Maximizing Intensive Care Bed Utilization While Maintaining Pediatric Patient Safety and Quality of Care
November / December 2006
Feature Article
Maximizing Intensive Care Bed Utilization While Maintaining Pediatric Patient Safety and Quality of Care
For clinical leaders involved in decisions about patient flow and intensive care bed utilization, these are challenging times. Capacity management and the allocation of intensive care unit beds are frequently debated as hospitals grapple with how many beds are needed for the provision of clinical care characterized as high quality, safe, cost effective, and efficient.
In 2001, the demand for pediatric intensive care services continued to exceed the supply of the 18-bed pediatric intensive care unit (PICU) at Children’s Hospital Boston. As a result, a significant number of critically ill transports were being diverted to other hospitals including in different states. In 2001, 234 critically ill patients were diverted, followed by 203 patients in 2002 and 171 patients in 2003.
Three problems required immediate solutions: 1) inability to admit surgical patients for specialized procedures, 2) consistent bottleneck of patients in the PICU because of length of stay, and 3) insufficient bed capacity for the number of patients needing pediatric intensive care services. Since structural limitations prevented an expansion in PICU beds, senior leadership approved a proposal to increase the bed capacity of the neonatal intensive care unit (NICU) from 20 to 24 beds and to use these 4 beds, referred to as the PICU satellite, to care for PICU overflow patients jointly by the PICU and NICU staff.
In the fall of 2003, the Intensive Care Unit Governance Committee revised the patient criteria for admission to the NICU to include patients up to 6 months of age. This organizational change required that the NICU staff become primarily responsible for this new patient population. Along with the recreated NICU admission criteria, guidelines for the care of these infants within the NICU were developed to assist with communication and management of these patients, and increased nursing resources were implemented.
The purpose of this study was to evaluate patient safety, quality of nursing care, resource utilization, and patient outcomes for these infants cared for in the NICU, in comparison to both historical and contemporaneous infants eligible for admission to the NICU, but admitted to the PICU. A secondary purpose was to continue to monitor the number of critically ill transports that were being diverted to other hospitals during this time of reorganized care.
Methods
Setting
Children’s Hospital Boston is a 325-bed tertiary academic hospital with three intensive care units including an 18-bed PICU, a 24-bed neonatal intensive care unit (NICU), and a 23-bed cardiovascular intensive care unit (CICU). The reorganization in care only involved the PICU and NICU. The NICU is a Level III referral center for critically ill premature newborns and infants requiring complex medical and surgical care. The PICU is a pediatric intensive care unit caring for patients ranging in age from newborn to early adult with a variety of medical and surgical needs.
Description of Reorganization
ICU admission data were reviewed for the period of 8/1/02 to 7/31/03 to identify all patients less than 6 months of age, cared for in any of the institution’s ICUs. It was estimated that approximately 1,400 patient days of ICU care could be transferred to the expanded NICU.
New admission criteria for the NICU were created. Newborns and patients less than 6 months of age with surgical diagnoses were included, as well as infants with most medical diagnoses. The only exclusion criteria were newborns with congenital diaphragmatic hernia or meconium aspiration requiring extracorpeal membrane oxygenation (ECMO). Guidelines for the care of patients within the NICU were developed to assist with physician communication and patient management. Daily meetings of the interdisciplinary staff from each unit were convened and a competency-based nursing education plan was implemented to ensure skill acquisition and competence of the NICU nurses.
Study Design
Comparing process and outcomes of care for patients admitted to the NICU to similar patients admitted contemporaneously and historically to the PICU, a retrospective cohort study was performed.
Patient Selection
Patient inclusion reflected the population that was relocated from the PICU to NICU. Patients ≤7 months of age directly admitted to the NICU from 1/1/04 to 4/30/04 acted as the cases. Patient admissions to the PICU age ≤7 months were considered the controls; the PICU historical controls (control group 1) were admitted from 1/1/03 to 4/30/03 and the contemporaneous controls (control group 2) were admitted from 1/1/04 to 4/30/04. All term infants, or if premature > 44 weeks corrected gestational age, up to 7 months of age were included. Patients who were admitted to PICU and then later transferred to NICU were excluded. Patients admitted with congenital diaphragmatic hernia were also excluded, as patients with this diagnosis are not cared for in the NICU.
Data source
All data elements were extracted from the electronic medical record (Eclipsys®) or from the Virtual Pediatric Intensive Care Unit (VPICU Performance System®), a quality-monitoring database that contains comprehensive information on a child’s entire patient experience in a pediatric intensive care unit, from admission through discharge. Using primary admitting diagnoses, patients were grouped by medical or surgical diagnoses, and further characterized as acute or chronic condition. In addition, secondary acute conditions as well as history of prematurity were also noted (Table 1).
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Nursing Activity and Nursing Quality of Care
Nursing workload in both units was described using the Nine Equivalents of Nursing Manpower Use Score (NEMS). As listed in Table 2, there are nine identified patient treatments, which are used as surrogates of nursing activity. The NEMS, a simplification of the Therapeutic Intervention Scoring System, is a well-validated instrument used to predict the workload and planning of nursing staff allocation at the individual patient level (Miranda, Nap, de Rijk, Schaufeli, & Iapichino, 2003; Reis Miranda, Moreno, & Iapichino, 1997). However, this instrument has not been widely used in neonatal and pediatric populations.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The quality of nursing care was determined by the documentation of 12 nursing assessment and care elements in the first 24 hours of admission. These elements are considered a standard of care within the institution and include:
- documentation of the patient’s weight, length, and head circumference, skin assessment, and pain history on admission;
- positional changes, pain assessment using an appropriate pain instrument Face, Legs, Activity, Cry, and Consolability (FLACC), and sedation assessment using the Modified Motor Activity Assessment Scale (MMAAS) every 4 hours;
- nutritional plan, gastrointestinal prophylaxis if NPO, note of parent presence, and a nursing progress note within the first 24 hours of admission.
Also assessed was the time to critical intervention; specifically the time interval between the first identification of patient distress to the time of the first intervention to manage that distress. Eight phenomena were assessed: fever, hypotension, desaturation, pneumothorax, respiratory distress, gastric aspiration, low urine output, and seizures.
Patient Safety
Patient safety was assessed by the presence of institutional incident reports filed within the identified study time periods. Incidents were assessed for three levels of seriousness:
- Not serious — results in minimal transient impairment of a body function or damage to body structure and does not require any intervention than monitoring;
- Moderately serious — results in moderate transient impairment of a body function or transient damage to a body structure and requires intervention, such as administration of medication to prevent permanent impairment of a body function or damage to a body structure; and
- Serious — considered to be life-threatening, resulting in permanent impairment of a body function or permanent damage to a body structure or necessitates significant intervention, (such as major surgery, to prevent permanent impairment of a body function or permanent damage to a body structure), whether a treatment was required as a result of the incident, and whether the patient suffered any sequelae as a result of the incident.
Any incident identified for a case or control was described, categorized by event type, and linked to patient outcomes.
Resource Utilization, Disposition at Discharge, and Patient Outcome
Resource utilization was measured by length of stay, defined as stay in hospital from admission to discharge, and number of clinical services utilized. Attending service of record, general surgery, or pediatrics was also examined. Disposition at discharge and patient outcome was determined by discharge location or disposition from the intensive care unit. Responses were pediatric floor, home, transferred to another institution, or death.
Diversion of critical care transports
The number of critical care transports diverted due to unavailable intensive care beds were tabulated yearly from 2000 to 2004.
Risk Adjustment
Patient mortality risk was estimated using the Paediatric Index of Mortality Index (PIM2). PIM2, first developed in 1997 as PIM and then revised in 2002, is the only open-sourced risk prediction algorithm available using data collected in the most recent decade in pediatric intensive care. PIM2 is a mandatory field in the VPICU (Ruttimann et al., 2000). Patient outcome is predicted using 10 clinical data points examined within the first hour of a patient’s arrival to the intensive care unit. These include systolic blood pressure, pupils >3mm and fixed, PaO2/ FiO2, base excess, mechanical ventilation in 1 hour, elective admission, recovery from surgery/procedure, post cardiopulmonary bypass, admitting diagnosis high risk (i.e. post cardiac arrest, severe combined immune deficiency), or admitting diagnosis is low-risk (i.e. asthma or croup). This index has been used and validated in multiple pediatric intensive care settings in Australia and the United Kingdom (Slater et al., 2003; Shann et al., 1997) but not in the United States.
Statistical Analyses
Patient characteristics, resource utilization, patient outcomes, incident reports, and measures of quality of nursing care were compared for cases and the two control groups using Fisher’s exact test for categorical variables and the nonparametric Kruskal-Wallis test for continuous variables. Logistic regression analysis was used to explore the relationship between case/control status and death, using cases as the reference group; adjusting for case mix severity using the PIM2 score. Similarly, linear regression analysis was used to examine the association between case/control status and the continuous outcome NEMS, adjusting for PIM2 score.
Results
Summary of Cases and Controls
Of the 137 patient admissions examined (NICU: 53, PICU historical group: 51, and PICU contemporaneous group: 33), there were no significant differences between the NICU cases and PICU cohorts regarding gender, median age at ICU admission, admitting or acute diagnosis type, and diagnosis of prematurity (Table 1). However, patients admitted to the NICU were noted to have a significantly greater percentage of acute secondary (40% vs. 22%, and 3%, p<0.001) and chronic (66% vs. 27%, and 42%, p<0.001) diagnoses recorded as compared to the PICU admissions. The median PIM2 scores were comparable for NICU cases and PICU controls, ranging from 0.26 to 0.30, with an overall minimum among the 3 groups of 0.02 and a maximum of 99 (Table 1).
Nursing Activity and Quality of Nursing Care
Using the Nine Equivalents of Nursing Manpower Use Score (NEMS) to evaluate nursing activity revealed no significant difference between the cohorts (Table 2). The median NEMS score ranged from 15-18 with a minimum of 0 and a maximum of 45. The relationship between NEMS score and patient status was further examined adjusting for risk of mortality using the cases as the reference group. Although the NEMS was higher for the controls, this was not found to be statistically different (Table 3).
|
|||||||||||||||||||||||||||||||||||||
Quality of nursing care was evaluated by documentation of 12 elements of nursing assessment and care in the first 24 hours of admission. Of the 12 elements examined, measurement of patient length and head circumference and pain history and assessment were documented more frequently in the NICU versus the PICU (Table 4).
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Only 7 episodes of critical intervention were identified. Specifically there were 2 episodes of respiratory distress, 2 episodes of seizures, 2 episodes of desaturation, and 1 episode of fever reported. The response time for episodes of respiratory distress, seizures, and desaturation were all within 5 minutes. The response time for the episode of fever was within 1 hour. There were no reported incidents linked to these episodes.
Patient Safety
Twenty incidents were reported among the three cohorts (20/137, 15%). Seven categories were identified. Overall, incidents related to pharmacy dispension were found to be the most frequent (n=5) followed by incidents involving IV fluids (n=4) and self-extubation (n=3) of patients. The cohorts did not differ in the type or frequency of incident reports. All reported incidents were assessed to be mild to moderately serious and not found to result in any patient sequelae.
Resource Utilization
Median length of stay was noted to be 4 days in the NICU group (range 1-89 days) versus 2 days in each PICU group (range 0-90 days) (P<0.001). The median number of clinical services utilized by NICU was 2 (range 1-8) and was also 2 for the PICU historical cohort (range 0-6), while PICU contemporaneous cohort used a median of 1 service (range 1-5) (p=0.02) (Table 5).
|
|||||||||||||||||||||||||||||||||||||||||||
Disposition at Discharge and Patient Outcome
Disposition at discharge was found to be significantly different between among the cohorts (p<0.001), with approximately half of the NICU cases discharged directly to home while most of the PICU infants were transferred to a floor setting before discharging to home. It was also noted that a higher percentage of NICU infants (15% vs. 2%, and 6%) were transferred back to originating facilities as compared to the PICU infants (Table 5).
There was no significant difference found in the occurrence of death among the three groups (Table 5). In the 9 patients where death was recorded as disposition at discharge, death was not found to be related to care received or as a result of an adverse event but instead related to admitting diagnosis or secondary chronic diagnosis.
Intensive Bed Utilization
Critical care transport diversions continued to fall during this time of reorganized care from 171 in 2003 to 112 in 2004.
Discussion
Over the past decade, pediatric intensive care units have experienced a dramatic increase in the number of available beds nationally (Ruttimann et al., 2000). Decisions about the appropriate number of intensive care beds for a single institution are complex and involve micro- and macro-economics as well as justification of the need for expensive resources. Past evaluation of ICU bed utilization in this population of pediatric and neonatal patients has focused on cost containment, resource utilization, and efficiency of care (Pollack et al., 1988; Pollack, 1994). Researchers have concluded that resource use and efficiency of care are directly related to severity of illness or risk of mortality (Pollack, 1994; Pollack & Koch, 2003; Pollack et al., 2000; Pollack & Patel, 2005). Currently no data are available for comparing the setting of care in a NICU or PICU.
The results of this study highlight one institution’s strategy for evaluating a reorganization of care and the safety and quality of nursing care surrounding this change. Evaluation of resource utilization, clinical diagnosis and status, and patient outcomes revealed no clinical differences in the care received between the two units.
Intensive care unit monitoring is more frequent, invasive, and technologically complex. Our assessment using the NEMS score to evaluate nursing activity revealed no significant difference between the NICU and PICU groups. In the groups of patients studied, the median NEMS score ranged from 15 to18 with a minimum of 0 and a maximum of 45. This finding reveals a group of patients requiring a low to moderate level of nursing manpower hours, as the maximum score one can achieve is 56. This finding of low nursing manpower is further validated by the PIM2 scores suggesting low risk overall for mortality.
Critically ill infants receive care from a large multidisciplinary team. Outcomes for this population cannot be singularly credited to any one member but instead depend on the efforts of the whole team. In particular, there is a paucity of information describing nursing workforce characteristics and nursing-sensitive outcomes for these infants. We evaluated nursing quality of care in terms of documentation of selected elements available in the electronic medical record charting system. Although our findings revealed few differences between the two units, documentation was noted to be higher in the NICU areas of growth and pain assessment. While assessment of patient growth is inherent to the practice of neonatal care, pain assessment is common in both types of intensive care units. Improvements in the pain documentation between the PICU cohorts can be attributed to a major institution effort aimed at improving pain assessment and management between 2003 and 2004.
Nursing assessment, otherwise termed patient surveillance or vigilance, has been identified as a factor consistently lowering patient mortality and increasing quality of care when the relationship between organizational structures/processes and patient/mortality/ adverse events has been examined (Mitchell & Shortell, 1997). Ongoing nursing surveillance has been described as a function to detect early changes in patient status or the advent of an adverse event whereby the nurse initiates activities to “rescue” the patient and restore health. “Failure to rescue” is said to occur when the patient changes or adverse events are not detected. This concept of failure to rescue has been tested and validated as an indicator of the quality of care delivered for surgical patients in acute care hospitals (Silber et al., 1992). However, testing of this concept of “failure to rescue” in a pediatric population revealed inaccuracies and were recommended not to be used to estimate pediatric quality of care and assessment of preventable adverse events (Sedman et al., 2005).
Data were collected on 8 phenomena that would potentially result in patient distress and require a timely critical intervention to prevent further patient decline. Due to the small sample size of those identified as having the need for critical intervention, we were unable to make any conclusions as to whether this was related to a nursing assessment failure. Identification of “failure to rescue” indicators in an acutely ill infant population would be an important tool in assessing quality of care of patients in an acute care setting.
Our evaluation of patient safety using hospital incident reports did not reveal any differences in type or frequency of occurrence between the patient groups. Overall for our sample size of 137 this was noted to be 15%. Of note, pharmacy/medication related events accounted for the most of the events reported, which is supported in the literature by others evaluating the nature and frequency of errors in the pediatric population (Slonim et al., 2003; Kaushal et al., 2001).
Resource utilization between the two units was found to be significantly different with respect to median length of stay and clinical services utilized. The NICU cases had a length of stay approximately twice as long as the PICU controls. However, this finding is related to the institution practice of NICU patients remaining in the NICU until they are able to be discharged to a community hospital nursery or to home, whereas the PICU patients are more frequently transferred to a non-ICU setting prior to discharge home. Although more clinical services where utilized in the NICU, the NICU cases were found to have more acute and chronic diagnosis reported.
Patient outcomes, specifically occurrence of death, were all found to be related to the severity of admitting diagnosis and not a result of documented care or incident. Use of risk-adjustment was useful in the analysis; as noted, in most cases the risk for mortality using the PIM2 was quite low. However, the odds ratios of both control groups doubled once adjusting for risk of mortality showing a trend toward significance. These findings might be due to our limited sample size of available cases meeting inclusion criteria.
Critical care transports diversions continued to decrease during this period of reorganized care, suggesting that the institution attained its goal of improving ICU capacity.
Limitations
Study limitations include the retrospective analysis of data contained within the medical record. The work of direct-care nursing includes both visible and invisible activities (Star & Strauss, 1999). Visible activities such as assisting patients with physical activity, administering medications and treatments, and educating patients and their families about disease and therapies were captured using the electronic medical records. However, invisible or cognitive work nurses perform, which is learned through formal education and developed through mentoring and clinical experience, is the most critical component of ensuring quality care for patients. Unfortunately our data source did not allow for this critical assessment and should be an area of future consideration.
Using the identified eligibility criteria, our sample size was limited in restricting our proposed analysis of “time until critical intervention.” Despite these limitations, our analyses of patient safety and quality of care yielded important information.
Conclusion
Although documentation of nursing assessment was higher in the NICU, no clinical differences in care received or patient outcomes were noted between the NICU cases and PICU controls. This strategy of ICU bed reorganization was considered to meet both quality of nursing care and safety as defined in this analysis. Critical care transport diversions continued to decrease and were not adversely impacted by this reorganization of care. Studies focusing on the development of valid and reliable nurse-sensitive indicators specific to infant and pediatric populations are needed to better assess resource utilization, patient safety, quality care, and patient outcomes.
Jean Connor is the co-director for Cardiology Clinical and Regulatory Group, nurse scientist for the Cardiovascular Program, and a faculty member for the Program of Patient Safety& Quality at Children’s Hospital Boston. She may be contacted at Jean.Connor@cardio.chboston.org.
Martha Curley is the director of Cardiovascular and Critical Care Nursing Research, Children’s Hospital Boston, and assistant clinical professor of anesthesia at Harvard Medical School.
Kathy Jenkins is a senior associate in cardiology, Children’s Hospital Boston, and associate professor in pediatrics, Harvard Medical School. She is also the director for the Cardiology Clinical Research and Regulatory Group and director for the Program for Patient Safety and Quality Children’s Hospital Boston.
Kimberlee Gauvreau is a research associate in cardiology, Children’s Hospital Boston, and assistant professor of biostatistics, Harvard School of Public Health. She is also a faculty member for the Program for Patient Safety & Quality at Children’s Hospital Boston.
Patricia Hickey is the vice president for Cardiovascular and Critical Care Services at the Children’s Hospital Boston and oversees all clinical nursing services for each of the three ICUs. She is also the co-chair of the Program for Patient Safety & Quality Implementation Committee charged with implementation of all major quality initiatives.
References
Reference
Kaushal, R., Bates, D. W., Landrigan, C., McKenna, K. J., Clapp, M. D., Federico, F., et al. (2001). Medication errors and adverse drug events in pediatric inpatients. JAMA, 285(16), 2114-2120.
Miranda, D. R., Nap, R., de Rijk, A., Schaufeli, W., & Iapichino, G. (2003). Nursing activities score. Crit Care Med, 31(2), 374-382.
Mitchell, P. H., & Shortell, S. M. (1997). Adverse outcomes and variations in organization of care delivery. Med Care, 35(11 Suppl), NS19-32.
Pollack, M. M. (1994). Cost containment: the pediatric perspective. New Horiz, 2(3), 305-311.
Pollack, M. M., Katz, R. W., Ruttimann, U. E., & Getson, P. R. (1988). Improving the outcome and efficiency of intensive care: the impact of an intensivist. Crit Care Med, 16(1), 11-17.
Pollack, M. M., & Koch, M. A. (2003). Association of outcomes with organizational characteristics of neonatal intensive care units. Crit Care Med, 31(6), 1620-1629.
Pollack, M. M., Koch, M. A., Bartel, D. A., Rapoport, I., Dhanireddy, R., El-Mohandes, A. A., et al. (2000). A comparison of neonatal mortality risk prediction models in very low birth weight infants. Pediatrics, 105(5), 1051-1057.
Pollack, M. M., & Patel, K. M. (2005). The volume-outcome relationship in pediatric intensive care units: a conceptual advance or a statistical novelty? Pediatr Crit Care Med, 6(2), 227-228.
Pollack, M. M., Ruttimann, U. E., & Getson, P. R. (1988). Pediatric risk of mortality (PRISM) score. Crit Care Med, 16(11), 1110-1116.
Reis Miranda, D., Moreno, R., & Iapichino, G. (1997). Nine equivalents of nursing manpower use score (NEMS). Intensive Care Med, 23(7), 760-765.
Ruttimann, U. E., Patel, K. M., & Pollack, M. M. (2000). Relevance of diagnostic diversity and patient volumes for quality and length of stay in pediatric intensive care units. Pediatr Crit Care Med, 1(2), 133-139.
Sedman, A., Harris, J. M., 2nd, Schulz, K., Schwalenstocker, E., Remus, D., Scanlon, M., et al. (2005). Relevance of the Agency for Healthcare Research and Quality Patient Safety Indicators for children’s hospitals. Pediatrics, 115(1), 135-145.
Shann, F., Pearson, G., Slater, A., & Wilkinson, K. (1997). Paediatric index of mortality (PIM): a mortality prediction model for children in intensive care. Intensive Care Med, 23(2), 201-207.
Silber, J. H., Williams, S. V., Krakauer, H., & Schwartz, J. S. (1992). Hospital and patient characteristics associated with death after surgery. A study of adverse occurrence and failure to rescue. Med Care, 30(7), 615-629.
Slater, A., Shann, F., & Pearson, G. (2003). PIM2: a revised version of the Paediatric Index of Mortality. Intensive Care Med, 29(2), 278-285.
Slonim, A. D., LaFleur, B. J., Ahmed, W., & Joseph, J. G. (2003). Hospital-reported medical errors in children. Pediatrics, 111(3), 617-621.
Star, S. a. S., A. (1999). Layers of silence, arenas of voice: the ecology of visible and invisible work. Computer Supported Cooperative Work, 8, 9-30.