Organizing Regional Perinatal Quality Improvement

After completing this article, readers should be able to: This case study provides information on key decisions and critical steps for those implementing regional perinatal quality improvement (QI) programs. Designers and developers should be cognizant of the considerable evidence base and theoretical work that can guide program development. Such evidence and theory are contained in sidebar reviews and commentary that accompany this description of the development of the California Perinatal Quality Care Collaborative (CPQCC). The authors recognize that others might and should act on the evidence differently according to local circumstances.Perinatal-neonatal leaders in California have been addressing perceived concerns relating to the quality of care for decades. From the pioneering work of Williams (1) to more recent Perinatal Profiles of risk-adjusted hospital level perinatal-neonatal mortality, (2) significant hospital-specific variations in mortality have persisted. Such differences persist even after adjusting for level of care and case mix across institutions. (3) In addition, both Wirtschafter and associates (4) and Wilson and colleagues (5) have described variation within hospital networks operated by large California health maintenance organizations. Increasingly, California’s leaders became aware that both private and public payers were concerned about how and who should address the issue. (See Sidebar #1)The first step in building a regional program is to identify and involve as many stakeholders as possible to ensure program support. CPQCC has had many parents, drawn by shared goals but with differing issues. Employers and payers, represented by the Pacific Business Group on Health (PBGH), sought to move beyond rating hospital perinatal services simply by their cesarean section rates. The David and Lucille Packard Foundation (DLPF), an early sponsor of the Vermont-Oxford Network (VON), was interested in attracting Californians to the potential they saw in the VON. State Health Department officials were concerned with developing a system to monitor the quality of care based on morbidity rather than mortality data and were stimulated to revisit how their regional perinatal programs might be redirected toward improving perinatal outcome after a briefing on the regional perinatal programs of the Southern (6) and Northern California Kaiser-Permanente Health Care Plan in 1994. Most importantly, the leaders and membership of the then embryonic California Association of Neonatology (CAN) placed QI as their foremost goal. (7) The membership of CAN also was concerned about the capability of detecting the impact of tightening reimbursement on the quality of care and, if report cards were issued, that they be credible.The Executive Committee of CAN worked with DLPF to identify and gather potential partners from California’s Department of Public Health (Maternal and Child Health [MCH] Branch, California Childrens’ Services [CCS] Branch, and the Vital Records Branch); the Office of Statewide Planning and Development; leaders from the employer-payer sector (PBGH); and leaders from the Perinatal Medicine Section, District IX, American Academy of Pediatrics (AAP) and the California Chapter of the American Academy of Obstetrics/Gynecology (ACOG). Their work was guided by the principle that potential organizers and participants would align their own “responsibility-based organizations” (8) with the proposed goals, providing a market-based solution rather than depending on a regulatory approach.Discussions reflected a maturation of relationships from competition to collaboration and from individual visions to a comprehensive program. We promoted organizational linkages by engaging and including representation from as many interested providers and entities as possible. We used smaller initial meetings to overcome communication barriers engendered by past conflict or suspicion between some of the parties. Planning committees addressed organizational structure, provider oversight, database definition, information processing, and QI. Significant effort was required to conceptualize the contractual needs of the regional network and to draft replacements to the existing contracts between hospitals and VON. (See Sidebar #2)Organizers need to address the local mix of problems that their MCH agencies face in enhancing perinatal QI programming for both public and private sectors. (9) In California, the interest and enthusiasm of MCH was tempered by the challenges of ensuring the scientific validity, clinical credibility, and professional confidentiality of the CPQCC data collection, analysis, and distribution functions as well as having a QI effort that was clearly responsive to clinician needs.Local database development and operation start with building on existing efforts to ensure compatibility with the parent system and should proceed by complementary rather than competitive pathways. The leadership of CPQCC was intent on developing a collaborative relationship with VON, even as it became the local (ie, California) intermediary for hospital membership in VON. The leadership of VON encouraged and helped shape the relationship. Contracts addressed issues of member confidentiality and responsibility, data transfers and report distribution, funds transfers, and continuing participation in research and QI projects. The contracts also had to address the local issues raised by collecting data on larger infants and by providing for local reporting.CPQCC required a “home” institution to assure appropriate fiscal oversight and management functions; provide administrative, personnel, and legal services; and ensure the physical safety and confidentiality of patient data. In choosing a home, and given the past or potential history for competition between community and academic-based neonatal intensive care units (NICUs), it is important that the home institution be neutral and not derive any special advantage by assuming the role. After considering several alternatives, Stanford University was given these responsibilities.The major challenges are to obtain initial funding and then to develop a “business” plan to ensure stable funding. The DLPF generously funded CPQCC’s start-up for 5 years. Additionally, California’s MCH branch has supported the data center’s infrastructure. Initially, member fees (identical to those of VON) were passed through to VON for support of data processing for very low-birthweight (VLBW) infants. An additional fee was charged members who chose to submit data on larger infants. As of 2004, a portion of the VLBW fee will be retained by CPQCC to reflect the work performed locally rather than at the VON data center. The CCS decision to require certified NICUs to submit their annual activity reports using the CPQCC process will increase membership fee income by approximately 40%. Additional grants have been received for special studies (eg, support from the California Health Care Foundation to study the effects of cesarean section rate variation on neonatal morbidity).Goals and objectives guide both short- and long-term activities as well as foster participants’ commitment and involvement. The goal of CPQCC is to foster benchmark performance by all of its members. The vision evolved to include the following capabilities: 1) link the VON dataset to other relevant (and population-based) data sources (eg, California’s linked birth-death certificate file, hospital discharge abstracts (UB92), and neonatal transport file); 2) provide flexible and innovative report generation; 3) provide clinically relevant and credible indicators, appropriately risk-adjusted; 4) foster an interactive perinatal-neonatal community; 5) and organize and aid QI efforts.There are three arms to the project: the data center, the QI panel, and the research unit. The data center supports member data production and submissions—VLBW data that are uploaded to VON and “big baby” (birthweights >1.5 kg meeting certain acuity criteria) data that address California-specific objectives. A combined data set serves to satisfy the annual CCS reporting requirement. The data center maintains a unique Web-based reporting engine that enables users to generate and download hundreds of user-defined reports and graphs (test site available at cpqcc.org).The Perinatal Quality Improvement Panel (PQIP) serves primarily as the guiding body for organizing and implementing improvement projects. The PQIP is composed of neonatologists, perinatologists, and nurses from California’s Regional Perinatal Systems Programs, chosen to assure geographic and organizational (“town” and “gown”) representation.The research unit is responsible for analyzing and reporting on the database and QI experience to further both the understanding of factors that affect perinatal health and QI programming.Both the Executive Committee of CPQCC and the CAN membership spent considerable time developing a consensus on a policy relating to “public release.” Tension is inevitable between those favoring public release as a means to define a QI agenda and those concerned that it too narrowly focuses QI initiatives or is unrepresentative of actual clinical practice because of poor or insufficient data. (See Sidebar #3) CPQCC resolved to designate a process or outcome as an “indicator” if it could be collected unambiguously and efficiently, showed significant variation across units, was recognized as reflecting a clinically important process or outcome, and was capable of performance improvement. Prospective indicators are proposed by the CPQCC QI panel, reviewed by a committee that includes both clinical leaders and statisticians, and presented for final approval to the Executive Committee. CPQCC selected antenatal steroids prior to VLBW delivery as its first “public release” indicator. PBGH, CPQCC’s employer-payer partner, guided, counseled, and eventually hosted the results on its Web site. (To access, go to http://www.healthscope.org/ and select: 1) hospitals; 2) LA-westside; 3) View Ratings; 4) Get Good Results; and 5) High Risk Pregnancy Care.)PQIP members undertook a self-development process in which they considered the “measure-to-improve” paradigm, reviewed various approaches to change clinician behavior, and considered implications of and boundaries for guidelines development. (See Sidebars #4, #5, #6, #7, #8) The PQIP elected to address “targets with improvement potential” if their status could be ascertained with currently available VON data elements, the processes related to clinically important events, and the recommended processes met the higher levels of evidence-based medicine (EBM) criteria. (10) Commensurate with available time and resources, this selection process has been applied successively (see section on Progress To Date).The PQIP adopted a QI model primarily informed by a meta-analysis of randomized, controlled trials that addressed how to change physician performance (11) and sociological descriptions of implementing change. (12) Additional perspectives and more recent evaluations are discussed separately. (See Sidebars #5, #6, #7) The CPQCC model uses multiple strategies rather than a single one.The first strategy to influence physicians addresses setting the stage for them to be predisposed to change a specific practice. Change must be based on authoritative EBM reviews. We use regional opinion leaders (PQIP) to identify practices that have the potential for improvement, using as criteria the availability of indicator data, demonstrated variability in current practice, authoritative reviews indicating that the recommended process is efficacious, and indications that the process addressed has an important impact on outcome. Generally, PQIP only “recommends” care processes based on single or multi-institutional randomized, controlled trials. In a few instances, relating to nasal continuous positive airway pressure use and several nosocomial infection prevention practices, lower-level evidence-based practices were noted simply as “commended” by others. (See Sidebar #8)Bringing recommended practices effectively to practitioners is and continues to be a challenge. CPQCC uses workshops (often in conjunction with CAN’s Annual Scientific Meetings), newsletters and e-mails, and “academic detailing” by Regional Perinatal Program staff to develop awareness and gain attention. “Webcasting” (scheduled access to a presentation at a Web site during which viewers can interact with the speakers via phone and e-mail) is another technique that will be implemented soon and is modeled on the process initiated by VON’s iNICQ Safety Collaborative Series.The second strategy addresses practice-enabling techniques. Based on our experiences with the highly successful AAP/AHA Neonatal Resuscitation Program (13) and the NIC/Q2000 Collaborative, (14) we initiated two training and implementation tools: Toolkits and Quality Improvement Workshops. (15)The toolkit serves as a “stand-alone” QI “software” aid that contains all of the components for a perinatal-neonatal unit to address a topic without painstaking and lengthy preparation or attendance at the Quality Improvement Workshop. Toolkits contain several sections. These include EBM practice reviews with relevant citations and reprints when available for distribution; benchmarking data with instructions for an individual unit to review and compare its own data to the benchmark; problem identification worksheets for guiding chart reviews; a general structure for QI (16) followed by specific examples on how to organize teams; and appendices filled with example policies and procedures, teaching aids, evaluation tools, and other materials. (Toolkits are available at cpqcc.org.)The Quality Improvement Workshop is designed to “jump-start” unit teams (we request participating units to send at least three physicians, nurses, or clinicians from other disciplines) in working with the toolkits. Participants are sent exercises before the workshop that introduce them to using the problem identification worksheets relevant to workshop topics. This ensures some awareness of actual unit practice and facilitates what psychologists term “cognitive dissonance.” Typically, 1-hour sessions are broken into four quarters: 1) didactic presentation by the topic specialist (authoritative presentation); 2) question-and-answer session with the speaker and often a panel of speakers from related disciplines if applicable (opportunity for the audience to address their concerns and skepticism relating to the recommended process); 3) small facilitated break-out sessions, usually with the group from one unit (targeted to identify the unit’s performance issues based on the premeeting work, address barriers to adopting the recommended process, examine means to overcome these barriers, and initiate the practice of collaborating for change); and 4) a report session, moderated by the speaker, designed to promote collaboration between the units. The format is a compromise between an emphasis on fixing a specific care process and addressing unit collaboration processes; the intent is to help attendees realize the importance of the latter process while their interest has been piqued by the former.Audit and feedback strategies involve primarily the use by CPQCC members of standard reports produced by VON and supplemented by special reports from CPQCC’s data center that enable a closer evaluation of the recommended practice. (See test center reports at cpqcc.org.) For example, CPQCC calculates rates of antenatal steroid use prior to VLBW delivery in accord with the ACOG/AAP guideline, (17) addressing only newborns from 24 0/7 weeks through 33 6/7 weeks gestational age. The intent is to enhance the clinical relevance and credibility of the benchmark process.The benchmark process generally uses the private report processes of CPQCC and VON. In special circumstances, a public release “indicator” can be developed and distributed. CPQCC chose to designate “benchmark” performance as equal to or better than that of the best quartile within its membership during the baseline year.(See Sidebars #9, #10)All privately and publicly reported indicators are used to evaluate both individual unit progress and overall program effects. The minimum evaluation cycle time is 5 years: Year 01 for baseline data collection, Year 02 for fourth quarter report out of Year 01 data, Year 03 for QI efforts, Year 04 for collection of evaluation data, and Year 05 for fourth quarter report out of Year 04 data and report out of a “private” or “public release” indicator.The organizational structure, funding, and data center for CPQCC are well established. Membership has grown on a voluntary basis from the initial 28 units in 1998 to 66 in 2003. Approximately 67% of VLBW births in California currently are tracked through CPQCC. The CCS mandate that all of its member units report their annual statistics through CPQCC will take effect in 2004 and likely will increase membership to more than 90 units.QI initiatives have targeted chronic lung disease prevention/management (antenatal steroid administration during 1999, surfactant use during 2000 and 2001, promotion of consistent practices as applied to mechanical and nasal ventilation during 2003, and abandonment of postnatal steroid administration during 2002) and nosocomial infection prevention during 2002. The initial membership cohort has completed the full evaluation cycle for the antenatal steroid intervention, including public release of data, and a report on the program’s effects is in preparation. In 2004, we will address nutrition support of the VLBW infant and prevention of early-onset sepsis.Local implementation is a prerequisite to achieving the global goals of clinical perinatology. We have described the steps, and the reasoning behind them, used to de-velop the CPQCC program. We believe that our results demonstrate the potential for similar types programs and look forward to assisting others.