Inhaler education, hospital mergers, and more

Summaries from ACP Hospitalist Weekly.


Virtual instruction for inhaler technique nearly as effective as in-person instruction

A virtual, patient-directed intervention for improving inhaler technique worked nearly as well as a time-intensive, in-person strategy in a recent study of adults hospitalized with asthma or chronic obstructive pulmonary disease (COPD).

Researchers randomized adults with asthma or COPD admitted to general inpatient wards to receive either virtual (n=59) or in-person (n=59) educational interventions for improving inhaler technique. Virtual education was a module delivered by handheld tablet with narrated video demonstration of the correct technique, as well as self-assessment questions before and after demonstration (up to three rounds were repeated as needed). Participants in the in-person education group received iterative rounds of inhaler technique assessment and education by trained staff. Investigators and research assistants were blinded to interventions.

Study enrollment, assessments, and delivery of the interventions occurred in the hospital, and participants returned at 30 days for a follow-up research visit. Researchers tested the virtual education for noninferiority, based on whether the groups achieved an equal percentage with correct inhaler technique after education (>9 of 12 steps correct), against an a priori noninferiority limit of −10%. They also adjusted for differences in baseline technique and health literacy. Results were published on Jan. 3 by JAMA Network Open.

Of 118 participants, most were black (n=114 [97%]) and female (n=76 [64%]), with a mean age of 54.5 years. Compared to before education, correct technique increased when assessed at discharge in both the virtual group (from 2% to 69%; difference, 67%) and the in-person group (from 17% to 83%; difference, 66%). While the difference after intervention exceeded the noninferiority limit (−14%; 95% CI lower bound, −26%), the difference was equivalent to the noninferiority limit (−10%; 95% CI lower bound, −22%) after adjustment for baseline technique. Among 100 (85%) participants who completed the 30-day follow-up visit, inhaler performance declined by a similar amount in both groups, with 26 of 49 (53%) in the virtual group and 32 of 51 (63%) in the in-person group using correct technique. After adjustment, the difference in the rates was −4% (95% CI lower bound, −18%).

Limitations of the study include the fact that participants were primarily urban, underserved, and black patients, and that there was an imbalance in baseline inhaler misuse between the two groups, the authors noted. The findings should be validated in future studies across diverse settings and populations with similar baseline rates of misuse, they said.

While technological interventions are not suitable for all patients, a potential advantage of virtual inhaler instruction is the identification of patients who may benefit from additional in-person training, they said, adding that the virtual education may be repeated at home as necessary. Larger studies are needed to determine whether the intervention will improve long-term inhaler technique skills and/or patient outcomes, they added.

“This virtual approach has potentially important implications for increasing access to high-quality education because the virtual intervention likely has significantly lower costs and time constraints in real-world settings compared with the costs of training and delivering in-person education in hospital and at home,” the authors wrote.

Hospital mergers, acquisitions don't appear to improve care

Patient experience declined modestly at hospitals acquired by another hospital or hospital system, and outcomes did not appear to improve, a recent study found.

Researchers used Medicare claims and data from Hospital Compare from 2007 through 2016 to perform difference-in-differences analyses that compared changes in performance of acquired hospitals from before acquisition to after acquisition with changes in control hospitals that did not change ownership. The study focused on 2009 through 2013 as the dates of merger or acquisition. Data on a composite of clinical process measures, a composite of patient experience measures, 30-day mortality rate after admission, and rate of readmission 30 days after discharge were analyzed. The study results were published Jan. 2 by the New England Journal of Medicine.

Two hundred forty-six hospitals that were acquired or merged from 2009 to 2013 and 1,986 control hospitals that were not acquired and did not merge from 2007 to 2016 were included in the study sample. Hospital acquisition was associated with a modest differential decrease in the patient experience measure (adjusted differential change, −0.17 SD; 95% CI, −0.26 to −0.07; P=0.002) but no significant differential change in 30-day readmission rates (adjusted differential change, −0.10 percentage point; 95% CI, −0.53 to 0.34 percentage point; P=0.72) or 30-day mortality (adjusted differential change, −0.03 percentage point; 95% CI, −0.20 to 0.14 percentage point; P=0.72).

For the composite of clinical process measures, a significant differential improvement was seen for acquired hospitals (0.22 SD; 95% CI, 0.05 to 0.38; P=0.03), but because some of this effect occurred before acquisition, it is unclear whether it could be attributed to the change in ownership, the researchers said.

The researchers noted that they were unable to examine all dimensions of quality and that difference-in-differences analyses of voluntary exposure to treatment can be affected by selection bias, among other limitations. Patient experiences involve aspects of quality that patients can observe and may therefore be most likely to be affected by reduced competitive pressure, while the effect of acquisitions on clinical process measures or other outcomes may be less visible, the authors wrote. They concluded that hospital mergers and acquisitions are associated with modest decreases in patient experience; small, nonsignificant changes in readmission and mortality rates; and inconclusive effects on clinical process measures. “These findings challenge arguments that hospital consolidation, which is known to increase prices, also improves quality,” they wrote.

Patients increasingly discharged alive from inpatient palliative care, referred to home or outpatient programs

From 2013 to 2017, inpatient palliative care teams saw more patients with diagnoses other than cancer and discharged more patients alive, a recent study found.

The retrospective cohort study used data gathered from 2013 through 2017 at 88 hospitals in which palliative care teams voluntarily participated in the Palliative Care Quality Network, a national quality improvement collaborative. A total of 135,197 patients were studied (51.0% female; mean age, 71.3 years). Patient characteristics and process of care were compared over time and among hospitals. Results were published by JAMA Network Open on Dec. 6, 2019.

The patients' mean Palliative Performance Scale score was 34.7% (range, 14.9% to 56.8%). Cancer was the most common primary diagnosis (32.0%; range, 11.3% to 93.9%), but it was less common in 2017 than 2013 (odds ratio [OR], 0.84 [95% CI, 0.79 to 0.91]; P<0.001). Pain and other symptoms (including anxiety, nausea, and dyspnea) were common and improved significantly while patients were receiving inpatient palliative care. Most patients were discharged alive (78.7%; range, 44.7% to 99.4%), and this percentage increased from 2013 to 2017 (OR, 1.36 [95% CI, 1.27 to 1.46]; P<0.001). Discharge referral to clinic-based (OR, 4.00 [95% CI, 2.95 to 5.43]; P<0.001) and home-based palliative care (OR, 2.63 [95% CI, 1.92 to 3.61]; P<0.001) both increased significantly over time, while referrals to hospice decreased (OR, 0.56 [95% CI, 0.51 to 0.62]; P<0.001).

The study also found significant variability among hospitals. For example, the percentage of patients discharged from the hospital varied more than twofold and the percentage of a team's patients who had heart disease varied from 0% to 25%. These variations suggests room for improvement, according to the study authors. “Identifying these best practices, implementing them broadly to evaluate their generalizability, and disseminating the most effective strategies could result in significant improvements in the quality and consistency of care provided by PC [palliative care] teams nationally,” they wrote.

The changes found in the study are consistent with guidelines recommending earlier palliative care for all patients with serious illnesses, the study authors said. The observed improvements in symptoms indicate that palliative care succeeded in improving patients' quality of life, they added. Limitations of the study include that palliative care programs participating in a voluntary quality improvement network might differ from those that are not, and that patients who were readmitted might have been included more than once in the study data.

Calls to poison control centers for gabapentin, baclofen increased in recent years

Gabapentin and baclofen misuse, use in suicide attempts, and associated hospital admissions among U.S. adults have increased since 2013, a retrospective study found.

Image by Getty Images
Image by Getty Images

Researchers assessed trends in adult exposures to gabapentin and baclofen reported to U.S. poison control centers and entered in the National Poison Data System. They analyzed demographics, common co-ingestions, medical outcomes, and geographic distribution. The study analyzed exposures to the two drugs with or without co-exposure to other compounds. Isolated exposures were defined as those only involving gabapentin or baclofen. Results were published on Dec. 1, 2019, by Clinical Toxicology.

There were 74,175 gabapentin exposures from 2013 to 2017 (mean age, 44.6 years; 59.7% female). Gabapentin exposure increased by 72.3% overall during this period, while isolated exposures increased by 67.1% and isolated abuse/misuse exposures increased by 119.9%. There were 15,397 baclofen exposures from 2014 to 2017 (mean age, 39.9 years; 63.4% female). Baclofen exposure increased by 36.2% overall during this period, while isolated exposures increased by 35.0% and isolated abuse/misuse exposures increased by 31.7%.

Co-ingestions of sedatives and opioids were common for both medications. Admission to a health care facility was required in 16.7% of isolated gabapentin exposures and 52.2% of isolated baclofen exposures. Critical care unit admission was required in a majority of these cases (53.9% for gabapentin and 73.5% for baclofen). Intentional suspected suicide attempts with isolated gabapentin exposures increased by 80.5% over a five-year period and by 43.0% for isolated baclofen exposures over a four-year period. All states saw increases in gabapentin exposures, and most saw increases in baclofen exposures, gabapentin misuse/abuse, and baclofen misuse/abuse.

Among other limitations, the study relied on voluntarily reported exposure cases and may not represent the total incidence of national exposures, and confirmatory testing for the presence of gabapentin, baclofen, and/or additional co-ingestions was not consistently documented, the authors noted. Increased exposures to gabapentin and baclofen coincide with reductions in opioid prescribing and may represent an unintended consequence of the ongoing need for effective pain management and migration away from opioid use, they concluded.

“While the risks of opioids have been widely publicized, medication alternatives to opioids also carry risks that need to be better understood, described, and disseminated so that providers and patients can make decisions regarding the role of these medications in their pain management based upon an evidence-informed risk-benefit analysis,” the authors wrote.

Model including functional impairments may help predict death in older adults after acute MI

A risk model that includes information on functional impairments may help predict six-month mortality rates after acute myocardial infarction (MI), according to a recent study.

Researchers used data from the SILVER-AMI study, a prospective observational study of patients ages 75 years or older who were hospitalized for acute MI at 94 U.S. hospitals, to develop the risk model. Patients in the SILVER-AMI study received a baseline assessment at hospitalization that include demographic characteristics, symptoms, health status, functional impairments (e.g., cognition, mobility, muscle strength, vision, and hearing), and other conditions, such as unintended weight loss, that are common with age. The primary outcome of SILVER-AMI was death within six months of hospital discharge. The derivation cohort included 2,004 participants, and the validation cohort included 1,002 participants. Results of the study were published Dec. 10, 2019, by Annals of Internal Medicine and appeared in the Jan. 7 issue.

All participants were ages 75 years or older and were discharged alive after hospitalization for acute MI. Mean age was 81.5 years, 44.4% were women, and 10.5% were of nonwhite ethnicity. Two hundred sixty-six participants (8.8%) died within six months of hospital discharge, 184 in the derivation cohort and 82 in the validation cohort. Patents who died were more likely to have cognitive impairment (33.2% vs. 15.6%), unintentional weight loss (38.8% vs. 20.9%), and impaired mobility (83.7% vs. 64.0%). Fifteen factors were associated with six-month mortality in the multivariable model, including hearing impairment, mobility impairment, unintended weight loss, and lower patient-reported health status, and were included in the final model. Model calibration and discrimination were good (area under the curve, 0.84 for the validation cohort), and the addition of functional impairments appeared to significantly improve model performance.

The researchers noted that their model was not validated externally, that vision and hearing impairment rates in their study were lower than those seen in other published studies, and that functional impairments are not routinely assessed during hospitalization, among other limitations. However, they concluded that their novel risk model, which included information on traditional risk factors, functional impairment, and other aging-related variables, helped predict risk for death in older patients within six months of acute MI hospitalization and could be used for prognostication at hospital discharge. A Web-based calculator for six-month mortality after MI was developed based on the results from the multivariable model.

The authors of an accompanying editorial said that the model could potentially be used to improve clinical decision making once the current results are replicated, if the added measures on function could be routinely collected and if clinicians are informed about what do with the information. “Routine use of risk prediction models is valuable only if they actually enhance clinical decision making in a way that improves outcomes for older adults at the highest risk,” the editorialists wrote, noting that it is not clear whether the functional predictors should be targeted for intervention or whether they are markers for other features that require attention.

The editorialists noted that frail elderly patients who have recently had MIs are at particularly high risk for death and that the current study makes an important contribution to the literature by developing a model that helps quantify that risk. “If the ultimate goal of a health system is to improve health, we will increasingly need to be willing to use prediction models to identify our highest-risk patients and develop, test, and deploy interventions to help address this increased risk,” the editorialists wrote.

Geographic cohorting may increase patient interactions, but with potential risks

Geographic cohorting, which assigns a hospitalist to a single inpatient unit, may increase clinicians' interaction with patients but could also negatively affect workload, according to a recent observational study.

Researchers at an academic hospital in the Midwestern U.S. studied hospitalists who worked in geographic cohorts and those who did not. The hospital had implemented geographic cohorting in 2012, and teams aimed to maintain at least 80% of their patients on their assigned unit. The study used geotracking to determine the time that 17 hospitalists spent in direct patient care (defined as time in patient rooms) and indirect care (defined as other locations). In addition, four hospitalists who worked in geographic cohorts and four who did not were each observed in person for a workday.

The researchers used multilevel modeling to analyze associations between time spent in direct and indirect care, as well as characteristics of teams and workdays. Results of the study were published Nov. 20, 2019, by the Journal of Hospital Medicine.

Of the 17 hospitalists in the study, seven were observed only in the geographic cohorting model, one was observed only in the nongeographic cohorting model, and nine were observed in both. A total of 10,522 direct care episodes were noted via geotracking, with an average duration ranging from 4.1 to 5.8 minutes. An association was seen between geographic cohorts and longer patient visits, as well as between increasing patient loads and shorter visits. Geographic cohorts, increasing patient loads, and increasing unit visits were associated with time spent in indirect care.

In the 3,032 minutes of data obtained via in-person observations, geographic cohort hospitalists spent 56% of their day in computer interactions compared with 39% for nongeographic cohort hospitalists (P<0.005). Time spent multitasking in each group was similar (18% vs. 14%; P>0.05). Hospitalists in each group were frequently interrupted, but the highest rate, once every eight minutes in the afternoon, was seen in the geographic cohort group, which had a total of 139 interruptions versus 102 for the nongeographic cohort group.

The researchers noted that their study was done at a single center and did not address clinical outcomes or clinician, patient, or nursing satisfaction, among other limitations. They concluded that while geographic cohorting may increase hospitalists' interactions with patients, these gains may be offset by suboptimal patient loads and ineffective cohorting structure.

“Our work underscores the importance of paying careful attention to specific components and monitoring for unintended consequences in a complex intervention such as cohorting to allow subsequent refinement,” the authors wrote. They called for additional studies to examine the interplay between models of care and their effect on such factors as interruptions, errors, and burnout.

Bedside optic nerve ultrasonography may be useful for measuring intracranial pressure

Optic nerve ultrasonography at the bedside may be able to quickly and accurately diagnose increased intracranial pressure, according to a recent study.

Researchers reviewed studies published through May 2019 to determine the accuracy of optic nerve ultrasonography for diagnosing increased intracranial pressure in both adults and children. Optic nerve ultrasonography, also called optic nerve sheath diameter sonography, involves applying a layer of gel to the closed eyelid with the patient in supine position and measuring the optic nerve sheath using a linear array transducer 3 mm posterior to the orbit, usually in the coronal plane. Prospective studies of the diagnostic accuracy of this test were selected from 13 databases, as well as reference lists and meeting proceedings. The results of the systematic review and meta-analysis were published Nov. 19, 2019, by Annals of Internal Medicine and appeared in the Dec. 17, 2019, issue.

Image by Getty Images
Image by Getty Images

Seventy-one studies involving 4,551 patients were eligible for inclusion. Of these, 61 included adults and 35 were considered to have a low risk for bias. Eighteen studies included patients with traumatic brain injury, 26 included patients with nontraumatic brain injury, and 27 included both. In patients with traumatic brain injury, the pooled sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio of optic nerve ultrasonography were 97% (95% CI, 92% to 99%), 86% (95% CI, 74% to 93%), 6.93 (95% CI, 3.55 to 13.54), and 0.04 (95% CI, 0.02 to 0.10), respectively. For patients with nontraumatic brain injury, these values were 92% (95% CI, 86% to 96%), 86% (95% CI, 77% to 92%), 6.39 (95% CI, 3.77 to 10.84), and 0.09 (95% CI, 0.05 to 0.17). Estimates of accuracy were similar when stratified by patient age, operator specialty and training, reference standard, sonographer blinding, and cutoff value. The researchers found that 5.0 mm was the optimal cutoff.

Most of the included studies were small and had imprecise summary estimates, publication bias was possible, and effect of optic nerve ultrasonography on outcomes was not evaluated. However, despite these limitations, the authors concluded that optic nerve ultrasonography at the bedside is a noninvasive, quick, easy-to-use test. “A normal reading on optic nerve sheath ultrasonography (<5.0 mm) with high sensitivity and low negative likelihood ratio may rule out increased intracranial pressure, whereas an elevated reading (≥5.0 mm) with high specificity and positive likelihood ratio may indicate increased intracranial pressure and require confirmatory diagnostic tests,” they wrote.

The authors of an accompanying editorial stressed that the 5-mm cutoff described in the study should be interpreted and applied with caution, since the clinical significance of one bedside measurement is limited and cutoffs varied markedly in the 71 included studies. They agreed that bedside ultrasonography of the optic nerve sheath can decrease time to definitive management of brain injury but recommended that future studies should focus on developing a standardized, multidisciplinary approach to this test, as well as a cutoff that considers sex and other characteristics. “Once an agreed cutoff and protocol are determined, novice sonographer-clinicians could be supported by ultrasound experts for remote integration of bedside ultrasonography,” the editorialists wrote.

Survey finds shortcomings in assessment of neurologic function after cardiac arrest

Many physicians use outdated practices to assess the neurological function of patients after cardiac arrest, according to a recent survey of specialists.

The web-based survey gathered data from 762 members of the Neurocritical Care Society, Society of Critical Care Medicine, and American Academy of Neurology to assess current practices in neuroprognostication. The respondents came from 22 countries, although the majority were in the U.S., and all reported caring for successfully resuscitated unconscious cardiac arrest patients. Results were published by Critical Care Medicine on Nov. 18, 2019, and appeared in the February issue.

More than a third of the respondents said they used absent corneal reflexes (33.5%) and absent pupillary reflexes (36.2%) at 24 hours to assess patient prognosis. Current guidelines recommend waiting 72 hours. Most of the respondents (87%) considered absent motor response or extensor posturing to be very or critically important prognostic indicators. The study authors noted that although these used to be considered robust predictors, more recent research has found unacceptably high false-positive rates, so the use of these indicators alone is no longer recommended.

Asked to choose which tests were very or critically important, the most respondents chose neurologic examination (97.7%), followed by myoclonic status epilepticus (80.3%), brain MRI (70.2%), electroencephalography (68.4%), head CT (62.7%), and median nerve somatosensory evoked potentials (SSEP, 53.5%). The study authors noted that SSEP have been found to have a higher predictive value than myoclonic status epilepticus or imaging. The survey also found that up to 25% of the respondents would give a poor prognosis and definitive recommendations sooner than is supported by the evidence.

“Neuroprognostic approaches to hypoxic-ischemic encephalopathy vary among physicians and are often not consistent with current guidelines. The overall inconsistency in approaches and deviation from evidence-based recommendations are concerning in this disease state where mortality is so integrally related to outcome prediction,” the authors concluded. They called for educational efforts to inform clinicians about the latest evidence and guidelines.