Executives should be aware of the operational issues associated with a PACS deployment in an EMR environment.

By Cary D. Negley and Rod Walker, FACHE, FHFMA

In recent years health care organizations seeking to generate improvements in patient care have taken on challenges of implementing PACS and EMR systems. When combining both systems in a single deployment as an integrated solution for clinical care areas, they've discovered the collective issues are significant. An organization that fails to recognize the operational challenges associated with such a deployment will not generate the required return on investment (ROI) from a successful implementation.

This article identifies and addresses some of the fundamental issues and challenges frequently overlooked by hospitals and health care organizations early on in an implementation initiative.

Integration and interoperability

Historically, PACS and EMRs existed as two distinct and separate systems, often requiring unique hardware and in many cases separate (dedicated) viewing stations. As the manufacturing of monitors has improved in the area of total screen size and resolution quality, and the bandwidth of networks has increased in conjunction with the decreasing price of servers, fiber and storage media, we have seen significant gains in interoperability at the clinician desktop for integration of images and text. Most PACS systems can support the storage of the textual findings associated with images as part of their core product. In addition, most EMRs have integrated PACS images into the EMR offering or have partnered with PACS vendors to support interconnected images and text presented at the EMR desktop. All of this integration is good news to the clinician for decision support. However, there is a price to pay operationally to ensure the level of integration assumed when PACs and EMRs were purchased.

Patient record management

PACS systems are usually subordinate to both EMR and registration systems for ADT data, yet patient record creation, record merging and collapsing of records has significant clinical consequences in both PACS and EMR systems where an electronic record is part of the clinical decision support process. Classically, within a paper-based operational environment, many of the activities associated with registration management were seen as back-office responsibilities (e.g., resolving duplicate patient record issues and new visits added to the wrong patient record, as well as collapsing patient records). In the past, these activities were not seen as generating negative operational impacts to clinical processing if not caught and addressed immediately. This is absolutely no longer the case once a hospital bridges the gap to a paperless world of automation. At an IT level, errors that used to be monitored and resolved out of the interface engine on a daily basis can no longer be so cavalierly addressed due to the real-time nature of an electronic medical record.

The two largest benefits of PACS systems are the immediate communication of current images to appropriate clinical staff and the instantaneous retrieval of prior images for comparison purposes. The same two benefits exist in EMRs when dealing with the clinical chart. Failures in the ADT process, as described in the following section, severely restrict or totally remove any of the clinical benefits of these systems.

In order to effectively deploy PACS and EMR systems as part of a combined project effort, patient identification and record creation must be accurate and timely with minimal rework after the fact. This assertion establishes a high threshold for quality and thoroughness for front-office operations involved in the registration and scheduling process. The following scenarios present three operational challenges created when front-office practices are not tuned to address stable and accurate registration requirements within a PACS/EMR initiative:

Duplicate patient record created (MPI). Although this is the easiest challenge to resolve of the three cited, it can hinder clinical decision support by creating a situation in which no clinical history is available within the newly created (duplicate) EMR. In this scenario, due to duplicate record creation, there would be no pre-retrieval of pre-existing records or prior image review within the PACS system. Note that this scenario can never be totally removed due to the need to register "John Doe" patients in emergent situations when no patient identification is possible.

New encounter added to wrong patient record. This scenario creates the potential for more serious patient care implications by corrupting the clinical decision support process. Undetected, incorrect historical clinical data are presented to clinicians in the form of somebody else's images and clinical data appearing on the wrong record. When evaluating the various evils in the ADT process, "no data" is usually preferable to "wrong data."

Duplicate encounters/visits created. Although this scenario appears similar to the MPI scenario, it is slightly different. If a new patient visit creates duplicate encounter/visit records, the clinical documentation and PACS images could be stored in a disparate and non-cohesive manner by splintering their contents across multiple separate and distinct encounter records — despite the fact that they were entered under the correct MPI number. The patient record and clinical processes would be corrupted by not having all of the clinical actions generated and performed in a single encounter or visit. This would create incomplete views of clinical status for the clinician treating the patient and could lead to redundant testing or missed access to clinically relevant test results.

Once you have identified potential problems, you need to craft solutions. While solutions may appear reasonable and straightforward, sometimes the cure is worse than the disease. Look carefully at the resolution model, as well as the ability of any interfaces to seamlessly accommodate the prescribed remediation efforts and any clinical data with the potential of being "orphaned" or destroyed by the remediation steps. The following considerations must be addressed as part of developing a resolution path:

1)When an encounter is added to the wrong patient's record, can all downstream systems effectively accommodate the movement of the encounter without losing clinical data if the error is fixed in the originating ADT system? You would be surprised at the number of sites where the answer to this is "no." You would also be surprised at the number of systems in which an error at the ADT level — if not caught immediately and fixed — causes serious problems for all downstream and subsequent processing. Also surprising are the number of EMRs that require all orders entered in the wrong visit to be canceled and reordered in the correct visit. In such cases, you have to ask yourself, What does that do to my clinical results for the chest X-ray that was already documented but has now been canceled and must be re-documented in the correct visit? What does that do to the accession number that was tied to the image and acts as the connection point between PACS and the EMR?

2)What is the impact on any system in which clinical decision support was previously set into motion by virtue of clinical pathways, treatment recommendations or clinical criteria — originally based on what is now recognized as incorrect information relating to prior clinical history from the incorrect patient? What mechanisms are in place to annotate and correct mislabeled medical record numbers on images and forms?

3)When duplicate patient records are collapsed together, who determines the impact, if any, on permanent clinical data? For example, when nursing staff documents patient allergies in the original patient's EMR and then documents new (and potentially different) allergies in the new patient's record, who determines which data survives the merge? Is it a system-based decision relying upon the medical record number determined to be correct, according to HIM staff? Or, does a medical records clerk or clinician determine which documented information should survive?

The decisions made on how to resolve the aforementioned situations must be determined early on in order to mitigate the risks and complexity of reworking the enterprise project plan or design. It should suffice to say that these problems are exacerbated when the EMR or PACS is deployed across a large enterprise with multiple venues of care and many points for patient/encounter creation.

Timeliness and ADT considerations

Both EMRs and PACS systems are highly time-sensitive regarding ADT transactions. If ADT transactions have not been provided by the registration system due to the link being down each night for backup and day-end processing, then records will not be created in either the EMR system or the PACS system. This causes significant problems for PACS and EMR systems. If no patient record or encounter file exists in the subsystems, no repository exists for clinical data or images. Most EMRs and PACS have the ability to add their own patients and encounters, but such additions create a larger potential risk for duplicate records once the interface link with the ADT side comes back up. This situation creates the same types of challenges already identified in this article. Having a centralized registration area may minimize duplicate records due to increased standardization controls. However, centralized registration areas have a tendency to create a bottleneck in patient flow that delays a clinician's ability to treat the patient by using tools in either a PACS or EMR. This significantly impacts ROI in the use of PACS and EMRs. Such a situation can negate the potential benefits of avoiding duplicate records if the solution simply results in creating a greater problem for the care delivery process.

The timeliness of processing transfer and discharge transactions is also a major issue. Some PACS configurations manage image pre-retrieval and interim storage based on patient location. For example, the emergency department and coronary care unit may have dedicated servers to expedite retrieval and viewing of images by clinicians for patients in those specific locations. If discharges are not performed in a timely fashion, images will not be archived in a timely fashion, reducing the efficiency of the PACS primary data storage. If transfers are not completed in a timely manner, images will not be stored on the correct interim server to support fast retrieval_r(pre-fetching). Depending on the product or skill level of the clinician, it may appear to the user that the records are unavailable or may take additional time to retrieve, which compromises timely delivery of care in a critical care setting.

In addition to PACS implications, the timeliness of processing transfer transactions also impacts EMRs. Many EMRs tie significant clinical decision support, administrative functionality and user access to patients based on the patient's location; therefore, it is equally important to process the transfer transactions in a timely manner. Getting copies of lab reports to print to the correct location, generating the correct orders sets and having the patient show up on the correct unit census are all affected by the timeliness of the transfer process.

Workflow questions

The decision to deploy a PACS system in conjunction with an EMR can be riddled with many challenges that do not have a readily available or obvious solution. You should always begin with the end in mind from an overall project vision perspective. In doing so, you should be prepared to tackle all of the previously referenced issues and others that will surely surface during your path to EMR/PACS deployment. The key is to use workflow as the basis of both your analysis and decisions. Assume that you are deploying EMR/PACS systems to improve workflow in order to improve care delivery processes and outcomes. If a system is not going to improve workflow overall, then it begs the question, Why deploy it? Listed in the following section are examples of additional workflow-related questions that should be addressed by your project team before making any final deployment decisions. These questions will help your organization solidify the vision for the final project plan:

1) How will clinicians use the images? Will they leverage the system prospectively for rendering care, or will they only use it as a quality assurance tool after care delivery?

2) Is there a need to combine PACS images with other clinical data from the EMR to enhance decision support?

3) How will images be related to clinical data if the images come from separate systems?

4) What process will ensure the relevance of information combined between PACS and EMR systems so they will accurately present the patient record?

5) Where are images most likely to be used operationally, and what will be the impact on the HIS infrastructure? Will additional pre-retrieval infrastructures be needed and funded?

6) EMRs have significant security measures for controlling access to individual pieces of clinical data. How will rights to PACS data be configured and managed to ensure they are consistent with policy and rules governing an EMR environment?

The deployment of both PACS and EMR systems offers significant clinical benefit to health care organizations willing to integrate these tools into the care delivery process. As a general rule of thumb, the higher the integration into the care delivery process, the higher the ROI for these types of systems. The concomitant axiom is that the higher the level of integration, the higher the need to ensure that the systems accurately reflect and support the care delivery process. For example, a process that does not support the third shift emergency department registration clerk can have a large impact on the success or failure of a PACS/EMR implementation. Likewise, a process that does not support interoperability at the clinician level lessens the overall value of the installed systems. A project plan that addresses these issues will increase the overall success of your implementation.