Agent/Broker Portal
Forgot your User ID
or Password?
Sign Up Now
Contact Us: 844.466.7225

EHR Integration Problems Contribute to Delayed Diagnosis of Lung Cancer

June 22, 2017

Assuming an EHR automatically decreases medical record errors is risky. While some medical record errors may diminish, others that are unique to EHR use may arise. It is important to identify these potential risks and put systems in place to catch and correct errors before they cause patient injuries.

In the following case, a portion of one patient’s radiology report was transcribed into another patient’s report. Although the case was litigated for close to a year, it was never entirely clear how this error occurred. In addition to the primary defendants in the case (the individuals deemed most likely to have caused the report error), the radiologist who signed the report by proxy and the primary care physician who received the report had clear opportunities to catch the mistake but did not. Had more effective patient safety protocols and procedures been in place, the delayed diagnosis could have been avoided.

Allegation

Failure to communicate the abnormal results of a chest CT scan caused a one-year delay in the diagnosis of lung cancer.

Case File

The following case study involves radiologists and radiology software and computer systems, but the risk management/patient safety lessons can be applied to any practice that uses integration software to connect the various aspects of an EHR system, e.g., e-prescribing, computerized physician order entry (CPOE), billing, scheduling, dictation and transcription.

Initial Visits

Data and Metadata from the EHR

On a Friday afternoon in 2009, a female patient presented to a radiology practice for a follow-up chest X-ray (CXR) for a nodular density in her left lung. The task of interpreting the images was assigned to Radiologist 1. One hour later, a male patient presented to the radiology practice for a CT scan with contrast because an earlier CXR had shown a pulmonary nodule in his right lung. The task of interpreting the images was also assigned to Radiologist 1. Later that day, Radiologist 1 reviewed the female patient’s CXR films (which were normal), dictated a report and then faxed a preliminary report to the female patient’s family physician (FP). An hour later, she brought the male patient’s CT scan images up in the picture archiving and communication system (PACS), dictated her interpretation and impressions of the male patient’s CT scan and entered the completed report into the system. On Saturday, she left for a weeklong conference.

Patient Details and Characteristics

Patient 1

Patient 2

  • Female
  • Accession number 123456
  • Chest CXR
  • Left lung nodule
  • 2 images
  • Male
  • Accession number 123356
  • Chest CT scan with contrast
  • Right lung nodule
  • 50 images

(These patient details and characteristics may seem very different when compared side by side, but when the information in the reports was inadvertently combined, three different individuals who reviewed the mixed-up report did not recognize its inconsistencies.)

The following Wednesday, a CT scan report for the male patient was transcribed. Because Radiologist 1 was unavailable, Radiologist 2 signed the report by proxy. The report was then transmitted to the male patient’s FP. His FP noted the “normal impression” and called the patient to inform him that his CT scan was negative.

The following Monday, the radiology information system (RIS) prompted the transcription service to create a report for the female patient’s CXR. A different transcriptionist typed the female patient’s accession number into the transcription software program, which brought up the female patient’s CXR dictation. When the transcription was finished, the report was automatically added to Radiologist 1’s RIS work list. The report was reviewed and signed by Radiologist 1, who had returned from her conference. The final report was emailed to the female patient’s FP.

Two weeks after Radiologist 1 dictated the reports for the female patient and the male patient, the dictation files were automatically deleted by the transcription software because they were marked as transcribed.

One Year Later

Information from EHR Data and Defendant Testimony

At his annual exam in 2010, the male patient was still complaining of a chronic cough. His FP ordered another chest CT scan at the radiology practice.

Radiologist 3 was tasked with interpreting the CT scan. He searched the male patient’s file for the 2009 CT report for comparison purposes. He discovered a report with the correct date, name, accession number and procedure label, but realized the rest of the document related to a different patient’s CXR. Radiologist 3 then went into the PACS, where he found the male patient’s 2009 CT scan images. A right lung mass was easily identifiable, and comparison with the new images confirmed it had grown significantly over the past year.

Radiologist 3 called the male patient’s FP and told him the current CT scan showed a mass in the right lung. He also told him the patient’s 2009 CT report contained the interpretation and impression of a different patient’s CXR. He further informed the male patient’s FP the 2009 images showed a right lung mass. Radiologist 3 then called Radiologist 1 and Radiologist 2 to describe the confusing circumstances. Neither radiologist had any memory of the male patient and could not fathom why a CXR interpretation and impression was in the male patient’s 2009 CT scan report. Radiologist 3 then called the transcription service manager to determine how the mistake had occurred. The manager surmised Radiologist 1 didn’t dictate a report for the male patient’s 2009 CT scan. When Radiologist 3 asked to listen to the 2009 dictation files from the date in question, he was told they were automatically deleted pursuant to the transcription service’s normal protocol.

Shortly after the 2010 CT scan, the male patient was diagnosed with terminal lung cancer. He and his wife filed a malpractice lawsuit against his FP, Radiologist 1, Radiologist 2, the radiology practice and the transcription service.

Discussion

Because there was a one-year delay in the treatment and diagnosis of the patient’s cancer due to a negligent communication error, and the patient’s cancer had advanced, it was agreed that settlement of the claim was the best option. Assigning apportionment for liability was very complicated because there was no obvious evidence of how the error occurred. During litigation, it became apparent the delay in diagnosis was caused by a combination of EHR integration problems, human error and failure to fully take advantage of the redundant processes in the system.

Dueling Defense Theories

The target defendants in the case (Radiologist 1 and the transcription service) developed different theories to explain how the female patient’s CXR interpretation and impression got into the male patient’s CT scan report. Each defendant placed the primary blame on the other. The transcription service claimed Radiologist 1 opened the male patient’s file in the RIS, but instead of dictating her interpretation of the male patient’s CT scan, she dictated her interpretation of the female patient’s CXR. Alternatively, Radiologist 1 claimed she dictated two separate reports, one for the female patient and one for the male patient. She believed the transcriptionist opened the male patient’s file on the RIS work list and mistakenly entered the female patient’s accession number into the transcription program, which called up the dictation file for the female patient, not the male patient. The transcriptionist then typed the female patient’s CXR interpretation into the CT scan template that had opened for the male patient. Radiologist 1 asserted the transcriptionist never accessed her dictation for the male patient. This finger pointing complicated settlement negotiations.

Human Error and Lack of Patient Safety Policies

By the time this case was litigated, the defendants had no memory of working on the male patient’s 2009 CT scan report. Therefore, their defenses relied heavily on custom and practice testimony. The descriptions of the customs and practices of the defendants revealed multiple patient safety and liability risks, including:

  • None of the defendants had written policies and procedures for creating reports and delivering results. Had written policies and procedures been in place and enforced, the erroneous report might have been caught and corrected.
  • There were inconsistencies between Radiologist 1’s customary practice and the outcome. For example, Radiologist 1 testified she would attempt to contact the ordering physician by telephone and/or fax if the studies showed an abnormality. However, there was no evidence the male patient’s FP was advised in 2009 of the abnormal CT scan images. This raised questions about whether the radiologist followed her own stated custom and practice.
  • Instead of scanning the code on the patient’s file, which would have brought up the patient’s information in the RIS, Radiologist 1 manually typed accession numbers into the system. She was not in the practice of double-checking the accession number against a patient’s other identifying information. This manual entry, without double-checking the numbers, would have allowed for Radiologist 1 to switch the electronic files of the female and male patient, who had very similar accession numbers.
  • When Radiologist 2 signed reports by proxy, he focused on the impression. He did not look at the images and only skimmed the body of the report. He admitted there were very few reports he had ever questioned or refused to sign. Experts believed Radiologist 2 should have reviewed the images before signing off on the report. Had he done so, he probably would have realized the majority of the report he signed referred to a woman’s left lung CXR, which only involved two images and no contrast.

    The American College of Radiology (ACR) recommends against proxy signing radiology reports for various reasons, including the Centers for Medicare and Medicaid Services requirement that radiologists sign their own reports.1

  • When the male patient’s FP received radiology reports, it was his typical practice to review the impression section of the report and skim through the interpretation section. Although experts believed the FP should be able to depend on a radiologist to provide accurate impressions of radiological images, they were uncomfortable with the FP’s failure to recognize the majority of his patient’s CT report referred to a woman’s CXR on an opposite lung.
  • The transcriptionist overlooked numerous “red flags” that should have alerted her that something was wrong. For example, she typed a CXR report into a CT template. She also would have seen there were 50 images associated with the CT and contrast was used, but she went ahead and transcribed a CXR impression.

It can be argued that physicians should be able to depend on others to do their jobs correctly. The problem with this approach is that it doesn’t prevent patient injury and it doesn’t insulate the physician from being sued. During litigation, Radiologist 1 admitted that long before this adverse event occurred, she began having serious doubts about the safety of the radiology practice’s process of dictating, transcribing, allowing proxy signatures, and receiving and communicating unexpected results. She also believed the transcription service company was aware of the risk that a mistake could be made or a report could fall through the cracks. Unfortunately, neither she nor anyone else attempted to speak up about the issue or remedy the situation. In this case, each person relied on someone else to catch a mistake and fix the problem, resulting in a patient’s injury and a subsequent lawsuit involving everyone in the chain of this patient’s healthcare encounter.

Computer Software Problems

Although this wasn’t apparent before the litigation revealed it, the RIS and the transcription program were not fully integrated or optimized to ensure patient safety. For example:

  • The transcription program did not alert a transcriptionist when she was transcribing a dictation file that had already been transcribed.
  • The transcription software automatically deleted all “completed” dictation files after two weeks. This policy resulted in deletion of a dictation that had not yet been transcribed.
  • The transcription program did not have a process whereby the transcriptionist could confirm the accession number she typed into the transcription software matched the patient accessed in the RIS.

System-to-system interface error (communication breakdown among software applications) is a recognized issue in EHR liability and patient safety risk management. Interface errors have many causes. Sometimes the problems may be difficult for the EHR user to detect. One method of managing this risk is to be on alert for signs of an interface error, for example:2

  • Orders or test results are missing.
  • Interface “error logs” contain orders or test results that are failing to transmit between different components of the EHR system.
  • Lab reports are missing information (e.g., reference ranges, comments).
  • Patients receive medications that have not been ordered.
  • There are errors/inconsistencies between structured data and free-text comment fields.
  • Clinician comments are not transferring from system to system.

Clinicians should pay attention to clues that an EHR is not working the way it should. It is important to report or otherwise act on suspicions of unsafe EHR processes.

Risk Management Recommendations

Getting the most out of an EHR requires the coordinated effort of clinicians, staff, administrators, IT personnel and vendors. Consider the following recommendations:

  • Identify key staff and/or key physician EHR “champions” for each office.
    • Champions should be people who are IT savvy and are enthusiastic about using electronic health records. Champions should meet regularly to discuss issues and problems and assist in identifying best workflow practices. This team should be available to identify and discuss staff concerns and develop plans for addressing them.
  • Appoint an IT project manager for your practice and ensure that this person oversees EHR optimization.
    • The IT project manager should work with the IT champion groups and the vendor to review, identify and outline in writing best practices and workflows for the EHR system.
  • Identify any system or practice workflow issues that need to be changed, or any templates or other EHR-related documentation issues that need to be addressed.
    • Keep a list of identified EHR-related issues.
    • Track how individuals are using the system (or not using it) and ascertain who will need additional training on the EHR.
    • Communicate in a defined centralized process the issues, the follow-up and the resolution of the issues.
    • Ensure all staff and clinician users are knowledgeable and compliant with EHR best practices and workflows.
  • Take advantage of redundancies in the patient care delivery system, e.g., carefully reviewing transcribed reports, using two different modes of identifying a patient when entering programs or moving between them.

Regardless of the size of your practice, it is important to understand enough about your EHR to know if it is functioning safely. Work with your software vendor, IT department, EHR champions or other physicians and staff to get the most out of your EHR system and make it work for you.

This content from Claims Rx

References

1. American College of Radiology. “Medicare Regulation FAQ - Authentication of Reports.” (accessed 7/14/2017)

2. Sittig DF, Singh H. A red-flag-based approach to risk management of EHR- related safety concerns. J of Healthcare Risk Mgmt. 2013;33(2):21–26.

Filed under: Digital Health, Electronic Health Records, Radiology, Medical Records & Documentation, Case Study, Practice Manager, Physician

 Topics 

 Specialties 

Interested in NORCAL Group?

Contact Your Agent/Broker or call 844.4NORCAL today