Our August 21, 2007 Patient Tip of the Week “Costly Complications About to Become Costlier” provided estimates of the costs to hospitals of some common potentially preventable complications. Our January 22, 2008 Patient Safety Tip of the Week “More on the Cost of Complications” provided further insights into such costs.
Now two new studies have looked at the cost of complications. The first (Shreve 2010), done for the Society of Actuaries, estimates the cost of medical errors in 2008 to have been $19.5 billion. Of this $17 billion was in direct medical costs and the rest in indirect costs.
That study was based on medical claims data and used ICD-9 codes to identify possible cases of medical error. It used assumptions based on prior studies to come up with the methodology. It assumed 7% of inpatients have some type of medical injury, resulting in 6.3 million medical injuries and, of these, 1.5 million are associated with medical errors. They then compared the costs against those of a cross-matched control sample. They note these estimates do not include things like malpractice settlements and pain and suffering and suspect these represent underestimates and should be looked at as a lower limit to the actual cost. The 5 categories with the highest measurable costs were decubitus ulcers, post-operative infections, medical complications of devices, implants or grafts, post-laminectomy syndrome, and hemorrhagic complications of procedures.
The second study (Lucado 2010) was an AHRQ HCUP study looking at the influence of hospital associated infections from 2000-2007, using a methodology that captured AHRQ Patient Safety Indicator PSI #7 and applying some risk adjustment for factors such as comorbidities. Comparing the cost of cases with such infections to those without infections, they found HAI’s were associated with 19.2 days longer lengths of stay and average costs of $43,000 more. The incidence of HAI’s peaked in 2004 and 2005 at 2.3 cases per 1000 admissions, then declined to 2.03 cases per 1000 in 2007. Though the incidence increased with age, the highest occurrence rate was in the age population 45-64 years old.
Note, however, that in our January 22, 2008 Patient Safety Tip of the Week “More on the Cost of Complications” we mentioned a study by Kilgore et al (Kilgore 2008) that pointed out an important confounding factor in most prior studies on cost of complications: the length of stay confounding variable. In many cases, excess LOS is due to the nosocomial infection. However, in others excess LOS may simply reflect severity of illness and indicate a patient who is also at risk for nosocomial infections. So they attempted to correct for confounding variables, particularly for LOS, and found the “attributable” costs (i.e. attributable to the infection) was considerably less than other methodologies would have shown.
Nevertheless, the 2 new studies do provide us with some reasonable estimates of the potential costs of a number of different iatrogenic occurrences.
This month’s Pennsylvania Patient Safety Advisory also has a timely article “Demonstrating How To Demonstrate Return On Investment For Infection Prevention And Control" that walks you through the steps to develop a business case for infection control programs and a link to a sample business plan.
Shreve J, Van Den Bos J, Gray T, et al for the Society of Actuaries. The Economic Measurement of Medical Errors. Milliman June 2010
Lucado J, Paez K, Andrews R, Steiner C. Adult Hospital Stays with Infections Due to Medical Care, 2007. AHRQ. H-CUP Statistical Brief #94. August 2010
Kilgore ML, Ghosh K, Beavers,C M, Wong DY, Hymel PA, Brossette SE. The Costs of Nosocomial Infections. Medical Care 2008; 46(1):101-104
Pennsylvania Patient Safety Authority. Demonstrating Return on Investment for Infection Prevention and Control. Pa Patient Saf Advis 2010; 7(3): 102-7
Our January 15, 2008 Patient Safety Tip of the Week “Potentially Inappropriate Medication Use in Elderly Hospitalized Patients” we focused on an article by Rothberg et al (2008) that sheds some light on use of such medications on the inpatient side. They found that almost half of all inpatients over the age of 65 in a widely representative hospital sample received at least one medication meeting criteria as a “potentially inappropriate medication” (PIM) from the Beers classification. 49% received at least one PIM, 38% received at least one PIM with a high severity rating, and 6% received 3 or more PIM’s.“ focused on prescribing of medications on Beers List that may be potentially inappropriate for elderly patients. That column discussed the origin and updating of Beers List and ways to improve medication management in the elderly. Most of the published literature on Beers List has focused on outpatient or nursing home settings. Then in our June 2008 What’s New in the Patient Safety World “
Note that the study, which was based on review of administrative data, was not able to determine which PIM’s might have been continuation of medications prescribed prior to admission vs. medications newly started in the hospital. The process of medication reconciliation, which is now done on admission and discharge and should be done at all transfers of care, offers a good opportunity to minimize use of PIM’s.
As we implement CPOE (computerized physician order entry) with clinical decision support, alerts can be programmed to trigger when a PIM is ordered on a patient over a specified age. To minimize the number of alerts a physician may encounter (to avoid “alert fatigue”), we usually try to use standardized order sets wherever possible. However, the Rothberg article raises the possibility that sometimes the standardized order set, which often fails to take into consideration the age of the patient, might actually promote use of certain PIM’s. Therefore, organizations would be wise to review their standardized order sets to see whether any modification for Beer’s list medications is desirable.
Though we have been strong advocates of incorporating alerts to Beers list medications in CPOE (computerized physician order entry) there had been no studies delineating the efficacy or the unintended consequences of such alerts. Now, researchers at the Beth Israel Deaconess Medical System in Boston (Mattison 2010) have published results of such a system. Cognizant of the real potential for alert fatigue, these researchers carefully chose a subset of “potentially inappropriate drug (PIM’s)” from Beers list drugs to which to attach computerized warnings. They were able to demostrate approximately a 20% reduction in prescribing of flagged drugs. That decrease was noted immediately after implementation and was sustained over time. As a “control” they noted no reduction in those other Beers list drugs that were not flagged. Of specific drugs from the list, diphenhydramine had been the most often prescribed drug and showed the biggest reduction.
The study is important in that it confirms that alerts during CPOE can reduce prescription of potentially inappropriate drugs in the elderly. But it also contains other very useful lessons:
Don’t expect as great a reduction in prescribing of Beers list drugs when you incorporate alerts into e-prescribing on an outpatient basis. In our previous column, we mentioned that when a system in the outpatient setting notifies physicians that one of their patients is on a drug that is on the Beers List, the physician seldom discontinues the flagged drug in that individual patient. However, the number of new prescriptions for that drug diminishes in the patient population cared for by that physician.
Some continue to argue that the Beers List concept is not truly “evidence-based”. It is true that the list was originally conceived by an expert consensus panel and there are no randomized controlled trials regarding the safety in the elderly for each of the drugs on the list. However, there is ample evidence in the literature that each of those drugs may be associated with untoward effects in the elderly and the List has proven to be quite useful now in a variety of healthcare settings. We remain strong advocates of using clinical decision support tools to alert healthcare providers of such potentially inappropriate drugs in the elderly. The Mattison paper provides very helpful advice on how to implement such clinical decision support.
Update: See also our Patient Safety Tips of the Week for January 15, 2008 “ and October 19, 2010 “ ” and September 22, 2009 “Psychotropic Drugs and Falls in the SNF” and our What’s New in the Patient Safety World columns for June 2008 “Potentially Inappropriate Medication Use in Elderly Hospitalized Patients”.
Rothberg MB, Pekow PS, Liu F, Korc-Grodzicki B, Brennan MJ, Bellantonio S, Heelon M, Lindenauer PK. Potentially Inappropriate Medication Use in Hospitalized Elders. Journal of Hospital Medicine 2008; 3: 91-102 http://www3.interscience.wiley.com/journal/118860229/abstract
Mattison MLP, Afonso KA, Ngo LH, Mukamal KJ. Preventing Potentially Inappropriate Medication Use in Hospitalized Older Patients With a Computerized Provider Order Entry Warning System. Arch Intern Med. 2010; 170(15): 1331-1336
Since our June 22, 2010 Patient Safety Tip of the Week “Disclosure and Apology: How to Do It”, there have been a couple very relevant papers published on disclosure and apology. We had noted in our previous columns the accumulating evidence that malpractice costs and settlement costs may be lower when a policy of disclosure and apology is adopted. Now we have a new study (Kachalia 2010) that gives us a good estimate of the magnitude of the potential savings that results from a transparent disclosure and apology program. The University of Michigan Health System implemented such a system in 2001. It was established not as a means of saving money but to help establish a culture of patient safety. That system included identification and reporting of medical errors, disclosure of harmful medical errors to patients and offers of compensation to patients affected by such errors. They then did a before/after analysis of claims and settlements and found roughly a 60% reduction in average monthly rate of lawsuits, average monthly costs for liability, patient compensation, and non-compensation-related legal costs. The median time for claim resolution also diminished from 1.36 to 0.95 years. The authors do note that there was a trend in Michigan in the latter part of the study toward fewer malpractice claims so such confounding factors cannot be ruled out as having a role in these results. Nevertheless, the findings are quite impressive and further support the concept that disclosure and apology is not only the right thing to do from an ethical standpoint, but it is also the right thing to do from a fiscal standpoint.
The second publication (Dudzinski 2010) deals with disclosure of adverse or potentially adverse events involving multiple people. We’ve previously discussed such events (see our June 16, 2009 Patient Safety Tip of the Week “Disclosing Errors That Affect Multiple Patients”), focusing on an article appeared in the Canadian Medical Association Journal on “ ” (Chafe et al 2009).The new paper discusses in detail many of the dilemmas faced in the decision to disclose in such cases, noting that the issues are not uniform in all large-scale adverse events. There may be differences in the likelihood of harm, the ability to detect such harm, the presence or absence of potential treatment, etc. They even discuss the fact that in some cases the potential psychological harm caused by disclosure itself may be more serious than the risk of potential physical harm. But in almost every scenario they discuss, the bulk of the arguments favor transparency and disclosure.
Perhaps most importantly, they suggest every health care organization should have some preparation (whether it be a formal policy or not) for how to deal with potential large-scale adverse events. Our July 24, 2007 Patient Safety Tip of the Week “model policy put in place by the Veterans’ Health Administration. That policy includes convening of an advisory board that is multidisciplinary and representative of multiple constituencies to help frame all the relevant questions and guide the disclosure process.” provides an example of the sort of tool you need to have available in your organization to guide the response to serious incidents. But it does not include specifics in dealing with large-scale adverse events. The Dudzinski paper points out that without such guidance the responses of healthcare organizations to such events are typically slow and haphazard. They also cite a good
We’d like to refer you back to the CMAJ paper we highlighted in our June 16, 2009 Patient Safety Tip of the Week “Disclosing Errors That Affect Multiple Patients”. That paper nicely spells out the steps necessary in an investigation into multiple-patient events and outlines the many issues in setting up an effective communication strategy. The authors nicely describe the problems involved in the first step – timely identification of the error – because a representative sample of records to review requires an estimate of the potential number of patients affected. This step requires identifying the time period to review, scope of the review, training teams to do the reviews, and developing reliability checks. Then it needs to be determined whether the error affected clinical decision making. Then the physicians need to follow up with patients who may require changes in management after the review is done.
The decision about who and when to notify is complex. Especially when the number of patients to be notified is large, the decision is even harder. The authors voice their opinion that disclosure on a scale this large should be led by a physician other than the one(s) directly involved in the error. A formal disclosure plan should be developed, complete with dates and plans for specific disclosure to all potential stakeholders, including the public and the media. They point out that sensitivity must be used, particularly since some patients may have died in the interim (regardless of whether that was related to the incident or not). They provide some specific examples where timely disclosure to the public helped mitigate the response to errors and stress the need for continued honest updates to the public.
Their guidelines for public disclosure are well thought out. They stress early disclosure and note that it is best if the public hears about the event from you, not the media or other parties. As such, disclosure even before all the details are known may be appropriate, though initial disclosure should avoid making assumptions or identifying specific individuals involved. Patients need to be contacted individually. But using a website or dedicated phone line for keeping everyone up to date may be useful. The results of the investigation of the event need to be made public along with a description of the steps that were taken to prevent similar occurrences in the future. Overall, this is an excellent guideline to help any organization that must deal with an untoward event affecting or possibly affecting multiple patients.
The Dudzinski and Chafe papers plus the VHA directive should help your organization develop guidelines for what to do if you are ever faced with such circumstances.
Kachalia A, Kaufman SR, Boothman R, et al. Liability Claims and Costs Before and After Implementation of a Medical Error Disclosure Program.
Ann Intern Med 2010; 153: 213-221
Dudzinski DM, Hébert PC, Foglia MB, Gallagher TH.
The Disclosure Dilemma — Large-Scale Adverse Events. NEJM 2010; 363(10): 978-986
Chafe R, Levinson W, Sullivan T. Disclosing errors that affect multiple patients. CMAJ 2009 180: 1125-1127
Veterans Health Administration. Disclosure of adverse events to patients. VHA directive 2008-002. Washington, DC: Department of Veterans Affairs, January 18, 2008. http://www1.va.gov/vhapublications/viewpublication.asp?pub_id=1637
The UK NPSA (National Patient Safety Agency) has issued an alert “Reducing treatment dose errors with low molecular weight heparins”. The report and alert is based on over 2700 reported incidents in the UK involving dosing of low molecular weight heparins (LMWH’s). The dosage errors were most often related to failure to calculate the dosage based upon patient weight, renal function or clinical indication. Both underdosing and overdosing were seen. Failures in communication at transitions of care also contributed to adverse outcomes.
Failure to record accurate weights was problematic. Often the initial weight in the chart was an estimate. Patient estimates of their weight were usually better than estimates made by healthcare workers but both were often substantially different than the actual measured weight. Some of the problems in inaccurate recording of patient weight may be broken equipment, failure to use special scales in patients too ill (or too big) for conventional scales, limited staffing, and failure to appreciate the importance of accurate weight measurement. And the weights need to be readily available to all involved in the prescribing, preparation, and administration of LMWH’s (for example, weights need to be recorded on medication administration records or in the electronic medical record). While underdosing is common in obese patients, it is surprising how often overdosing is seen in patients whose actual weight is less than their estimated weight. Errors in dose calculation may be seen when pounds and kilograms are mixed up. Weights should be accurately in the chart or the electronic medical record using a consistent unit of measurement (usually kilograms). If the initial weight recording is an estimate, there should be some sort of automated reminder for a responsible individual to get an accurate actual weight measurement at some later date. Also, patients who are hospitalized for long periods may have substantial changes in weight so it is important to reassess weight on a regular basis.
Because most LMWH’s are renally excreted, dosing changes may be necessary based upon the patient’s renal function. Most commonly, overdosing with subsequent risk of hemorrhage occurs when LMWH’s are given to patients with impaired renal function.
Dose of LMWH’s may also vary depending upon the clinical indication. For example, different dosing is used for DVT prophylaxis, treatment of actual DVT or PE, acute coronary syndrome, myocardial infarction, etc.
Using standardized order sets or protocols or nomograms for ordering LMWH’s is advisable and such should take into consideration the factors of patient weight, renal function, and clinical indication. The supporting information for the NPSA alert includes examples of LMWH prescribing tools and dose calculation tools.
Good medication reconciliation is essential at all transitions of care. The NPSA alert gave examples of harm from LMWH’s that were continued long after they should have been discontinued.
All heparins should be considered high-alert drugs. All too often we have a tendency to be more complacent when using LMWH’s than unfractionated heparin. We somehow think they are safer because we do not have to routinely monitor coagulation parameters. But, clearly, adverse events related to use of LMWH’s are not uncommon and you need to have good systems in place to minimize those risks.
NPSA (UK). Reducing treatment dose errors with low molecular weight heparins
July 30, 2010
rapid response reportt