Increased mortality after drug entry system installed

Tags: A Data ICU Information iS US

16 Dec 2005

Researchers in the US have found a surprise rise in the mortality rates in a children's hospital since the implementation of a software system designed to reduce adverse drug effects.

The computerised physician order entry system (CPOE) was installed over a period of six days in the Children's Hospital in Pittsburgh. The study was undertaken during an eighteen month period before and after the system was put in place. Researchers at the University of Pittsburgh found that the unadjusted mortality rate rose from 2.8%, before the installation, to 6.57% afterwards.

Possible reasons for the unexpected increase given by researchers include the increased amount of time needed to administer drugs through the system, the time spent by nurses away from patients at the computer and the centralisation of the pharmacy services.

"Although CPOE technology holds great promise as a tool to reduce human error during health care delivery, our unanticipated finding suggests that when implementing CPOE systems, institutions should continue to evaluate mortality effects, in addition to medication error rates, for children who are dependent on time-sensitive therapies," recommends the study.

Researchers expressed surprise and concern at the apparent rise in mortality rates, while noting that so far "no study has actually reported a direct association between CPOE and reduced mortality". The increase in mortality was despite the system successfully reducing the number of adverse drug effects in the hospital.

The CPOE system in question, from Cerner, was installed in the hospital in October 2002, and was designed to offer decision support and warnings of contraindications to doctors ordering drugs. Users securely log in, identify the patient, input all drug information (dose, frequency, administration method and length of treatment) and the system checks the order and sends it to a nurse to be activated before it reaches the pharmacy.

Mortality rates were found to have increased most in those patients who had been transferred to the hospital from elsewhere, and those in intensive care. The study suggested that the extra time needed for entering data into the system and waiting for approval may not have mattered in a general ward but could "have significant patient care consequences" in intensive care.

"The interactions between ICU team members have remained fundamentally altered," say the researchers, adding that before the system was put in place, antibiotics and other drugs were generally given to patients within government-approved targets. "After CPOE implementation, we have found that fewer than half of the patients received critical antibiotics and vasoactive infusions within these timelines."

Each member of staff was trained in the use of the system over a period of three months before implementation, by way of a compulsory three-hour training session, the study stressed.

The study is published in the latest issue of Pediatrics, the official journal of the American Academy of Pediatrics.

Links

Abstract of article at Pediatrics

Discussion of article on HISTalk blog

Readers Comments

Wow!!

16 Dec 05 14:32

I'm sure Cerner will criticise technical aspects of the study but there is no getting away from the fact that, unless they have robust data to the contrary, patient safety is not guaranteed simply by the introduction of fancy US technology.

What is the NHS doing, apart from their weak 'clinical certification' and non-delivering 'Common User Interface' processes to ensure User Interfaces are slick, safe and delivering the highest level of encoded error prevention? Surely this is a must if we are to follow this electronic dream?

What does the Wirral think of Cerner's prescribing system now that they are inside this exclusive circle?

System benefits do not simply appear

colin@kineticconsulting.co.uk

19 Dec 05 10:01

Though this study in Pediatrics is rather uneven (unequal periods before and after implementation, for example) it does illustrate an important point that I have been repeating for years: benefits from healthcare IT do not simply appear post-go live.

Benefits must be identified, planned, monitored and managed. Many pay lip service to this principle, few paractice it.

Dangers of rigid standardisation

20 Dec 05 10:14

Add the complexities of RBAC, the technical infrastructure and contractural framework of NPfIT. In the Southern Cluster, overlayed on the complexities of the Cerner billing product being shoehorned into a clinical system, procured in a rush and implemented faster.

I would suggest the prospects for a sucessful, clinically safe and beneficial result are significantly lower than having a white christmas .

Securing safe NHS IT standards

20 Dec 05 10:56

Doesn't the NHS have an Information Standards Board (isb.nhs.uk) with complex (and presumably costly) processes to ensure that NHS IT is safe? And I'd put the safety odds much longer (along the lines of elephants flying), though I suspect that the ISB doesn't have the clout to guarantee either airborne elephants, white Christmases or safe NHS IT systems!

Benefits?

raf_99@hotmail.co.uk

20 Dec 05 18:34

Colin....I appreciate that workflow and efficiency benefits take time to emerge from all IT implementations but this is no ordinary IT implementation. You are correct....time and measurement is required.

I certainly would not want to go to a hospital knowing my chances of morbidity/mortality were 4% higher in the 5 months following the introduction of a new computer system. Would anybody?

This is worrying research for all patients and should either be confirmed or disproved...not explained away by some intellectual arguements about study setup parameters.

Isn't this self-evident?

21 Dec 05 09:59

Can anybody tell me the basis for the hypothesis that introducing a computerised process will lead to any reduction in either mortality or morbidity? Put another way, just what is the *clinical* problem that IT is supposed to alleviate?

Personally - having spent much of my professional life working in theatre and intensive care - the fact that more patients in high-dependency situations die when their clinical carers are distracted into interacting with computers come as no surprise at all.

Isn't this self-evident?

21 Dec 05 10:59

Benefits proven allegedly include reduction in drug:drug interaction; reduction in wrong doses but at the end of the day, computers are not in control and the user - the clinician, can still make mistakes with a computer just as much as without one. Anyone who has sucessfully implemented electronic prescribing, (and there aren't that many in UK), always tell us about just how difficult it is, but that the benfits are worth while. Well, let us see that evidence and let us see contradictory evidence and then lets make an informed comment and/or judgement. At the moment I a m not sure I have seen enough evidence either way.

computerised process will lead to any reduction in either mortality or morbidity?

george.brown4@btinternet.com

21 Dec 05 11:08

1. Unambiguous prescriptions....reducing the chance of the wrong medication/dose being adminstered to the patient. Handwriting legibility issues are a constant problem with manual prescribing and drug administration.

2. Decision Support.....appropriateness of dug/dose selection.

For example - in my past as a Pharmacy Technicain (a previous career), I've intervened plenty of times with regards to allergy issues - such as the patient who is allergic to penicillin being prescribed Amoxicillin - with a potentially fatal outcome.

The benefits of e-prescribing over the traditional manual prescribing methods in not in doubt - study after study has been done over the past few years and proven the benefits.

Was there any businesss process review done before the system was installed? Was no pilot done ? Was the system developed with clinical input from clinicians/ nurses/ pharmacists etc?

A reference please?

21 Dec 05 14:02

If George can supply a reference to a single scientifically robust study (with appropriate randomisation of subjects, a validated methodology for evaluation, appropriate controls for other variables, and some evidence of reliability) that confirms - taking into account benefits *and* drawbacks of both manual and electronic approaches - then I for one would be very interested to read the paper.

Health informatics research was described some time ago in the BMJ as a "descriptive feast but an evaluative famine". And the hard (evaluative) evidence that we are repeatedly told underpins most the the CfH programme's activities is actually quite difficult to find, even in these days of evidence-based practice.

Re: A reference please...

colin@kineticconsulting.co.uk

22 Dec 05 08:17

This is academic thinking taken too far. The poster refers to the same BMJ that stated that medical errors kill 20 000 to 30 000 people a year (see: http://news.bbc.co.uk/1/hi/uk/682000.stm).

Do I really need scientifically robust study with appropriate randomisation of subjects, a validated methodology for evaluation etc. to tell me that a tired junior doctor will make better prescribing decisions when supported by an information system that contains comprehensive information on drugs and their interactions?

I suspect, similary sweeping statements could be made about many fields.

Evidence-based academic studies alone do not guarantee or prove efficacy. What is needed is a scientific approach with a solid dose of practicality and common sense.

Wirral in "the exclusive circle"

Pete.marsh@whnt.nhs.uk

22 Dec 05 13:45

Wirral have been looking at Cerner amongst other suppliers for a number of years and we beleive their are significant patient benefits in safety and quality with CPOE and Eprescribing and administration. We have demonstrated many before and after effects in the past in open days, articles, conferences in UK and world wide. Over this period of reviewing replacement suppliers, most Wirral staff would prefer to stay with the same solution set they have now, but the environment has changed and support for the current application precludes this decision. Looking at the paper referred to in this article, some of the discussions revolved around being divorced from care giving, delays in electronic admission so they could start treatment. There is a suggestion now, they have resolved some of these, maybe doing insurance and credit asessment in USA delays the admission process, it doesnt in a UK hospital (yet anyway). Wirral always have, and will have a basis of treat the patient first and worry about the paper/computer later, I cant imagine anything else being the case with the technology currently available. One could look at the rollout plan - big bang over 6 days and doing ITU in the early phase certainly is not part of Wirral's strategy for transitioning to another supplier. Double Blind trials? - whover did that for the use of a telephone in A&E or ITU ? - decision support certainly should be ratified in terms of patient safety.

Wirral plans for Cerner eprescribing are different to that from the current installs. Wirral have always beleived in proactive decision support, i.e. get the right product, dose,route,schedule first time and not fire alerts at the end. A work flow that educates and influences the physicians behaviour has been the moethod of choice and the most effective in our local setting. Having Star Wars switch on bells and whistles at the end of the physician workflow has never (in my opinion) been the way to do it. They are busy people and want to do it once, quickly and accurately and according to best practice agreed locally.

On another note Wirral have yet to sign any contract with Cerner/Fujitsu.

But when lives may be at risk .....

22 Dec 05 19:27

In response to the last two comments - just a few thoughts.

I'm all for practicality and common sense. However, given that there is now some published data suggesting that using computers in high-dependency situations actually puts lives at risk, then there must surely be a strong argument for more science here. Taken at face value, Colin's description of e-prescribing associated with information on drug interactions, decision support, alerts etc does seem like a safer option for tired junior doctors (and others): as long as all other things are equal. But this latter point is the real issue and the findings of the Pittsburgh study suggest that there are some detrimental factors resulting from the adoption of computerised approaches that need to be more fully explained. And it'd be highly unfortunate if the widespread introduction of computers merely introduced a whole new set of fatal error types rather than preventing whatever proportion of the 20 - 30,000 annual deaths (not clear from the BBC report) result from the limitations of paper-based prescribing.

Pete's comment expresses "beliefs" about significant benefits and mentions a "suggestion" that some issues have been resolved. I've actually seen the Wirral system, it looked good and we've been told it delivers benefits. Now let's see if we can turn the local "beliefs and suggestions" into something approaching a "generalisable proof" that we can confidently apply NHS-wide.

Lastly, this is not about sterile academic debate nor about setting up inappropriate "double blind" trials (which was never suggested and is a little hard to imagine in the prescribing situation). Rather, it is about acknowledging that new technology carries risks (and the suggestion here is that it may kill more people); about taking a balanced and objective approach to its introduction into an often hostile and sceptical clinical environment; about accepting good evaluation as an essential element of best practice in both medicine and IT; and in establishing collaborative working between the coal-face "doers" and the academic "thinkers" so that the optimum result is delivered for patients.