Managing Acute Pain in Critically Ill Non-communicative Palliative Care Patients

Overview

The purpose of this project is to test an innovative method for managing pain in acutely ill hospitalized patients who are not able to report their pain verbally to health care professionals. Nurses will use a Pain Assessment and Intervention for the Non-communicative (PAIN) Algorithm to guide assessment of pain, selection of pain medications, and management of medication side effects. The researchers will evaluate whether patients who are managed with the PAIN Algorithm have less severe pain and increased use of pharmacologic pain management strategies than those who are not managed with the PAIN Algorithm. The study design is a non-randomized quasi-experimental cohort design with two cohorts who will be sequentially studied. In phase 1, patients will comprise the usual care group (UCG), or control cohort, defined as receiving pain assessment and management practices that nurses are currently performing on the study units. In phase 2 the PAIN Algorithm coupled with analgesic order sets will be introduced to nurses and physicians on all participating units as the intervention. Patients enrolled in this phase will be considered the intervention group (IG), also called the experimental cohort. Nurses will be enrolled from the participating inpatient units to provide data on the clinical utility of the PAIN Algorithm

The Multiple Principal Investigators (MPIs) are Dr. Deborah McGuire (VCU) and Dr. Carl Shanholtz (University of Maryland. University of Maryland Medical Center (UMMC) delivers palliative care to acutely ill persons across a range of ages, medical illnesses, and traumatic injuries for whom death is not a foregone conclusion. Data will be collected on eight inpatient acute care nursing units, which are grouped into services defined by medical specialties and patients: Medical Intensive Care Unit, Medical Intermediate Care Unit (IMC), Neuro-Trauma IMC Unit, Neuro-Trauma Critical Care Unit, Multi-trauma Critical Care Unit, Select Trauma IMC Unit, Select Trauma Critical Care, and Surgical Intensive Care Unit. Average lengths of stay are highly variable but are generally a week or less. This is a one site, quasi-experimental cohort control group design, conducted in two sequential phases with repeated measures in two different but comparable cohorts. The primary aim is to test whether a pain algorithm that incorporates the Multi-dimensional Pain Assessment Tool (MOPAT) and an analgesic order set improves pain severity and use of pharmacologic pain management strategies in critically ill non-communicative palliative care patients who are hospitalized on medical, surgical, and trauma intensive care units when compared to patients without the algorithm. The secondary descriptive aims are to: (S1) compare pain severity and use of pharmacologic pain management strategies in patients with and without concurrent pain-related conditions, (S2) describe the pattern of patients' pain over time, and (S3) evaluate nurses' perceptions of clinical utility of the pain algorithm. This study has two samples: non-communicative palliative care patients who have acute pain and nurses who will use the PAIN Algorithm and order set to manage their patients' pain. Patients will be 300 critically ill adults who are non-communicative for a variety of reasons (intubation, neurological impairment, etc.) and have conditions that are known to produce acute pain. Patient data collection will be collected for 7 days (day 1 to 7) or until the patient dies, regains the ability to self-report pain, is transferred to a non-participating unit, or is discharged, whichever comes first. This timeframe is based on average lengths of stay for non-communicative patients on the units, quality of care benchmarks for pain relief in self-reporting patients, and practical considerations for research in acutely ill individuals. Volunteer Staff Nurse data will be collected at baseline and monthly over 24-36 months. Study Phase 1 will occur before introduction of the pain algorithm which occurs in Phase 2. Prior to starting this phase, nurses will be trained to use the MOPAT and it will be incorporated into the pain standard of care and the electronic medical record (EMR), permitting comparison of pain assessment data between Phases 1 and 2. The MOPAT will replace the Checklist of Nonverbal Pain Indicators (CNPI), which is currently used to assess pain in non-communicative patients but has multiple limitations. Following a 6 week run-in period in which nurses' appropriate and universal use of the MOPAT is assessed and assured, data collection will begin. Patients in this cohort will comprise the usual care group (UCG), or control cohort. Nurses will also be enrolled from the participating inpatient units. Usual care is defined as all pain assessment and management practices that nurses are currently performing on the study units, including inconsistent use of existing algorithms and protocols. Data will be collected on patients and nurses until we accrue 150 patients. Study Phase 2 will begin following the completion of Phase 1. Prior to starting this phase, our pain algorithm, christened the Pain Assessment and Intervention for the Non-communicative (PAIN) Algorithm, will be introduced to nurses and physicians on all participating units and they will be trained in its use. The PAIN Algorithm couples the MOPAT with an analgesic order set. The final algorithm will have specific numerical cut points derived from a consensual process. In addition, orders in the order set, and adaptations to accommodate patient demographic and medical variables will be finalized through a detailed collaborative process. After a 6 week run-in period to assure appropriate use of the PAIN Algorithm, data collection will begin. Patients enrolled in phase 2 will be considered the intervention group (IG), also called the experimental cohort. We will collect the same patient and nurse data as in Phase 1, with the addition of nurse perceptions of clinical utility of the PAIN Algorithm, and will continue until we have accrued 150 patients. Considerations regarding potential risks to patients are as follows. Because the algorithm includes opioids, and patients targeted in this study will be critically ill with numerous pathological processes that could be adversely affected by these drugs, there is concern about adverse opioid-related side effects such as sedation and respiratory depression. To deal with this concern, the PAIN Algorithm order set will include orders for routine monitoring of these side effects and interventions to manage them, for example, titrating the opioid dose or using a reversing agent in the case of severe respiratory depression. To ensure that these orders conform to standards for safe practice, the interdisciplinary panel that develops the analgesic order set in conjunction with the researchers will build in drug side effect assessment, treatment, and reassessment. Another area of potential risk, however, remains of concern. Since the MOPAT instrument is a relatively new pain assessment tool, and medication decisions will be made using MOPAT Behavioral Dimension scores, there is the possibility that treatment decision may under-medicate, or overmedicate a patient's pain. Thus, even with careful attention to drug side effects, the study is probably greater than minimal risk. There are no alternative treatments and procedures, since the PAIN Algorithm will be integrated into clinical practice as the standard of care while the patient is on study. There are no anticipated potential risks to the volunteer study nurses who consent to participate in the study. Since the PAIN Algorithm with the analgesic order set will be incorporated into standard of care all nurses on the inpatient units will use it whether or not they consent to participate in the study. Those nurses who consent will provide demographic and practice data, and provide monthly appraisals of usefulness of the MOPAT in Phase 1, and the PAIN Algorithm (including MOPAT) in Phase 2 and will not be affected by the study intervention. Because there is always a risk of opioid-induced side effects when opioids are used to manage pain, the PAIN Algorithm analgesic order set includes provisions for monitoring, detecting, and managing such side effects thus making the algorithm potentially safer than usual care which does not generally include side effect monitoring. There are no physical or psychosocial risks to nurses who consent to participate in the study. For both patient and nurse subjects, protection of confidentiality is a concern. Therefore, all patient and nurse data collection forms will contain only the unique study identification numbers, with the master list kept in a locked file cabinet in a locked project office. When the study is completed, the list will be destroyed and all data will be identified only by study identification numbers. Only the PI and project manager will have access to these lists during the study. Standard operating procedures for analysis activities, such as recruitment, data collection, data management, data analysis, reporting of adverse events, and change in management are outlined in the study manual of operating procedures. Source data include: electronic medical records, patient's paper charts, data collection "teleforms" . The manual of operations also includes standard operating procedures for data entry, transfer, and quality assurance. "Teleforms" have a built in quality checks including predetermined rules for range and consistency with other study data fields. The plan for statistical analysis is as follows.The primary aim testing that there is a decrease in pain severity for the intervention group as compared to the usual care group, will use the interaction F-test from 2 x 4 repeated measures analysis. If test assumptions are not met, or if there is missing data, mixed linear modeling will be used that does not require sphericity or complete data. Specific comparisons also will be made between the groups at both 2 and 4 days using interaction contrasts. The number and total dosage of pharmacologic interventions will be computed for each time period using the American Pain Society equi-analgesic table.To test for an increase in the number of pharmacologic agents used for pain and an increase in the total equi-analgesic dosage of pharmacologic agents used for pain a 2 X 3 ANOVA or mixed linear modeling approach will be used depending on whether assumptions are met. To address covariates and control for unit-based effects, first a comparison of units to identify unit-level differences, such as type of service or ambient environment, that might be related to patient outcomes and then include them in a hierarchical linear model will be conducted. Patient demographic variables such as age and gender and clinical variables such as pain-related conditions, type of pain and co-analgesics, and Glasgow Coma Scale score will be examined for their relationship to outcomes and incorporated as factors or covariates depending on the level of measure and strength of bivariate relationship. These variables can be incorporated into a hierarchical model or, if unit-level differences are not relevant, then into analysis of covariance or multiple regression models. The secondary aims will be analyzed using descriptive approaches. For (S1) patients will be categorized on the basis of whether they have pain-related conditions using the Clinical Classification System and a 2 x 4 repeated measures analysis will be used to compare these two groups on pain severity and a 2 x 3 repeated measures analysis to compare the number and types of drug categories and total dosage of pharmacologic agents. If there is a significant interaction effect, the groups will be compare at each time point as post-hoc analyses. For (S2) a descriptive pattern of patients' pain over time will be examined. Of interest is whether there are sub-groups of patients who demonstrate different patterns. A graphic techniques (e.g., spaghetti plots) will be used to identify sub-groups and attempt to describe them based on clinical and demographic variables. In addition, we will compute the change in pain between the time periods and plot the cumulative distribution function at each time point as an initial step in identifying potential cut points to define a clinical important change. To evaluate nurse perceptions of the clinical utility of the PAIN Algorithm and the MOPAT (S3), we will examine the frequency distribution for responses to each item on the Clinical Utility Questionnaire (CUQ) and the overall summated response. For CUQ items that are used in evaluating MOPAT utility in both phases, the responses will be compared using t-tests. In addition, the percentage of nurses who agree or strongly agree with each statement and make comparisons based on nurse demographic and practice characteristics will be done.