44% of hospitals reported to HHS that they can delete the contents of their EHR audit logs whenever they'd like?

Modern Healthcare published an article "Feds eye crackdown on cut-and-paste EHR fraud" on Dec. 10, 2013 by Joe Carlson.

The article is about federal efforts to reduce the amount of clinician cut-and-paste from prior notes of a patient - which can even be done between charts of different patients.  This practice can result in overbilling for work not actually performed.  The practice can also result in no-longer-accurate data being carried forward; I have been consultant to cases where that phenomenon, in my opinion, contributed to grave patient injury in cases that have settled out of court.

It is at this link:  http://www.modernhealthcare.com/article/20131210/NEWS/312109965/feds-eye-crackdown-on-cut-and-paste-ehr-fraud?utm_source=articlelink&utm_medium=website&utm_campaign=TodaysHeadlines#

Subscription required, but googling the article title may allow reading it in its entirety.

The article begins:

Federal officials say the cut-and-paste features common to electronic health records invite fraudulent use of duplicated clinical notes and that there is a need to clamp down on the emerging threat. That concern is enhanced by the fact that it's too easy to turn off features of EHR systems that allow tracking of sloppy or fraudulent records.

In an audit report released Tuesday morning (PDF), [HHS Office of Inspector General, "NOT ALL RECOMMENDED FRAUD SAFEGUARDS HAVE BEEN IMPLEMENTED IN HOSPITAL EHR TECHNOLOGY"], HHS agencies confirmed that they are developing comprehensive plans to deter fraud and abuse involving EHRs, including guidelines for cut-and-paste features. The issue arises at a time when critics say federally subsidized digital patient record systems are sometimes being used inappropriately by providers to drive up reimbursement.

“Certain EHR documentation features, if poorly designed or used inappropriately, can result in poor data quality or fraud,” according a report from HHS' Office of the Inspector General.

None of this is a surprise to me, and to readers of this blog.

However, the real "money quote" in the article, I believe, is this:

"In addition, only 44% of hospitals' “audit log” systems could record whether cut-and-paste was used to enter data, and an identical percentage of hospitals reported [to OIG] that they can delete the contents of their internal audit logs whenever they'd like."

From page 11 of the HHS OIG Report linked above (http://www.modernhealthcare.com/assets/pdf/CH92135129.PDF):

[In 2006, ONC contracted with RTI International (RTI) to develop recommendations to enhance data protection; increase data validity, accuracy, and integrity; and strengthen fraud protection in EHR technology.]

... Hospitals' control over audit logs may be at odds with their RTI- recommended use as fraud safeguards:

RTI recommends that EHR users not be allowed to delete the contents of their audit log so that data are always available for fraud detection, yet nearly half of hospitals (44 percent) reported that they can delete their audit logs. Although these hospitals reported that they limit the ability to delete the audit log to certain EHR users, such as system administrators, one EHR vendor noted that any software programmer could delete the audit log.

RTI recommends that the ability to disable the audit log be limited to certain individuals, such as system administrators, and that EHR users, such as doctors and nurses, be prevented from editing the contents of the audit log because these actions can compromise the audit log's effectiveness. Hospitals reported they have the ability to disable (33 percent) and edit (11 percent) their audit logs, although they reported restricting those abilities to certain EHR users, such as system administrators or EHR vendors. All four EHR vendors we spoke with reported that the audit logs cannot be disabled in their products, but one vendor again noted that a programmer could disable the audit log.

I further note that, being voluntarily provided, i.e., not part of a formal investigation of any specific organization, those numbers are likely low, perhaps very low considering this issue.

An audit log or audit trail is an automatically-generated dataset, invisible to most users, containing items such as who viewed records, the date/time/location of viewing, and indication of actions they may have performed on the records such as editing/changes/additions/deletions, etc.

As an EHR itself is a collection of magnetized or optically encoded bits on some computer storage medium, it cannot be authenticated as complete and free from alteration by humans.

The audit trail is the only way to authenticate an EHR printout, however (as well as EHR screenshots or any other electronic data turned into a tangible form from those bits) as complete and free from alteration.

If an EHR printout cannot be authenticated as complete and free from alteration, its trustworthiness and perhaps even court admissibility as a business record under an exception to the hearsay rules regarding evidence may be damaged or invalidated.

My concern is that, if true, and considering the conflict of interest a hospital has regarding hiding potential fraud or malpractice that could cost them millions of dollars, a capability to "delete the contents of their internal audit logs whenever they'd like" and to edit audit trails (which based on the capabilities of relational databases also implies an ability to delete sections of audit logs selectively and/or to substitute false data) is simply alarming.

I don't think the EHR pioneers intended EHRs to be used for purposes of allowing evidence spoliation without traceability ...

-- SS

Dec. 13, 2013 Addendum:

I received the following reply from EHR compliance expert Dr. Reed D. Gelzer.  Re-posted with permission:

Good morning Dr Silverstein,

Thank you yet again for the illumination that you bring to matters of truth in Healthcare Information Technology.

Regarding the OIG report’s source document, the 2007 report to the ONC, I was the Fraud Prevention Workgroup Chair for that project, working under Principal Investigators Dr. Don Simborg and Susan Hanson, former Chair of AHIMA. 

For anyone who is interested in this subject matter, I would recommend that you go to the source document and, among other things, review the list of contributors.  These were all individuals who volunteered time to attempt to mitigate harms of defective HIT, in their capacities of records management systems, nearly 8 years ago now.   Many have gone on into leadership roles in related organizations and domains, some still working towards trustworthy health information technology systems.

I believe that I can say that none of those working on the report then would have believed that it was conceivable that even our most basic recommendations regarding the fitness of audit functions would remain "novel" in the industry in 2013.  One cannot be surprised at the low level of authenticity supports in hospitals’ EHRs systems given that fitness as record management systems for patient care has, to date, been either neglected or presumed, not tested or attested.   I am gratified that our 2007 work was utilized for the OIG report to illuminate the deplorable state of integrity supports in these patient care information systems.  This will undoubtedly spur interest in supportive resources such as the HL7 EHR System Functional Model Standard and the HL7 Records Management and Evidentiary Support Profile Standard.

All of us who worked on that ONC report are, I hope, as gratified as I am that the OIG removed our work product from its designated obscurity.   We developed the guidelines via methods that were more qualitative than quantitative, entirely intended to guide initial implementation backed by more methodical research.   We represented the most informed at that time, including those like myself and my ADIC associate Patricia Trites who had performed compliance testing on over 30 among the leading EHRs at the time and found extraordinary ranges of deficiencies, including audit functions that could be disabled at will.   Standards and tools existed then to support mitigation of risks and those Standards and tools have expanded since.  Now that the events and ONC decisions that led to inactions on the report are now in the past, we can more rapidly achieve the potentials nascent in HIT by rendering it more trustworthy, usable, and safe.

Thank you again for your ongoing vigilance.

Sincerely,

Reed D. Gelzer, MD, MPH, CHCC
Trustworthy EHR, LLC
Co-Facilitator, HL7 Records Management and Evidentiary Support Workgroup

To this I add that I also would not have found it conceivable that my concerns about bad health IT and the risks of patient harm it poses, as well as common healthcare IT project mismanagement, of which I started writing about in 1998 (http://cci.drexel.edu/faculty/ssilverstein/cases/) would remain "novel" ideas in the industry in 2013.

The Obamacare healthcare exchange website debacle has made the latter issue mainstream.  The former issues still need more sunlight.

-- SS

2/4/14 addendum:

HHS is apparently starting to pay attention to the importance of robust and secure EHR audit trails.

I note in the HHS document "Meaningful Use Stage 2, 2014 Edition EHR CERTIFICATION CRITERIA 45 CFR 170.314", page 7, regarding audit trails, available at this writing at http://www.healthit.gov/sites/default/files/meaningfulusetablesseries2_110112.pdf:

§170.314(d)(2) Auditable events and tamper-resistance.

(i) Record actions. EHR technology must be able to:

(A) Record actions related to electronic health information in accordance with the standard specified in § 170.210(e)(1);
(B) Record the audit log status (enabled or disabled) in accordance with the standard specified in § 170.210(e)(2) unless it cannot be disabled by any user; and
(C) Record the encryption status (enabled or disabled) of electronic health information locally stored on end-user devices by EHR technology in accordance with the standard specified in § 170.210(e)(3) unless the EHR technology prevents electronic health information from being locally stored on end-user devices (see 170.314(d)(7) of this section).

(ii) Default setting. EHR technology must be set by default to perform the capabilities specified in paragraph (d)(2)(i)(A) of this section and, where applicable, paragraphs (d)(2)(i)(B) or (d)(2)(i)(C), or both paragraphs (d)(2)(i)(B) and (C).

(iii) When disabling the audit log is permitted. For each capability specified in paragraphs (d)(2)(i)(A), (B), and (C) of this section that EHR technology permits to be disabled, the ability to do so must be restricted to a limited set of identified users.

(iv) Audit log protection. Actions and statuses recorded in accordance with paragraph (d)(2)(i) must not be capable of being changed, overwritten, or deleted by the EHR technology.

(v) Detection. EHR technology must be able to detect whether the audit log has been altered. 

From a posting at http://healthcaresecprivacy.blogspot.com/2012/09/meaningful-use-stage-2-audit-logging.html:

... The Information that needs to be recorded: § 170.210(e)(1)(i):  These rules [in a column I did not show here - ed.] identify “sections 7.2 through 7.4, 7.6, and 7.7 of the standard specified”. This is simply the list of attributes that an audit log entry should contain that ASTM E2147 says are mandatory, and excludes the values it outlines as important but not mandatory. Below is about 90% of what is in section 7, I didn't want to copy all of it out of respect for the copyright. But, the part missing is just a one-line definition of each item, nothing more than that.

7. Audit Log Content
7.1 Audit log content is determined by regulatory initiatives, accreditation standards, and principles and organizational needs. Information is needed to adequately understand and oversee access to patient identifiable data in health information systems in order to perform security oversight tasks responsibly.
Logs must contain the following minimum data elements:
7.2 Date and Time of Event
7.3 Patient Identification
7.4 User Identification
7.5 Access Device (optional)
7.6 Type of Action (additions, deletions, changes, queries, print, copy)
7.7 Identification of the Patient Data that is Accessed(optional)
7.8 Source of Access (optional unless the log is combined from multiple systems or can be indisputably inferred)
7.9 Reason for Access (optional)
7.10 If capability exists, there should be recognition that both an electronic “copy” operation and a paper “print” operation are qualitatively different from other actions.

I am not sanguine about the "optional" components, especially 7.7 - the actual data that was accessed and acted upon.

I also note it is stunning that these audit trail 'rules' have only been promulgated recently.  It will be interesting to see how rigorous the EHR "certification" process will be regarding audit trails.

-- SS

Manipulation of 12,000 Medical Records Made Easy by EHR

This from a hospital in Canberra, Australia using a common ED EHR in that part of the world, iSOFT:

Canberra Hospital embroiled in data scandal
SBI Magazine (Secure Business Intelligence)
Jul 5, 2012 

A Canberra Hospital executive has admitted to manipulating Emergency Department records to make wait times and stays appear shorter than they were.

The executive told the Director-General of the Health Directorate they had made "approximately 20 to 30 changes to hospital records" a day from "late 2010" onwards.

ABC [Australian Broadcasting Corp.] News reported that the matter has been referred to police, while the executive has been suspended without pay.

Though the data manipulation was initially said to be motivated by concerns over job security, changes in 2011 and early 2012 were said to have been made due to "managerial pressure" to improve publicly-reported performance statistics.

This raises the issue that data manipulation might have been performed not just to improve reported statistics, but to cover up medical error, computer related or not, and thus deny injured patients or their heirs the right to legal redress.

"The only thing that worked to achieve benchmark targets was to alter the data," the executive later told investigators at PricewaterhouseCoopers (PwC), which was engaged by Health to perform a forensics analysis. The analysis is detailed in a new Auditor-General report (pdf).

In total, PwC found 11,700 performance records - about six percent of all records stored in the hospital's iSOFT emergency department information solution (EDIS) - had been altered.

It is believed more staff at Canberra Hospital altered records than the executive that has so far admitted responsibility.  "While an executive has admitted to changing EDIS records, it is probable that EDIS records have also been manipulated by other persons with access to the system," the federal auditor-general noted overnight.

This is another area where electronic records make possible tasks that are probably impossible with paper.  Altering 11,000+ records would be hard in paper charts, as the alterations would likely stick out in a pronounced manner.

"The executive’s admission to Audit does not appear to account for all of the changes to EDIS records that have been made to improve timeliness performance."

For example, changes to EDIS records, albeit a much smaller number, appear to have been made on days when the executive was on leave (seven days in total in 2010-11 and early 2011-12). 

I am saddened to note, a proper term for this activity might indeed be "conspiracy":  a conspiracy is an agreement between two or more persons to break the law at some time in the future.

User access control, IT security failures

Poor controls such as generic logins and inadequate user and password security made it easy for insiders to game the data.

While EDIS was on approximately 259 workstations across the hospital and 253 users had permission to run the software, there were only 23 user accounts.

Of these user accounts, only eight were in regular use, including four named administrator accounts (specific to administrative staff) and four generic user accounts: CLERK, NURSE, DOCTOR and BEDMAN.

The generic accounts could be used by personnel across the hospital, not just within the Emergency Department.

Passwords for the four generic user accounts were "very poor" and had "never been changed". Password expiry was set at a default 999 days.

Audit logs were equally poor, not proactively checked and unreliable.

The proper term for these arrangements might be "gross mismanagement" of clinical information technology.

"A feature of the logging record is that it logs the changed field in EDIS and a number of other fields simultaneously, while not identifying which field was changed and what its original value was," auditors noted.

"Audit also notes that the logging record is also ineffective, because every entry in EDIS is logged from “Workstation 14”.  

"Although EDIS has been disseminated widely throughout the Canberra Hospital each of these users logs into EDIS using the common “Workstation 14”.

"This practice, combined with the use of generic user accounts, makes the EDIS logging information useless for investigations of unauthorised activity."

Furthermore, it was possible to edit EDIS records up to 72 hours after a patient’s treatment, providing a generous window for later unauthorised changes to the records.

These "features" sound like seller misdesign with regard to the metadata (logging records).

Noticing anomalies

It was only in April this year that a full inquiry was commissioned after "anomalies" in performance figures were spotted by the Australian Institute of Health and Welfare (AIHW).

The AIHW found an unusually high number of emergency patients that were reported to have been seen at exactly within the required time for their illness category.

For example, there was an unusually high number of patients who were reported to have been seen at exactly 30 minutes or 60 minutes.

In addition, an unusually high number of people checked out of the Emergency Department precisely 240 minutes after their recorded arrival.

If you're going to engage in this type of activity, at least be competent at it...instead of setting up a red flag bigger than the flag that used to fly over the Kremlin.

The records that were manipulated mean that publicly reported information relating to the timeliness of access to the Emergency Department and overall length of stay in the Emergency Department have been inaccurately reported.

The report could not ascertain the level of over‐estimation due to the lack of a clear audit trail identifying what were legitimate and what were fabricated entries in patients’ records.  

Timelines can be critical to proving medical negligence in court.  Further, if time data could have been manipulated, it seems clinical data could have been manipulated as well.

EHR data manipulation is of unknown magnitude worldwide, but I can imagine if it's easy to do and the benefits potentially substantial, electronic records could possibly be less trustworthy than paper records.

-- SS

Addendum:  while on the topic of clinical IT Down Under, there's also this:

Coast medical records system 'dangerous'
Stephanie Bedo
Goldcoast.com.au

July 6, 2012

SENIOR doctors say Gold Coast Heath's new multimillion dollar electronic medical record system is 'inadequate and dangerous' and could put patients' lives at risk.

Doctors have complained about the system, saying some patient documents are missing, it has log-in problems and 10-minute delays in accessing critical information.

Gold Coast Health was the first region in the state to move to electronic record-keeping, rolled out progressively from October last year.

Queensland Health spent about $200 million on the electronic medical record roll-out last year, which was delayed by 12 months because of problems with the software provider.

... Hospital cardiologist Dr Greg Aroney raised concerns about the system at a Griffith University forum on the future of health on the Gold Coast this week.

"Our system is totally inadequate and dangerous," Dr Aroney said.

Read the whole story at this link:   http://www.goldcoast.com.au/article/2012/07/06/429621_gold-coast-news.html

A similar story from the states where the doctors' complaints were actually ignored is at my Sept. 2011 post "Blake Medical Center (Bradenton, Fla.) Ignores Health IT Warning Letter From 100 Staff Physicians." 

Let's hope the Australian physicians' complaints are taken more seriously.

-- SS