PEDIATRICS Vol. 107 No. 1 January 2001, pp. 23-29

Economic Implications of Neonatal Intensive Care Unit Collaborative Quality Improvement

Jeannette A. Rogowski, PhD^*, Jeffrey D. Horbar, MD^, , Paul E. Plsek, MS^§, Linda Schuurmann Baker, MPH, Julie Deterding, MBA^¶, William H. Edwards, MD^#, James Hocker, MD^**, Anand D. Kantak, MD, Patrick Lewallen, MD^§§, William Lewis, MD^||, Eugene Lewit, PhD, Connie J. McCarroll, RN^¶¶, Dennis Mujsce, MD^##, Nathaniel R. Payne, MD^***, Patricia Shiono, PhD, Roger F. Soll, MD, and Kathy Leahy, RN, NNP

From ^* Rand Corporation, Washington, DC;  Vermont Oxford Network, Burlington, Vermont; ^§ Paul E. Plsek & Associates, Inc, Roswell, Georgia;  David and Lucile Packard Foundation, Los Altos, California; ^¶ Wesley Medical Center, Wichita, Kansas; ^# Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire; ^** The Children's Hospital of Illinois, Peoria, Illinois;  Children's Hospital Medical Center of Akron, Akron, Ohio; ^§§ Legacy Emanuel Hospital and Health Center, Portland, Oregon; ^|| Parkview Memorial Hospital, Fort Wayne, Indiana; ^¶¶ Miami Valley Hospital, Dayton, Ohio; ^## Milton S. Hershey Medical Center, Hershey, Pennsylvania; ^*** Children's Health Care-Minneapolis, Minneapolis, Minnesota; and  University of Vermont College of Medicine, Burlington, Vermont.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objective.  To make measurable improvements in the quality and cost of neonatal intensive care using a multidisciplinary collaborative quality improvement model.

Design.  Interventional study. Data on treatment costs were collected for infants with birth weight 501 to 1500 g for the period of January 1, 1994 to December 31, 1997. Data on resources expended by hospitals to conduct this project were collected in a survey for the period January 1, 1995 to December 31, 1996.

Setting.  Ten self-selected neonatal intensive care units (NICUs) received the intervention. They formed 2 subgroups (6 NICUs working on infection, 4 NICUs working on chronic lung disease). Nine other NICUs served as a contemporaneous comparison group.

Patients.  Infants with birth weight 501 to 1500 g born at or admitted within 28 days of birth between 1994 and 1997 to the 6 study NICUs in the infection group (N = 2993) and the 9 comparison NICUs (N = 2203); infants with birth weight 501 to 1000 g at the 4 study NICUs in the chronic lung disease group (N = 663) and the 9 comparison NICUs (N = 1007).

Interventions.  NICUs formed multidisciplinary teams which worked together to undertake a collaborative quality improvement effort between January 1995 and December 1996. They received instruction in quality improvement, reviewed performance data, identified common improvement goals, and implemented "potentially better practices" developed through analysis of the processes of care, literature review, and site visits.

Main Outcome Measures.  Treatment cost per infant is the primary economic outcome measure. In addition, the resources spent by hospitals in undertaking the collaborative quality improvement effort were determined.

Results.  Between 1994 and 1996, the median treatment cost per infant with birth weight 501 to 1500 g at the 6 project NICUs in the infection group decreased from $57 606 to $46 674 (a statistical decline); at the 4 chronic lung disease hospitals, for infants with birth weights 501 to 1000 g, it decreased from $85 959 to $77 250. Treatment costs at hospitals in the control group rose over the same period. There was heterogeneity in the effects among the NICUs in both project groups. Cost savings were maintained in the year following the intervention. On average, hospitals spent $68 206 in resources to undertake the collaborative quality improvement effort between 1995 and 1996. Two thirds of these costs were incurred in the first year, with the remaining third in the second year. The average savings per hospital in patient care costs for very low birth weight infants in the infection group was $2.3 million in the post-intervention year (1996). There was considerable heterogeneity in the cost savings across hospitals associated with participation in the collaborative quality improvement project.

Conclusion.  Cost savings may be achieved as a result of collaborative quality improvement efforts and when they occur, they appear to be sustainable, at least in the short run. In high-cost patient populations, such as infants with very low birth weights, cost savings can quickly offset institutional expenditures for quality improvement efforts.  Key words:  economics of collaborative quality improvement, neonatal intensive care unit treatment costs.

Collaborative quality improvement has been successfully applied in a number of clinical contexts for a diverse set of clinical conditions, ranging from adult critical care to asthma.^1-7 However, there is little information on the effects of such efforts on costs of patient care or on the resources required by hospitals to undertake these types of quality improvement efforts. Organizations may be unwilling to invest in the human and financial resources required to participate in such efforts without a better understanding of both their clinical value and their potential for reducing patient care costs.

In this article, we describe the economic implications of a collaborative quality improvement effort for very low birth weight infants, referred to as the NIC/Q (Neonatal Intensive Care Collaborative Quality) Project of the Vermont Oxford Network (VON). In 1995, 10 hospitals in this project began a 2-year collaborative quality improvement project with the goal of making measurable improvements in the quality and cost of neonatal intensive care. Six hospitals chose to make measurable improvements in the rates of nosocomial infection and 4 in the outcomes associated with chronic lung disease (CLD). The participating institutions formed multidisciplinary teams that worked together under the direction of a trained facilitator over a 3-year period beginning in January 1995. They received instruction in quality improvement, reviewed performance data, identified common improvement goals, and implemented potentially better practices developed through analysis of the processes of care, literature review, and site visits. The intervention is described in detail in the companion article.¹⁰

In this article, we describe the effects of the NIC/Q effort on treatment costs. In addition, we document the resources used by participating hospitals over the 2-year course of the study and the infrastructure costs associated with supporting this effort. Overall cost savings in patient care are contrasted to the resources expended in undertaking the collaborative quality improvement interventions. The results should shed important new light on the economic implications of collaborative quality improvement efforts.

	METHODS

Top Abstract Methods Results Discussion References

The VON

The VON is a voluntary group of health professionals committed to improving the quality of medical care for newborn infants and their families. The Network facilitates a coordinated program of research, education, and quality improvement.^8,9

To support this program, the Network maintains a database for infants weighing 401 to 1500 g at birth who were born at or transferred to participating neonatal intensive care units (NICUs) within 28 days of birth. The database includes information concerning medical practices and patient outcomes such as morbidity, mortality, and length of stay.¹¹ Data on infants are collected even after transfer from the participating NICU to other hospitals until discharge to home to determine survival status and total hospital length of stay. The Network Database was supplemented by the collection of data on treatment costs to evaluate the results of the quality improvement interventions on the costs of patient care.

Recruitment of Participating NICUs and Control Hospitals

Before the inception of the project, written descriptions of the proposed collaborative quality improvement project and applications to participate were mailed to all NICUs in the VON. The project was discussed at the Network Annual Meeting and was the subject of an article in the Vermont Oxford Network Newsletter. The overall goals and structure of the project were described. It was explained that the participating institutions would be responsible for funding all travel expenses for their team members and for supporting the internal staff time required by the project. Eleven institutions applied and were accepted. One institution withdrew before the initial project meeting because of funding difficulties.

Six hospitals participated in the nosocomial infection subgroup of the NIC/Q Project. These were a self-selected group of hospitals interested in undertaking a collaborative quality improvement project. They chose as their goal to improve the rates of nosocomial infection for infants 501 to 1500 g to the 25th percentile for NICUs in the Network with 70 or more very low birth weight admissions per year. Analysis of the nosocomial infection rates in 1994 demonstrated that 25% of the eligible NICUs had rates of 15% or less. Nosocomial infection was defined as the occurrence of 1 or more infections after the third day of life with either coagulase-negative staphylococcus or another bacterial pathogen from a predefined list included in the Database Manual of Operations.¹¹ The specific organisms that are considered bacterial pathogens for the VON Database represent organisms unlikely to represent contaminants.

Four hospitals participated in the CLD group. These hospitals chose as their goal an absolute reduction in the rate of death or oxygen supplementation at 36 weeks' postconceptional age by 10% for infants 501 to 1000 g with gestational ages of 34 weeks or less. The characteristics of the patient population subject to the quality improvement efforts at each of the infection and CLD subgroups in 1994 is presented in Table 1.

Nine hospitals that participated in the Cost and Resource Utilization Study of the VON served as a control group. The Cost and Resource Utilization Study was a project whose goals were to measure the costs of neonatal intensive care and study its determinants. The hospitals in the latter study were a self-selected group of hospital interested in measuring and monitoring patient care costs. All 10 NIC/Q hospitals participated in this study as well. The characteristics of the patient population at the control group hospitals are also presented in Table 1.

There was a small amount of missing data in each of the intervention and control groups attributable to cost data that was either missing or of poor quality. Overall, for the CLD subgroup the missing data rate was 4% for the intervention and 7% for the control group. For the infection subgroup it was 3% for the intervention and 4% for the control group. There was no systematic pattern of missing data by birth weight for each of these groups in either the preintervention or postintervention periods.

For the infection subgroup, 18% of infants had birth weights between 501 g and 750 g, 27% between 751 g and 1000 g, and 1001 g and 1250 g respectively, and 28% in the 1251-g to 1500-g range. The control group had a somewhat lower proportion of infants with birth weights in the lowest range (13%) and a higher proportion in the highest range (37%). The NIC/Q sample was primarily white (85%) while an additional 11% of infants were black. In contrast, the control group had a lower proportion of white infants (68%) and a higher proportion of black infants (25%). A large fraction of infants (81%) in the infection subgroup hospitals received assisted ventilation and 12% had a major surgical procedure during their stay. The rates in the control group were similar.

For the CLD subgroup, the sample of infants is restricted to those with extremely low birth weights. In both the CLD subgroup hospitals and in the control group 41% had birth weights between 501 and 750 g and 59% had birth weights between 751 and 1000 g. The majority of infants in the CLD subgroup were white (78%) while an additional 17% were black. In contrast, the control group had a lower proportion of white infants (58%) and a higher proportion of black infants (33%). Virtually all infants in the CLD subgroup had assisted ventilation (98%) and 26% had a major surgical procedure during their stay. In contrast, the infants at the control group hospitals were somewhat less likely to have received assisted ventilation (95%) or to have had major surgery (19%). The infants in the CLD subgroup were also somewhat less likely to have died on or before the third day of life (8%) compared with the control group (11%).

Patient-Level Analytic Measures

Data on treatment costs was obtained through primary data collection. Data were obtained for hospital costs only. Physician costs were not determined as it was not possible to collect the universe of physician claims associated with the care of these infants. For 1994, detailed hospital bills were obtained for all infants at the 10 study hospitals as well as at the 9 control group hospitals. For 1995 through 1997, UB-92 data were obtained for these infants. UB-92 data are aggregations of the detailed hospital bills, and thus the cost estimates generated by the 2 different types of data are identical. This was validated in a subsample of hospitals that submitted both types of data. The charges from these hospital bills were converted to measures of treatment cost using Cost Reports from the Health Care Financing Administration ([HCFA] Form 2552) following established methods.¹² Cost Reports were used in each year to create departmental-level cost-to-charge ratios. Charges were mapped into departments and then multiplied by the appropriate cost-to-charge ratio to create measures of treatment costs. All costs were converted to 1996 constant dollars using the medical component of the consumer price index.¹³ To measure geographic differences in hospital input costs, we used the HCFA wage index. This index, reported each year in the Federal Register, is used in setting hospital payments under Medicare's Prospective Payment System.

Clinical information on the infants in the study was obtained from the VON Database. The Network maintains a Database for infants weighing 401 to 1500 g at birth who were born at or transferred to participating NICUs within 28 days of birth. The Database includes information concerning medical practices and patient outcomes such as morbidity, mortality and length of stay.^8,9,11,14 Data on infants are collected even after transfer from the participating NICU to other hospitals until discharge to home to determine survival status and total hospital length of stay.

Database elements used in this study include information on infant characteristics (birth weight, gender, race), treatments received, location of birth, transfer status, and death. Infants are defined to have assisted ventilation if they are on respiratory support at any time after leaving the delivery room, either through conventional or high frequency ventilation. Information on the presence of major surgery was obtained from 2 different sources. For 1996, the presence of major surgery was obtained from the Vermont Oxford clinical database. This is defined as the presence of either a patent ductus arteriosus ligation, retinopathy of prematurity surgery, necrotizing enterocolitis surgery or other major surgery. The latter includes a list of major surgical procedures. In 1994, the database did not collect data on major surgery. The presence of major surgery in 1994 was determined from the procedure codes on the hospital bills. The procedure codes included are comparable to the types of surgery measured in the 1996 survey. Procedure codes were not available in the 1996 data. Thus, it was not possible to create identical measures of major surgery in the preintervention and postintervention year.

Several measures for the characteristics of the stay were also constructed from the VON database. These include the location of birth (inborn versus outborn) as well as whether the infant was transferred out of the unit. Finally, because many infants with birth weights under 1000 g die within the first few days of life, we constructed a measure for whether the infant died on or before the third day of life.

Measures of the Resources Used by Hospitals in the Quality Improvement Effort

Data on the resources used by hospitals in the quality improvement effort were self-reported and were determined by primary data collection for 9 of the 10 NIC/Q hospitals. This included a combination of weekly diaries and a follow-up survey. Resources were measured for the entire course of the collaborative quality improvement effort, January 1, 1995 to December 31, 1996. Three hospitals maintained weekly diaries of all activities. Three additional hospitals maintained weekly diaries for the initial portion of the study period (4, 5, and 5.5 months of 1995). Finally, 3 sites did not maintain any weekly diaries because of the reporting burden. For the 6 sites without complete weekly information, a follow-up survey was administered to determine aggregate levels of effort over the study period. One hospital did not provide data.

The weekly diaries included information for each staff member who regularly participated in the NIC/Q Project on the type and duration of their project activities for the week. This included time spent in benchmarking visits or external meetings, conference calls, question and answer, data collection and analysis, policy review, internal analysis, policy reformulation, education, testing, reporting, internal meetings, individual work, clerical time, travel time, and resource measurement time (filling out the coding sheets). In addition, information on nontime resources, such as travel costs, was also collected for each week. The follow-up survey collected information for the entire study period on the time and nontime costs associated with project activities. These included: external benchmarking visits and meetings, team meetings, conference calls, and data collection and analysis. One individual at each site responded and estimated the average number of participants of each staff type that were involved in each activity over the two year period. Data on the frequency of conference calls as well as average duration of calls was collected separately for 1995 and 1996. In addition, travel time and costs were ascertained by date of travel in the follow-up survey. Thus, it was possible to allocate these to individual years of the study. Both types of data collection activities, weekly diaries, and the follow-up survey collected information on the entire set of activities encompassed by the collaborative quality improvement effort. The follow-up survey was somewhat less detailed than the weekly survey to maximize respondent recall and to ensure high participation rates.

For both the weekly and the summary surveys, the value of staff time is estimated by multiplying reported staff time for participants of different staff types (physicians, nurses, administrators, therapists, other staff) by the average hourly wage for the particular staff type. Wages were determined via a literature review and by a brief survey of project participants to check the validity of the results. The value of travel and other costs were reported in dollar amounts by participants. For hospitals with partial weekly data, the resources used in the remainder of the study period were imputed based on the data provided in the follow-up survey, as well as the separately-collected information on travel and the duration and frequency of conference calls.

Statistical Methods

The statistical analyses compare treatment costs in 1996, the first postintervention year, with those occurring in 1994, the baseline year before the initial meeting of the project. These analyses include pre-post comparisons in the project subgroups (infection and CLD) with those at the control group of hospitals. The models control for patient mix as well as other factors known to affect treatment costs. They are estimated using ordinary least squares regression. Statistical analyses were performed using SAS (SAS Institute, Inc, Cary, NC, Version 6.12). The number of infants in the regressions for the infection intervention was N = 2536 and for the chronic lung intervention N = 855.

The dependent variable in the regression models is the log of treatment cost per infant in 1996 constant dollars. The log is modeled according to the long right tail in measures of treatment cost. The models include dummy variables for the intervention subgroup (infection or chronic lung subgroups), as well as for the postintervention period and the interaction of the intervention subgroup with the postintervention period. The significance of the latter interaction is used as a test of whether the change over time in treatment costs was significantly different between the NIC/Q subgroup of interest and the comparison group, after controlling for potentially confounding differences attributable to changes in other covariates.

Control variables in the cost models include patient characteristics (birth weight, gender and race). In addition, other factors known to affect treatment costs such as the presence of assisted ventilation and major surgery are also included in the models. Because slightly different definitions of major surgery were used in 1994 and 1996, separate variables for each year were created. In addition, because infants that are inborn as well as those that are transferred out of the NICU have shorter lengths of stay and thus lower costs, controls for these factors are also included. Finally, the log of the HCFA wage index is included to control for geographic differences in the cost of hospital input prices. Because mortality rates in the first few days of life are high for infants with extremely low birth weights (1000 g and under), in the regressions to evaluate the chronic lung initiative we also control for death on or before the third day of life.

	RESULTS

Top Abstract Methods Results Discussion References

Effect of Collaborative Quality Improvement Effort on Treatment Costs

The time trends in median treatment cost per infant (in constant 1996 dollars) for the NIC/Q subgroups and the control group are presented in Table 2. These figures do not control for patient mix, but instead represent the actual cost experience of the hospitals. For hospitals in the nosocomial infection subgroup, treatment cost per infant with a birth weight of 501 to 1500 g (and hospitalized for >3 days) declined from $57 606 in the base year (1994) to $46 674 in the postintervention year (1996). In contrast, median treatment cost per infant at the control group hospitals rose from $60 771 to $73 020 at the control group hospitals. When patient mix and other factors known to affect treatment costs are controlled for in a multivariate regression, costs at the nosocomial infection subgroup hospitals declined significantly relative to the control group (P < .0001). The R² for this regression was 0.50. 

For hospitals in the CLD subgroup, median treatment cost per infant with birth weight 501 to 1000 g and a gestational age at birth of 34 weeks or less) also declined between the base year and the postintervention year. In 1994, median treatment cost per infant was $85 959. This had declined to $77 250 by 1996. In contrast, in the control group, median treatment cost per infant rose from $91 969 to $98 570 over the same period. Controlling for patient mix as well as other factors that affect treatment costs, this difference was not, however, statistically significant (P = .7980). The R² for this regression was 0.49.

There was heterogeneity in the cost experience of subgroup hospitals. Four of the 6 hospitals in the infection subgroup had decreases in treatment cost for very low birth weight infants (birth weight 1500 g) between the base year and the postintervention year. Cost decreases ranged from $8800 to $18 500 per infant (in 1996 constant dollars). Two hospitals had no change in treatment costs as measured in 1996 dollars. In the CLD subgroup, all hospitals had decreases in median treatment costs for extremely low birth weight (birth weight 1000 g) infants. These ranged from $4800 to $37 800 per infant (as measured in 1996 dollars).

The declines in treatment cost per infant at the NIC/Q subgroup hospitals were maintained into 1997, a year after the intervention. For hospitals in the infection subgroup, the medical treatment cost per infant in 1997 was $45 874 in constant 1996 dollars. For the CLD subgroup it was $75 084. For the control group hospitals, there was a decline in median treatment cost per infant between 1996 and 1997. Relative to the base year (1994), however, treatment costs at the control groups rose for infants with very low birth weights but declined for those with extremely low birth weights. These figures are not adjusted for patient mix, but instead reflect the actual cost experience of hospitals in those years. When patient mix and other factors known to affect treatment costs are controlled for in a multivariate regression, treatment costs in 1997 at hospitals in the nosocomial infection subgroup declined significantly (P = .0037) relative to the control group. For hospitals in the CLD subgroup, however, there was no significant difference in treatment costs (P = .8230), a finding similar to the postintervention year (1996) experience.

Cost Savings

The cost regressions were used to determine the cost savings, if any, associated with participation in the collaborative quality improvement effort. The actual treatment costs for the cohort of infants in the postintervention year (1996) were compared with the predicted cost if the hospital had not participated in the NIC/Q Project, based on the cost regression. Because costs were only found to have significantly decreased in the infection group, controlling for patient mix, cost savings were only computed for hospitals in the infection subgroup. The average cost saving per hospital for hospitals in the nosocomial infection subgroup was $2.3 million dollars. The savings varied by hospital, ranging from half a million dollars to $4.5 million.

Resources Used in Undertaking Collaborative Quality Improvement

The resources expended by hospitals in undertaking the NIC/Q Project are reported in Table 3. The average expenditure by hospitals in the NIC/Q Project over the 2-year period of the study (1995-1996) was $68 206. The expenditures were comparable for hospitals in the nosocomial infection subgroup ($67 744) and those in the CLD subgroup ($68 784). There was considerable heterogeneity in the resources expended by hospitals. These ranged from $53 775 to $88 385 for hospitals in the CLD subgroup and from $42 298 to $78 611 for hospitals in the nosocomial infection subgroup. There was a large decline in resources expended over time on the collaborative quality improvement project, with two-thirds of expenditures occurring in the first year of the project and one-third in the second year. The average first year expenditure was $44 055 compared with $24 152 in the second year.

The collaborative approach to quality improvement involves travel to project meetings and for benchmarking purposes as well as work within the hospital itself. Table 3 demonstrates that resources spent on this approach to quality improvement are approximately equally distributed between internal and external efforts. On average, $35 584 were expended on external efforts. This included $15 494 for staff time associated with travel and $20 090 in travel costs. On average, the internal effort resulted in $32 622 in resources expended over the 2-year course of the study. Of this, approximately half ($16 893) was accounted for by the cost of physician time. One-third of internal resources, or $9899, were associated with the cost of nurse time for participation in the quality improvement collaborative. Finally, administrators accounted for an additional 8% of internal resources ($2593), respiratory, and physical therapists for 3% ($881) and other staff for 7% ($2357).

In addition, the costs associated with maintaining the infrastructure to support the collaborative quality improvement effort over that time period was $820 000. This was funded through a grant from the Center for the Future of Children of the David and Lucile Packard Foundation. Ideally, the costs associated with hosting the 3 site visits by hospitals outside of the project would also have been included in this figure, but this data were not collected.

	DISCUSSION

Top Abstract Methods Results Discussion References

This article provides some of the first information available on the economic implications of collaborative quality improvement in clinical settings. Information on the effects of the collaborative approach to quality improvement on both patient care costs and the resources expended by hospitals participating in a quality improvement effort for infants with very low birth weights is reported. Economic evaluations of quality improvement efforts based on collaborative learning are particularly valuable because hospitals may be unwilling to undertake such initiatives without a clear understanding of the resource costs or the potential savings in terms of patient care costs.

The results presented here demonstrate that in high-cost patient populations, such as infants with very low birth weights, the cost savings associated with collaborative quality improvement can largely overshadow the expenditures that hospitals undertake in improving the quality of care. For instance, in the NIC/Q Project, hospitals in the nosocomial infection subgroup expended an average of $68 206 to undertake the project over a 2-year period, but had savings in patient care costs in the postintervention year of $2.3 million. Cost savings were also maintained in the year after the study. Thus, cost savings appear to be sustainable, at least in the short run. However, it would be necessary to determine the cost experience of study hospitals over a longer period of time to accurately determine the time profile of treatment costs.

The hospitals in the NIC/Q Project were a self-selected group of institutions that were highly motivated. These hospitals participated in both the NIC/Q and Cost and Resource Utilization Projects. They were thus highly interested in both improving quality and reducing costs. Because of this, caution should be applied in generalizing these findings to a wider universe of NICUs. Organizations chosen at random and forced to participate in a collaborative quality improvement effort might have a much less favorable cost experience. Thus, the cost savings observed here may not be achievable in a broad sample of hospitals. It should also be noted that the control group of hospitals was also a self-selected group of hospitals that was interested in measuring and monitoring treatment costs. In a randomly selected control group, the cost experience over time might have been less favorable than that observed here. Cost savings with a randomly selected group of control hospitals might therefore have been larger than those reported here.

There was considerable heterogeneity in the cost experience of hospitals participating in the study. This is not surprising, however, given the multitude of factors that contribute to a successful quality improvement intervention. The level of commitment and motivation of quality improvement staff will influence the outcome of the intervention and thus also the observed effect on treatment costs. In addition, organizational characteristics may influence the observed level of success. With the small number of sites and a small number of patients within each site, we lack the statistical power to justify a center-by-center analysis. We cannot exclude the possibility that our findings could result from a few centers with the largest improvement in treatment costs. We will be able to explore the determinants of observed variations in the cost experience of centers more fully in a follow-on study of 34 NICUs that is currently underway.

From a global perspective, it is of interest to determine how the resources expended toward these quality improvement efforts compare with the cost savings generated in patient care. Across all 10 hospitals that participated in the study, the total annual savings in patient care costs for the 1996 caseload was $14 million. The total resources expended by hospitals over a 2-year period (1995-1996) was $680 000. An additional $820 000 was required to maintain the infrastructure to support the effort. Some costs were also incurred by hospitals hosting the 3 site visits but these were not measured. In aggregate, each dollar invested in the improvement effort yielded $9 in cost savings for patient care. Because the hospitals participating in this study were a self-selected group of highly motivated organizations, it is unlikely that these cost savings would be achievable in a randomly-selected group of hospitals forced to undertake a collaborative quality improvement intervention. As with the cost experience of hospitals, cost savings also varied widely, with some hospitals experiencing no cost savings and others substantial savings.

How cost reductions affect hospital revenue depends, of course, on the nature of the reimbursements received from insurers. In a fully capitated system, all cost savings accrue to the hospital. Most prospectively-based reimbursement systems, such as diagnosis-related groups also enable hospitals to capture some, if not all, of their cost savings. In contrast, in a pure fee-for-service environment, where hospitals are paid based on full charges, reductions in cost will decrease hospital revenue. However, the latter reimbursement systems are becoming increasingly rare. There is instead a strong trend toward managed care in the health care marketplace, particularly among state Medicaid programs who are among the largest insurers of this patient population. Reductions in the cost of care will further enable hospitals to compete more effectively for managed care contracts and thus have positive long range financial implications for hospital revenue by affecting patient volume. As shown in this study, in high-cost patient populations, the reduction in patient care costs from collaborative quality improvement may far outweigh the institutional expenditures for the effort. In an era of managed care and cost containment pressures, such efforts are particularly worthwhile, because they both improve patient outcomes as well as reduce costs.

	ACKNOWLEDGMENTS

This study was supported by a grant from the Center for the Future of Children, David and Lucile Packard Foundation.

We thank all of the staff members at participating hospitals for their commitment and dedication to improvement and for their hospitality during the site visits. We especially thank the staffs at those institutions that served as Benchmark sites and hosted visits by the NIC/Q Project teams: Sutter Memorial Hospital, Sacramento, California; Tacoma General Hospital, Tacoma, Washington; and University of Kentucky Children's Hospital, Lexington, Kentucky.

	APPENDIX

NIC/Q Project Investigators

JEFFREY D. HORBAR, MD

Principal Investigator

JEANNETTE A. ROGOWSKI, PHD

Co-Principal Investigator

KATHY LEAHY, RN,NNP

Project Coordinator

Steering Committee

EUGENE LEWIT, PHD

JEROLD F. LUCEY, MD

PAUL E. PLSEK, MS

PATRICIA SHIONO, PHD

Staff

RANDALL HIRSCHER

BOICHI SAN

NICU Teams

Children's Hospital Medical Center of Akron, Akron, Ohio: Lela Bartley, RNC, BSN, Anand Kantak, MD, Joann Lindeman, RNC, and Judy Ohlinger, RN.

Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire: William Edwards, MD, Karen Shea-Ricci, RN, and Karen Vergura, RNC.

Emanuel Children's Hospital, Portland, Oregon: Jan Genke, RN, MSN, Patrick Lewallen, MD, Pat Seifert, RN, and Karen Waske, RN.

Fletcher Allen Health Care, Burlington, Vermont: Marcia Patterson, RN, Roger F. Soll, MD, and Greg Ward, RRT.

Miami Valley Hospital, Dayton, Ohio: Beatrice Harris, RN, William Johnson, and Connie McCarroll, DO.

Milton S. Hershey Medical Center, Hershey, Pennsylvania: Cindy Banta, RN, Jane Ebersole, NNP, Dennis Mujsce, MD, and Moira Winstanley, NNP.

Minneapolis Children's Medical Center, Minneapolis, Minnesota: Jo Crosby, RNC, MS, Kathy Johnson, RN, Carol Miller, RN, Kristin Nelson, RN, MS, and Nathaniel R. Payne, MD.

Parkview Memorial Hospital, Fort Wayne, Indiana: Eva Fish, RRT, Bill Lewis, MD, and Nancy Nicholas, RNC.

St Francis Medical Center, Peoria, Illinois: Howard Cohen, MD, Elaine Hartmann, NNP, Jim Hocker, MD, Nancy Huber, RRT, Connie McConnell, RN, and Julie Otterstrom, NNP.

Wesley Medical Center, Wichita, Kansas: Barry T. Bloom, MD, Paula Delmore, MSN, Julie Deterding, Anita Dorf, PhD, and Cindy Harmon, RN.

	FOOTNOTES

Received for publication Apr 21, 2000; accepted Aug 16, 2000.

Reprint requests to (J.A.R.) Rand, 1333 H St, NW, Suite 800, Washington, DC 20005. E-mail: jar{at}rand.org

	ABBREVIATIONS

NIC/Q, Neonatal Intensive Care Collaborative Quality; VON, Vermont Oxford Network; CLD, chronic lung disease; NICU, neonatal intensive care unit; HCFA, Health Care Financing Administration.

	REFERENCES

Top Abstract Methods Results Discussion References

1. Plsek PE Collaborating across organizational boundaries to improve the quality of care. Am J Infect Control. 1997; 25:86-95
2. Nolan T, Schall MW, Berwick DM, Roessner J. Breakthrough Series Guide: Reducing Delays and Waiting Times Throughout the Healthcare System. Boston, MA: Institute for Healthcare Improvement;1996
3. Weiss KB, Mendoza G, Schall MW, Berwick DM, Roessner J. Breakthrough Series Guide: Improving Asthma Care in Children and Adults. Boston, MA: Institute for Healthcare Improvement;1996
4. Rainey TG, Kabcenell A, Berwick DM, Roessner J. Breakthrough Series Guide: Reducing Costs and Improving Outcomes in Adult Intensive Care. Boston, MA: Institute for Healthcare Improvement;1996
5. Leape LL, Kabcenell A, Berwick DM, Roessner J. Breakthrough Series Guide: Reducing Adverse Drug Events. Boston, MA: Institute for Healthcare Improvement;1996
6. O'Connor GT, Plume SK, Olmstead EM, A regional intervention to improve the hospital mortality associated with coronary artery bypass graft surgery. JAMA. 1996; 275:841-846 [Abstract]
7. Plsek PE Quality improvement methods in clinical medicine. Pediatrics 1999; 103:203-214
8. Horbar JD The Vermont Oxford Network. Integrating research and clinical practice to improve the quality of medical care. Semin Perinatol. 1995; 19:124-133 [CrossRef][Medline]
9. Horbar JD The Vermont Oxford Network. Collaborative quality improvement for neonatology. Pediatrics 1999; 103:350-359
10. Horbar JD, Rogowski JA, Plsek PE, Collaborative quality improvement for neonatal intensive care. Pediatrics. 2001; 107:14-22 [Abstract/Free Full Text]
11. Vermont Oxford Network Database Manual of Operations (Release 3.3). Burlington, VT: Vermont Oxford Network; 1997
12. Rogowski JA Measuring the cost of neonatal and perinatal care. Pediatrics. 1999; 103:329-335
13. US Bureau of Labor Statistics. Monthly Labor Review: January. Washington, DC; 1998
14. Vermont Oxford Network Database 1997 Annual Summary. Burlington, VT: Vermont Oxford Network; 1998