Health IT: Avoiding potential pitfalls

ECRI report names potential safety hazards

We look at technology as a tool that has given us an edge — over weather and darkness, over toil and disease. It is here to make our lives easier, better, safer. And there is a lot of technology that does just that. But there are also technological advances that are right now causing harm to patients in the best-regarded hospitals. It’s not that the technology was bad, but that there are often unintended consequences that come with advances.

ECRI, an independent health research organization based in Plymouth Meeting, PA, released a list of 10 health technologies that can — and do — harm patients. The list includes:

  • alarm hazards;
  • medication administration errors using infusion pumps;
  • unnecessary exposures and radiation burns from diagnostic radiology procedures;
  • patient/data mismatches in EHRs and other health IT systems;
  • interoperability failures with medical devices and health IT systems;
  • air embolism hazards;
  • inattention to the needs of pediatric patients when using “adult” technologies;
  • inadequate reprocessing of endoscopic devices and surgical instruments;
  • caregiver distractions from smart phones and other mobile devices;
  • surgical fires.

Over the next few months, Hospital Peer Review will look at these risks and talk to experts who can point you to the best information on the potential danger and help you develop strategies for minimizing the risk. This month, we look at issues related to data errors in electronic health records.

Ross Koppel, PhD, a sociology professor at University of Pennsylvania and its associated school of medicine, didn’t expect to find his wife dehydrated and shivering near a hospital elevator hours after she should have returned to her room from surgery. After all, it was one of the best hospitals in the country. But that’s where he found her. Computers at the nurses’ station said she was in her room, but she never arrived, and it wasn’t until Koppel frantically searched for her that she was found.

Lost patients aren’t supposed to happen with bar-coded wrist bands that are scanned at every step. But they do. Another story Koppel — whose book First Do Less Harm examines many of the technology-driven risks to patients — tells is about a man who was told he was dying one day, only to be told it was all a mistake the next. Someone put the wrong information into his electronic health record. There are legions of such stories.

The promise of health information technology (HIT) was that it would make hospitals more efficient, care more uniform (and thus better), and prevent common errors such as errors in medication dosing. But for every technology designed to help patients and providers alike, there is an unforeseen consequence that leads to some off-the-books work around, which leads to potential (and actual) patient harm, says Koppel.

A study by Koppel and colleagues in the July 2008 issue of the Journal of American Medical Informatics Association (see box of further reading, below) lists 15 instances they found of providers circumventing bar-coding technologies related to medication administration and dozens of potential kinds of harm that could arise as a result. They often have great reasons for not using the systems as intended — for example, misplaced stickers that cover up patient bar codes, dead batteries in the scanners, or computer carts that won’t fit in patient rooms. But that doesn’t make it any better.

Further reading

The results can be horrific — like a case in Illinois where a patient was given 60 times the recommended medication dose. The patient died, the family sued, and the payout was in the millions of dollars, says Kimberly Reich, MBA, MJ, PBCI, RHIA, CPHQ, FAHIMA, the compliance and case manager for Lake County Physicians Association in Waukegan, IL. “It’s frighteningly easy to give a wrong dose,” she says. “The devices don’t stop those kinds of errors.” That’s particularly true if providers turn off irritating alerts that might warn of a bad dose or wrong medicine, she says. It could be an error input into an infusion pump that leads to over- or under-dosing. Or it could be a provider just not using a system the way it should be used.

How often do hospitals check the devices they use with patients to make sure they are functioning properly, Reich asks? How easy is it to override safety features?

In her 30-plus years in healthcare, Jean S. Clark, RHIA, CSHA, has seen a vast influx of new technology, all promising to make things better. Until her January retirement, Clark was the accreditation director at Roper St. Francis Medical Center in Charleston, SC, and she remembers the days when there was a single master patient index that had one medical record number for each patient. Everything that happened to that patient was tied to that number.

Electronic records were supposed to make things faster and easier, but Clark says people would come in with a new last name, a different spelling, or a nickname. The addition of a middle name or initial could lead to duplicates of records. Some departments might use the patient name as the main identifier, while others might look patients up based on address, phone number or Social Security number, says Clark.

Koppel points to an article in which he outlined the 24 different ways a patient name Jonah Noel Tobias could be entered into a system. Clark all but calls out a hallelujah. “There is a great potential for all of those different names to not map to the right record, the right person,” she says. “You have to regularly audit and see if records with the same name share other things — addresses, patient identifying numbers. If so, those multiple records might be rightly a single one.” Do this at least every week, and if your system has this built in, there’s no reason not to do it daily. “Not having the right information on the right person at the right time can be extremely dangerous.”

A lot of issues would be solved if there was a single universal medical identity number for each person. “So you have a system where a specimen might have one number on it, but it might be based on fraud from someone stealing his grandmother’s Social Security number. Or a patient comes into the hospital and you find he has seven different patient records.” And then there are typos. There are about 2,000 data entry errors per day at a typical hospital, he says. Most don’t cause harm, but some could, some do.

Add to that the fact that your typical hospital might have as many as 117 devices that can be used in the intensive care unit, or that there are between 30 and 800 different IT systems in use in every hospital. None of these necessarily interact with the electronic health record a particular hospital chose, never mind the one that a hospital in another state used when that patient was there last year. There is no sharing of medical information in a meaningful way because there are no overarching data standards, no requirement that different systems talk to each other.

Koppel says that after spending $200 million on a new EHR system, and spending 5 years and maybe another $800 million to implement it, hospitals aren’t interested in hearing that what they have doesn’t work because the cost of adding something on at the end or doing the kind of revamping that might make it safer and more useful is prohibitive. So they make do, and making do makes for mistakes. “There are 40 different ways of recording blood pressure. There are eight major suppliers of EHRs for enterprise systems. We have hundreds of data standards and lexicons in use, which is as bad as none. It’s a thousand towers of Babel.”

There was a time in centuries past when it just took a strong president to look at other Babel-like systems such as railroads, which had about a dozen different gauges in the country, and tell them that on such a date, they would all use the same one. There were 115 time zones in America once, Koppel says. “Imagine if we were able to have just one system. Think of the savings in duplicated tests and doubled doses. Think of the lives saved if you could show up in any ED in any state and have your complete medical record available.”

Not going to happen, though, Koppel says. So providers write on their hands and sticky notes, because information that should be contiguous in a record isn’t and they have to click through a dozen times and scroll way down to the bottom to find what they are looking for. “Bad interfaces obscure information,” he says.

And should you want to take your vendor to task for bad design, you can’t, because most of the vendors of EHRs have required buyers to sign hold-harmless and non-disclosure agreements. The Institute of Medicine and the American Medical Informatics Association call such contracts unethical, and Koppel says the first thing a hospital can do is stop saying OK when a vendor asks you to sign one. If you’re spending the kind of money you have to spend for a system, make sure the vendor knows who works for whom. Make them work out unified standards for your institution and your allies. Insist that there is a code book and dictionary associated with crosswalks between the various parts of the systems — those places where you call something by one name and another department or hospital in your system calls it something else. Push for interoperability.

Send out the clones

Meanwhile, QI and patient safety professionals have to be diligent about making sure that providers are using systems as they ought. Koppel says getting a good nurse or sociologist to study in real time how HIT is being used and ask questions of users about what they do and why they do it is something worth the investment. Residents are happy to talk to people who aren’t senior physicians or attendings, and they do most of the orders in a typical hospital.

Be on the lookout for cloned data, too — that is, information on a single patient’s record that is identical from one day to the next. The Office of Inspector General will be looking for that kind of cutting and pasting and may refuse to pay if it’s not evident that additional work was done. Perhaps scarier is the fact that there are some instances of cloning of notes from one patient to another. While Koppel acknowledges that it’s possible two similar patients might have the exact same progress in their relative conditions, it’s also possible that a provider is cutting and pasting from one record to another inadvertently, or on purpose as a way to save time.

It’s another unintended consequence of electronic records: Rather than thinking holistically about the patient, a provider is thinking in terms of boxes to fill out, says Bob Wachter, MD, professor at UC San Francisco and head of the hospitalist program at the institution’s hospital. “The systems make it easy to be lazy,” he says. Writing a new note every time you see a patient is a way to get the juices flowing about that patient’s particular set of problems. That said, there’s another problem about making notes so long and in-depth that you lose vital information in the morass of verbiage.

It’s not just a lot of stuff to read that’s the problem. There is the real chance of harming a patient, Wachter notes. If a doctor cuts and pastes information from day one — where there is mention of the potentiality of a particular problem — into the space for day four, the new attending may view that potentiality as an actuality, he says.

Wachter says it needs to be easier to find the information that is relevant to the care you give to the patient today, while also providing immediate access to information that might impact what a physician does — allergies, drug interactions, what the patient’s specialist thinks the plan of action should be. And the copy and paste function is too useful to get rid of, but only if providers know that what they paste should just be the prompt for more thorough and current notes on patient care.

Koppel says that the early claims about the potential for electronic records was “grotesquely overblown,” and that the “HIT emperor is mostly naked now.” But that doesn’t mean that what we have already invested trillions of dollars in should be chucked onto the rubbish heap. “It’s still better than paper. But it is not as good as it needs to be or as good as it could be. Universal medical IDs would solve a lot of the problems. A coherent system for drug nomenclature would be good. Residents at my hospital came up with 100 ways to spell acetaminophen. I was rather proud.” He notes that in the Netherlands and several Australian states, there is a single drug/drug interaction database run by pharmacists and doctors. “It is beloved by everyone.”

Until some of these things change, there will be cases like one gentleman who retired to be near his daughter. He had a cardiac issue and was in the ED 17 times in about 18 months. The decision support program said the man was a good candidate for a pneumococcal vaccine. It said that every single time he went to the ED. So he had 17 vaccines. “It was a great fancy system that could correctly identify relevant characteristics of this patient, but it had no memory.”

For more information on this topic, contact:

  • Ross Koppel, Ph.D., FACMI, Professor, University of Pennsylvania School of Medicine and Department of Sociology and Harvard University School of Medicine. Philadelphia, PA. Telephone: (215) 576-8221.
  • Jean S. Clark, RHIA, CSHA, Former Accreditation Director, Roper St. Francis Medical Center, Charleston, SC. Email:
  • Kimberly Reich, MBA, MJ, PBCI, RHIA, CPHQ, FAHIMA, Compliance and Case Manager, Lake County Physicians Association, Waukegan, IL. Email: Telephone: (847) 360-2883.
  • Robert M. Wachter, MD, Professor and Associate Chairman, Department of Medicine, Chief, Division of Hospital Medicine, UC San Francisco. Telephone: (415) 476-5632.