Healthcare Consulting Services
PO Box 1230
Grantham, NH 03753
ph: 716-550-1106
btruax
April 4, 2017
Deprescribing in Long-Term Care
AHRQ has just added to their collection of patient safety primers a new one on patient safety in long-term care (AHRQ 2017). It separates long-term care into long-term acute care hospitals, inpatient rehabilitation hospitals, and skilled nursing facilities (SNF’s). Almost 40% of Medicare patients are discharged to some form of long-term care facility following hospital discharge. The term “post-acute care” generally refers to care in a setting where care following an acute hospital discharge is rendered. Such patients represent an increasing proportion of the overall SNF population. One study found that 22% of Medicare SNF patients had an adverse event, about half of which were preventable. A similar study (Levinson 2016) found that 29% of Medicare patients admitted to a post-acute rehabilitation facility (rehab units in acute care hospitals were excluded) experienced either an adverse event or temporary harm event (see our September 2016 What's New in the Patient Safety World column “Adverse Events in Rehab Facilities”). Almost half (46%) of these were deemed to be likely preventable. These event rates are really quite similar to rates the OIG has found for Medicare patients in acute hospitals (27%) and simply demonstrate that all the factors which contribute to adverse events in hospitals are not unique to acute care hospitals but also occur in almost all healthcare settings.
The new AHRQ primer points out that common hazards in older patients in these settings include medication errors, healthcare-associated infections, delirium, falls, and pressure ulcers. Timely in this regard are some new studies on medication issues in long-term care settings.
A recent systematic review of medication errors found that medication errors occurred in 16-27% of SNF residents in studies looking at all types of medication errors (Ferrah 2017). And 13-31% of SNF residents had medication errors in those studies looking at transfer-related medication errors. And 75% of SNF residents received at least one potentially inappropriate medication (PIM). However, the authors also found that serious outcomes were relatively infrequent and death related to medication errors was rare in this setting, though they speculated that serious outcomes might be underreported.
It should not be surprising that medication errors and adverse drug events (ADE’s) are near the top of the list of adverse events in long-term care. The AHRQ primer notes that many of the safety tools we use in acute hospitals (eg. barcoding, CPOE) are not as widely implemented in long-term care. In addition, patients in long-term care often have multiple comorbidities and are on numerous medications. So two interventions in long-term care are critical in reducing ADE’s: medication reconciliation and deprescribing. We’ve done numerous columns on deprescribing (see the list at the end of today’s column). Liu and Campbell recently offered some tips on deprescribing in long-term care (Liu 2016).
The Liu article includes an excellent tip sheet for deprescribing in the nursing home. It builds upon the excellent 5-step protocol to aid the deprescribing process described by Scott and colleagues (Scott 2015) that we discussed in our July 2015 What's New in the Patient Safety World column “Tools for Deprescribing”. But it adds a few considerations that focus on patients in long-term care. The Liu tip sheet emphasizes that deprescribing should be considered any time the patient’s medication list is reviewed, which should include any time there is a change in a patient’s condition or a new symptom, before prescribing any new medication, and at every visit. It then recommends the following steps before discontinuation or tapering of any medication:
Once you’ve identified a medication to discontinue, begin taper process and monitor and reassess:
And then they add another important point that we have harped upon in many columns (most recently in our March 2017 What's New in the Patient Safety World column “Yes! Another Voice for Medication e-Discontinuation!”): document all about discontinuation of the medication! That includes documenting the reason for discontinuation, the plan for monitoring and follow-up, the patient/family communication, other communications, and outcomes).
A couple points deserve special comment. Particularly important is consideration on how the intended benefit of any medication relates to the overall goals of care, life expectancy, and disease trajectory (and length of time to benefit of preventive medications). The article also has tables with symptoms that are common side effects of various medications and common examples of the prescribing cascade. For example, a patient might develop urinary retention while taking amitriptyline. But if it is not recognized as a side effect of the amitriptyline the patient may be started on tamulosin. The article also has a nice list of medications that are commonly associated with discontinuation syndromes. The article also has a nice discussion about some of the barriers to deprescribing that may come from patients, families, collaborative partners, and nursing staff.
We think you’ll use this tip sheet often if you care for patients in long-term care. Quite frankly, all the recommendations apply equally well to elderly patients in almost any setting.
And, of course, one of the items on the American Geriatrics Society list of things to avoid in the Choosing Wisely® Campaign is “Don’t use antipsychotics as the first choice to treat behavioral and psychological symptoms of dementia.” Antipsychotics have often been used in long-term care not just in patients with delirium but also in patients with dementia and agitation or aggressive behavioral issues. We’ve often discussed the efforts to reduce use of antipsychotics in long-term care patients (see our February 3, 2015 Patient Safety Tip of the Week “CMS Hopes to Reduce Antipsychotics in Dementia”).
Use of antipsychotics in patients with dementia has long under fire because of limited efficacy and occurrence of serious adverse effects, such as an increase in stroke and mortality (Corbett 2014). They may also cause sedation, extrapyramidal signs, and some may produce orthostatic hypotension. The latter may all contribute to falls and fractures, as reported recently in patients 65 years and older who were started on an atypical antipsychotic medication as an outpatient (Fraser 2015). That study found that antipsychotic use increased the risk of serious falls by 52% and the risk or nonvertebral osteoporotic fracture by 50% compared to a matched control group, regardless of the specific agent used.
Despite guidelines and warnings against their use, antipsychotics continue to be used often in nursing homes and long-term care settings. In 2012 CMS challenged LTC and SNF facilities to reduce use of antipsychotics by 15% and between the end of 2011 and the end of 2013, the national prevalence of antipsychotic use in long-stay nursing home residents was reduced by 15.1 percent. CMS developed its National Partnership to Improve Dementia Care in Nursing Homes, a public-private coalition of CMS and several other partners, with a national goal of reducing the use of antipsychotic medications in long-stay nursing home residents by a further 25 percent by the end of 2015, and 30 percent by the end of 2016 (CMS 2014). In a commentary on those goals Leonard Gellman, MD pointed out that most antipsychotic medications are not actually prescribed by the nursing homes (Frieden 2014). Rather they are often started when the patient is in a hospital and continued upon discharge or they may have been started by the patient’s primary care physician. Once they have been started, facilities and patients’ families are reluctant to discontinue or reduce them. Since the start of the CMS National Partnership, there has been a decrease of 27 percent in the prevalence of antipsychotic medication use in long-stay nursing home residents, to a national prevalence of 17.4 percent in Fiscal Year (FY) 2015 Quarter 3 (CMS 2016a).
Our February 3, 2015 Patient Safety Tip of the Week “CMS Hopes to Reduce Antipsychotics in Dementia” further described several interventions and programs that had successfully reduced the use of antipsychotic drugs in long-term care settings. The updated CMS web page for the National Partnership also provides access to numerous resources (CMS 2016b).
One underappreciated consideration regarding medications in the long-term care population is the association between frailty and impaired renal function. The percentage of patients with frailty in this population is very high and, because of sarcopenia in the frail, measuring renal function with creatinine and eGFR is problematic. Therefore, Ballew and colleagues (Ballew 2017) looked at cystatin C in addition to creatinine and urine albumin in community-dwelling men and women 66 years or older. Of almost 5000 subjects, 7% were determined to be frail. Among frail subjects, prevalence of eGFR < 60 calculated using creatinine and cystatin C were 45% and 77%, respectively. After adjustment, frailty showed a moderate association with eGFRcr but a strong association with eGFRcys and albumin-creatinine ratio. Moreover, hyperpolypharmacy (defined as taking ≥10 classes of medications) was more common in frail individuals (54% vs 38% of nonfrail), including classes requiring kidney clearance (eg, digoxin) and associated with falls and subsequent complications (eg, hypnotic/sedatives and anticoagulants). While this was not a long-term care population, we don’t doubt that the same findings would apply to the long-term care population since the prevalence of frailty there is so high. These results suggest we need to be assessing renal function in this population using cystatin C to make more rational judgments about usage and dosing of certain drugs in this population.
Transitions from acute care to long-term care or other post-acute care setting are particularly vulnerable to error. Medication reconciliation at the time of hospital discharge provides a good opportunity to identify medications that might be appropriate for discontinuation or dose reduction (see our May 2015 What’s New in the Patient Safety World column “Hospitalization: Missed Opportunity to Deprescribe”). One group of clinicians implemented a brown bag medication reconciliation process in the hospital setting to decrease medication discrepancies by encouraging evaluation of medication adherence, side effects, and monitoring at posthospitalization follow-up (Becker 2015). After implementation, a 7% decrease in reportable errors was noted.
And several times we have mentioned a specific type of error that may occur in patients discharged back to an SNF. The classic example is methotrexate. Methotrexate for most non-oncologic indications is dosed on a weekly basis. But sometimes patients returning from a hospital or ER or specialty clinic erroneously get started on daily methotrexate at the SNF, often with dire consequences. In our July 2011 What's New in the Patient Safety World column “More Problems With Methotrexate” we noted that the patient in a long-term care facility may be especially vulnerable. In such cases, the original order for methotrexate is usually written by a specialist. The patient is then followed in the LTC facility typically by a primary care physician who may be less knowledgeable about the particular use of methotrexate for that condition. Also, the LTC patient may not be seen by a physician for periods as long as a month. And many LTC patients have cognitive impairments that might prevent them from understanding issues about their medications. So if a medication reconciliation error has occurred and a patient intended for once weekly dosing is now on daily dosing, the opportunity for toxicity is greatly increased. So LTC facilities should take steps to ensure that any of their residents taking methotrexate get the same level of supervision and protections that non-LTC patients would get. Those SNF’s with CPOE should always use weekly dosing as the default and require hard stop overrides for any attempt to order daily methotrexate.
And, of course, the classic example of medications that are no longer needed are proton pump inhibitors that were begun during an ICU admission but never discontinued. If the patient gets sent back to the long-term care facility on PPI’s they are likely to be continued indefinitely, even though they are no longer necessary, unless someone does a careful review of the medication list with deprescribing in mind.
Lastly, one anecdote we love to tell. The fall rate at one SNF fell dramatically one month. So we asked “how did you do it?”. It turned out that there had been a disruption in the relationship the SNF had with a consulting psychiatrist. As a result, multiple patients did not have renewals of their orders for a variety of antidepressants and antipsychotic agents. That inadvertent cessation of these medications had led to a striking reduction in falls!
Some of our past columns on deprescribing:
Some of our past columns on patient safety issues in long-term care settings:
Some of our past columns on Beers’ List and Inappropriate Prescribing in the Elderly:
Our prior columns related to methotrexate issues:
References:
AHRQ PSNet. Patient Safety Primer. Safety in Long-term Care. AHRQ 2017
https://psnet.ahrq.gov/primers/primer/39
Levinson DR (Office of the Inspector General. Adverse Events in Rehabilitation Hospitals: National Incidence Among Medicare Beneficiaries. July 2016
https://oig.hhs.gov/oei/reports/oei-06-14-00110.asp
https://oig.hhs.gov/oei/reports/oei-06-14-00110.pdf
Ferrah N, Lovell JJ, Ibrahim JE. Systematic Review of the Prevalence of Medication Errors Resulting in Hospitalization and Death of Nursing Home Residents. J Am Geriatr Soc 2017; 65(2): 433-442
http://onlinelibrary.wiley.com/doi/10.1111/jgs.14683/full
Liu LM, Campbell IG. Tips for Deprescribing in the Nursing Home. Annals of Long-Term Care 2016; 24(9): October 17, 2016
http://www.managedhealthcareconnect.com/article/tips-deprescribing-nursing-home
Scott IA, Hilmer SN, Reeve E, et al. Reducing Inappropriate Polypharmacy. The Process of Deprescribing. JAMA Intern Med 2015; 175(5): 827-834
http://archinte.jamanetwork.com/article.aspx?articleid=2204035
Choosing Wisely® Campaign. American Geriatrics Society list.
http://www.choosingwisely.org/clinician-lists/#parentSociety=American_Geriatrics_Society
Corbett A, Burns A, Ballard C. Don’t use antipsychotics routinely to treat agitation and aggression in people with dementia. BMJ 2014; 349 doi: http://dx.doi.org/10.1136/bmj.g6420 (Published 03 November 2014)
http://www.bmj.com/content/349/bmj.g6420
Fraser L-A, Liu K, Naylor KL, et al. Falls and Fractures With Atypical Antipsychotic Medication Use: A Population-Based Cohort Study. Research Letter. JAMA Intern Med 2015; Published online January 12, 2015
http://archinte.jamanetwork.com/article.aspx?articleid=2089230
CMS. National Partnership to Improve Dementia Care exceeds goal to reduce use of antipsychotic medications in nursing homes: CMS announces new goal. CMS Press Release September 19, 2014
CMS. Update Report on the National Partnership to Improve Dementia Care in Nursing Homes. June 3, 2016
CMS. National Partnership to Improve Dementia Care in Nursing Homes. Last updated 12/1/2016
Frieden J. Antipsychotics for Dementia: CMS Says Use Less. Medpagetoday 2014; September 22, 2014
http://www.medpagetoday.com/Geriatrics/Dementia/47781
Ballew SH, Chen Y, Daya NR, et al. Frailty, Kidney Function, and Polypharmacy. The Atherosclerosis Risk in Communities (ARIC) Study. Am J Kidney Dis 2017; 69(2): 228-236
http://www.ajkd.org/article/S0272-6386(16)30525-X/fulltext
Becker D.Implementation of a Bag Medication Reconciliation Initiative to Decrease Posthospitalization Medication Discrepancies. Journal of Nursing Care Quality 2015; 30(3): 220-225
Print “Deprescribing in Long-Term Care”
April 11, 2017
Interruptions: The Ones We Forget About
We admit it. We’ve fallen into the same trap in our columns that we fall into every day in the office or hospital. All our columns on the impact of interruptions and distractions have focused on interruptions to healthcare workers. Now a new “Piece of My Mind” in JAMA reminds us about the impact of interruptions on our patients (Mauksch 2017).
For several decades now, when teaching students and residents about listening skills, we have quoted from a study by Beckman and Frankel that physicians interrupt their patients on an average of 18 seconds (Beckman 1984). Actually, the timeframe might be even shorter. A study of resident interactions with patients showed the interruption occurred on an average of 12 seconds (Rhoades 2001).
But in the recent piece Larry Mauksch points out that the interruptions are not always negative and provides some great examples of how to interrupt your patient tactfully so that you do not appear rude yet help focus on things that are important to the patient. He notes that some interruptions may build rapport, offer support, and express cooperation.
He suggests adhering to “The Triple E” when interrupting patients:
Mauksch suggests the following script for redirecting patients who have a tendency to jump from one symptom/topic to another: “Excuse me, your back pain sounds distressing, but we were only part way through addressing your asthma. How about we finish the asthma and then see if there is time for your back pain?”
We all have a few patients who tend to move from topic to topic and are difficult to get to focus on key points. Most of us have learned that, to be fair to both those patients and other patients we are seeing the same day, thoughtful scheduling is important. If we schedule such patients early in an office/clinic session we will always be pressured by time and workload and all patients get short-changed. So we tend to schedule the more “loquacious” patients toward the end of an office/clinic session.
But what if you are a patient? When we do our patient safety for patients courses we tell our audience they should have a checklist of the symptoms/topics they hope to discuss with their physician or other healthcare provider. The items should be roughly prioritized with those the patient considers most important near the top of the list. Giving the physician a copy of the checklist at the beginning of the session may also help him/her get an idea about time allocation.
Mauksch also points out that the physician can also do agenda setting early in a session to help avoid late “oh, by the way” issues. For example, he suggests the following script: “Excuse me for a moment. Your knee has been painful. Before we talk further about this pain, I’d like to know if you have something else important to address today. This way you and I can figure out how to make the best use of our time.”
Read Mauksch’s article. It’s short but to the point. And his suggestions are valuable not only for your patient interactions but apply equally to many of the everyday interactions you have with multiple other people! They are an important part of treating individuals with respect and you’ll find in the long run that will make your life easier.
Actually, another group of clinicians recently took agenda setting to a new level: collaborative agenda setting in the electronic visit note. Anderson et al. (Anderson 2017) had patients attending a large primary care safety-net clinic type their agendas into the electronic visit note before seeing their clinicians. Patients and clinicians agreed that the agendas improved patient-clinician communication (patients 79%, clinician 74%), and wanted to continue having patients type agendas in the future (73%, 82%).
Are there implications for patient safety? You bet! One of the examples provided by Anderson et al. was a patient typing on his agenda “lumps on my lungs”. The clinician explained that a pulmonary nodule had been identified on a CT scan on an ER visit that the clinician might have missed if the patient had not put it on his agenda.
Prior Patient Safety Tips of the Week dealing with interruptions and distractions:
References:
Mauksch LB. Questioning a Taboo. Physicians’ Interruptions During Interactions With Patients. JAMA 2017; 317(10): 1021-1022
http://jamanetwork.com/journals/jama/fullarticle/2610340
Beckman HB, Frankel RM. The effect of physician behavior on the collection of data. Ann Intern Med 1984; 101(5): 692-696
http://annals.org/aim/article/699136/effect-physician-behavior-collection-data
Rhoades DR, McFarland KF, Finch WH, Johnson AO. Speaking and interruptions during primary care office visits. Fam Med 2001; 33(7): 528-532
http://www.stfm.org/FamilyMedicine/Vol33Issue7/Rhoades528
Anderson MO, Jackson SL, Oster NV, et al. Patients Typing Their Own Visit Agendas Into an Electronic Medical Record: Pilot in a Safety-Net Clinic. Ann Fam Med 2017; 15(2): 158-161
http://www.annfammed.org/content/15/2/158.full
Print “Interruptions: The Ones We Forget About”
April 18, 2017
Alarm Response and Nurse Shift Duration
Wow! Two of our all-time favorite topics – alarm fatigue and healthcare worker fatigue – in the same column!
Researchers have just published a very revealing study on response times to alarms in a children’s hospital (Bonafide 2017). They used video monitoring to assess how long it took nurses to respond to physiologic monitor alarms and assessed the factors associated with those response times. Of 11,745 alarms among 100 children, 50 (0.5%) were actionable. That probably should not come as a surprise since previous studies have also demonstrated very high volumes of alarms that trigger for events for which no action is indicated, though the percentage of actionable alarms in this study is one of the lowest we’ve seen. In our February 23, 2010 “Alarm Issues in the News Again” where we discussed alarm fatigue we noted a study by Siebig and colleagues (Siebig et al 2010) in medical intensive care units that found only 15% of alarms were considered clinically relevant. Our November 2010 What’s New in the Patient Safety World column “Alarms in the Operating Room” noted another study (Schmid 2011) on 25 consecutive cardiac surgery cases. They noted an average of 1.2 alarms per minute and noted that approximately 80% of the alarms had no therapeutic consequences, a figure remarkably similar to that found in the Siebig study mentioned above. The 2013 Joint Commission Sentinel Event Alert on alarm safety cited the statistic that between 85% and 99% of alarm signals do not require clinical intervention. And in our August 16, 2016 “How Is Your Alarm Management Initiative Going?” we noted the study from researchers at UCSF (Drew 2014) in which 88.8% of the 12,671 annotated arrhythmia alarms were false positives. So the large number of non-actionable alarms in the Bonafide study should not surprise anyone.
But the Bonafide study went further and assessed factors that were associated with responses to alarms. Alarm responses were assessed only when the nurse was outside the room when the alarm triggered (note that the alarm system in this hospital sent a text message to responsible nurses for some of the more clinically significant alarms). The observed (unadjusted) median response time to out-of-room alarms was 7.0 minutes (range 5.2-8.8 minutes). The adjusted median response time among nurses was 10.4 minutes but varied considerably.
Do the long alarm response times in the Bonafide study surprise you? They shouldn’t. We previously noted a hospital found it took an average of 9.5 minutes for a clinician to respond to high-priority alarms before it implemented better communications systems to dramatically improve the response times (TJC 2011). We suspect if you were to do your own observational study (either direct observation or video recording) you may find similarly long response times.
But perhaps the most striking finding in the Bonafide study was that each hour that elapsed during a nurse’s shift was associated with a 15% longer response time, In hour #2 the response time was 6.1 minutes on average, compared to 14.1 minutes in hour 8.
Though the authors noted that some nurse work shifts were 12 hours, they did not report results for shifts longer than 8 hours. It also would have been interesting to see if those nurses having rest breaks or nap breaks had shorter alarm response times.
Other factors were also important:
Interestingly, the number of non-actionable alarms to which the nurse was exposed in the preceding 120 minutes was not associated with response time.
It would have been interesting to see whether the number of consecutive days worked had any influence on nurse response time. Prior studies in other industries have demonstrated that more errors are made on each consecutive day worked, a phenomenon we refer to as the “consecutive day phenomenon”. A study on shift workers in fields other than healthcare (Folkard 2003) showed that the risk of incidents increased each consecutive day worked. For example, on average for night shifts risk was 6% higher on the second night, 17% higher on the third night, and 36% higher on the fourth night (for morning/day shifts the corresponding risks were 2%, 7% and 17%). Extrapolating, one might suspect that we might see fewer errors if you only have to work 3 straight days rather than 4 or 5. The Folkard study demonstrated that risks are not uniform throughout the day but are greater at certain times, especially at night, and even vary based upon temporal relationship to breaks. Those authors stress that all these factors (number of successive night shifts, length of night shifts, and provision of breaks) must all be considered in combination. They note it is conceivable that a 12-hour night shift with frequent rest breaks might well prove safer than an 8-hour shift with only one mid-shift break. So we’d be very interested to see if alarm response times increase each day consecutively worked.
Overall, the Bonafide study suggests that nurses apply heuristics to guide how they respond to alarms. The results also suggest that chronic alarm fatigue, the result of long term exposure to nonactionable alarms during a nurse’s career, may be a more important predictor of response time than short term exposure to non-actionable alarms. But the most important point is that mental or physical fatigue as nurses are further into their shifts may significantly impact alarm response times. The authors note that this and the nurse:patient ratio are likely the two most modifiable factors they found impacting alarm response times.
The accompanying editorial (Sowan 2017) called for using the perspective of nurses on the causes for alarm fatigue and long response times and soliciting their suggestions for improvement. Those authors had previously surveyed nurses in a transplant/cardiac ICU and found nurses believed that existing monitoring devices are complex, questioned the ability and adequacy of the new monitoring systems to solve alarm management issues, pointed to the lack of prompt response to alarms, and indicated the lack of clinical policy on alarm management (Sowan 2015).
The Bonafide study is really a great contribution to our growing knowledge base on alarm management. It also demonstrates the importance of direct observation techniques (video monitoring in this case) in your alarm management initiatives to better understand both the nature of alarm responses and the factors contributing to those alarm response times.
There are several takeaways from the Bonafide study. First, as you do your inventory of alarms to determine which ones are truly needed, you should also do direct observation (either physical observation or video recording) to assess response times to the various alarms. That will help you prioritize your alarms and eliminate those that are clinically non-actionable. Our many columns listed below will help you identify those that may simply be nuisance alarms that just foster alarm fatigue. But you’ll probably find the most useful information in our October 2014 What's New in the Patient Safety World column “Alarm Fatigue: Reducing Unnecessary Telemetry Monitoring” and our Patient Safety Tips of the Week for July 2, 2013 “Issues in Alarm Management” and August 16, 2016 “How Is Your Alarm Management Initiative Going?”.
Moving lower priority alarms from audible status to messaging status can go a long way to reduce alarm fatigue and excessive noise in a variety of settings. Just keep in mind that things can still go wrong. See our February 9, 2016 Patient Safety Tip of the Week “It was just a matter of time…” for such an example.
The causes for prolonged response times may differ in your facility. The only way you’ll know is by collecting the information about those response times, soliciting input from your frontline staff about how to eliminate or mitigate the factors contributing to prolonged response times.
And you need to consider how you address the problem of healthcare worker fatigue, as we’ve discussed in the many columns below on the issue. Maybe when you do a detailed observational study you can also determine the impact of shifts longer than 8 hours and the impact of “the consecutive day phenomenon” and see whether interventions like allowing napping have any impact on alarm response times.
Prior Patient Safety Tips of the Week pertaining to alarm-related issues:
Some of our other columns on the role of fatigue in Patient Safety:
November 9, 2010 “12-Hour Nursing Shifts and Patient Safety”
April 26, 2011 “Sleeping Air Traffic Controllers: What About Healthcare?”
February 2011 “Update on 12-hour Nursing Shifts”
September 2011 “Shiftwork and Patient Safety
November 2011 “Restricted Housestaff Work Hours and Patient Handoffs”
January 2012 “Joint Commission Sentinel Event Alert: Healthcare Worker Fatigue and Patient Safety
January 3, 2012 “Unintended Consequences of Restricted Housestaff Hours”
June 2012 “June 2012 Surgeon Fatigue”
November 2012 “The Mid-Day Nap”
November 13, 2012 “The 12-Hour Nursing Shift: More Downsides”
July 29, 2014 “The 12-Hour Nursing Shift: Debate Continues”
October 2014 “Another Rap on the 12-Hour Nursing Shift”
December 2, 2014 “ANA Position Statement on Nurse Fatigue”
August 2015 “Surgical Resident Duty Reform and Postoperative Outcomes”
September 2015 “Surgery Previous Night Does Not Impact Attending Surgeon Next Day”
September 6, 2016 “Napping Debate Rekindled”
References:
Bonafide CP, Localio AR, JH, et al. Video Analysis of Factors Associated With Response Time to Physiologic Monitor Alarms in a Children’s Hospital. JAMA Pediatr 2017; Published online April 10, 2017
http://jamanetwork.com/journals/jamapediatrics/fullarticle/2614074
Siebig S, Kuhls S; Imhoff M, et al. Intensive care unit alarms - How many do we need? Critical Care Medicine 2010; 38(2): 451-456
Schmid F, Goepfert MS, Kuhnt D, et al. The Wolf Is Crying in the Operating Room: Patient Monitor and Anesthesia Workstation Alarming Patterns during Cardiac Surgery. Anesth Analg 2011; 112(1): 78-83; published ahead of print October 21, 2010
The Joint Commission. Medical device alarm safety in hospitals. The Joint Commission Sentinel Event Alert 2013; 50: 1-3 April 8, 2013
http://www.jointcommission.org/assets/1/18/SEA_50_alarms_4_5_13_FINAL1.PDF
Drew BJ, Harris P, Zègre-Hemsey JK, et al. Insights into the Problem of Alarm Fatigue with Physiologic Monitor Devices: A Comprehensive Observational Study of Consecutive Intensive Care Unit Patients. PLOS One 2014; Published: October 22, 2014
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0110274
The Joint Commission. Sound the Alarm: Managing Physiologic Monitoring Systems. The Joint Commission Perspectives on Patient Safety 2011; 11(12): 6-11 December 2011
http://www.jointcommission.org/assets/1/6/Perspectives_Alarm.pdf
Folkard S, Tucker P. Shift work, safety and productivity. Occupational Medicine 2003; 53: 95-101
Sowan AK, Reed CC. Alarm Fatigue in Adaptive Healthcare Systems. JAMA Pediatr 2017; Published online April 10, 2017
http://jamanetwork.com/journals/jamapediatrics/fullarticle/2614070
Sowan AK, Tarriela AF, Gomez TM, et al. Nurses’ perceptions and practices toward clinical alarms in a transplant cardiac intensive care unit: exploring key issues leading to alarm fatigue. JMIR Hum Factors 2015; 2(1) e3
http://humanfactors.jmir.org/2015/1/e3/
Print “Alarm Response and Nurse Shift Duration”
April 25, 2017
Dialysis and Alarm Fatigue
We publish details about significant incidents because if an event occurs at one facility, a similar event may well occur at another facility no matter how unusual the circumstances surrounding the event were. One of our earliest introductions to alarm fatigue happened in a dialysis unit. In fact, you’ve heard us state that we often bet hospital CEO’s that we will find a certain number of alarms disabled or otherwise tampered with when we do a hospital visit. One of the areas we know to go to for finding such alarms is the dialysis unit.
Though the incident occurred much earlier, we first discussed our dialysis alarm incident in our March 26, 2007 Patient Safety Tip of the Week “Alarms Should Point to the Problem”. An ESRD patient was having his regularly scheduled dialysis session. Since he would be in the dialysis center for several hours, he was in a comfortable lounge chair that tipped back. Also, since it was somewhat cool, he was offered a blanket to keep warm. Midway through the dialysis session, the low-pressure alarm rang. The nurse turned off the alarm and eyeballed the patient and saw no blood. Nothing further was done. Soon thereafter the low pressure alarm triggered again. This time it was recognized that the dialysis catheter had become dislodged and the patient had, in fact, had considerable blood loss. It had not been appreciated immediately because the blanket had been covering up the catheter site and the blood, rather than being visible on the floor, had been pooling in the webbing of the lounge chair.
Anyone who has ever spent time in an ICU or other high tech medical environment knows that the usual first response to an alarm is to turn the alarm off. Proper design of medical equipment therefore should force the responder to focus on the source of the problem. In the case at hand, the equipment and alarm were on the side of the patient opposite from the involved limb so that the visual attention of the responder was not directed immediately to the site the alarm was drawing attention to.
Unfortunately, you all have lots of equipment that have alarms that don’t make the responder focus directly at the problem. Faulty response to alarms is one of the “big 3” problems encountered in many root cause analyses of sentinel events. Performing FMEA (Failure Mode and Effects Analysis) is a good way to help anticipate events that might arise in your critical settings.
Fast forward to the present. The California Department of Public Health (CDPH) just published its most recent batch of statements of deficiencies (SOD’s) and plans of correction (POC’s) regarding incidents in California hospitals. One involved a case with striking similarities to that in our previous column (CDPH 2017). A patient admitted with an MI suffered deterioration of his chronic renal disease and was begun on continuous renal replacement therapy (CRRT) with a femoral catheter for dialysis access (technically, it was continuous veno-venous hemofiltration or CVVH). For comfort, the patient was covered with a blanket, which obscured the catheter access site. Unfortunately, at some point the return line became loose and disconnected from the femoral catheter which caused massive blood loss and cardiac arrest. He received CPR and blood transfusions and was resuscitated but died several days later.
Review of the CRRT Machine Data History print-out showed that several alarms had been triggered. First was an alarm warning that return pressure was dropping. The print-out indicated that the alarm was cleared 13 seconds after it was issued. The CRRT Machine manufacturer indicated that at the time of that alarm the screen had the sign, "WARNING: Return Pressure Dropping" on top of the screen in red color. On the left side of the screen was written in bold, "Possible leakage, or disconnection of return line or catheter. Patient is moving or being moved. Action: 1. Make sure return catheter is securely connected to both the return line and the patient. 2. To resume treatment press CONTINUE." The Prismaflex screen had touch screen buttons for EXAMINE ALARMS, DISCONNECT, bell icon with X {means MUTE), CONTINUE and HELP. Roughly 3 minutes later another alarm with the warning “Access Extremely Negative” was issued and this alarm automatically stopped the machine.
The “complainant” told the CDPH investigator that family members were at the bedside when the dialysis machine alarmed with a "warning message about return pressure". The complainant stated a nurse "silenced the alarm without checking the patient and walked out of the room for several minutes". The family member called the nurse and when the nurse came back, the patient looked like he was having seizure and was in cardiac arrest. The complainant stated when the nurse pulled the blanket at the time of the cardiac arrest there was pool of blood on the bed and the line was disconnected.
As we typically see in serious incidents with adverse patient outcomes, there was a cascade of events that each contributed to the unfolding crisis. The nurse had earlier temporarily stepped outside the patient’s room to take a call from the laboratory about a critical value (high lactate) for that patient and was looking for the physician to tell him/her about that critical value. The patient’s nurse at the time was also a “break nurse”, covering for the primary nurse who was on break.
We discussed salient distracting features in our January 14, 2014 Patient Safety Tip of the Week “Diagnostic Error: Salient Distracting Features”. Salient distractions obviously can apply to more than just diagnosis. In the current case, the lab had just called back a critical value on this patient status post MI. His serum lactate was high (indicating likely inadequacy of tissue perfusion) and that merited prompt intervention. So the nurse had to leave the room to first take the phone call from the lab and then locate and inform the physician about that critical value. The physician then ordered dobutamine and a venous blood gas. And the physician came and talked to the patient’s family about the lab results and medications. The nurse also had to respond to those orders. It’s not clear what role those events may have played in distracting the nurse from responding correctly to the alarm. Perhaps there was even an element of confirmation bias in that the physician had just been in the patient’s room and had not noticed anything unusual. It is not clear from the CDPH document why the nurse stepped out of the room immediately after clearing the alarm. Perhaps it was to carry out the orders related to the critical value.
The exact timeline is a bit sketchy in one regard. The CDPH document notes that the “break nurse” was expecting to cover for a 30-minute break. That nurse apparently took over at about 7:00 PM. The time of the first “Return Pressure Dropping” warning alarm was 7:54 PM. The CDPH document notes that the primary nurse “returned early” from break because the patient was coding. So we might wonder whether anticipation of a change in nursing coverage played any role.
The thrust of our March 26, 2007 Patient Safety Tip of the Week “Alarms Should Point to the Problem” was that alarm system setups should focus visual attention to that part of the system where the problem originates. The current incident would seem to indicate that does not work. The readout on the screen seems to clearly indicate where the responder should focus and even indicates what actions should be taken.
But there are at least 2 factors important in that regard. First is the location of the alarm screen. In the incident we described in our prior column the alarm system was positioned on the side opposite the dialysis access site, causing the responder to be visually focusing away from the site of the problem. We don’t know where the alarm screen was positioned in the current incident. But given that the low pressure alarm is probably the most important alarm (because you can bleed out if the problem is not promptly corrected) wouldn’t it make sense to position the alarm screen in a place that would force the responder to visually inspect the access site?
The second consideration is the complexity of the alarm screen. We don’t know what that screen looked like in the current incident. While the description of the readout sounds straightforward, we don’t know that it was the only element visible on the screen. Some of the screens of monitoring equipment can be incredibly complicated. That complexity may lead to responders simply taking the easy way out and clearing the alarm, particularly if they anticipate the alarm will be re-triggered if something serious is really going on.
But the best interventions are forcing functions. We’d suggest that these alarm systems program in a “hard” stop for this particular alarm that requires the responder to verify that he/she has inspected the access site. That verification should then become part of the medical record. Note that in the current event, there was no section in the electronic health record for documenting the monitoring of dialysis access and blood lines.
There were several other deficiencies or lessons learned in the CDPH case:
In the immediate aftermath of a serious event it is important to sequester any equipment and materials that may have been involved in the incident. That is important for two reasons: (1) to identify and mechanical defects that may have contributed and (2) to prevent using potentially defective equipment on other patients. Every facility should have ready access to a checklist for responding to serious incidents. Typically, with a serious event like the one described today, the nursing supervisor or the administrator on-call should be immediately notified. That individual would then access the checklist. The checklist should specify who needs to be contacted and what action steps need to be taken, including sequestering equipment. See our sample Serious Event Response Checklist.
The facility’s plan of correction included the following:
One of the elements in the facility’s POC was to stop use of blankets in patients receiving CRRT. Frankly, we don’t know how practical that is, let alone how patient unfriendly that may be. It should not be difficult to create blankets that would provide both warmth and privacy for patients, yet leave critical dialysis access sites uncovered. Note that the facility’s response technically did not ban blankets per se but just says blankets cannot cover the dialysis access site.
We often use lessons learned from these CDPH documents because events they describe might be replicated at other hospitals. The fact that this current event almost duplicates one we described many years ago is evidence that the circumstances, root causes, and contributing factors present at one facility may well also be present at other facilities. We hope that our readers will look at today’s cases and evaluate their own facilities to see if their practices might render them vulnerable to similar events.
Prior Patient Safety Tips of the Week pertaining to alarm-related issues:
References:
CDPH (California Department of Public Health). Complaint Intake Number: CA00471877; posted 4/20/2017
http://www.cdph.ca.gov/certlic/facilities/Documents/2567_Kaiser220012544_IJAP_SanFrancisco.pdf
Print “Dialysis and Alarm Fatigue”
May 2, 2017
Anatomy of a Wrong Procedure
Our May 2017 What's New in the Patient Safety World column “Another Success for the Safe Surgery Checklist” discusses the utility of the Safe Surgery Checklist (also known as the Surgical Safety Checklist) in improving patient outcomes. It stresses the importance of not just the checklist but also active engagement of and buy-in by staff in roll out and adoption of the checklist.
Today’s column illustrates how having such a checklist without an appropriate culture of safety can lead to adverse patient outcomes.
In its most recent batch of statements of deficiencies (SOD’s) and plans of correction (POC’s) regarding incidents in California hospitals the California Department of Public Health (CDPH) included a case in which a patient’s ovaries were incorrectly removed during a surgical procedure (CDPH 2017). A patient with symptomatic uterine leiomyomata was intended to have a total laparoscopic hysterectomy with bilateral salpingectomy and incidental appendectomy. Intent was to preserve her ovaries. However, the ovaries mistakenly removed during surgery.
The surgical booking form listed as the procedure “TCH (total complete hysterectomy), BSO (bilateral salpingo oophorectomy), appy (appendectomy) ... " and the roster of surgical cases scheduled for that day indicated for this patient "Procedure: laparoscopic hysterectomy, BSO (bilateral salpingo oophorectomy, appendectomy".
But the consent indicated the (handwritten) procedure: "Laparoscopic hysterectomy-removal of both fallopian tubes –appendectomy" and the preop H&P signed by the surgeon indicated under "Plan” “A total laparoscopic hysterectomy with bilateral salpingectomy is planned. We will save the ovaries and incidental appendectomy will be performed.”
The surgeon led the surgical timeout and no one noted the disparity between the booking form and surgical case roster vs. the H&P and consent. The surgery was performed and both ovaries were mistakenly removed. The following day the patient apparently informed the surgeon that she was not supposed to remove her ovaries. The surgeon admitted she had made a mistake and removed the ovaries.
Root cause analysis revealed numerous contributing factors/root causes. One contributing factor for wrong patient/site/side/procedure incidents is surgical scheduling or booking. In our October 30, 2012 Patient Safety Tip of the Week “Surgical Scheduling Errors” we discussed how errors made during booking or scheduling are commonly made. But we noted that the vast majority of such errors never reach the patient because they are intercepted during several opportunities before the surgery actually takes place. In this case those several opportunities failed to intercept the error. In addition to the faulty immediate surgical (verification) timeout in the OR, there were opportunities to identify the error when the patient was admitted and when the patient was transferred to the preprocedure area. Also missing from the CDPH documents is any mention of a presurgical huddle/briefing. The presurgical huddle is an excellent opportunity for the surgeon, anesthesiologist, and OR nurse to get together and not only confirm what is to be done but also discuss any special needs for the procedure and plan for possible contingencies (see the list of our prior columns on briefings and debriefings below). For example, we feel that one topic to be discussed during the preop briefing is what surgical specimens are expected to be sent to the pathology lab. Had that been discussed here it probably would have been recognized that the ovaries were not to be removed.
The facility apparently did utilize the WHO Surgical Safety Checklist but, as we noted above, there were several features in the local culture that rendered that checklist inadequate. One nurse stated "My signature was for the patient identification only, the doctor MD led the time out..". Another nurse stated: "My signature only means I was part of identifying the correct patient and date of birth, not the right procedure". Neither nurse saw the consent or the H&P or other portions of the medical record. Hospital policy required the timeout to be led by the circulating nurse, not the physician. Yet both nurses and the Director of Quality Services indicated this physician leads her own timeout. The physician said “…she is not perfect and she forgot the correct procedure ... " yet did not seem to acknowledge that use of a checklist is to help avoid forgetting such items. And it is not clear from the CDPH document whether the facility was auditing compliance with the timeout/verification process and the checklist before this event took place. Even regarding the surgical scheduling, the response was "That was just a request for a time slot from the doctor's office ... " rather than acknowledging the importance of scheduling in avoiding wrong patient/site/laterality/procedure. The old saw “culture trumps policy every time” was clearly in effect here.
This case reaffirms the problems associated with surgical scheduling/booking. For details about all aspects of surgical scheduling that impact patient safety see our October 30, 2012 Patient Safety Tip of the Week “Surgical Scheduling Errors”. Another point not elaborated upon in the CDPH document is use of the abbreviation “BSO” for bilateral salpingo-oopherectomy. We’ve cautioned that in scheduling forms and documents you should avoid both abbreviations and acronyms. The full name of the procedure should be written out and abbreviations avoided. In particular, use of “R” or “L” or “B” for laterality should be avoided. Sometimes an “R” gets misinterpreted as a “B” or vice versa when indicating laterality of the procedure. We would also wonder here whether the person calling from the physician office simply called in “BSO” out of habit when only a bilateral salpingectomy was intended (without the oophorectomy). Often it is a non-clinical person calling in the case from the office or clinic and just as often (as appears to have happened in this particular case) a non-clinical person is receiving the request at the facility and entering it into the schedule.
The case also clearly illustrates the importance of having primary source documents available in the OR. During the verification process all parties need to make sure that the procedure and laterality are corroborated in the H&P and consent.
One critical issue is the availability of the H&P at the time of surgery. Particularly since most patients having elective surgery are admitted on the day of surgery or are having same day surgery, the H&P must be available in advance. We’ve seen cases where the surgeon dictates the H&P on the day of admission and a readable copy may not be available for all the OR players to read. Therefore, it is imperative that your OR require the H&P from the surgeon’s office be available prior to the day of surgery (and remember it must be appropriately updated when it is done in advance). Having a “surgical home” is a good way to ensure this, whether the “surgical home” is staffed by surgeons, anesthesiologists, or preferably a multidisciplinary group. The other way, as described in our October 30, 2012 Patient Safety Tip of the Week “Surgical Scheduling Errors” is to cancel any elective cases for which a copy of both the informed consent and the H&P are not available at least a couple days in advance of the scheduled procedure.
It is extremely common in academic settings and even in community or rural hospitals for surgeons to obtain the informed consent in the preoperative area. Our June 5, 2007 Patient Safety Tip of the Week “Patient Safety in Ambulatory Surgery” noted that ambulatory surgery is particularly vulnerable to missing documents because those documents are usually in the surgeon’s office rather than at the hospital. That is why you need to be firm in your requirement for such documents before cases are scheduled. And, yes, we still continue to encounter some surgeons who look at the document as some sort of regulatory requirement foisted upon them! We hope you’ll also go back to our September 10, 2013 Patient Safety Tip of the Week “Informed Consent and Wrong-Site Surgery” to see problems associated with informed consent.
And don’t forget about imaging studies. We have recommended that copies of relevant imaging reports also be available. While copies of actual images are often present in the OR, many of the personnel who need to participate in site/side verification may not be familiar with interpretation of such images. Therefore, they should look at the imaging reports to verify site/side.
There was also deviation in this case from the facility policy that the circulating nurse lead the timeout. We also like the “Minnesota Timeout” concept in which someone other than the surgeon leads the time out process. That helps prevent team members from simply agreeing with the surgeon. The timeout is supposed to be an active rather than passive process and there should not be undue deference to the surgeon. Every member of the team needs to speak up and not be afraid to challenge any aspect.
There are several topics we hope the facility addressed in its own RCA that are not mentioned in the CDPH document. An obvious one is whether there were any time pressures that may have contributed to shortcuts taken. Did this surgeon/team have multiple cases scheduled that day, particularly cases that may have been similar and actually included removal of ovaries.
One issue we usually ask early in an RCA of an adverse event is whether indications for surgery were appropriate. For example, in this case we’d ask not only whether the hysterectomy was clinically indicated but also whether the incidental appendectomy was indicated. “Incidental” appendectomy has been a controversial topic for many years. There are some guidelines, albeit based on lower grades of evidence, regarding incidental appendectomy in gynecologic procedures (ACOG 2016, Tam 2013).
Though a site marking may not be indicated in patients having bilateral procedures, don’t forget that involving the patient in the sort of preop setting where a site marking would ordinarily be performed is also another opportunity to verify with the patient the procedure to be performed. So there should be a “site marking ceremony” even when no site marking is indicated.
The POC (plan of correction) from the facility reeducated all relevant staff regarding use of the signed consent form and H&P to correctly verify the patient, surgical procedure, and site/side/level. It acknowledged the fallibility of using scheduling documents for the purposes of verification of patient/procedure/site/laterality and indicated the schedule should not be used for verification. We’re not so sure we’d dismiss the schedule outright. Having a clinical person do the scheduling and review the required documentation (consent, H&P) provides one more opportunity to flag a possible mistake.
The facility’s POC also reiterated that their policy precludes the timeout being led by the surgeon.
The attending surgeon and the facility’s Risk and Patient Safety Manager met with affected patient promptly following surgical event and agreements were reached with the patient regarding appropriate clinical monitoring and provision of appropriate medication therapy.
As part of the POC a checklist/tracer tool was developed, implemented and revised to audit correct use of the checklist and verification process. We’ll also add our commentary here that simply auditing a checklist is not enough. We’ve seen too many times all items checked on such checklists even if they have not been done. Therefore, we recommend any audit for compliance include some form of random observation of the timeouts (by either direct observation or review of video monitoring of procedures). You will be surprised at how many hospitals have implemented the Safe Surgery Checklist (or equivalent) and assumed it was being used properly, only to find out compliance was poor when they have an adverse event.
Yes, errors during surgery booking/scheduling contributed to the incorrect procedure in this case but the real root causes were likely more related to issues related to culture and imperfect adoption of the surgical safety checklist. This is a good reminder that faulty adoption of a safe surgical checklist without ensuring an appropriate culture of safety is in place may lead to a false sense of security.
Some of our prior columns related to wrong-site surgery:
September 23, 2008 “Checklists and Wrong Site Surgery”
June 5, 2007 “Patient Safety in Ambulatory Surgery”
July 2007 “Pennsylvania PSA: Preventing Wrong-Site Surgery”
March 11, 2008 “Lessons from Ophthalmology”
July 1, 2008 “WHO’s New Surgical Safety Checklist”
January 20, 2009 “The WHO Surgical Safety Checklist Delivers the Outcomes”
September 14, 2010 “Wrong-Site Craniotomy: Lessons Learned”
November 25, 2008 “Wrong-Site Neurosurgery”
January 19, 2010 “Timeouts and Safe Surgery”
June 8, 2010 “Surgical Safety Checklist for Cataract Surgery”
December 6, 2010 “More Tips to Prevent Wrong-Site Surgery”
June 6, 2011 “Timeouts Outside the OR”
August 2011 “New Wrong-Site Surgery Resources”
December 2011 “Novel Technique to Prevent Wrong Level Spine Surgery”
October 30, 2012 “Surgical Scheduling Errors”
January 2013 “How Frequent are Surgical Never Events?”
January 1, 2013 “Don’t Throw Away Those View Boxes Yet”
August 27, 2013 “Lessons on Wrong-Site Surgery”
September 10, 2013 “Informed Consent and Wrong-Site Surgery”
July 2014 “Wrong-Sided Thoracenteses”
May 17, 2016 “Patient Safety Issues in Cataract Surgery”
July 19, 2016 “Infants and Wrong Site Surgery”
September 13, 2016 “Vanderbilt’s Electronic Procedural Timeout”
May 2017 “Another Success for the Safe Surgery Checklist”
See our prior columns on huddles, briefings, and debriefings:
References:
CDPH (California Department of Public Health). Complaint Intake Number: CA00477434
http://www.cdph.ca.gov/certlic/facilities/Documents/2567_SequoiaHospital_IJAP_SanMateo.pdf
WHO Surgical Safety Checklist
http://www.who.int/entity/patientsafety/safesurgery/tools_resources/SSSL_Checklist_finalJun08.pdf
ACOG (American College of Obstetricians and Gynecologists). Committee Opinion Number 323. Elective Coincidental Appendectomy. November 2005 (reaffirmed in 2016).
Tam T, Harkins G. Elective laparoscopic appendectomy in gynecologic surgery: When, why, and how. OBG Manag 2013;25(3): 42-49
Print “Anatomy of a Wrong Procedure”
May 9, 2017
Missed Nursing Care and Mortality Risk
Many studies have demonstrated a relationship between nursing staffing levels and patient mortality and complications. High levels of nursing staffing are associated with lower mortality and lower levels are associated with higher mortality. In our November 26, 2013 Patient Safety Tip of the Week “Missed Care: New Opportunities?” we also noted that adverse patient outcomes may be related to missed nursing care (also known as “care left undone”). Now a new study appears to connect the dots and demonstrate that much of the excess mortality related to poor nursing staffing is due to missed nursing care. Jane Ball, in a thesis using data from the RN4Cast study, found that a 10% increase in the amount of care left undone by nurses was associated with a 16% increase in the likelihood of a patient dying within 30 days of admission following common surgical procedures (Ball 2017).
Lack of time is the major reason that care gets left undone. This, in turn, may be due to nursing staffing shortages or to disparities between nursing staffing levels and complexity of care required or overall task load.
In several of our columns we’ve discussed findings that have come out of the RN4Cast study. That is a collaborative study of nurse staffing at hospitals in fifteen European countries, though the data in the Ball study came from a subset of hospitals in England, Sweden, and nine countries.
In the Ball study missed care (aka care left undone) was measured using the “Basel Extent of Rationing of Nursing Care (BERNCA)”instrument (Schubert 2009). Nurses were asked ‘On your most recent shift, which of the following activities were necessary but left undone because you lacked the time to complete them?’ Respondents were presented with a list of 13 nursing care activities and asked to tick all that applied. The list included activities such as timely medication administration, skin care, oral care, comforting patients, care documentation, pain management, changing a patient’s position, care planning, discharge preparation, patient surveillance, and patient/family education.
It was not uncommon for necessary nursing care to be left undone by RNs on a shift due to lack of time. Ball found that 86% of RNs surveyed in England and 74% in Sweden reported that they left some care undone on their last shift.
Interestingly, higher support worker staffing levels (eg. nursing aides) were not associated with better outcomes. But see our comments below about the potential relationship with non-nursing staffing.
Nurse-rated quality of care and patient safety environment scores were also significantly related to differences in care left undone. But the striking finding was that a 10% increase in the amount of care left undone by nurses was associated with a 16% increase in the 30 day mortality rate.
In our November 26, 2013 Patient Safety Tip of the Week “Missed Care: New Opportunities?” we noted that the concept of missed care as a potential contributor to adverse patient events can largely be attributed to Beatrice Kalisch, RN, PhD. In 2006 (Kalisch 2006) she first brought examples of commonly missed nursing care that have been associated with adverse patient outcomes. (Make no mistake: the root causes of missed nursing care extend well beyond nursing and those factors put nurses in the position of having to prioritize care, leaving some care undone or delayed). Prior to 2006 there was virtually nothing in the literature about missed nursing care and its occurrence was described as “undiscussable” (Kalisch 2009a). Kalisch did qualitative studies of hospital nursing staff using focus group interviews and developed a tool, the MISSCARE survey, to measure missed nursing care (Kalisch 2006, Kalisch 2009a). We are not talking here about occasionally missed or delayed nursing care but rather regularly missed nursing care. We refer you back to our November 26, 2013 Patient Safety Tip of the Week “Missed Care: New Opportunities?” for a discussion of the individual elements of missed care and the themes as to the reasons for missed care.
While understaffing is an obvious root cause for missed care or care left undone, don’t just look at nurse:patient staffing ratios. Even when nurse:patient ratios are “acceptable” there may be additional factors that prevent nurses from carrying out all regular aspects of nursing care. In our numerous columns on “the weekend effect” we’ve noted the many additional activities nurses get stuck doing on weekends because of inadequate non-nursing staffing. Sometimes the nurses end up doing tasks such as transporting patients or even mopping floors. There is also less dietary and nutrition support, pharmacy and imaging services, physical therapy, patient teaching, and social services. They may spend more time on the phone trying to track down doctors on weekends. So nurses end up doing many more tasks that they do not normally perform during regular “day” hours and they do not have as much time to do patient care and bedside nursing.
It’s interesting that the Ball study did not find use of higher support worker staffing levels to be of any mortality benefit. We have often recommended having dedicated “teams” of appropriately trained individuals for regular turning of patients at risk for decubiti, or for ambulating patients, or for feeding patients. These could perhaps free up nurses to perform some of the other activities that require more professional backgrounds. Such may be more practical at larger hospitals and may not be feasible at small hospitals.
But the Ball study did not find evidence that the availability of nursing support staff increased the ability of RNs to complete their work. They also note some prior studies have shown that higher support staff levels in certain contexts may even be associated with increased mortality rates. So the jury is still out on what, if any, specific activities currently done by nurses might be done by support personnel to free nurses up for those activities more closely tied to outcomes.
Kalisch has also pointed out that experience levels of staff may vary from shift to shift and that issues with orientation and handoffs may also be contributing factors. In a subsequent concept paper (Kalisch 2009b) Kalisch and colleagues developed a Missed Nursing Care Model which highlights teamwork and communication issues as one of three major antecedents to missed care.
Missed nursing care has also recently been implicated in some disparities of care in the US (Brooks-Carthon 2016). Looking at older patients admitted with acute myocardial infarction, the researchers found that unmet nursing care was associated with a higher risk of 30-day readmission for older black patients but not older white patients. Older black patients were 18% more likely to experience a readmission after adjusting for patient and hospital characteristics and more likely to be in hospitals where nursing care was often left undone. Factors identified as contributing to this phenomenon were when nurses were unable to talk/comfort patients, complete documentation, or administer medications in a timely manner.
Missed nursing care or care left undone is still a relatively new concept in the patient safety world. The striking finding by Ball that a 10% increase in the amount of care left undone by nurses was associated with a 16% increase in mortality is a wakeup call that tells us we must begin to address the issue. Using tools like the BERNCA instrument or the MISSCARE survey to identify what aspects of care are not being completed, trending them over time and, most importantly, identifying and ameliorating the root causes could result in significant improvement in patient outcomes.
References:
Ball JE. Nurse Staffing Levels, Care Left Undone, & Patient Mortality in Acute Hospitals. Karolinska Institutet; Stockholm 2017
RN4CAST Study.
Schubert M., Clarke SP, Glass TR, et al. Identifying thresholds for relationships between impacts of rationing of nursing care and nurse- and patient-reported outcomes in Swiss hospitals: a correlational study. International Journal of Nursing Studies 2009; 46(7): 884-893
http://www.sciencedirect.com/science/article/pii/S0020748908002800
Kalisch BJ. Missed Nursing Care: A Qualitative Study. Journal of Nursing Care Quality 2006; 21(4): 306–313
Kalisch BJ, Williams RA. Development and Psychometric Testing of a Tool to Measure Missed Nursing Care. J Nurs Admin 2009; 39(5): 211-219
Kalisch BJ, Landstrom GL, Hinshaw AS. Missed nursing care: a concept analysis. Journal of Advanced Nursing 2009; 65(7): 1509–1517
http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2648.2009.05027.x/pdf
Brooks-Carthon JM, Lasater KB, Rearden J, et al. Unmet Nursing Care Linked to Rehospitalizations Among Older Black AMI Patients: A Cross-Sectional Study of US Hospitals. Medical Care 2016; 54(5): 457-465, May 2016
Print “Missed Nursing Care and Mortality Risk”
May 16, 2017
Are Surgeons Finally Ready to Screen for Frailty?
In a recent interview we did with HCPro’s Patient Safety Monitor (Ward 2017) we emphasized our long-advocated view that screening for frailty is one of the three most important activities that should be done pre-operatively (the other two being screening for obstructive sleep apnea and delirium risk). We discussed this in our August 17, 2010 Patient Safety Tip of the Week “Preoperative Consultation – Time to Change” and the numerous columns on frailty listed at the end of today’s column.
Frailty is associated with both increased complications following surgery and a number of other undesirable outcomes, such as discharge to destinations other than home. Add to the many prior studies a new one that demonstrates frailty is associated with unplanned readmission among military veterans following surgery (Wahl 2017). Wahl and colleagues used the Modified Frailty Index (mFI) to assess adult patients who underwent surgery and were discharged alive from Veterans Affairs hospitals for orthopedic, general, and vascular conditions. Unadjusted rates of overall 30-day readmission, postdischarge emergency department visit, any predischarge or postdischarge complication, and postdischarge mortality varied by frailty in a dose-dependent fashion. Thus, the mFI was associated with poor surgical outcomes, including readmission, primarily due to impaired functional status. The authors suggest that targeting potentially modifiable aspects of frailty preoperatively, such as improving functional status, may improve perioperative outcomes and decrease readmissions.
That latter suggestion has always been somewhat problematic in that there has been scant evidence that programs aimed at modifying frailty (eg. physical therapy or “pre-hab”) before planned surgery impact outcomes.
But now evidence is beginning to accumulate demonstrating improved outcomes related to identification of frailty before surgery. Hall and colleagues compared mortality and complications before and after implementation of a program of screening for frailty prior to surgery (Hall 2017). The Risk Analysis Index (RAI) was used as the tool to screen for frailty (Hall 2016). If a patient had an RAI score ≥21 clinicians from surgery, anesthesia, critical care, and palliative care were notified of the patient’s frailty and associated surgical risks; if indicated, perioperative plans were modified based on team input. Overall 30-day mortality decreased from 1.6% to 0.7% after frailty screening implementation. Improvement was greatest among frail patients (12.2% to 3.8%) although it also improved in “robust” patients (1.2% to 0.3%). Improvement was even greater among frail patients at 180 (23.9% to 7.7%) and 365 days (34.5% to 11.7%). This was a single site study and, though prospective, was an observational study rather than a randomized controlled trial. The observed improvement cannot be causally linked to the frailty screening initiative with certainty. Further study in multiple centers, perhaps with a randomized trial design, would be helpful.
Another RCT (randomized controlled trial) of standard preoperative assessment or preoperative comprehensive geriatric assessment and optimization was also recently published (Partridge 2017). The patients aged at least 65 years were scheduled for elective aortic aneurysm repair or lower-limb arterial surgery. We’re not told how many had frailty but certainly some percentage of frail patients would be expected in this population. Randomization was stratified by sex and surgical site (aorta/lower limb). Geometric mean length of stay was 5.53 days in the control group and 3.32 days in the intervention group. There was a lower incidence of delirium (11% vs. 24%), cardiac complications (8% vs. 27%), and bladder/bowel complications (33% vs. 55%) in the intervention group compared with the control group. Patients in the intervention group were also less likely to require discharge to a higher level of dependency (4.7% vs. 13.2%).
Another interesting finding comes from a retrospective, population-based cohort study using linked administrative data in Ontario, Canada (McIsaac 2017). They identified all adult major, elective noncardiac surgery patients who were frail according to the validated Johns Hopkins Adjusted Clinical Groups (ACG®) frailty-defining diagnoses indicator. They then compared mortality rates across hospitals, stratified by the volume of frail patients seen at those hospitals. The thirty-day mortality rate in the lowest volume quintile was 1.1% compared to 0.9% in the highest and, after adjustment for a variety of demographic and clinical variables, found a significant association between frailty volume and improved survival (hazard ratio 0.51 for highest volume vs. lowest volume hospitals).
McIsaac et al. and Wang and Wunsch in an accompanying editorial (Wang 2017) surmise that hospitals caring for higher volumes of frail patients likely have teams and protocols that recognize the increased vulnerabilities of frail patients and address them both before and after scheduled surgery. And, just as we recommend certain surgeries be preferably performed in high volume hospitals, perhaps frail patients should be preferably treated in hospitals with high volumes of frail patients.
There are, of course, a variety of tools used for screening for frailty, varying from simple to complex. Many are described in our May 31, 2016 Patient Safety Tip of the Week “More Frailty Measures That Predict Surgical Outcomes” and the other columns listed at the end of today’s column. We’ve noted some of the simpler ones have looked at gait speed, the timed up-and-go test, handgrip strength, and others. A recent study looked at the trajectory of handgrip strength in a community-dwelling population of 70-90 year olds (Stessman 2017). As expected, handgrip strength decreased with age. They found associations between low handgrip strength and poor functional measures, lower educational and financial status, smoking, and diabetes mellitus. After adjustment for a number of variables, low handgrip strength predicted subsequent activity of daily living dependence from age 78 to 85 (OR = 2.68) and 85 to 90 (OR = 2.31) and low handgrip strength was associated with significantly higher mortality. Subjects in the lowest age-specific quartile of handgrip strength had a higher risk of subsequent functional decline and mortality.
Another Canadian study looked at individual components of the Fried frailty phenotype measures (gait speed, hand-grip strength as measured by a dynamometer, and self-reported exhaustion, low physical activity, and unintended weight loss) in a primary care setting (Lee 2017). The researchers found that individual criteria all showed sensitivity and specificity of more than 80%, with the exception of weight loss. The positive predictive value of the single-item criteria in predicting the Fried frailty phenotype ranged from 12.5% to 52.5%. When gait speed and hand-grip strength were combined as a dual measure, the positive predictive value increased to 87.5%. They conclude that, while use of gait speed or grip strength alone was found to be sensitive and specific as a proxy for the Fried frailty phenotype, use of both measures together was found to be accurate, precise, specific, and more sensitive than other possible combinations and that assessing both measures is feasible within the primary care setting.
The bottom line: screening for frailty is not time consuming and can be easily performed in an office or clinic setting prior to anticipated surgery using either one of the formal frailty scores or one of the simple tests noted above. Patients identified as frail by these methods not only need closer surveillance for complications post-operatively but may benefit from a multidisciplinary comprehensive geriatric management program prior to surgery.
Some of our prior columns on preoperative assessment and frailty:
References:
Ward B (ed.). Study: Postoperative complications are a major concern for hospitals. HCPro Patient Safety Monitor, March 1, 2017
Wahl TS, Graham LA, Hawn MT, et al. Association of the Modified Frailty Index With 30-Day Surgical Readmission. JAMA Surg 2017; Published online May 3, 2017
http://jamanetwork.com/journals/jamasurgery/article-abstract/2623724
Hall DE, Arya S, Schmid KK, et al. Association of a Frailty Screening Initiative With Postoperative Survival at 30, 180, and 365 Days. JAMA Surg 2017; 152(3): 233-240
http://jamanetwork.com/journals/jamasurgery/article-abstract/2587479
Hall DE, Arya S, Schmid KK, et al. Development and Initial Validation of the Risk Analysis Index for Measuring Frailty in Surgical Populations. JAMA Surg 2016; 152(2): 175-182
http://jamanetwork.com/journals/jamasurgery/article-abstract/2586340
Partridge JSL, Harari D, Martin FC, et al. Randomized clinical trial of comprehensive geriatric assessment and optimization in vascular surgery. British Journal of Surgery 2017; 104(6): 679-687
http://onlinelibrary.wiley.com/doi/10.1002/bjs.10459/full
McIsaac DI, Wijeysundera DN, Huang A, et al. Association of the Hospital Volume of Frail Surgical Patients Cared for with Outcomes after Elective, Major Noncardiac Surgery: A Retrospective Population-based Cohort Study. Anesthesiology 2017; 126: 575-576
http://anesthesiology.pubs.asahq.org/article.aspx?articleid=2602421
Wang HT, Wunsch H. Hospital Experience Caring for the Frail: A New Concern for Surgical Patients (editorial). Anesthesiology 2017; 126: 575-576
http://anesthesiology.pubs.asahq.org/article.aspx?articleid=2602424
Stessman J, Rottenberg Y, Fischer M, Hammerman-Rozenberg A, Jacobs JM. Handgrip Strength in Old and Very Old Adults: Mood, Cognition, Function, and Mortality. J Am Geriatr Soc 2017; 65:(3): 526-532
http://onlinelibrary.wiley.com/doi/10.1111/jgs.14509/full
Lee L, Patel T, Costa A, Bryce E, Hillier LM, Slonim K, et al. Screening for frailty in primary care. Accuracy of gait speed and hand-grip strength. Can Fam Physician 2017; 63: e51-7
http://www.cfp.ca/content/63/1/e51
Print “Are Surgeons Finally Ready to Screen for Frailty?”
May 23, 2017
Trolling the RCA
Hard to believe it’s been almost a year since we had to defend the root cause analysis (see our July 12, 2016 Patient Safety Tip of the Week “Forget Brexit – Brits Bash the RCA!”). But now we’re at it again!!
The May 2017 issue of BMJ Quality and Safety has several articles questioning whether RCA’s are good or bad for patient safety (Kellogg 2017) (Trbovich 2017) (Peerally 2017). One of the articles (Peerally 2017) is actually the article published online ahead of print that triggered our 2016 column mentioned above but these articles provide the opportunity to again discuss what’s good and what’s bad about RCA’s in healthcare.
The RCA is not dead. We’ve always had a strong conviction that the RCA (root cause analysis) is probably the most important learning tool that an organization with a good culture of safety has at its disposal. We encourage organizations to do RCA’s not just on events with bad patient outcomes but on any event that had the potential to induce harm (near-misses). The key problems with RCA’s are:
That last problem is a major failure in our healthcare system. How many times do we need to see a problem uncovered in one hospital occur in other hospitals simply because our system does not promote sharing of problems and solutions?
The new study was a review of RCA’s from a major academic medical center (Kellogg 2017). Researchers reviewed 302 RCA’s over an 8-year period, 106 of which proposed solutions. The average number of solutions proposed per RCA was 4.7. But a large proportion of solutions offered were what we would deem “weak” interventions (eg. training 20%, process change 19.6%, policy reinforcement 15.2%). Moreover, multiple event types were repeated during the study period despite repeated RCA’s.
Kellogg and colleagues, of course, are correct that many or most RCA’s fail to implement strong interventions as solutions. That echoes a criticism in the second article (Peerally 2017), one with which we strongly agree, that solutions and action plans are poorly designed or implemented. In our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s” we noted prior studies in the VA system (Hughes 2006) which analyzed action items from RCA’s and found that 30% were not implemented and another 25% were only partially implemented. Stronger action items were more likely to be implemented. Actions that were assigned to specific departments or people were more likely to be implemented than those assigned to general areas. And they found that the patient safety manager plays a critical role in RCA action implementation.
In our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s” we emphasized the importance of tracking whether recommended action steps were implemented following an RCA, whether they were effective, and whether there were any unintended consequences. All too often action steps never get implemented at all or consist solely of “weak” action steps and organizations are then surprised when a similar adverse event occurs in the future. Moreover, even the most well intentioned and well planned action steps sometimes lead to consequences that were never anticipated. We typically see weak actions like education and training or policy changes as the sole actions undertaken rather than strong actions like constraints and forcing functions. We discussed strength of actions in our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s”. In that column we included an analogy to the effectiveness of signs and tools used to try to get drivers to slow down in construction zones on highways. We put them together in pictures with RCA action items and now incorporate them in our webinar presentations on doing good RCA’s. Click here to see them. Remember: images are more likely to be remembered than words!
One of the biggest issues we see in hospitals related to RCA’s is failure to follow up and close the feedback loop. In fact, probably the majority of hospitals lack formal procedures for ensuring the corrective actions recommended in an RCA are actually carried out (or barriers to their implementation identified and alternative steps taken). In our March 30, 2010 Patient Safety Tip of the Week “Publicly Released RCA’s: Everyone Learns from Them” we discussed an incident at a hospital in which a similar incident had occurred several years prior. After the first incident an extensive root cause analysis was done and multiple recommendations were made, including key recommendations that should have prevented the second incident. But all those recommendations had never been fully implemented. Importantly, the recommendations were communicated back to those individuals deemed to be in the “need to know” but not widely disseminated to middle for front line management nor to front line staff.
We recommend you keep a list or table of such identified action items from all your RCA’s to discuss at your monthly patient safety committee or performance improvement committee meetings. Action items should remain on that list until they have been implemented or completed. Only that sort of rigorous discipline will ensure that you did what you said you were going to do, i.e. that you “closed the loop”. And don’t forget you need to monitor your implemented actions for unanticipated and unintended consequences. For example, you might take the strong action of removing a drug from a particular setting, only to realize later that there were circumstances where that drug was needed in that setting.
Take a look at the cases discussed in some of our recent columns. Though these were not technically the RCA’s performed by the hospitals, their plans of correction (POC’s) did include the actions taken. And those actions were relatively weak. In our April 25, 2017 Patient Safety Tip of the Week “Dialysis and Alarm Fatigue” the actions taken by the hospital were primarily education and policy changes, both of which are “weak” interventions. They missed the opportunity to implement stronger interventions. One would have been to redesign the alarm system to focus the responder’s eyes to the site indicated by the alarm. A second would have been to add a hard stop to the alarm. The best interventions are forcing functions. We’d suggest that these alarm systems program in a “hard” stop for this particular alarm that requires the responder to verify that he/she has inspected the access site. That verification should then become part of the medical record.
As above, constraints and forcing functions are the strongest of actions. In our May 20, 2014 Patient Safety Tip of the Week “Ophthalmology: Blue Dye Mixup” and September 2014 What's New in the Patient Safety World column “Another Blue Dye Eye Mixup” we discussed unfortunate cases where methylene blue was used during cataract surgery rather than trypan blue. It is very clear there is a huge system issue here. The system actually put those healthcare workers and the patient in a vulnerable position that allowed the mistake to happen. It is very much akin to the concentrated potassium chloride issue of the past in which nurses accidentally administered fatal doses of concentrated KCl to patients. There was little reason for nurses to have access to vials of concentrated KCl yet we placed them on nursing units and it was simply a matter of time until someone unwittingly drew up a syringeful and administered a fatal dose. Our eventual system fix was to remove vials of concentrated KCl from floor stock on nursing units. If you are a facility that only does eye cases, you probably have no need for methylene blue and therefore should not stock it at all. In other facilities where you may have a legitimate need for methylene blue (for example, it is used to help identify leaks in some surgeries or to help identify tissue in need of debridement in others) you clearly need to store the two blue dyes separately. If you have a dedicated “eye” room and can store all the medications and materials for eye surgery there (or in an automated dispensing cabinet dedicated to ophthalmology) make sure that methylene blue is not in those areas.
But sometimes it is difficult to implement a strong intervention following an RCA. In our May 2, 2017 Patient Safety Tip of the Week “Anatomy of a Wrong Procedure” the major problem was a poor culture of safety. In their recent editorial Trbovich and Shojania (Trbovich 2017) put at the top of their hierarchy of effectiveness “culture change” as one of the “strong” actions. Of course, it is undoubtedly the strongest of all actions. You’ve often heard us use the phrase “culture trumps ________” (fill in the blank with words like policy, procedure, strategy, tactics, vision, etc). In fact, “Culture trumps…Everything!”. But, unfortunately, changing the culture is a long-term process, tough to implement and difficult to measure. So while it’s something that desperately needs work, very few RCA’s include it as an action. One possible stronger action to take in that case, and we would consider it an action of only intermediate strength, would have been to cancel any elective case in which copies of the consent and H&P are not available several days prior to the scheduled procedure and then ensure those documents are available in the OR at the time the procedure is actually done. But even those are weak actions because there is no guarantee that people will use them. The case also illustrates that use of a stronger action (implementation of the WHO Surgical Safety Checklist) failed because of poor implementation and verification of its use.
But that gets us back to our most significant point. Hospitals often fail to include strong actions because they don’t know what those strong actions are. Our inability to disseminate lessons learned at other facilities and their solutions is, in our minds, the single biggest barrier to improving patient safety. Peerally and colleagues also lament a lack of dissemination of lessons learned and lack of aggregation of similar events. Ironically, the example Peerally and colleagues used to illustrate lessons not learned was implantation of incorrect intraocular lenses. In several of our columns (most recently in our May 17, 2016 Patient Safety Tip of the Week “Patient Safety Issues in Cataract Surgery”) we’ve described that very issue as the one that led us over 20 years ago to develop one of the earliest surgical timeout protocols that subsequently served as a model for subsequent state and national timeout protocols! In those columns we describe how cases of incorrect intraocular lens (IOL) implantation occurred singly (or occasionally multiply) in many hospitals yet those cases and their contributing factors were never shared widely. The same concept, of course, was seen with cases of fatal overdoses from inadvertent injection of concentrated potassium chloride. Those typically occurred as single isolated events in many hospitals and it was only years later that the widespread occurrence of this unfortunate incident was appreciated and steps taken to remove concentrated potassium chloride from floor stocks.
Even in organizations capable of wider dissemination of lessons learned there is a tendency to wait until several cases have been aggregated before sharing those lessons. But some isolated solitary cases also need to be shared because the circumstances leading to those cases is very likely replicated at multiple other venues. Such an example is that ophthalmological incident with the inadvertent use of methylene blue dye instead of trypan blue (see our prior columns of May 20, 2014 “Ophthalmology: Blue Dye Mixup” and September 2014 “Another Blue Dye Eye Mixup”). When we discussed the first case, we said “we can’t believe this is the first time this has happened”. Then, shortly thereafter a second case was reported. In fact, the second case anteceded the first. Perhaps with better dissemination of lessons learned the subsequent case might have been avoided.
In reality, such failure to share is a societal problem. The various legal and public relations consequences of sharing lessons and aggregating similar cases have been among the biggest barriers to implementation of sound patient safety practices. PSO’s (Patient Safety Organizations) have the potential to help disseminate lessons learned and solutions but to date have had a limited impact since their work is shared only with their individual member organizations.
See our July 12, 2016 Patient Safety Tip of the Week “Forget Brexit – Brits Bash the RCA!” for discussion of several of the other points raised by Peerally et al. such as the problem of “many hands”, political hijacking of the RCA process, challenges of getting unbiased information, timeline issues, conduct of and participation in an RCA, having the right expertise available, figuring how to fit patients and families into the RCA process, and more.
In their recent editorial Trbovich and Shojania (Trbovich 2017) warn that jumping to corrective actions on the basis of a single case can be problematic. That’s one reason we also recommend periodically reviewing all your RCA’s to cull out recurrent themes. It’s often such collective reviews that make you realize the increased importance of root causes that were identified in individual RCA’s.
One of the most important things in making your organization’s use of RCA’s is to have a culture that understands the focus is to uncover system issues that can be fixed to prevent subsequent similar events. That means people should not fear reporting events or speaking openly in RCA interviews. James Bagian does a nice job in an AHRQ interview on RCA’s (AHRQ 2016) explaining that you have to educate people that, when they report in the safety system, they will be held harmless but that doesn't mean people get a free pass. If there was a “blameworthy act”, the case would be placed on an administrative route where the facts of the event have to be "rediscovered" by the administrative system, which could culminate in punitive action. But if it was not blameworthy, under no circumstances would there ever be punitive action.
In our July 12, 2016 Patient Safety Tip of the Week “Forget Brexit – Brits Bash the RCA!” we noted that perhaps the toughest nut to crack is the complicated issue of blame. The beauty of the RCA is that it stresses identification and remediation of system defects that are generally more amenable than human behaviors. And it is clear that system defects may put individuals at risk of committing human errors that then result in adverse patient outcomes. One key tenet of the National Patient Safety Foundation’s RCA2 Guidelines (NPSF 2015) is that it only addresses system issues and should not address or focus on individual performance (see our July 14, 2015 Patient Safety Tip of the Week “NPSF’s RCA2 Guidelines”). In fact, NPSF recommends that all organizations should define “blameworthy” events and actions that fall outside the purview of the safety system and define how and under what circumstances they will be handled separately. Of course, we would emphasize that system issues that lead to or facilitate improper individual performance must be addressed under the RCA2 process. For example, workarounds are (often) improper individual actions that almost always have a system issue that led to their use. Another example is “normalization of deviance” where the culture of the system led to acceptance of a certain deviation from proper practice as being “normal” and allowed that deviation to be performed by many individuals.
Peerally and colleagues acknowledge that a “no-blame” approach is not always possible or appropriate and may impede thorough incident investigation and we often see failure to place blame when placing blame may be appropriate. They note that most of us have adopted the “Just Culture” approach but that tools such as algorithms and decision tools (eg. the “culpability tree”) have flaws of their own. Also, don’t forget that in parallel to your RCA process you need to ensure your organization has a means to address the “second victim” or healthcare workers involved in such incidents (see also our December 17, 2013 Patient Safety Tips of the Week “The Second Victim” and August 9, 2016 “More on the Second Victim”.)
Alternatives to the RCA are available. Gupta and Lyndon (Gupta 2017) recently reviewed many of the issues raised in the Peerally article and in the RCA2 documents (see our July 14, 2015 Patient Safety Tip of the Week “NPSF’s RCA2 Guidelines”) and note several other review tools and techniques that can be applied effectively in certain situations, often with less time and effort. The FMEA (Failure Mode and Effects Analysis) is also a great tool to identify system issues that may lead to patient safety events. But a FMEA is time consuming and most organizations can do only one or two in a year. They also deal with a lot of theoretical issues or “what ifs” that staff may consider to be unlikely to occur. On the other hand, the RCA is typically done after an actual event or near miss that grabs everyone’s attention and hammers home that we need to make changes to avoid another event. The best case, however, is seeing the RCA from someone else’s facility and saying “Wow! That could happen here!” and implementing changes at your facility before you have an untoward event.
See our Patient Safety Tips of the Week for July 14, 2015 “NPSF’s RCA2 Guidelines” and July 12, 2016 “Forget Brexit – Brits Bash the RCA!” for many other recommendations to include in your RCA process.
So our message to all the detractors of RCA’s…stop trolling the RCA! The RCA is a great learning tool. The problem is doing it right and what you do with it. Instead, help fix the barriers that prevent the sharing of lessons learned and the solutions that utilize strong actions to help prevent patient safety incidents in all healthcare settings.
Some of our prior columns on RCA’s, FMEA’s, response to serious incidents, etc:
July 24, 2007 “Serious Incident Response Checklist”
March 30, 2010 “Publicly Released RCA’s: Everyone Learns from Them”
April 2010 “RCA: Epidural Solution Infused Intravenously”
March 27, 2012 “Action Plan Strength in RCA’s”
March 2014 “FMEA to Avoid Breastmilk Mixups”
July 14, 2015 “NPSF’s RCA2 Guidelines”
July 12, 2016 “Forget Brexit – Brits Bash the RCA!”
References:
Kellogg KM, Hettinger Z, Shah M, et al. Our current approach to root cause analysis: is it contributing to our failure to improve patient safety? BMJ Qual Saf 2017; 26(5): 381-387
http://qualitysafety.bmj.com/content/26/5/381
Trbovich P, Shojania KG. Root-cause analysis: swatting at mosquitoes versus draining the swamp. BMJ Qual Saf 2017; 26(5): 350-353
http://qualitysafety.bmj.com/content/26/5/350
Peerally MF, Carr S, Waring J, Dixon-Woods M. The problem with root cause analysis. BMJ Qual Saf 2017; 26(5): 417-422
http://qualitysafety.bmj.com/content/26/5/417
Hughes D. Root Cause Analysis: Bridging the Gap Between Ideas and Execution. VA NCPS Topics in Patient Safety TIPS 2006; 6(5): 1,4 Nov/Dec 2006
http://www.patientsafety.va.gov/docs/TIPS/TIPS_NovDec06.pdf#page=1
Weak vs. Strong Responses to an RCA (Power Point presentation).
http://patientsafetysolutions.com/docs/RCA_strong_vs_weak_responses.ppt
AHRQ. Perspectives on Safety. In Conversation With... James P. Bagian, MD, PE. Root Cause Analysis: What Have We Learned? AHRQ PSNet 2016; Published December 2016
https://psnet.ahrq.gov/perspectives/perspective/211/in-conversation-with--james-p-bagian-md-pe
NPSF (National Patient Safety Foundation) RCA2. Improving Root Cause Analyses and Actions to Prevent Harm. NPSF 2015
http://c.ymcdn.com/sites/www.npsf.org/resource/resmgr/PDF/RCA2_first-online-pub_061615.pdf
Gupta K, Lyndon A. Perspectives on Safety. Annual Perspective 2016. Rethinking Root Cause Analysis. AHRQ PSNet 2017; January 2017
https://psnet.ahrq.gov/perspectives/perspective/216/rethinking-root-cause-analysis
Print “Trolling the RCA”
May 30, 2017
Errors in Pre-Populated Medication Lists
When we did our first electronic medical record implementation over 10 years ago we were excited. We thought we’d have the medication reconciliation problem solved. We had vendors lined up who had medication fill data from the insurers and third party pharmaceutical middlemen to add to data from our own hospital and clinic records. We thought all that data would give us not only accurate current medication lists but also tell us a lot about patient compliance/adherence with their medications. Boy, were we wrong!
We found there were significant gaps in the data. Drugs provided as free samples in physicians’ offices were not on the list and any drug a patient might have paid for out-of-pocket was not on the list. Drugs provided as part of clinical trials were not included. And OTC drugs, including important ones like aspirin, were not on the lists. Moreover, we often saw drugs on the lists that our patients were not taking. Drugs that had been discontinued since last dispensed remained on the list (see our March 2017 What's New in the Patient Safety World column “Yes! Another Voice for Medication e-Discontinuation!” regarding the need for an electronic way to discontinue medications). In some cases there were simply errors in the data. We even occasionally saw instances where a patient was getting a drug that was actually intended for their (uninsured) family member or friend.
We also saw cases where doses of medications were erroneous. For example, some patients were utilizing pill-splitting so they were really only taking half the dose that appeared on the pre-populated forms. In other cases, the dose had been adjusted since the last dispensed amount appearing on the forms.
So there were several dangers in using such pre-populated lists. An important medication might be omitted when a patient was admitted to a hospital. Or a medication that had been discontinued or never intended to be used might be inadvertently started on admission to a hospital. In other cases, incorrect doses were given.
Now, a formal study on the errors on such pre-populated medication lists expands upon our experience. Canadian researchers (Stockton 2017) analyzed data on hospitalized patients where the “best possible medication history” lists were pre-populated with data from a Canadian medication dispensing system and records of drug dispensing from other outpatient dispensing facilities. They found that 47% of the 151 patients in their study were exposed to medication errors on admission. Of 112 medication errors identified, 85 (75.9%) were categorized as unexplained medication discrepancies. The majority of these were inappropriate discontinuations (38%) and omissions (28%). But they also found 24% were errors of “commission”, including 10 cases of continuation of medications that were contraindicated and 17 cases where previously discontinued medications were reordered. 15% of the medication discrepancies were deemed to have the potential to cause moderate harm and 1 case had the potential to cause serious harm. Errors of commission were especially likely to potentially cause harm. Analyzing multiple variables they found taking 8 or more medications and presence of cognitive impairment were factors associated with unexplained medication discrepancies.
The authors felt that there may be an overreliance on dispensing data by prescribers, leading to less rigorous attempts to take careful medication histories and otherwise verify the lists. They also note that it is quite easy to click boxes on the prepopulated forms.
Obviously, we are not surprised at the findings in the Stockton study. It clearly points out the pre-populating medication lists is fraught with dangers. That doesn’t mean pre-population should never be used. But it emphasizes the need for careful verification, using multiple sources, of any and all medications during medication reconciliation.
Also, that one unexpected issue we encountered (medications intended for someone other than the patient) puts the physician in a very uncomfortable position of wanting to help his/her patient but having uncovered an instance of healthcare fraud.
References:
Stockton KR, Wickham ME, Lai S, et al. Incidence of clinically relevant medication errors in the era of electronically prepopulated medication reconciliation forms: a retrospective chart review. CMAJ Open 2017; 5(2): E345-E353 May 5, 2017
http://cmajopen.ca/content/5/2/E345.full
Print “Errors in Pre-Populated Medication Lists”
June 6, 2017
NYS Mandate for Sepsis Protocol Works
In 2013 the New York State Department of Health mandated that hospitals begin using protocols to help with early identification and treatment of sepsis. Hospitals began implementing these protocols in 2014 and outcomes related to this project were reported in 2 recent publications (NYSDOH 2017, Seymour 2017). By the third quarter of 2016, 84.7% of adult patients and 85.3% of pediatric patients with severe sepsis or septic shock were treated using protocols (NYSDOH 2017). Adult in-hospital mortality fell from 30.2% in early 2014 to 25.4% by late 2016. Pediatric mortality rates were more variable without clearcut trend. After adjusting for patient factors, the NYSDOH analysis of the data showed that the odds of dying were 21% less for adult patients who received protocol-driven treatments compared to patients who do not receive protocol-driven treatments. The time frame for management was also critical. After adjustment, the NYSDOH analysis showed that the odds of dying were 27% less for adult patients who received all of the recommended treatments within three hours compared to patients who did not receive all of the recommended treatments.
Seymour et al. dove deeper into the data in an article published in the New England Journal of Medicine (Seymour 2017). Protocols were required to include a 3-hour bundle consisting of receipt of the following care within 3 hours:
Protocols were also required to include a 6-hour bundle, consisting of:
Individual facilities could further customize the protocols as they wished.
Of over 49,000 eligible patients in the emergency department at 149 hospitals, 82.5% had the 3-hour bundle completed within 3 hours, with a median time to the completion of the
3-hour bundle of 1.30 hours. Median time to the administration of broad-spectrum antibiotics was 0.95 hours and median time to the completion of the initial bolus of intravenous fluids was 2.56 hours.
Supporting the importance of early treatment, they found that each hour of time to the completion of the 3-hour bundle was associated with higher mortality (odds ratio of death until completion of 3-hour bundle, 1.04 per hour). Patients who had the bundle completed during hours 3 through 12 had 14% higher odds of dying in the hospital than those whose bundle was completed by 3 hours. Those same odds (1.04 per hour) were seen for time to administration of antibiotics and in-hospital mortality and patients who received first dose of antibiotics during hours 3 through 12 had 14% higher odds of dying in the hospital than those receiving antibiotics by 3 hours. However, the researchers found no association between the timing of the fluid bolus and mortality.
But the authors caution against concluding that early administration of a fluid bolus is not important. Early fluid resuscitation is likely important. In fact, another recent study on pediatric sepsis patients noted that the nature of fluid resuscitation in sepsis may be important. The Surviving Sepsis Campaign guidelines updated in 2016 (Rhodes 2017) did not consider the evidence strong enough to recommend balanced fluids over unbalanced fluids (“balanced” fluids are crystalloids such as lactated ringers, while chloride-rich fluids such as normal saline are not balanced). But Emrath and colleagues (Emrath 2017) recently compared outcomes in children with pediatric severe sepsis receiving balanced fluids for resuscitation in the first 24 and 72 hours of treatment to those receiving unbalanced fluids. After propensity matching, they found the 72-hour balanced fluids group had lower mortality (12.5% vs 15.9%), lower prevalence of acute kidney injury (16.0% vs 19.2%), and fewer vasoactive infusion days (3.0 vs 3.3 days) compared with the unbalanced fluids group.
Overall, the New York State experience re-emphasizes the importance of early recognition and treatment of sepsis in reducing mortality. We have one additional comment on the New York State studies. They used the Sepsis-2 consensus criteria for diagnosis. Those include use of the SIRS criteria in making a diagnosis of sepsis. We’ve done numerous columns on the pitfalls of the SIRS criteria and have praised the newer criteria for sepsis that do not use the SIRS criteria (see our What's New in the Patient Safety World columns for March 2016 “Finally…A More Rationale Definition for Sepsis” and February 2017 “Yes, the New Sepsis Criteria Fit the Bill”). We doubt, however, that the key conclusions reached in the NYS studies would be altered if the newer Sepsis-3 criteria (Singer 2016) had been used.
In a “Perspective” accompanying the Seymour article, Hershey and Kahn (Hershey 2017) note that, even though 82.5% of hospitals in the New York State experience did complete the 3-hour bundle within 3 hours, there was still considerable variation across hospitals. There are several potential reasons.
One possible reason may relate to how busy the emergency departments are. In a study just presented in abstract form (Peltan 2017), researchers found that patients received antibiotics within three hours in 83 percent of cases in uncrowded ERs, but only 72 percent of the time when the ER was crowded (exceeded the ERs’ licensed beds).
The New York State studies reported only the relationship of mortality to the timing of the first dose of antibiotics. But subsequent administration of antibiotics may also be important. Another recent study (Leisman 2017) found that major second antibiotic dose delays were common, especially for patients given shorter half-life pharmacotherapies and who boarded in the emergency department. They also observed an association between major second dose delay and increased mortality, length of stay, and mechanical ventilation requirement. In fact, in their multivariable analysis, major delay was associated with a 61% increased odds of hospital mortality. Interestingly, they found that major delays in second doses were paradoxically more frequent for patients receiving compliant initial care. So the moral of this study is that we can’t pat ourselves on the back when we meet the first 3 hours goals.
Those second dose delays should not be surprising. In our January 29, 2013 Patient Safety Tip of the Week “A Flurry of Activity on Handoffs” we noted that back in the 1990’s we realized that patients with community-acquired pneumonia at some renowned organizations were not getting their first dose of antibiotics for up to 18 hours. That, of course, was related to bottlenecks in moving patients from the ED to the floor and problems with handoffs (relating to both information transfer and transfer of responsibility for managing the patients). Fortunately, we had performance improvement projects that focused on ensuring timely administration of the antibiotics regardless of physical location of the patient. But we likely still see ambiguities of coordination and responsibility that occur in between-unit transfers that need to be resolved in the handoff. The finding in the Leisman study that delays in a second antibiotic dose were paradoxically more frequent in those patients who had received a first antibiotic dose in a timely fashion suggests problems in that coordination of care.
Some of the nuances of CPOE and standardized medication administration times may also contribute to such delays in antibiotic administration once the patient reaches the inpatient unit. For example, some facilities consider an order for “drug X q6hours” will be given at the facility’s standard times of 6AM, 12PM, 6PM, and 12AM. If the patient arrives on that inpatient unit at 12:01 PM, he/she may not get that next dose until 6PM even when the intent of the ordering physician was for earlier administration of the next dose.
Given the magnitude of the increase in in-hospital mortality (61%) seen with major delays in second dose of antibiotics in the Leisman study we would wholeheartedly endorse adding the timing of the second antibiotic dose as a quality parameter to be measured in patients with sepsis.
Our other columns on sepsis:
References:
NYSDOH (New York State Department of Health). New York State report on sepsis care improvement initiative: hospital quality performance. March 2017 https://www.health.ny.gov/press/reports/docs/2015_sepsis_care_improvement_initiative.pdf
Seymour CW, Gesten F, Prescott HC, et al. Time to Treatment and Mortality during Mandated Emergency Care for Sepsis. NEJM 2017; Online First May 23, 2017
http://www.nejm.org/doi/full/10.1056/NEJMoa1703058?query=featured_home
Rhodes A, Evans LE, Alhazzani W, et al. Surviving Sepsis Campaign: International Guidelines for Management of Sepsis and Septic Shock: 2016.
Critical Care Medicine 2017; 45(3): 486-552 March 2017
Emrath ET, Fortenberry JD, Travers C, et al. Resuscitation With Balanced Fluids Is Associated With Improved Survival in Pediatric Severe Sepsis. Critical Care Medicine 2017; Published Ahead of Print Post Author Corrections: April 21, 2017
Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016; 315(8): 801-810
http://jama.jamanetwork.com/article.aspx?articleid=2492881
Hershey TB, Kahn JM. State Sepsis Mandates - A New Era for Regulation of Hospital Quality. (Perspective). NEJM 2017; Online First May 23, 2017
http://www.nejm.org/doi/full/10.1056/NEJMp1611928
Peltan ID, Bledsoe JR, Oniki TA, et al. Increasing ED Workload Is Associated with Delayed Antibiotic Initiation for Sepsis. Abstract 5505. 2017 American Thoracic Society International Conference. Presented May 21, 2017
http://www.atsjournals.org/doi/abs/10.1164/ajrccm-conference.2017.195.1_MeetingAbstracts.A1155
Leisman D, Huang V, Zhou Q, et al. Delayed Second Dose Antibiotics for Patients Admitted From the Emergency Department With Sepsis: Prevalence, Risk Factors, and Outcomes. Critical Care Medicine 2017; 45(6): 956-965, June 2017
Print “NYS Mandate for Sepsis Protocol Works”
June 13, 2017
Layering Up
In our May 23, 2017 Patient Safety Tip of the Week “Trolling the RCA” we make our case that there is nothing wrong with the RCA (root cause analysis). Rather, the problem is in what you do with it or don’t do with it. The basic problems are failure to implement solutions that are strong actions, failure to monitor how effective those solutions are (also watching for unintended consequences), and failure to more widely share lessons learned.
So timely is an ISMP Medication Safety Alert that beautifully illustrates how such failures can allow recurrences of serious adverse events (ISMP 2017). We hope you’ll go to the ISMP alert for details but, basically, the events described involve cases where IV infusions of substances in similar bags, each covered with brown overwraps and with similar looking pharmacy-applied labels, were transposed and hooked up to the wrong smart pumps so each medication was infused at the wrong rate. The second event occurred within months of an RCA following the first event, despite the action plans put in place to prevent such events.
The action plan put in place after the first event relied largely on education and policy change. In our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s” we noted that all too often action steps never get implemented at all or consist solely of “weak” action steps and organizations are then surprised when a similar adverse event occurs in the future. We typically see weak actions like education and training or policy changes as the sole actions undertaken rather than strong actions like constraints and forcing functions. (We discussed strength of actions in our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s”. In that column we included an analogy to the effectiveness of signs and tools used to try to get drivers to slow down in construction zones on highways. We put them together in pictures with RCA action items and now incorporate them in our webinar presentations on doing good RCA’s. Click here to see them. Remember: images are more likely to be remembered than words!)
The action after the first event was educating nursing staff on changes to the infusion policy which included:
In the second event the IV lines were labeled but the labels were between the pumps and the patient, not between the pumps and the infusion bags. The nurse did not trace the lines back from the patient to the infusion bags (and apparently did not even notice the labels on the tubing). The infusion bags again looked very similar (same size, both covered with light-protective overwraps, and with pharmacy-applied labels that looked very similar).
The ISMP alert has several valuable lessons:
The ISMP alert calls heavily upon David Marx’s “3 dice” principle (Marx 2017). Everyone in patient safety is well familiar with James Reason’s “swiss cheese” theory of error occurrence and understands that multiple defenses prevent many more errors from occurring. Marx (whom you all recognize of “Just Culture” fame) says we should use the mathematical probabilities of rolling dice to help us construct effective interventions to prevent errors. In the Marx article he gives the example of a person being run over by a vehicle that was backing up. Rolling a “one” on one roll of the dice puts the person behind the vehicle. Doing a second roll of the dice by doing a pre-drive safety walk around still could roll another “one” but now the odds of rolling “one” on two rolls has increased. Doing a third roll by using a rearview camera would further increase the odds against having a bad accident. (Do the math: chance of each die coming up “one” is 1/6 so chance of all three coming up “one” is 1/6 x 1/6 x 1/6 = 1/216). So Marx recommends you “play with three dice, when you can”. Of course, if you could do even more dice rolls you could do even better. (And P.S.: don’t ever ask Google what the odds are of getting three ones when you roll 3 dice. You’d be amazed at how many different answers you get!)
Redundancy is a great approach to safety in any field – as long as you don’t over-rely on it! For example, in one of our recent columns (our May 30, 2017 Patient Safety Tip of the Week “Errors in Pre-Populated Medication Lists”) we noted that physicians may over-rely on pre-populated medication fields and fail to perform a thorough medication reconciliation. And you’ve often heard us say that improperly performed “double checks” may have higher individual error rates for both the initial checker and the person doing the second check. (See our October 16, 2012 Patient Safety Tip of the Week “What is the Evidence on Double Checks?” but note that we still consider double checks to be an action of intermediate strength worth using in multiple situations. But they need to be truly “independent” double checks.) And make sure when you are backing up your vehicle that you roll all 3 dice and don’t skip the middle one because you over-rely on your backup camera!
The cases also illustrate both failure to address all causal factors identified and addressing some factors that were not causal. Using an action for one vulnerability uncovered during the first RCA (the barcode pre-scanning) that actually was not a causal factor in the event was exemplified by the change in policy regarding barcode scanning at the bedside even though the smart pump system had not yet been integrated with the electronic health record. So the new barcoding policy simply ensured the correct medication and correct patient but did not ensure the correct infusion pump hookup and, hence, the correct rate of infusion.
But a big causal factor not adequately addressed was the similarity of the infusion bag preparations. As above, they were the same size, were covered with the same brown overwrap, and had labels that looked very much alike.
The last ISMP recommendation (need to measure implementation and effectiveness of action plans) is one we’ve echoed over and over. In our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s” we noted prior studies in the VA system (Hughes 2006) which analyzed action items from RCA’s and found that 30% were not implemented and another 25% were only partially implemented. Stronger action items were more likely to be implemented. Actions that were assigned to specific departments or people were more likely to be implemented than those assigned to general areas. And they found that the patient safety manager plays a critical role in RCA action implementation.
In our March 27, 2012 Patient Safety Tip of the Week “Action Plan Strength in RCA’s” we emphasized the importance of tracking whether recommended action steps were implemented following an RCA, whether they were effective, and whether there were any unintended consequences. Even the most well intentioned and well planned action steps sometimes lead to consequences that were never anticipated. One of the biggest issues we see in hospitals related to RCA’s is failure to follow up and close the feedback loop. In fact, probably the majority of hospitals lack formal procedures for ensuring the corrective actions recommended in an RCA are actually carried out (or barriers to their implementation identified and alternative steps taken). In our March 30, 2010 Patient Safety Tip of the Week “Publicly Released RCA’s: Everyone Learns from Them” we discussed an incident at a hospital in which a similar incident had occurred several years prior. After the first incident an extensive root cause analysis was done and multiple recommendations were made, including key recommendations that should have prevented the second incident. But all those recommendations had never been fully implemented. Importantly, the recommendations were communicated back to those individuals deemed to be in the “need to know” but not widely disseminated to middle or front line management nor to front line staff.
We recommend you keep a list or table of such identified action items from all your RCA’s to discuss at your monthly patient safety committee or performance improvement committee meetings. Action items should remain on that list until they have been implemented or completed. Only that sort of rigorous discipline will ensure that you did what you said you were going to do, i.e. that you “closed the loop”. And don’t forget you need to monitor your implemented actions for unanticipated and unintended consequences. For example, you might take the strong action of removing a drug from a particular setting, only to realize later that there were circumstances where that drug was needed in that setting.
We’d really like to emphasize one other point in the ISMP alert. They commented specifically about avoiding changing two infusions at the same time. For many years we have cautioned against trying to do more than one thing at a time. In our June 19, 2007 Patient Safety Tip of the Week “Unintended Consequences of Technological Solutons” we wrote about an unfortunate case in which a patient died as the result of transposition of telemetry units with another patient. One day, right around nursing change of shift, two patients were admitted to the remote floor and telemetry was ordered on both. The nurse took two transmitters with him and hooked the patients up, then called the CCU monitoring nurse to tell her about the two patients just hooked up. About an hour later the CCU monitoring nurse called the remote floor because one of the patients was in ventricular fibrillation. A code was called and the floor staff and code team ran to the patient’s room, only to find him sitting in bed, watching TV and eating a meal. Only after several minutes of fiddling with his EKG leads and talking to the nurse in the CCU did anyone realize that the patient several rooms down the hall was really the one in ventricular fibrillation. The transmitters obviously had been transposed! (This is a variation of the “two in a box” phenomenon we talked about in the April 23, 2007 Patient Safety Tip of the Week “Predictable Errors”.) And, of course, the system was poorly designed in that it allowed the first nurse to take out two remote telemetry transmitters at the same time.
ISMP recommends that each infusion bag should be changed independently, completing the process for one bag before bringing the next bag to the pump. Sometimes we don’t even think about the little things we do that might contribute to such errors. For instance, did the pharmacy deliver both infusion bags to the site(s) at the same time? Perhaps a pharmacist should deliver the bags and wait with the second bag until the first bag has been appropriately hooked up.
Another feature worth our comment is that the second incident occurred in the cardiac cath lab. New bags of the two infusions were needed after the patient had finished a procedure in the cardiac cath lab. In our October 22, 2013 Patient Safety Tip of the Week “How Safe Is Your Radiology Suite?” we had a section about medication errors that occur when patients are in the radiology suite or MRI suite. The cardiac cath lab is no different. Many of the same circumstances that lead to errors in those other suites are also present in the cath lab.
ISMP recommends you ask 5 key questions in addressing a safety hazard or error:
As usual, ISMP has provided another extremely valuable contribution to our better understanding of patient safety and the complex issues involved. Using real-life scenarios makes us all wary that “Wow! That could happen here!” and take appropriate steps to minimize the risk of similar incidents in our organizations. That’s the last problem with RCA’s that we noted in today’s introduction: failure to more widely share lessons learned. ISMP is one of the few organizations that provides us with RCA’s of real events and examples of action steps that are likely to prevent similar occurrences in our organizations.
Some of our prior columns on RCA’s, FMEA’s, response to serious incidents, etc:
July 24, 2007 “Serious Incident Response Checklist”
March 30, 2010 “Publicly Released RCA’s: Everyone Learns from Them”
April 2010 “RCA: Epidural Solution Infused Intravenously”
March 27, 2012 “Action Plan Strength in RCA’s”
March 2014 “FMEA to Avoid Breastmilk Mixups”
July 14, 2015 “NPSF’s RCA2 Guidelines”
July 12, 2016 “Forget Brexit – Brits Bash the RCA!”
May 23, 2017 “Trolling the RCA”
References:
ISMP (Institute for Safe Medication Practices). Common missteps with medication safety: Rolling a single dice, ineffective strategies, and unexecuted action plans. ISMP Medication Safety Alert! Acute Care Edition. June 1, 2017
http://www.ismp.org/newsletters/acutecare/showarticle.aspx?id=1168
Marx D. Play with three dice, when you can. What We Believe. Outcome Engenuity 2017; 1(3): 1-2
https://www.outcome-eng.com/wp-content/uploads/2017/03/WhatWeBelieve_Issue3_030917.pdf
Hughes D. Root Cause Analysis: Bridging the Gap Between Ideas and Execution. VA NCPS Topics in Patient Safety TIPS 2006; 6(5): 1,4 Nov/Dec 2006
http://www.patientsafety.va.gov/docs/TIPS/TIPS_NovDec06.pdf#page=1
Weak vs. Strong Responses to an RCA (Power Point presentation).
http://patientsafetysolutions.com/docs/RCA_strong_vs_weak_responses.ppt
Print “Layering Up”
June 20, 2017
Dilaudid Dangers #4
A recent Canadian study described 8 cases of fatalities of inpatients in hospitals or long-term care related to morphine or hydromorphone (Lowe 2017). Though there were only 8 cases reported, they really run the gamut of the types of errors in every phase of the medication process that contribute to or cause these lethal mishaps. And, as belied in our numerous columns listed below, Dilaudid/HYDROmorphone is the central figure in most of these cases.
As we have seen so often in the past, conversion from one opioid to another presents several vulnerabilities. Of course, failure to recognize the difference in relative potency between morphine and hydromorphone was an issue in several cases. But in other cases there was a failure to discontinue morphine administration when hydromorphone was begun so the patient was receiving multiple opioid preparations simultaneously. In another case a patient was receiving both intravenous hydromorphone and oral codeine.
Problems reconciling the correct dosage with the concentration in the vials used for preparation were also prominent. In several cases nurses on the patient care units, rather than pharmacists in the pharmacy, prepared the doses and drew up a fluid amount they thought was the correct dose, not realizing that the vial contained a higher concentration of drug.
LASA (look-alike sound-alike) confusion also occurred. Even when using tall man lettering, many healthcare workers still confuse morphine and HYDROmorphone.
Failure to rescue was also noted. In several of the cases, because the patient had DNR (do not resuscitate) status, decisions were made not to use naloxone for reversal of the opioid effect. However, in one case in which naloxone was used in a patient receiving both morphine and hydromorphone, the patient’s vital signs normalized after the naloxone administration but an hour later the patient was found unresponsive with a low respiratory rate. This was likely an example of “renarcotization” where there was a disparity between the time of action of naloxone and that of the opioid(s).
Perhaps somewhat surprising was the relative lack of cases in which non-opioid drugs used in combination with opioids had an additive respiratory depression effect. Such co-administration has been an issue in other cases of opioid-related respiratory depression.
Monitoring was an issue. We are not told what, if any, electronic monitoring was being done on any of the patients. But one provides an example of a common error: use of respiratory rate by itself as a monitoring parameter. In that case, morphine was to be held “if the respiratory rate was less than 10”. Respiratory rate by itself is actually a poor parameter for early identification of opioid-induced respiratory depression. Actually one of this month’s What's New in the Patient Safety World columns (June 2017 “Masterpiece: Monitoring for Opioid-Induced Respiratory Depression”) has a nice discussion on some of the pitfalls of monitoring as well as the appropriate ways to monitor patients receiving opioids.
The Canadian study authors note that errors occurred in all stages of the medication process: prescribing, order processing and transcription, dispensing, administration, and monitoring. Moreover, for 7 of the 8 cases there were multiple (2 or more) possible intervention points. At least six cases could have been prevented by additional patient monitoring.
At the prescribing/ordering phase, use of clinical decision support tools can be very useful. For example, you might use an automatic warning any time hydromorphone is prescribed or ordered (eg. a reminder that hydromorphone is 5-7 times more potent than morphine on a mg basis, or a reminder that the initial dose of hydromorphone in opioid-naïve patients is 0.2 to 0.5 mg IV or limiting that initial dose to 1 mg with a “hard” stop). And you can use standardized order sets (electronic or paper-based) to minimize the risk of an order for too high a dose.
CPOE and medication administration systems need to be designed and programmed to prevent simultaneously prescribing or administering more than one opioid. While there may be very rare instances where use of more than one opioid is necessary, programming in a “hard stop” that requires specific action to override the order should be mandatory.
Limiting the number of opioid products available is also useful.
Pharmacist review of all orders and all medications is an important patient safety tool. In one of the cases described the order for a hydromorphone infusion came after pharmacy hours when no pharmacist was available for review.
On the dispensing side, in many of the cases reported in the Canadian study nurses, rather than pharmacists, prepared the medication to be administered. One problem the authors noted was that high-concentration products were readily available on patient care units, thus increasing the chance that an inappropriately high dose might be inadvertently prepared and administered. Removal of such high-concentration products from patient care areas (requiring that only pharmacy have such products) makes sense.
The authors note that independent double checks are important as a potential tool to help avoid medication errors with opioids. We concur but note that in one of the cases reported above an independent double check was performed but a nurse noted the discrepancy between the dosage and the concentration about 90 minutes after the infusion had started.
The authors also noted that 6 of the 8 deaths might have been prevented by additional patient monitoring. We, of course, have done multiple columns on monitoring for opioid-related respiratory depression (see list below). And, because it is so hard to predict which patients will suffer respiratory depression, it is not enough to simply have intensive monitoring for high-risk patients. We really need to be monitoring all hospitalized patients receiving opioid treatment. Note also that the PPAHS (Physician-Patient Alliance for Health & Safety) has just released a position paper calling for continuous electronic monitoring for all patients receiving opioids (PPAHS 2017). It emphasizes, as we have, that use of pulse oximetry alone is not sufficient. It states patient monitoring plans should provide continuous monitoring of multiple physiologic metrics, with the inclusion of capnography monitors alongside other methodologies. Most importantly, PPAHS notes that monitors are “not meant to remove clinicians from the equation; instead, monitoring technology should be a multiplying factor for hands-on, proactive care.”
To reiterate from our multiple columns on Dilaudid dangers, here are strategies you should consider to reduce the risk of Dilaudid/HYDROmorphone (and other opioid) adverse events:
Our prior columns on patient safety issues related to Dilaudid/HYDROmorphone:
Other Patient Safety Tips of the Week pertaining to opioid-induced respiratory depression and PCA safety:
References:
Lowe A, Hamilton M, Greenall J, et al. Fatal overdoses involving hydromorphone and morphine among inpatients: a case series. CMAJ Open 2017; 5: E184-E189; published online March 2, 2017
http://cmajopen.ca/content/5/1/E184.full.pdf+html
PPAHS (Physician-Patient Alliance for Health & Safety). Patients Receiving Opioids Must Be Monitored With Continuous Electronic Monitoring. PPAHS Position Paper. June 2017
http://www.ppahs.org/wp-content/uploads/2017/05/PPAHS-Continuous-Electronic-Monitoring-Position.pdf
Print “Dilaudid Dangers #4”
June 27, 2017
Texting - We Told You So!
Our May 24, 2016 Patient Safety Tip of the Week “Texting Orders – Is It Really Safe?” and our January 2017 What's New in the Patient Safety World column “Joint Commission Thinks Twice About Texting Orders” decried the use of texting for orders. We lauded The Joint Commission’s decision to reverse itself and maintain the ban on texting orders. We identified the following areas of concern regarding texted orders:
See those previous columns for details on each of those concerns.
Secure text messaging does have numerous opportunities to improve communication in healthcare and we advocate its use in certain situations (for example, it can be extremely useful in helping to prevent alarm fatigue, though keep in mind the unintended consequences we noted in our February 9, 2016 Patient Safety Tip of the Week “It was just a matter of time…”). But texting also presents problems beyond just orders.
A new study (Luxenberg 2017) looked at a sample of text messages on an internal medicine service at UCSF. They reviewed 575 text-page messages relating to 217 patients. Most of the text messages were from non-physicians to physicians. They identified 3 main problematic themes:
Lack of standardization was a big problem, particularly as it related to vital signs. For example, a text message might provide a blood pressure reading but no heart rate. Others might have actual blood pressure recordings whereas others simply state “hypotensive”. Very few of the text messages used a structured format, such as SBAR, that we so often use in handoffs and other communications.
The vast majority (93%) of text messages were non-urgent. But most did not indicate the degree of urgency or the expected response. For example, some text messages had a tag “FYI” (for your information) but actually asked a question in the body of the text message for which a response was expected.
Other features of texted messages left gaps in communication. For example, use of uncommon abbreviations is a concern (eg. “prn hydral given”). Also, some text messages were missing words or used odd syntax.
In 2015, Nguyen and colleagues did a systematic review of the literature on use of technology for urgent clinician to clinician communications (Nguyen 2015). The technologies included 2-way pagers, smartphones, email, and blogging. They did find high rates of satisfaction following implementation of the new technologies and smartphones were generally perceived more favorably than 2-way pagers. However, there was scant evidence for actual improvements in patient care. They also identified some barriers and downsides to use of these modalities, including:
In the editorial accompanying the Luxenberg study, Mandl and Khoon (Mandl 2017) note that “health care team members tend to overuse these modalities, even when nonsynchronous communication would suffice. Notably, this problem, recognized for decades within medicine, has been amplified throughout society, as people spend their days and nights interrupted by Facebook messenger, email, texts, and Snapchats.” They do, however, note that there may be help on the horizon. They discuss how technologies such as natural language processing (NLP) and voice recognition may improve the ability to make such communications both more structured and more relevant.
We’ve all seen how the 140 character limit on Twitter can result in communications taking place void of context. And there is no doubt that texting has become so easy that we often do it without adequately considering the consequences. So we need stricter rules about what sorts of information are appropriate for texting and platforms that provide some structure to those communications.
In our May 24, 2016 Patient Safety Tip of the Week “Texting Orders – Is It Really Safe?” we warned that some orders may be given without adequate background on the patient’s condition or lab values. One of the text messages in the Luxenberg study was “Advise; May I have a 24 h extension on patient’s IV?” Even assuming that the text somehow provided the correct identification of the patient, how can a physician safely answer that question without more detailed knowledge about that patient?
And one issue not addressed in any of the studies mentioned above is whether texting actually might increase telephone time. We suspect that the terse nature of text messages might actually lead to telephone tag since clarification of some questions and responses may require such phone communication.
We also suspect that the issue of dangerous abbreviations will become a significant issue if texting becomes widespread. The character limitations will undoubtedly lead to attempts to shorten words (for example, drug names) or otherwise use abbreviations that should not be used. Theoretically, a good secure texting platform might incorporate mechanisms to prevent some dangerous abbreviations but, frankly, there are simply too many possibilities for complete capture.
Lastly, don’t be surprised if rudeness and lack of respect begin to appear in texted messages. We certainly see that people say things online that they would never say in face-to-face interactions. And then there’s the issue of what becomes part of the medical record. Our take is that any text message handled in a hospital’s formal secure texting system would be captured and become part of the electronic medical record. So a physician’s annoyed response “why are you interrupting me with that!” now becomes part of the record that might later become discoverable during a malpractice action.
See some of our other Patient Safety Tip of the Week columns dealing with unintended consequences of technology and other healthcare IT issues:
References:
Luxenberg A, Chan B, Khanna R, et al. Efficiency and Interpretability of Text Paging Communication for Medical Inpatients. A Mixed-Methods Analysis. JAMA Intern Med 2017; Published online June 19, 2017
http://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2631559
Nguyen C, McElroy LM, Abecassis MM, et al. The Use of Technology for Urgent Clinician to Clinician Communications: A systematic review of the literature. Int J Med Inform 2015; 84(2): 101-110
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4272907/
Mandl KD, Khoon E. Pagers and Beyond in an Era of Microcommunications—What Is Old Is New Again. JAMA Intern Med 2017; Published online June 19, 2017
http://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2631558
Print “Texting – We Told You So!”
Click here to leave a comment on any of these tips.
To get "Patient Safety Tip of the Week "emailed to you, click here and enter "subscribe" in the subject field.
Click on the "Contact Us" button at the left to send us your comments on our "Patient Safety Tip of the Week" cases.
To get "Patient Safety Tip of the Week "emailed to you, click here and enter "subscribe" in the subject field.