Public health in the abstract has frequently been promoted as a potential benefit of HIE[ 7][ 21-23] If HIOs become part of the health care landscape, they could significantly accelerate efforts to automate public health activities. However, except for the research initiatives discussed in this article, there have been very few implementations of these ideas. Also, specific use cases through which public health can be improved with HIE have not yet been described comprehensively in the literature. We formally describe 11 potential use cases in which HIE can improve public health-related activities. The box on page 618 provides hypothetical scenarios illustrating the potential utility of these public health-specific use cases.
Mandated Reporting of Laboratory Diagnoses
Mandated reporting of a predefined list of notifiable diseases and conditions is the cornerstone of public health surveillance. However, paper and faxed reports can be costly to generate and process. Electronic laboratory reporting has been shown to improve the timeliness and completeness of reporting,[ 24-26] but laboratories and health departments have been slow to adopt this reporting method. In a 2007 survey, only 14 of 56 jurisdictions reported electronic laboratory reporting systems that were at least 50% operational.[ 27] Local laboratory test and result codes are often customized for billing purposes, and it can be time-consuming to map them to a standardized vocabulary (e.g., Logical Observation Identifiers Names and Codes)[ 28] and to maintain that mapping on an ongoing basis. Furthermore, if electronic reporting draws only from laboratory information systems, the quality of reported data (e.g., inclusion of patient addresses) may actually decline, and certain health department programs may be reluctant to automate the entry of electronic reports into surveillance databases.[ 4] Still, the electronic transmission of laboratory reports can increase the efficiency of public health surveillance for high-volume diseases and the timeliness of reporting for cases requiring immediate public health action.[ 29]
The technical and organizational infrastructure and standards implemented as part of emerging HIO activities could facilitate automated laboratory reporting. Although the legal requirement to report rests upon the laboratories and can not be delegated, the HIO could ensure that ( 1) all the data necessary for notifiable disease reporting would be integrated and mapped to standard vocabularies, ( 2) notifiable conditions would be identified according to a standard rule set (e.g., what constitutes positive syphilis serology), ( 3) a standards-based secure message could be sent to public health, and ( 4) an electronic log of transmissions would be maintained for audit purposes. This mandated reporting to public health agencies would include full patient identifiers and could trigger a public health investigation, including contact tracing.
Nonmandated Reporting of Laboratory Data
Not all infectious diseases of public health significance are legally required to be reported. For example, viral diseases that account for the majority of seasonal disease morbidity for respiratory illnesses (e.g., influenza, respiratory syncitial virus) and gastrointestinal illnesses (e.g., norovirus, rotavirus) are not routinely reportable in most jurisdictions. These cases will not require public health action on an individual basis, but knowledge of the disease patterns in the community can help stakeholders guide public health messages and rule out other, less innocuous outbreaks. Sentinel laboratories already conduct manual reporting of selected respiratory and enteric viral pathogens.[ 30] As laboratory assays for these diseases become more available in clinical laboratories, automated electronic reporting to public health becomes more feasible and useful. Rather than adding these conditions to notifiable disease lists, an alternative would be for laboratories or HIOs to voluntarily report them to public health.
Another example of non reportable laboratory data that could be very useful for public health monitoring is antimicrobial resistance patterns. In both of these examples, negative as well as positive tests could be reported (in contrast to traditional reporting), providing an understanding of denominators and allowing an estimation of sensitivity and resistance patterns of specific infectious agents to specific drugs. This information could be used in conjunction with a geographical information system to provide spatial-temporal displays[ 31] such as a community-wide antibiogram that could help clinicians be more selective in their choices of antibiotics and prevent unnecessary propagation of antibiotic resistance within a given community. This non mandated reporting would not require patient identifiers, although the ability to link results from the same patient would offer some benefits by reducing double counting. As in the use cases already described, the HIO would provide the local infrastructure needed to map required data elements (e.g., laboratory results) to standard vocabularies, identify individual cases of interest according to standardized rules, and securely report them to public health. De-identification of the cases (with or without a linking variable) would be an additional requirement of this use case.
Mandated Reporting of Physician-Based Diagnoses
The second arm of notifiable disease surveillance is independent named reporting from physicians, mandated by law. Electronic laboratory reporting does not cover reporting from physicians, who are separately required to report, and who may be the only sources of required clinical information (e.g., onset date) or risk factor information (e.g., occupation, travel). Physician compliance with reporting requirements is highly variable (9%-99%),[ 32] depending on the disease entity, the provider’s awareness of reporting requirements, and the provider’s perception of the benefits of reporting. A counterpart to electronic laboratory reporting would therefore be enhanced reporting of suspect or confirmed clinical cases on the basis of diagnoses, procedures, or medications entered into clinical information systems and available to the local HIO infrastructure. This enhanced report could be sent in an automated fashion, or it could produce an alert that prompts clinicians to efficiently comply with the reporting requirement in line with their clinical workflow, a mechanism which should improve both compliance and timeliness of reporting.
The Boston Center of Excellence in Public Health Informatics is developing such a model on the basis of reporting from each individual institutions, electronic medical records,[ 33] but the principle could be readily extended to standardized data repositories maintained for HIE. The HIO would need to ensure mapping of required data elements to standard vocabularies, implementation of a standard rules engine, secure messaging to public health, and an audit trail. An additional workflow that might be prudent would be to have such cases first trigger a review by a hospital or health system’s infection control practitioner prior to transmission to the health department This would reduce the number of erroneous reports caused by nonspecific diagnoses (e.g., rale-outs) or miscodings (e.g., “trichinosis” entered instead of “trichomonas”) and ensure that the surveillance reports contain all the needed information, some of which may require manual chart review (e.g., occupation, travel history).
Nonmandatory Reporting of Clinical Data
Syndromic surveillance systems are based on ongoing monitoring of nonreportable, nondiagnostic data from existing information systems that can nonetheless provide information on trends in community health. Examples include monitoring of emergency department chief complaints or discharge diagnoses[ 34][ 35] and Web search engine hits for topics such as flu-related illnesses[ 36][ 37] or sales of over-the-counter flu remedies.[ 38] Most such surveillance systems are also tied to sophisticated statistical algorithms for detection of temporal and spatial-temporal clusters that may require investigation to rule out disease outbreaks of public health significance, whether naturally occurring or manmade.[ 39][ 40] The HIO infrastructure could enable voluntary automated transmission of these data to public health authorities.
Such a system should also support the ability of public health to conduct investigations in the event of a disease cluster of concern. The HIO would need to enable standards-based mapping of a potentially large array of data elements,[ 41] the ability to filter for cases of interest, and secure messaging to public health. Because identifiers are not required or mandated, 2 general approaches could be employed. As with laboratory data, individual clinical cases of interest could be found according to a standardized rules engine and then deidentified or pseudonymized before reporting. Alternatively, events of interest could be aggregated and reported to public health as counts (e.g., number of total encounters with febrile respiratory illness). Public health investigation of clusters would require a protocol for using the audit mechanism (e.g., with exact time of transmission) to enable reidentification of individual cases when necessary.