Psychological Support Based on Positive Suggestions (PSBPS) on Mental Health Morbidity and Cognitive Function

Background:

Psychological morbidity is a major health problem in survivors of critical illness with no effective intervention to mitigate its burden. The increasing prevalence of critical illness combined with advances in intensive care medicine has led to a growing number of patients surviving to hospital discharge. However, resolution of critical illness does not equate with the end of a patient's struggles. The consequences of critical illness may persist long after the ICU stay and manifest in a number of new or worsening impairments in physical, cognitive, or mental health termed "post-intensive care syndrome" or PICS. Such impairments may persist for years and lead to increased re-hospitalization, health care costs, impaired quality of life, and inability to return to work. Thus, there is considerable interest in identifying causal factors and designing interventions to treat, and ideally prevent, PICS. Pioneering work has been done in the field of physical rehabilitation of ICU patients, demonstrating not only safety and feasibility, but improvement in physical and cognitive outcomes. Others have identified risk factors and interventional strategies for delirium and cognitive impairments. However, prevention of mental health morbidity remains unaddressed. Over a third of critical illness survivors manifest symptoms of anxiety, depression, and post-traumatic stress disorder (PTSD) 6 to 12 months post discharge. Reported rates for survivors of acute respiratory failure (ARF) have been even higher, with up to two thirds experiencing impairments in at least one psychiatric domain, most commonly in all three, persisting for years. There are no evidence-based interventions available to prevent or effectively treat mental health morbidity in such patients. Despite the staggering rates of mental health morbidity among critical illness survivors, research on targeting preventive measures to be employed in the ICU before development of psychologic symptoms has earned sparse attention. Most known risk factors such as female sex, alcohol abuse, unemployment, and premorbid psychopathology are not modifiable. Of those that can be modified, memories of frightening and delusional experiences from the ICU stay appear to be the strongest risk factor. Indeed, psychiatric symptoms such as anxiety, depression, and PTSD can be conceptualized as disorders of memory content, whereby the neural representation of the emotional component of the memory of a traumatic experience persists beyond any adaptive purpose. This causes dysregulation in the fear memory system, resulting in pathology associated with impaired cognitive, affective, and volitional functions. Once set, these emotionally charged delusional memories are difficult to treat.

To be effective, psychological support has to be provided early in the course of critical illness before the traumatic memories become permanent. Very little is understood about the nature of psychological dysfunction during an ICU stay, as the state of critical illness makes conventional psychological evaluation and intervention challenging. In critical illness, the thinking process changes, becoming literal and overly sensitive to both direct and covert meanings of communicated messages. In such altered states, people tend to accept suggestions without critical appraisal, becoming open to messages that help them interpret their situation, with the content of such suggestions accepted as subjective reality. Given this heightened suggestibility, Dr. Varga and colleagues in Hungary adapted therapeutic suggestion to the ICU setting, calling it psychological support based on positive suggestions (PSBPS). Therapeutic suggestion is a well-known intervention that has been used over a century and practiced in a wide variety of medical settings including anesthesia, surgery, palliative care, radiology procedures, chronic pain and somatization disorders, as well as obstetric and pediatric practice. Although a call to mindful choice of words with the critically ill given their highly suggestible state dates back to 1969, it wasn't until the work of Dr. Varga and colleagues that the effect of positive suggestion has been evaluated on outcomes of the critically ill. Authors felt that such patients may be experiencing panic, loss of control, and lack of privacy while being surrounded by complex medical technology and receiving mostly painful interventions and therefore experience psychological harm. Critically ill patients may also perceive staff that carries out painful interventions as a hostile adversary, which can further complicate care and impair outcomes. Such patients have also been seen to develop a state of learned helplessness due to their inability to control basic physiological needs. These issues could prevent patients from taking an active role during their crucial recovery phase. Therefore, any successful psychological intervention in an ICU setting has to incorporate the patient into the care team as a valued partner, tasked with assisting in recovery and actively participating in rehabilitation. Investigators have developed specific suggestions that could be used verbatim for patients on mechanical ventilation. The intervention has been associated with earlier ventilation weaning and decreased need for sedatives and analgesics. Subsequent analysis demonstrated that the earlier positive reframing was the most important factor, as the later longer communications did not really improve outcome. Indeed, research on the formation of fear memories demonstrates that one can interfere with the initial process of fear conditioning during the temporal window when initial memory consolidation occurs. Furthermore, if mitigating information about a traumatic event is introduced during the time following memory formation and its subsequent recall, called the reconsolidation window, the emotional experience of the memory can essentially be rewritten. Therefore, a potential approach to treatment may be to interfere with the initial process of fear conditioning by introducing mitigating information during the temporal window when the consolidation of the initial memory occurs and also following its initial recall. This can potentially weaken or prevent the formation of a traumatic emotional memory by positively shaping its consolidation and reconsolidation, because at this time memories are labile and susceptible to alteration; however, the longer a memory has been stored in the brain, the more difficult it is to destabilize and alter it, explaining the lack of effectiveness of outpatient interventions. For the critically ill this would mean introducing psychological support as soon as feasible following admission and providing it in parallel with medical care in the ICU. Studies on PSBPS provided intriguing preliminary data for further evaluation as this form of behavioral therapy can be performed with patients regardless of their level of participation and therefore can be started early in the ICU course including during sedation. Notably, semantic processing continues in an altered states of consciousness, with some patients later recalling things said to them even while they were comatose.

Methods:

Design. This will be a two arm single-center, prospective, randomized, controlled trial comparing the effect of PSBPS vs. no PSBPS on psychological and clinical outcomes in the critically ill population. PSBPS will be administered by trained ICU doulas to eligible patients.

Eligible patients will be randomized at the earliest possible time after meeting inclusion criteria. Patients will be randomized 1:1 to intervention by ICU doulas or a non-intervention arms stratified by ICU location. Both intervention and controls will receive standard medical care at the discretion of the ICU team. The intervention arm will receive PSBPS as soon as feasible and continue on a daily basis. ICU doulas will also liaise with family, the team, and with the nursing staff. Both intervention and standard of care arms will complete follow-up questionnaires measuring symptoms of anxiety, depression, acute stress and cognitive function after ICU discharge and at 3 months follow up by phone. These questionnaire results will not be entered into the medical record. The intervention will happen once a day. All the patients randomized to the intervention arm will receive the same intervention. Length of the intervention will depend on the clinical situation as in our prior experience it is sometimes possible to spend 20+ minutes at bedside, and sometimes patients have conflicting procedures and tests and the intervention may need to be shorter.

Retention strategy: We have minimized questionnaire length and frequency to reduce patient burden. We plan to employ additional successful retention strategies described in the literature, such as emphasizing study benefits, by having patients experience the importance of adding an early humanistic touch to a critical care setting, and having the benefit of a well-functioning, organized, and persistent anesthesia clinical research unit research team that would employ systematic methods of patient contact. The nature of our intervention would also allow patients to partner with the research staff. We will obtain all available phone numbers by which participant may be reached to facilitate follow up. Additionally, patients will receive financial compensation after completing the 3 months follow up questionnaires.

Sample Size and Power:

A 2-point difference in HADS-A would be meaningful and relevant to detect based on prior studies. In our preliminary study, we observed a mean baseline HADS-A score of 7.7 (SD = 3.9) post ICU discharge. Assuming a pooled standard deviation of 4.0, n=64 patients per arm would achieve 80% power to detect a difference of 2 units using a two-sided two-sample t-test at alpha=0.05. At follow up, we observed a mean HADS-A score of 5.3 (SD = 4.0); using the same pooled standard deviation, the sample size calculation is identical to the baseline HADS-A comparison. Given the volume of patients observed in ICUs at our center, anticipated doula availability, and timeline, we believe that a study targeting enrollment of 64 patients per arm (128 total) with 3-month follow up is feasible. To accomplish this, we plan to enroll until 182 patients (91 per arm) have been randomized and have completed the baseline post-ICU discharge HADS-A assessment (conservatively anticipating 30% dropout at 3 months, which we think may be reduced through the retention strategy). This sample size provides >80% power for the early post-ICU discharge comparison. Additional secondary endpoints include HADS-D, IES-R, and MoCA at baseline and 3 months. Clinical endpoints include ventilator-free days (through 14 days after randomization), ICU and hospital-free days.

Clinical endpoints will include all randomized patients, even if not discharged alive from the ICU. Overall, we expect this sample size will provide adequate power for the analysis of early post-ICU and 3-month clinical and psychological outcomes and, more importantly, will provide strong data to motivate a subsequent multicenter and well-powered study of long-term cognitive and clinical outcomes.

Analysis plan: Demographic characteristics will be summarized by mean, median, SD, and IQR (continuous variables) and counts and percentage (categorical variables). The effect of intervention on outcomes will be estimated using ANCOVA with adjustment for relevant pre-randomization characteristics identified in our earlier studies that would improve power estimates. A non-parametric Wilcoxon rank-sum test may be used if appropriate. In order to administer the intervention to subjects on each day while in the ICU, we anticipate most subjects in the intervention arm will have contact with >1 doula during their ICU stay. Thus, we do not plan to address possible correlation of patient responses arising from a shared intervention experience with the same intervention doula. However, in the event we observe most patients in the PSBPS arm receive their intervention from a single doula during the entirety of their stay, additional sensitivity analyses may be considered, such as a stratified analysis. Interactions with relevant variables, including sex and race/ethnicity, will be assessed to determine if the effect of intervention differs in key subgroups; however, this study is not specifically powered to detect such effects. We anticipate little missing data with respect to patient demographics and baseline characteristics abstracted from the electronic health record (EHR). Similarly, clinical outcomes including ventilator-free days and length of stay are well-collected in the EHR. For cognitive outcomes, the reference population is patients discharged alive from the ICU. Baseline assessments may be missing for subjects unable to complete them or refusing to complete them after ICU discharge and 3-month assessments missing for patient mortality and loss to follow-up. We anticipate missing baseline assessments will be rare, but it is possible that these missing data will be not missing at random (NMAR). The initial analysis will use complete case data and sensitivity analyses will consider imputation of missing baseline data. Imputation may include multiple imputation under the missing at random mechanism as well as worst-case imputation assuming NMAR as those with missing data may be more likely to have poor (unobserved) outcomes. Analyses of 3-month outcomes will proceed similarly, as missing 3-month assessments are also possibly missing not at random. The primary analysis will be based on the intent-to-treat principle. A two-tailed p<0.05 will be considered statistically significant.