PEG placement does not prolong life in this setting.

A large retrospective study assessed the association between PEG placement and survival in 36,492 patients with advanced dementia.¹ Patients with a diagnosis of dementia based on the Minimum Data Set (a national nursing home database) were eligible for inclusion at the point of severe deficits in decision-making capacity with total dependence for eating, determined from their Cognitive Performance Scale, a validated dementia functional score. Patients were excluded if they were comatose or had received PEG feeding in the previous 6 months. Multivariable survival analysis utilized inverse probability weighting (to address selection bias) and controlled for demographic variables,
advanced care plans, comorbidities, dietary and nutritional variables, functional status, and models of mortality. PEG placement was not associated with an increase in survival (HR 1.03, 95% CI 0.94-1.13; P-value not reported). Study caveats include the observational and retrospective nature and use of administrative data.

These findings are in line with a 2009 Cochrane review of enteral tube feeding among elderly adults with advanced dementia.² The review included experimental and observational studies comparing tube feeding (either PEG or nasogastric) to usual care/no intervention. Studies were included if >50% of patients were suffering from dementia. Patients were also required to be ≥50 years old with nutritional problems or difficulty eating or swallowing, with cognitive impairment assessed by any validated tool. There were two primary outcomes: mortality and quality of life. Secondary outcomes included nutritional endpoints (including body mass index [BMI], weight, and albumin), healing of pressure sores, and adverse outcomes like aspiration pneumonia and local or systemic complications.

The review included seven observational studies, each of which was deemed at high risk of bias. No studies found an improvement in survival associated with tube feeding after adjustment for comorbidity, while two studies showed an increase in mortality with tube feeding. There were no studies addressing quality of life, and several did not report on key adverse outcomes. Evidence for effects on pressure ulcers was mixed. One study found a higher rate of aspiration pneumonia in tube-fed patients. Limitations of this review include the use of observational studies, individual study bias, and inclusion of studies with small sample sizes and/or nondementia-related cognitive impairment.

Consistent with these findings, the AGS³ and the AAHPM⁴ do not recommend percutaneous tube feeding in advanced dementia.

Family members who are invited to be present during a patient’s CPR have lower subsequent rates of posttraumatic stress disorder (PTSD), depression, and complicated grief. Hospital family presence policies are not associated with harmful effects on CPR efforts or patient outcomes.

Investigators in France randomized 15 prehospital emergency medical service teams (including an emergency medicine physician) to either an intervention or usual-care strategy for all adults with cardiac arrest undergoing resuscitation at home from October 2009 to November 2011.⁵ Teams in the intervention group used a communication guide to routinely and explicitly invite a single adult first-degree relative to witness CPR. The usual-care teams did not specifically invite relatives to witness CPR but permitted it if requested. The primary outcome was the prevalence of PTSD symptoms (using the Impact of Event Scale as assessed by a blinded psychologist) among the 570 selected family members at 90 days after resuscitation. Those who could not complete the assessment due to emotional distress were deemed also to have PTSD symptoms. Secondary outcomes included rates of anxiety and depression (based on the Hospital Anxiety and Depression Scale, HADS),
resuscitation efforts (such as duration of CPR or number of shocks), medical team emotional stress (on a visual analogue scale), and medicolegal claims. To assess the impact of family members actually observing CPR, several outcomes were also analyzed based on whether the CPR was witnessed or unwitnessed regardless of patient group assignment.

The rate of PTSD symptoms was higher among family members of patients resuscitated by the usual-care group (OR 1.7, 95% CI 1.2-2.5; P = .004). Symptoms of anxiety (P < .001) but not depression (P = .13) were also higher among family members of patients resuscitated by the usual-care group. Family members witnessing CPR was associated with lower rates of symptoms of PTSD (P = .01), anxiety (P < .001), and depression (P = .009). There were no differences between witnessed and unwitnessed episodes in terms of duration of CPR, number of shocks, survival to hospital admission, or medical team stress. There were no medicolegal claims in any of the cases at median 20 months follow-up.

In a longer-term 1-year follow-up study of the same population,⁶ family members of patients treated by the intervention group teams had a lower prevalence of PTSD (P = .02) and depression (P = .003) symptoms as well as complicated grief (P = .005).

Caveats of these studies include limited generalizability to in-hospital arrests and non-French settings.

A 2015 observational study assessed the effects of a hospital policy permitting family presence during resuscitation on patient-level outcomes among 41,568 patients with cardiac arrest at 252 hospitals participating in a nationwide registry. Patients were included if they were ≥18 years of age and had cardiac arrest. Patients were excluded if they had an implantable defibrillator or if their arrest occurred in an undocumented or procedural (such as a delivery room, cardiac catheterization lab, or electrophysiology suite) area of the hospital. The primary outcomes were return of spontaneous circulation (ROSC) and survival to discharge. Secondary outcomes included favorable neurologic status at discharge (no or only moderate disability, as opposed to severe disability, coma or vegetative state, and brain death), time to defibrillation, and duration of resuscitation. Given expected skew of findings between those who survived and those who did not, a number of these outcomes were examined separately in survivors and nonsurvivors. Linear and logistic regression models were adjusted for variables such as initial arrest rhythm, patient demographics, comorbidities, presenting illness
category, whether arrest was witnessed or unwitnessed, time from identification of pulselessness to compressions, and hospital characteristics (such as location, size, and teaching status).