevention and control of nosocomial infections. MMWR Morb Mortal Wkly Rep1992;41:783–787.
5. Haley RW, Culver DH, White JW, et al. The nationwide nosocomial infection rate: a new need for vital statistics. Am J Epidemiol1985;121:159–167.
6. Ducel G, Fabry J, Nicolle L, eds. Prevention of hospital-acquired infections. 2nd ed. Geneva: World Health Organization, 2002. (www.who.int/csr/resources/publications/drugresist/en/whocdscsreph200212.pdf) accessed 10 April 2006.
7. Centers for Disease Control and Prevention. Healthcare-associated infections. (www.cdc.gov/ncidod/dhqp/healthDis.html) accessed 10 April 2006.
8. Harbarth S, Sax H, Gastmeier P. The preventable proportion of nosocomial infections: an overview of published reports. J Hosp Infect2003;54:258–266.
9. Roberts M. Hospital admission rates—1987 and 1998 (http://www.meps.ahrq.gov/mepsweb/data_files/publications/st9/stat09.pdf#xml=http://207.188.212.220/cgi-bin/texis/webinator/search/pdfhi.txt?query=hospital+admission+rates&pr=MEPSFULLSITE&prox=page&rorder=500&rprox=500&rdfreq=500&rwfreq=500&rlead=500&sufs=0&order=r&cg=&id=4663bc6ebb) accessed 10 April 2006.
10. U.S. Census Bureau. U.S. and world population clocks (www.census.gov/main/www/popclock.html) accessed 10 April 2006.
11. Berenholtz SM, Pronovost PJ, Lipsett PA, et al. Eliminating catheter-related bloodstream infections in the intensive care unit. Crit Care Med2004;32:2014–2020.
12. Centers for Disease Control and Prevention. Reduction in central line–associated bloodstream infections among patients in intensive care units—Pennsylvania, April 2001–March 2005. MMWR Morb Mortal Wkly Rep2005; 54:1013–1016.
13. Altman LK. Clinton surgery puts attention on death rate. New York Times6 September 2004;6:Sec A:1.
14. Jewett JJ, Hibbard JH. Comprehension of quality care indicators: differences among privately insured, publicly insured, and uninsured. Health Care Financ Rev1996;18:75–94.
15. Hochhauser M. Can consumers understand managed care report cards? Manag Care Interface1998;11:91–95.
16. Hibbard JH, Jewett JJ. Will quality reports cards help consumers? Health Aff1997;16:218–228.
17. Schauffler HH, Mordavsky JK. Consumer reports in health care: do they make a difference? Ann Rev Pub Health2001;22:69–89.
18. Marshall MN, Shekelle PG, Leatherman S, Brook RH. The public release of performance data: what do we expect to gain? a review of the evidence. JAMA2000;283:1866–1874.
19. The Kaiser Family Foundation, Agency for Healthcare Research and Quality, Harvard School of Public Health. National survey on consumers' experiences with patient safety and quality information (www.kff.org/kaiserpolls/upload/National-Survey-on-Consumers-Experiences-With-Patient-Safety-and-Quality-Information-Survey-Summary-and-Chartpack.pdf accessed 9 April 2006.
20. Schneider EC, Epstein AM. Influence of cardiac-surgery performance reports on referral practices and access to care: a survey of cardiovascular specialists. N Engl J Med1996;335:251–256.
21. Schwartz LM, Woloshin S, Birkmeyer JD. How do elderly patients decide where to go for major surgery? telephone interview survey. BMJ2005;331:821.
22. Boscarino JA, Adams RE. Public perceptions of quality care and provider profiling in New York: implications for improving quality care and public health. J Public Health Management Practice2004;10:241–250.
23. Kaiser Family Foundation. Trends and indicators in the changing health care marketplace (www.kff.org/insurance/7031/print-sec2.cfm) accessed 9 april 2006. [online] 2005 [cited 2006 April 9]. Available from: URL:
24. Jha AK, Epstein AM. The predictive accuracy of the New York state coronary artery bypass surgery report-card system. Health Affairs2006;25:844–855.
25. Mukamel DB, Mushlin AI, Weimer D, et al. Do quality report cards play a role in HMOs' contracting practices? evidence from New York state. Health Serv Res2000;35:319–332.
26. Mukamel DB, Weimer DL, Zwanziger J, Mushlin AI. Quality of cardiac surgeons and managed care contracting practices. Health Serv Res2002;37:1129–1144.
27. Erickson LC, Torchiana DF, Schnieder EC, et al. The relationship between managed care insurance and use of lower-mortality hospitals for CABG surgery. JAMA2000;283:1976–1982.
28. Werner RM, Asch DA. The unintended consequences of publicly reporting quality information. JAMA2005;293:1239–1244.
29. Hannan EL, Kilburn H Jr, Racz M, et al. Improving the outcomes of coronary artery bypass surgery in New York State. JAMA1994;271:761–766.
30. Chassin MR, Hannan EL, DeBuono BA. Benefits and hazards of reporting medical outcomes publicly. N Engl J Med1996;334:394–398.
31. Hibbard JH, Stockard J, Tusler M. Does publicizing hospital performance stimulate quality improvement efforts? Health Affairs2004;22:84–94.
32. Marshall MN, Romano PS, Davies HTO. How do we maximize the impact of the public reporting of quality of care? Int J Qual Health Care2004;16(suppl 1):i57–i63.
33. Moscucci M, Eagle KA, Share D, et al. Public reporting and case selection for percutaneous coronary interventions: an analysis from two large multicenter percutaneous coronary intervention databases. J Am Coll Cardiol2005;45:1759–1765.
34. Narins CR, Dozier AM, Ling FS, Zareba W. The influence of public reporting of outcome data on medical decision making by physicians. Arch Intern Med2005;165:83–87.
35. Schneider EC, Epstein AM. Influence of cardiac-surgery performance reports on referral practices and access to care: a survey of cardiovascular specialists. N Engl J Med1996;335:251–256.
36. Burack JH, Impellizzeri P, Homel P, Cunningham JN Jr. Public reporting of surgical mortality: a survey of New York State cardiothoracic surgeons Ann Thorac Surg. 1999;68:1195–1200.
37. Werner RM, Asch DA, Polsky D. Racial profiling: the unintended consequences of coronary artery bypass graft report cards. Circulation2005;111:1257–1263.
38. Hughes CF, Mackay P. Sea change: public reporting and the safety and quality of the Australian health care system. Med J Aust2006;184(10 suppl):S44–S47.
39. Davies HTO, Washington AE, Bindman AB. Health care report cards: implications for vulnerable patient groups and the organizations providing them care. J Health Polit Policy Law2002;27:379–399.
40. American Hospital Association. The state of America's hospitals—taking the pulse (http://www.aha.org/aha/resource-and-trends/trendwatch/2007chartbook.html) accessed 8 June 2006.
41. Braun BI, Kritchevsky SB, Kusek L, et al. Comparing bloodstream infection rates: the effect of indicator specifications in the evaluation of processes and indicators in infection control (EPIC) study. Infect Control Hosp Epidemiol2006;27:14–22.

P.811

42. Cardo DM, Brennan PJ, Peaden D Jr, Khabbaz R. Mandatory reporting of hospital-acquired infections: steps for success. J Law Med Ethics2005;33(4 suppl):86–88.
43. Gaynes R, Richards C, Edwards J, et al. Feeding back surveillance data to prevent hospital-acquired infections. Emerg Infect Dis2001;7:295–298.
44. Emori TG, Edwards JR, Culver DH, et al. Accuracy of reporting nosocomial infections in intensive-care-unit patients to the National Nosocomial Infections Surveillance System: a pilot study. Infect Control Hosp Epidemiol1998;19:308–316.
45. McKibben L, Horan T, Tokars JI, et al. Guidance on public reporting of healthcare-associated infections: recommendations of the Healthcare Infection Control Practices Advisory Committee. Am J Infect Control2005;33:217–226.
46. Miller PR, Johnson JC III, Karchmer T, et al. National Nosocomial Infection Surveillance System: from benchmark to bedside in trauma patients. J Trauma2006;60:98–103.
47. Berhe M, Edmond MB, Bearman G. Measurement and feedback of infection control process measures in the intensive care unit: impact on compliance. Am J Infect Control2006;34:537–539.
48. Romano PS, Chan BK, Schembri ME, Rainwater JA. Can administrative data be used to compare postoperative complication rates across hospitals? Med Care2002;40:856–867.
49. McCarthy EP, Iezzoni LI, Davis RB, et al. Does clinical evidence support ICD-9-CM diagnosis coding of complications? Med Care2000;38:868–876.
50. Moro ML, Morsillo F. Can hospital discharge diagnoses be used for surveillance of surgical-site infections? J Hosp Infect2004;56:239–241.
51. Sherman ER, Heydon KH, St. Johns KH, et al. Administrative data fail to accurately identify cases of healthcare-associated infection. Infect Control Hosp Epidemiol2006;27:322–337.
52. Naessens JM, Huschka TR. Distinguishing hospital complications of care from pre-existing conditions. Int J Qual Health Care2004;16(suppl 1):i27–i35.
53. Glance LG, Dick AW, Osler TM, Mukamel DB. Does date stamping ICD-9-CM codes increase the value of clinical information in administrative data? Health Serv Res2006;41:231–251.
54. Edmond MB, White-Russell MB, Ober J, et al. A statewide survey of nosocomial infection surveillance in acute care hospitals. Am J Infect Control2005;33:480–482.