Article Text

Download PDFPDF
Quality and safety in the literature: September 2020
  1. Nathan Houchens1,2,
  2. Ashwin Gupta1,2
  1. 1 Medicine Service, Veterans Affairs Ann Arbor Healthcare System, Ann Arbor, Michigan, USA
  2. 2 Internal Medicine, University of Michigan, Ann Arbor, Michigan, USA
  1. Correspondence to Dr Nathan Houchens, Medicine Service, Veterans Affairs Ann Arbor Healthcare System, Ann Arbor, MI 48105, USA; nathanho{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Healthcare quality and safety span multiple topics across the spectrum of academic and clinical disciplines. Keeping abreast of the rapidly growing body of work can be challenging. In this series, we provide succinct summaries of selected relevant studies published in the last several months. Some articles will focus on a particular theme, while others will highlight unique publications from high-impact medical journals.

Key points

  • In a retrospective cohort study, patients with chronic life-limiting illness and pre-existing Physician Orders for Life-Sustaining Treatment (POLST) who were hospitalised near the end of their lives frequently received care that seemed incongruent with their documented preferences. Potential discordance occurred more frequently in those admitted with traumatic injury and less frequently in those with a history of cancer and dementia. JAMA. 16 Feb 2020.

  • In a cross-sectional study of US long-stay nursing home residents with advanced illness, rates of transfer to acute care hospitals for all causes and potentially avoidable conditions declined from 2011 to 2016 without an increase in mortality. JAMA Intern Med. 30 Dec 2019.

  • In a retrospective cohort study, researchers established and used novel metrics to characterise the utility and appropriateness of electronic consultations in a variety of medical specialties. These e-consults frequently resulted in avoided physician visits. Ann Intern Med. 19 May 2020.

Association of Physician Orders for Life-Sustaining Treatment with ICU admission among patients hospitalized near the end of life

JAMA. 16 Feb 2020.

Provision of care that honours a patient’s values, goals and wishes is the foundation of medicine. Goals of care conversations offer patients with chronic or life-limiting illness an opportunity to outline their wishes and expectations for medical care. It has been 30 years since the passage of the United States Patient Self Determination Act, which requires hospitals, skilled nursing facilities, home health agencies, hospice programmes and health maintenance organisations to inform patients of their rights to make decisions regarding their medical care, periodically inquire as to whether a patient executed an advance directive and document the patient’s wishes, and ensure that legally valid advance directives and documented medical care wishes are implemented.1

Since that time, most states have adopted programmes that aim to capture and execute patients’ goals of care through Physician Orders for Life-Sustaining Treatment (POLST), portable medical orders specifying treatment limitations for emergency care.2 These orders typically cover not just the patient’s wishes with respect to resuscitative treatments but also wishes for transfers to the acute care hospital (eg, for nursing home residents), the receipt of curative intravenous treatments such as antibiotics, and acceptable forms of nutrition, among others.

Lee and colleagues sought to determine how often healthcare professionals provide care concordant with POLST orders. Specifically, their retrospective cohort study examined the association between POLSTs and concordance or discordance of actual care received among adult patients with chronic life-limiting illness who were hospitalised in the last six months of life.3 All patients had died in Washington State, USA, between 2010 and 2017, had been admitted to one of two affiliated academic hospitals, and had pre-existing POLSTs. In the study hospitals, POLSTs were widely recognised and were frequently used to guide treatment decisions for hospital care for patients who had them. They were generally composed of one of the following three options: (1) comfort-only measures, (2) limited additional interventions (eg, intravenous fluids or antibiotics) and (3) full treatment (including mechanical ventilation and cardioversion as indicated). The primary outcome of interest was admission to the intensive care unit (ICU) (excluding admissions only for symptom management), and the secondary outcome was administration of at least one pre-specified life-sustaining treatment: mechanical ventilation, vasoactive medication infusions, new renal replacement therapy or cardiopulmonary resuscitation.

A total of 1818 decedents met all inclusion criteria. The mean age was 70.8 years, 41% were female, and 76% were white. Only 23% of all participants had only one chronic life-limiting illness diagnosis with nearly one-third having four or more. Among all decedents, 36% had POLSTs for full treatment, 42% specified limited interventions and 22% comfort-only measures. Among those with limited interventions and comfort-only orders, 46% (95% CI 42% to 49%) and 31% (95% CI 26% to 35%) were admitted to the ICU, respectively.

Among patients with limited interventions and comfort-only orders, 41% (95% CI 38% to 45%) and 30% (95% CI 26% to 35%) received care discordant with POLSTs, respectively, and 20% (95% CI 17% to 23%) and 14% (95% CI 11% to 17%) received life-sustaining treatments discordant with POLSTs, respectively. Patients with a history of cancer were significantly less likely to receive care discordant with POLSTs, as were patients with dementia. Patients admitted for traumatic injury, by contrast, were significantly more likely to receive care discordant with POLSTs.

Lee and colleagues concluded that patients with POLSTs defining limitations on life-sustaining treatments who were hospitalised near the end of their lives frequently received care that seemed incongruent with their documented preferences. This may include care that was unwanted, unlikely to be of benefit and/or associated with high costs. The authors identified potential differences between care that is discordant with a POLST and care that is inappropriate or discordant with the patient’s goals. One explanation may be that conversations involving life-sustaining treatment decisions frequently occur on admission to the hospital, and any new orders may differ from previously documented POLSTs as a result of a change in the clinical circumstances or identification of an unexpected, acute and potentially reversible condition. Support for this conclusion is garnered from the fact that patients with traumatic injury (a less predictable prognosis) more often received care considered to be discordant with previously completed POLSTs, whereas patients with cancer and dementia (a more predictable prognosis) more often received care considered to be concordant with POLSTs. A major limitation arose from the fact that patients who chose to avoid hospitalisation near the end of life would not be included in the population, and thus, the incidence of discordant care among patients with POLSTs may be overestimated.

This study adds to the literature on advance care planning. Many individuals prefer to receive medical care and die at home, yet particularly in Western countries, dying often happens in the hospital setting.4–6 Most older patients prefer quality of life rather than artificial prolongation of life using technology.7 Yet, similar to the study by Lee and colleagues, the literature indicates patients often receive aggressive life-support measures in the final months of life,8 even when the patient or their loved one indicates a preference for comfort care only. The emotional, relational and financial costs of this problem can be substantial.9–11

How might we explain the discordance? One facet of this complex issue is volume: despite 30 years of care planning programmes and strategy, only 1/3 of adults have an advance directive.12 Equally important is the fact that providing care with reliable concordance to a patient’s goals is challenging for several reasons13: first, most individuals participating in goals of care conversations do not have the sophistication and background necessary for effective advance care planning. Second, patients may struggle to express preferences for future complex, emotion-laden hypothetical scenarios when they are healthy. Third, the future can change, in that a patient may become more comfortable with new realities and disabilities or a patient’s viewpoints on death may adjust when confronted with it. Fourth, patients may not want to discuss these topics out of fear, distress, inability to anticipate their wishes, or concern that care before the end of life will be altered. Finally, most healthcare systems do not provide financial incentive or reimbursement for effective advance care planning. Instead of traditional advance care planning, perhaps healthcare systems might train their providers to conduct earlier conversations aimed to prepare patients and their loved ones for more realistic ‘in-the-moment’ decision making to better equip them for this difficult task.14 15

Hospital transfer rates among US nursing home residents with advanced illness before and after initiatives to reduce hospitalizations

JAMA Intern Med. 30 Dec 2019.

In the USA alone, more than 2.4 million individuals reside in long-term care settings which comprise over 15 000 federally regulated nursing facilities and over 42 000 assisted-living communities.16 Transferring residents to acute care hospitals is often more burdensome than beneficial for residents in long-term care settings,17–19 yet hospitalisation rates remain high in these populations with advanced chronic illnesses.20 Enhancing care in nursing homes and thereby reducing unnecessary acute healthcare exposure have recently become even more of a priority, given the increased risks of entering acute care hospitals in the era of the SARS-CoV-2 (COVID-19) pandemic.

In their cross-sectional study of US long-stay nursing home residents with advanced illness, McCarthy and colleagues sought to understand how national quality initiatives to reduce hospital transfers and readmissions have affected rates of hospital transfer in this population over time.21 The authors targeted transfers for conditions that were considered potentially avoidable (eg, sepsis, pneumonia and heart failure) to assess the effects, if any, of policy initiatives harnessing financial implications (eg, the Hospital Readmission Reduction Program22 and alternative payment models) that incentivise nursing homes to reduce hospital transfers.23 They also examined transfer rates for serious bone fractures which were thought to be less affected or unaffected by policy initiatives. The study used Minimum Data Set assessments linked to Medicare data from 2011 to 2016, and the study population included six annual inception cohorts of residents aged 65 years or older with dementia, heart failure and/or chronic obstructive pulmonary disease (COPD), all of which must have been considered advanced in nature using prespecified criteria. Hospital transfers included emergency department visits and observation stays, in addition to acute hospitalisations.

All resident cohorts consisted mostly of white female residents who were eligible for both Medicare and Medicaid. The adjusted proportions of residents who experienced one or more hospital transfers for any cause in the 2011 and 2016 cohorts were 56.1% and 45.4% for those with advanced dementia; 77.6% and 69.5% for those with heart failure; and 76.2% and 67.2% for those with COPD, respectively. These decreases in transfer rates remained significant after adjustment for baseline characteristics and were nearly entirely attributable to decreases in acute hospitalizations. The mean number of hospital transfers per person-year alive for potentially avoidable conditions also decreased for each condition from 2011 to 2016: 2.4 vs 1.6 (absolute risk reduction (aRR) 0.73, 95% CI 0.65 to 0.81) for residents with advanced dementia, 8.5 vs 6.7 (aRR 0.72, 95% CI 0.65 to 0.81) for residents with heart failure, and 7.8 vs 5.5 (aRR 0.64 95% CI 0.57 to 0.72) for residents with COPD, respectively. Hospital transfers for serious bone fractures remained stable. Overall enrollment in hospice was low despite a mortality in excess of 50% within 1 year for all advanced illness groups. The authors concluded that, during the time when national initiatives to reduce transfers to acute care settings were implemented, transfers for all causes and potentially avoidable conditions decreased for long-stay nursing home residents without an appreciable change in mortality. The low rates of observation status seemed to diminish the theory that hospitals are shifting care from acute hospitalisation to observation simply to avoid financial penalty.

The seemingly encouraging results from McCarthy and colleagues must be examined with a few important caveats: First, evidence of care quality should be carefully assessed. For example, Fuller and colleagues identified that Nursing Home Compare Star Ratings provided by Medicare have limited correlation with rates of potentially preventable events.24 Additionally, Centers for Medicare and Medicaid Services data do not include assisted-living facilities and institutions not federally regulated,16 which may lead to underestimation of mortality and safety data. Second, while number of hospital readmissions (a measure analogous to transfers for acute hospitalisation) has declined, the number of per-capita admissions to the hospital has also declined,25 and there have been notable changes in coding practices.26 Combined, these changes may result in less discrete and accurate comparative data. Finally, the number of deaths occurring in long-term care facilities during the COVID-19 pandemic more recently have been high.27 This underscores the need for up-to-date data and adequate staffing and resources to care for this vulnerable population. While encouraging, the results from McCarthy and colleagues should be combined with further examination of root causes and contributing factors for ongoing challenges in effective, efficient care for long-term facility residents.

Utility, appropriateness, and content of electronic consultations across medical subspecialties

Ann Intern Med. 19 May 2020.

Electronic consultation (e-consult) services, interprofessional consultation generally between primary care providers (PCPs) and specialty providers and typically facilitated through the electronic health record, have been successfully implemented around the world.28–30 Use of these consultations has increased dramatically since introduction, and the number of specialties represented continues to grow.28 Previous studies have proposed potential benefits of e-consultation including decreased specialty provider wait times,31 avoidance of hospital referrals,32 and improved PCP education though e-consult exchanges.33 Data from focus groups across five academic medical centres in the USA suggested that primary care patients strongly support e-consults, particularly when employed by a trusted PCP.34 While potential benefits to patients and providers have been demonstrated, population-level effects on healthcare utilisation remain unclear. More specifically, no metrics to assess the appropriateness of e-consults and their ability to prevent downstream specialty care visits currently exist.

In their retrospective cohort study, Ahmed and colleagues sought to use novel metrics to understand utility and appropriateness of e-consults placed by PCPs to specialty clinics at a large, integrated healthcare system in the USA.35 Selected specialties (psychiatry, infectious disease, haematology, rheumatology and dermatology) were included after initial evaluation of the entire e-consult programme and selection for high e-consult volume and other factors. Appropriateness of e-consults was defined as meeting four criteria, which were determined in consultation with various stakeholders:

  • The question was not answerable by reviewing evidence-based summary sources.

  • The question was not requesting logistical information.

  • The consultation has appropriate clinical urgency.

  • The consultation has appropriate patient complexity.

Consults were deemed appropriate only if they met all four criteria. These criteria were established based on two authors’ review of 50 identical medical records and discussion with the senior author. Two researchers then reviewed 750 medical records (of which nine were excluded due to incomplete data), and inter-rater agreement was calculated. Utility of e-consults was assessed based on the rate of avoided visits or the absence of in-person visits to that specialty within 120 days.

Data on 6512 e-consults placed between 1 October 2017 and 30 November 2018 were included. Across specialties, a majority of e-consults were completed within 24 hours (range, 73.1% for psychiatry to 87.8% for infectious disease). Overall, 520 of 741 e-consults (70.2%) were considered appropriate. The proportion of e-consults felt to be appropriate ranged from 60.5% in rheumatology to 77.9% in psychiatry. Agreement was 94% with a kappa statistic of 0.57 (95% CI 0.36 to 0.79), indicating moderate agreement. The proportion of e-consults resulting in avoided visits ranged from 61.9% in dermatology to 92.6% in psychiatry.

Within their single integrated healthcare system, Ahmed and colleagues noted large variability in both appropriateness of e-consults and their ability to avoid downstream specialty in-person appointments. While these findings are interesting, more important are the novel metrics created by which to judge appropriateness and utility of e-consults. These metrics may both provide a foundation by which future studies on e-consult outcomes may be compared and provide local institutions with the opportunity to obtain necessary data to evaluate the efficacy and efficiency of their own e-consult programmes.

While use of e-consults and other modalities of virtual medicine has expanded organically, the COVID-19 pandemic has rapidly accelerated these trends. In efforts to preserve personal protective equipment, allow for social distancing or cover inpatient services during surges of COVID-19,36 access to subspecialty care has been pressured.37 E-consults may be instrumental in alleviating this backlog of patients awaiting specialty care appointments. Understanding the appropriateness and utility of these e-consults, however, will be imperative to ensure that healthcare resources are optimally utilised.



  • Twitter @nate_houchens

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Provenance and peer review Not commissioned; internally peer reviewed.