Multisite research: A systems approach

Multisite Research: A Systems Approach Elizabeth A. Bossert, DNS, RN Sean Evans, BA Lois Van Cleve, PhD, RN Marilyn C. Savedra, DNS, RN, FAAN

Multisite research is becoming increasingly common because of the need for an adequate sample size and for generalizability of results beyond a single facility. Collaboration in a research project poses unique challenges due to the number of persons and facilities involved. The use of a systems approach to structure the research process in a study of pain in children with leukemia is described, using the principles to structure, conduct, and conclude the multisite project. Copyright 2002, Elsevier Science (USA). All rights reserved.

T

HE USE OF MULTIPLE data collection sites for a single study has increased as nursing research has developed. As the drive to strengthen the scientific foundation of nursing practice has gained momentum, nurse researchers are engaging in increasingly rigorous clinical research studies. This movement has raised the standards and created a variety of challenges that affect the number of sites used for data collection. One major challenge is acquiring a sample size that is sufficient to meet statistical requirements (Friedman, Furberg, & DeMets, 1996; Grant et al., 1998). Many of the phenomena of interest to nurses, while of clinical importance, have only a small, or at best medium, effect on the patients, necessitating a sample larger than that available in many single facilities (Smith, 1988). Related to this is the challenge of collecting the data within a reasonable time frame (Bergstrom et al., 1984; Birenbaum, Hinds, Clarke-Steffen, 1995), necessary because of the difficulty of controlling confounding variables over time. An addi-

From the Loma Linda University School of Nursing, Loma Linda University Graduate School, Department of Psychology, Loma Linda, CA and the University of California, San Francisco School of Nursing, San Franicisco, CA. Supported by the NIH, NINR, Bethesda, MD, grant no. RO1NR04201 and the Oncology Nursing Society, Pittsburgh, PA. Address reprint requests to Elizabeth Bossert, DNS, RN, School of Nursing, West Hall, Room 1128, Loma Linda University, Loma Linda, CA 92350. E-mail: [email protected]. Copyright 2002, Elsevier Science (USA). All rights reserved. 0882-5963/02/1701-0006$35.00/0 doi:10.1053/jpdn.2002.30932 38

tional challenge is the need to generalize the results of the research to a broad population, providing results not constrained by facility-specific characteristics (Bratt, 2000; Friedman, Furberg, & DeMets, 1996; Grant et al, 1998; Hanson, 1988; Minnick, Kleinpell, Micek, & Dudley, 1996). As nurse researchers work to solve these problems, research teams are formed that have members in diverse geographical locations. For these reasons, multisite research is becoming increasingly common in nursing. Examples of multisite studies in the recent pediatric nursing literature include Broome, Rehwaldt, and Fogg’s study (1998) of temperament, coping, and pain during lumbar punctures, conducted in three oncology clinics; Carlson, Broome, and Vessey’s study (2000) of distraction during painful procedures, conducted in 13 hospitals; Froman and Owen’s study (1999) of physical and mental health selfefficacy in American and Korean adolescents, done in three high schools, one in New England and two in Korea; Koshar, Lee, Goss, Heilemann, and Stinson’s study (1998) of Hispanic teen mothers, conducted in three hospitals; and Krulik, TurnerHenson, Kanematsu, Al-Ma’aitah, Swan, and Holaday’s study (1999) of the stress of mothers’ of children with chronic illness, conducted in four countries—Israel, Japan, Jordan, and the United States. Additionally, Bratt (2000) in discussing collaborative institutional research mentioned four multisite National Association of Children’s Hospitals and Related Institutions (NACRI) studies involving 72, 11, 12, and 15 sites, respectively. Journal of Pediatric Nursing, Vol 17, No 1 (February), 2002

MULTISITE RESEARCH

Although conducting research at multiple sites increases the number of potential subjects available, shortens the time needed for data collection, and enhances the generalizability of the findings, other problems may develop that must be addressed. The more sites and people involved in a study, the more carefully the research must be planned to be appropriately implemented at each site; the greater the necessity to ensure each site is adhering to the plan so that the results will be clear; and the greater the need for clarity about the responsibilities and benefits of involvement so that all involved will receive appropriate outcomes from the experience. To examine issues relevant to the planning, conduct, and conclusion of a multisite research study, a systems approach will be used, supported by a description of the authors’ experience with a multisite study. Figure 1 shows the application of a systems approach to the key aspects of each of these phases of the study. DESCRIPTION OF STUDY The study, Children with Leukemia: Pain, Management and Outcomes (Van Cleve, Bossert, & Savedra, 1997), was a 3-year longitudinal study, examining the pain experience of children, ages 4 through 16 years during the first year after diagnosis of acute leukemia. A core group of three nurse researchers from two universities that had previously worked together in other situations, joined to form a research team focused on pediatric pain. All contribute equally to and are responsible for the science, process, integrity, and dissemination of

Figure 1. A systems approach to multisite research.

39

the research. However, for the purposes of communication with funding sources, one is identified as the principal investigator (PI) and the others as coprincipal investigators (co-PIs). This group has access to one tertiary children’s hospital. As the proposal developed, it became evident that additional facilities would be needed for data collection because of the demographics of children with leukemia. Subsequently, nurses with advanced clinical and research experience from two additional children’s hospitals collaborated on the project and served as the co-investigators (co-Is) for those facilities. The co-Is contributed to the refinement of the proposal and coordinated the necessary approvals and data collection at their respective facilities. At each site, research associates also joined the team as needed for data collection. This approach is a blend of the traditional model and unification model of collaborative research (Chenitz, Sater, Davies, & Friesen, 1990) and has worked well to accomplish the goals of the study. PRINCIPLES OF A SYSTEMS APPROACH The fundamental systems principles, that a system is more than the sum of its parts and that it functions as a whole, are the underlying reasons for multisite research. Not surprisingly, a systems approach has been identified as foundational in collaborative interactions (American Association of Colleges of Nursing [AACN], 1997; Minnick et al., 1996). Working together as a team, a group of researchers is able to achieve a level and complexity of research that is difficult in isolation. However, the members of the research team come from facilities with different boundaries, guidelines, and policies for the conduct of research. These can be viewed either as roadblocks or as providing structure to the research process at each facility, allowing discussion and interaction between sites that is helpful in facilitating the research. As nurses increasingly become involved in research, feedback will be given to the research and practice committees resulting in examination of current policies and possible adaptation to facilitate future work. Thus, systems involved in multisite research are open systems and will grow and change in their understanding and support of nursing research, always striving toward a balanced relationship, or steady state, between the needs of the system for continuity of process and enhancement of best care practices. Without input from other systems, both in the process of conducting and applying research to practice, an isolated or closed system will be-

40

BOSSERT ET AL

come increasingly dysfunctional and eventually cease growing and functioning. Finally, in every research project, there will be a hierarchical structure of the team. Although all team members function as a part of the system, multiple subsystems will be functioning to accomplish the goals of the study, such as the system of the PIs, the subsystem of team members at each facility, and the subsystems of those responsible for data collection (Fawcett, 1984; Putt, 1978; Tores, 1986; von Bertalanffy, 1972). STRUCTURING A MULTISITE RESEARCH PROJECT In a collaborative, multisite research project, a number of persons are involved with differing responsibilities resulting in various subsystems within the total system of the research project. There are many possible scenarios. For example, if the primary team is from multiple sites with sufficient populations of potential subjects, additional investigators and facilities may not be necessary. However, if additional sites are needed, such as in the Leukemia Pain Study, the research team must expand beyond the primary system to include other systems. Identification of Sites There are multiple factors to consider in determining the sites for data collection. The first consideration is identifying a facility with the needed population. Often the initial identification is fairly easy; for example, if the subjects are to be acutely ill children, a children’s hospital will be targeted; if the subjects are pediatric critical care nurses working with cardiac transplant patients, facilities doing that procedure will be targeted. A second and parallel consideration is identifying a qualified person willing to join the research team and make a commitment to achieving the study purposes (Ingersoll et al., 1995). Participation on a research team requires interest in the topic, knowledge of the research process, and flexible work responsibilities. Although some large centers have a nurse researcher on staff that may be interested in participating, in other facilities the nurse most likely to fit these criteria is a clinical nurse specialist or nurse educator. Nurses working in one of these advanced practice roles are likely to have in-depth knowledge and interest in the clinical area, a master’s degree which provides advanced knowledge of research, and a job description that encourages or requires participation in research. It is also possible that the best person for the multisite collaboration

will not be a nurse, but rather a psychologist, physician, pharmacist, or other member of the health care team, forming an interdisciplinary team that can bring a diversity of background and knowledge to the project. The best method of identifying both the facility and the collaborator is though personal contacts. Networking at professional meetings, consultations, professional policy development groups all provide opportunities to discover others interested in similar areas of research. When a potential collaborator is identified the next step is to determine whether the facility can provide the needed population for the study. It is important to discuss the general goals of the study and the likely inclusion/ exclusion criteria; then ask for a careful determination of the numbers of patients within the system who would have met the criteria during the last year. If sufficient numbers of patients are available to warrant the effort and cost of including the site, the next consideration is the policies of the system. Health care systems generally recognize the value of research in improving patient outcomes and therefore are supportive of research, but all have policies intended to protect their system, patients, and staff. Many facilities require that a staff member be on the research team if data are to be collected at the site. Some have more stringent requirements; for example they may require that the employee be listed as a co-I or even as the PI or co-PI. Other facilities stipulate that the facility investigator be a Ph.D. or M.D. Whatever the stipulations, it is important to consider the potential advantages and disadvantages of each system’s participation. For the Leukemia Pain Study, five potential facilities were considered for participation in the study. All were identified because of previous networking with nurses or psychologists in the facilities and were participating in standardized protocols for treatment of leukemia. After varying periods of communication with the sites exploring the feasibility of collaboration, one was not included because of lack of support for nursing research and another because of an inability to reach agreement on the ethical appropriateness of the protocol. At three facilities the estimated number of potential subjects was adequate, a clinical nurse researcher and two clinical nurse specialists were interested in participating, and the support for nursing research was strong. Interestingly, despite careful determination of potential subjects, at one of these facilities, no children were diagnosed with

MULTISITE RESEARCH

leukemia within the first year of data collection and the facility was withdrawn from the study. In the final count, three facilities, all children’s hospitals located in Southern California, participated in the Leukemia Pain Study. The title given to the key collaborator from each additional facility will vary (Bratt, 2000, Minnick et al., 1996), dependent on the study structure and the requirements of the facility. Whatever the label, this person is integral to participation of the site. Although not as heavily involved in the development of the science, he or she generally is an expert in the clinical area and plays an important role in the refinement of the study plan and ensuring that the protocol is appropriate for that site. Additionally, this person will be responsible for implementation of the study at the site and for the integrity of the data. Additional members of the team are determined by the needs of the study. Depending on the design of the study and the type of data to be collected, research associates may join each of the site teams for data collection and processing. Other members may include a statistician, project coordinator, secretary, and transcriptionist, as necessary to accomplish the work. In the Leukemia Pain Study, within the overall system of the study, there were four major subsystems: the subsystem of the three primary investigators, and three subsystems focused on data collection, one at each of the sites, each consisting of the PI/co-PI or collaborating co-I and several research assistants. Negotiation of Responsibilities and Benefits When potential collaborators and facilities have been identified, pertinent aspects of the study must be negotiated. Negotiation is best approached if it is not considered to be an adversarial interaction, but rather a “consideration of mutual interests and expanding available options” (Norton & Grady, 1996, p. 264). Applying the principles of Fisher and Ury (1991), collaborators within a research study have the common goal of engaging in a project for the purpose of discovering knowledge (Broome, 1999). This shared interest may be used to identify the principles of collaboration necessary to facilitate a successful study, identify various solutions to areas needing negotiation, and move to agreement. Objective criteria should be developed that will guide evaluation of achievement of study goals and identify any areas needing renegotiation. Areas likely to be negotiated include both responsibility in relation to the study and benefits to be derived from the study.

41

Negotiation of responsibilities. The responsibilities of each subsystem involved in the project must be clear before the study begins. Among the issues are finalization of the proposal, funding and site specific costs, Institutional Review Board (IRB) approval, preparation of materials, and selection and training of study personnel. Although the core research proposal should be developed before identifying collaborators, flexibility is necessary. The collaborators need to take the responsibility to carefully consider each aspect of the proposal and the feasibility and practicality of implementation at their facilities. Study procedures can be finalized only after careful examination of the policies and environment of each facility. This may result in modification of not only the procedure, but also refinement of the aims, hypotheses, and study questions. Negotiation may be necessary at this point as the vision of the principal investigators is closely critiqued by the input of the collaborators (Ingersoll et al., 1995). Remembering the common goal of the study and negotiating elements to determine the method most likely to accomplish the purpose is important to establish the collaborative group process. Research is expensive and funding must be negotiated. An early decision that must be made by the PIs is the source of support. Many studies are funded by small grants. Possible sources would include research support money from the facility; nursing organizations such as Sigma Theta Tau and the Society of Pediatric Nurses; associations with a specific focus such as the American Cancer Society; and drug or equipment companies. It must be negotiated early whether the principal investigative team will be responsible for obtaining all funding, if the collaborator at each site will take this responsibility, or another arrangement. It will also be necessary to decide who will be responsible for the general study costs such as preparation of data collection materials. It should be understood early in the negotiation that if small grant funding is used, only the basic costs of the study would be covered; other expenses, such as salaries and overhead, will be absorbed by the facility. Alternately, if funding is to be sought from a major source such as the National Institutes of Health (NIH) or other such organization, the primary team is responsible for the application for funding. This includes determining the direct costs (expenses related to the research project such as salary of the data collector and fee for statistical analysis) and the indirect costs or overhead (expenses of the institution relating to the research such as office space and

42

electricity) for the total study and each individual site. The direct costs must be reasonable and justified in the proposal. The indirect cost is predetermined by the institution and will vary between facilities, so must be ascertained, clearly stated, and used to determine the cost for each individual site when preparing the budget. A signed subcontract from each collaborating facility must be included with the proposal submission. After funding, the principal investigator is responsible for total budget management (Bergstrom et al., 1984); the site coinvestigators for managing and accounting for the funds allocated to the site. For the Leukemia Pain Study, both types of funding, the small grant and NIH funding were sought. The primary investigative team first submitted a proposal to the Oncology Nursing Society that was funded and used to collect data on a small sample at the primary site. This work showed that the protocol was feasible and could provide the data needed to answer the research questions. This early work provided support for submission of the protocol to the National Institute of Nursing Research, which was subsequently funded, providing support for the full study. A major factor when negotiating responsibility for aspects of the study is IRB approval (Bergstrom et al., 1984; Bratt, 2000; Koebler, Miller Vojir, Hester, Foster, 1997). The PI has the oversight responsible for the ensuring that the approvals have been obtained. Individual approvals from the IRB at each participating site are needed, except in the uncommon event that a facility signs an NIH certificate approving the IRB at the primary site to review for them. The coinvestigator at each site generally will take the responsibility for obtaining the IRB approval, as most facilities require that the person requesting permission be employed by the system. To facilitate the IRB submissions, the PI should provide a master copy of the proposal via e-mail or on disk to each coinvestigator, including sample consent forms. The information then may be reformatted as required by that IRB. It is important that coinvestigators allow for sufficient time to obtain all necessary approvals within their system. Letters of support are usually needed from the medical and nursing directors of the areas to be used. Additionally, it may be necessary to obtain approval from the Nursing Research Committee, corresponding medical committee, and the grants management office before submission of the protocol to the IRB. In some facilities approvals may be obtained concurrently, in others a definite order is stipulated. The proposal review and re-

BOSSERT ET AL

sponse from each of these individuals and committees may take a number of weeks or months. Data collection cannot begin until all necessary approvals are obtained from that facility. After approval, all involved in the study must be alert for any adverse events pertaining to the subjects that might be associated with the study and report these immediately to the PI for evaluation and possible protocol modification with IRB review. Additionally, the coinvestigator must continue to assume responsibility for any subsequent interactions with the IRB, such as the annual review for extension of the protocol, and for closure with the IRB when the study is completed. Copies of all IRB approvals must be kept in the PI’s study files and a system implemented to track all renewals. Because increasing the number of sites for data collection increases the possibility of variation in data collection, it is a responsibility of all involved to ensure that the study procedures are followed precisely (Bergstrom et al., 1984). Negotiation regarding the study procedure has already occurred, but further negotiation regarding adequate training for all study personnel is necessary. The complexity of the study and the instruments will determine the length and content of the training. To facilitate this process, the principal investigative team should develop a training manual to be used at each site (Bratt, 2000; Minnick et al., 1996; Smith, Salyer, Geddes, Mark, 1998). This manual should operationalize the protocol and include other pertinent information regarding the study (see Table 1). Dated addenda to the manual should be distributed as any changes or clarifications are needed during the process of the study. Ideally, the PI and co-PIs will conduct the initial training at each site, with the site coinvestigator managing any subsequent training needed. The training should emphasize integrity in all aspects of the study, and a method of checking the veracity of the data should incorporated in the design, such as audio tape recording of interviews, or random spot checks of clinical data. The results are generalizable only if subjects are truly representative of the desired population, the protocol is followed carefully for each subject, and the data are accurate. The PI has the responsibility for the preparation and distribution of all data collection forms to the sites, ensuring consistency of data collection. After data collection, a plan is needed for sending the data in a timely and safe manner to the site responsible for data analysis (Smith et al., 1998). The frequency of return mailing is dependent on the quantity of material in each packet and the rate of

MULTISITE RESEARCH Table 1. Suggested Contents for Training Manual Section 1: Administrative matters Role description of study personnel categories, IE PI, Co-I, research assistant List of study personnel, facility, address, phone, fax, e-mail Number of subjects to be gathered from each site Timeline for study Section 2: Study background and purpose Brief background of problem Purpose statement Aims, hypotheses, research questions Section 3: Protocol for data collection Identification of subjects Subject inclusion/exclusion criteria Frequency of checking for subjects (daily, weekly, and so on) Process for identifying potential subjects Informed consent process Script for initial contact with potential subject Copy of consent form for that site Script for presenting study protocol and answering questions Signatures and copies of consent form according to facility policy Assignment of study ID number Record of refusals, reason, and basic demographic data Data collection procedure Brief overview of process Number of data collection points Location of data collection Supplies needed (IE pen, tape recorder, blood pressure equipment, and so on) Instruments Description of each tool (purpose, brief psychometrics if available) Instructions for administration (researcher or selfadministered, and so on) Order of administration (with checklist to be used with each subject) Data collection process Subject ID number on each data form Script for data collection with directions and supplies for each tool Suggestions to facilitate process (positive reinforcement, keep directions simple, and so on) Script for closure expressing appreciation Data processing Guidelines for review of data for completion and accuracy Form for recording notes pertinent interpretation of data Procedure for mailing data Integrity of the study Scientific honesty/dishonesty statement Appropriate informed consent process Procedure for ensuring valid data Confidentiality of consent forms and data Secure storage of consent forms and data Section 4: Answers to frequently asked questions Anticipatory guidance for issues such as retention/withdrawal of subjects, etc

data collection. All data should be returned by a mailing system with a strong record for accurate delivery, minimal loss of packages, and if possible, a tracking system. If there are any doubts about

43

mailings, copies of the data should be made and kept at the site until delivery is confirmed. A recent alternative to mailing data is actual data entry by each site, necessitating shared access to software and specific guidelines. For the Leukemia Pain Study, the use of a training manual proved invaluable in ensuring that all participants had the same understanding about the study purpose and procedures. The actual training sessions, using a combination of didactic information and roleplaying, were a positive event, enhancing the sense of collaboration among team members. A mailing system with an on-line tracking system proved a reliable method of returning all data to the primary site. Negotiation of benefits. Possible benefits of the study include recognition for involvement and dissemination of study results. It should be known at each facility that is participating in the study at that site since approval for including research in the workload is needed; thus providing an internal mechanism for recognition. Beyond this, letters of recommendation may be provided to support promotion, awards, or academic applications written by the site coinvestigator or PIs. If appropriate, certificates of participation may be distributed on completion of the study. Additionally, project staff may be mentioned by name in the acknowledgement section of published work. Perhaps the most important negotiation point concerning benefits is the question of rights in dissemination of the study results. This issue must be discussed early during negotiations between sites and agreement reached (Bratt, 2000; Erlen, Siminoff, Sereika, & Sutton, 1997; Hagle et al., 1987; Hanson, 1988; Nail et al., 1998; Thiele, 1989). It is helpful to list probable reports that will be written. The PI generally takes authorship and responsibility for the primary report of the results. Co-PIs may take responsibility for reports focusing on a specific aspect of the study or process. Coinvestigators at the various sites may decide to focus on case studies or some other aspect that is site specific. On all reports based on the study, the PI (and possibly the co-PIs depending on the organization of the team) may be listed as coauthors. The person who has conceptualized and written the report is the first author and others are listed based on the degree of their involvement in the study and report. All persons listed as authors must take responsibility for reviewing the report, because it will reflect on their professional work and reputation. If a report is developed and written by the coinvestigator at a site, that person will be the first

44

BOSSERT ET AL

author and authorship is also given to any others at the site who were integrally involved in that aspect of the work and the members of the core team. For articles written by a coinvestigator, the PI needs to approve the idea and be involved in the process to ensure accuracy and appropriateness of the results. A time frame, suggested between 6 months (Dunn & Chadwick, 1999) and 2 years (Bergstrom et al., 1984) should be negotiated for the preparation of manuscripts. If not completed, the PI should discuss the delay with the person and, if necessary, use her authority as PI to reassign that article to another participant. The PI must be kept informed of the status of all manuscripts resulting from the study. One method of assuring this is to require that all papers are sent and returned to the journal editor through the PI. This provides a content and quality check of the study results. Authorship for any oral presentations or posters would follow the same principles. In all presentations, sources of funding must be acknowledged. These principles were followed in determining potential publications and presentations from the Leukemia Pain Study. CONDUCTING THE RESEARCH PROJECT Clearly, an extensive sequence of actions and planning need to occur before a multisite research project can be initiated. However, all of the preparation before a study’s initiation cannot completely anticipate the challenges that may occur during the actual conduct of the study. Therefore, after structuring the multisite research, the investigative team must strive to foster a productive and positive experience among those involved in the conduct of the study. The systems approach, particularly the concepts of input/output/feedback and hierarchies, provides a method of remaining in contact with every aspect of the study whereas working to accomplish the goals of the study. Specific considerations include communication, problem solving, and developing a positive collegial spirit. Communication Communication is the cornerstone of a successful project, especially when multiple sites are involved (Bratt, 2000; Broome, 1999; Grant et al., 1998; Hagle et al., 1987; Koebler, 1997; Nail et al., 1998). Communication requires a dynamic exchange of information involving input, processing, response and feedback. Clearly, to be a dynamic exchange, the communication must be multidirectional with all persons engaged in both giving and

receiving information. For this to have a positive outcome for the research project, an atmosphere of openness and trust must be fostered in which each person feels their input will be seriously considered in light of the needs of the study. The system that is engaged in this process is an open, living system, the interchanges providing nourishment and growth essential for the conduct of a successful multisite project. If good communication is not maintained, the exchange will become static and the various subsystems will move toward dysfunction and the research project will not succeed. To maintain good communication within the research team, appropriate methods of accessing other personnel must be established. Multiple methods may be used for communicating, each appropriate for different situations (Bergstrom, 1984; Bratt, 2000; Hagle, 1987). The gold standard of communication is face-to-face interaction allowing for both the verbal and nonverbal content to be exchanged, the latter accounting for 65% to 93% of total communication (Wood, 1996), with the spoken word only a small portion of total communication. However, the reality of multisite research is that the proximity necessary for faceto-face communication is not often feasible. Thus other forms of communication must be used. Telephone conversations allow direct exchange of information and include a secondary level of communication through the nuances of voice, but may need to be followed up with a paper trail detailing the discussion and decisions. A useful variation on the telephone call is the conference call, which allows direct, simultaneous communication among sites, avoiding the miscommunication possible when relaying messages between sites. For the best outcome this call should be approached as a meeting, chaired by the PI, with a specific agenda, and followed up with minutes. Postal mail allows for the transfer and sharing of original documents that must be distributed between sites, but provides a slower rate of delivery than other methods. Fax allows for rapid transfer of hard copies of documents but does not provide a high quality document or original signature. The Internet allows for a rapid exchange of information, incorporating, to a lesser degree, the nonverbal nuances of telephone conversation (e.g., emoticons), the paper trail of postal mail, and the rapid transfer of high-quality documents. However, care must be taken that the message sent is clear and not clouded by irony, puns, comic twists or other literary tools that may be misunderstood without nonverbal cues such as a lifted eyebrow or other body language indicating

MULTISITE RESEARCH

incongruity. Surveys and questionnaires offer an anonymous opportunity to safely disclose information, feelings, and frustrations that may be difficult to communicate directly. If used, follow-up should occur quickly for the input to be used positively for the study. For the Leukemia Pain Study, all of the above methods were used, varying according to the current need. Problem Solving It is not possible to predict every challenge that will develop during the conduct of the study, but several situations are not unusual. During the course of a project, study personnel may have a change in job description or leave the facility, terminating their participation. Anticipatory planning would include requesting early notification from the collaborator if this appears likely and having a list of possible replacements at each site that are qualified to participate. If a replacement is not found, it may be necessary to drop that site. This may lead to a second potential problem, difficulty obtaining the needed number of subjects. Despite careful assessment of the population at each site before beginning a study, it generally takes longer than projected to accrue subjects. Problem solving for this may include additional sites that could be added to the study (if the money can be shifted or increased) or lengthening the time allotted for data collection if acceptable to the funding agency. Another challenge that may arise is lack of follow-through by study personnel. When the PI becomes aware that a site is not performing as expected, immediate communication is necessary to determine the cause and to reestablish accountability (Broome, 1999). One common cause is that work responsibilities of those likely to engage in research are heavy and these must be balanced with the demands of the research project (Hagle, 1987). If this is the challenge, the PI needs to help the collaborators maintain their commitment to the project and find alternate, creative ways of accomplishing the research. There are many other possible causes of lack of follow through and the solution will vary with the problem: allowing some slack time if the collaborator is ill, attempting to reinstill motivation if interest has decreased, or as a last resort identifying replacement personnel for some unable to meet their commitment to the study. Finally, because systems are dynamic, conflicts are likely to arise during the study that could threaten the outcome. At different times the focus

45

may need to be placed on different parts of the system, such as institutional requirements, the coinvestigator’s challenges, or the needs of data collectors. Lack of awareness of the various levels of the system leaves open the possibility that project needs will not be met. Therefore, the PI must maintain a flexible perspective on the entire system and be keenly aware of the team’s needs and concerns at each moment. Examining current challenges and deciding which must be addressed immediately and which can be addressed later is crucial. For example, a site struggling with internal conflict between the research team and the institution’s administrative policies needs a call from the principal investigator immediately to discuss the problem and provide support and suggestions for problem solving. Ongoing communication is necessary to identify any problem early. Negotiation may be useful to resolve the situation in a manner that will enable the continued progress of the study. At other times the PI will need to determine the necessary course of action and work with the persons involved to accept and implement it. As with most studies, several situations developed during the Leukemia Pain study requiring problem solving. Acquiring sufficient subjects became a challenge from the start. One facility was dropped because no children diagnosed with leukemia were admitted during the first year of the study, additionally a second facility was late starting data collection because of a pending hospital accreditation site visit. To compensate for these problems, the period for inducting subjects was extended by 1 year and the budget adjusted accordingly. Another problem that emerged at one facility was a gradual change in data collection. The investigative team met with the collaborator and research assistants at the site, reviewed the protocol, worked together to find methods of following it given the clinical situation, and spent time together over lunch talking about their clinical experiences. After this interaction a review of the data showed that the problem had been successfully addressed. Developing a Positive Collegial Spirit Creating a positive experience is essential for multi-site research (Minnick et al., 1996). People are likely to engage in behaviors that are rewarding and to avoid those that are not. Additionally, creating a positive experience will enhance the unity of purpose needed to facilitate the research. A key to developing a positive collegial spirit among team members is to identify and value each per-

46

son’s expertise (Bratt, 2000). The multisite team will come together with varied research experiences, differing clinical expertise, diverse worldviews, and possibly differing disciplines (Grant et al., 1998; Hagle et al., 1987). Although the team’s wealth of expertise and knowledge will be beneficial, it may also result in differences that complicate the project. It is imperative that the PI develops and maintains organization in the midst of complexity, or else the collegial spirit will slowly begin to dissipate. It is the responsibility of the PI to provide motivation and organization, to maximize the talents, experiences and wealth of knowledge that the multisite team has to offer, in short to guide the group process so that the whole is truly greater than the sum of the parts. Just as within the system of an orchestra the conductor is needed to guide the performance, the multisite team needs the principal investigator to “conduct” the group in such a way that the potential of every member of the system is maximized and in a balanced relationship with the rest of the group. Implicit in this is thoughtful consideration of the potential contribution of each person before inviting him or her to become a team member and, after joining the group, working to ensure that the unique talents of that person are incorporated in the study. Respecting and valuing each member of the team will create a sense of self-worth and collegiality that will foster cohesiveness and mutual accountability in the team’s progress towards its goals. In the Leukemia Pain Study, the team members came with a variety of nursing expertise (researchers, clinical nurse specialists, and clinicians) and also from psychology and medicine. Each member brought unique abilities that contributed to the success of the study. For example, at one of the sites, graduate students from the university’s department of psychology were integral members of the team, participating in data collection and analysis. They brought their expertise in interpersonal communication, research methods, and the psychosocial needs of a chronically ill child to the study. At other sites, nurses with years of clinical experience in pediatric oncology were strong team members, contributing as collaborators or research associates. Although each group brought valuable assets to the study, each also had areas needing development: the psychology students in the specifics of the clinical situation and the nurses in aspects of the research process. Every study will require a slightly different utilization of the elements described earlier — ensuring effective communication, problem solving, and

BOSSERT ET AL

creating a collegial spirit. These elements are not mutually exclusive; rather, they should be viewed as interconnecting fibers, woven together to create an extraordinary tapestry, a multisite research study. CONCLUDING A MULTISITE RESEARCH PROJECT As the accrual of subjects approaches the desired number, each site will anticipate the announcement that enrollment is complete. When this occurs, the final phase of interaction between the various systems involved in the study begins. This phase includes issues pertaining to data, findings, dissemination of the results, and discussion of future research and collaboration. As the primary site processes the data, questions will often arise that need clarification, such as the interpretation of scores for data entry, missing data, or outlying scores that need verification. Continuing communication is necessary to resolve these points. It is important to remember that however meticulous the participants are at the sites, human error may result in discrepancies. Maintaining collegiality is essential to accomplish the study purposes as data issues are resolved. The results of the data analysis will be eagerly anticipated by all involved in the study. Because there may be considerable time between the end of data collection and the completion of data analysis and interpretation, coinvestigators need to be kept aware of the status of the work. As soon as possible the study results should be shared with the various facilities. A telephone call to each coinvestigator shows collegial courtesy and professional respect. These persons have contributed considerable thought and time to the project and deserve early notice of the results. However, all must agree to keep the results confidential until the findings have been formally released. During this interim period, each coinvestigator should be given a copy of the data collected at that site, including both the raw data and site-specific analyses, both on disk and hard copy. This will allow internal examination of the data by the system as well as comparison with the overall study results. Each site retains the right to the confidentiality of its data. Comparisons between sites, other than basic analyses done to determine appropriateness of pooling data, may be made only if this has been agreed to by all systems involved. As noted earlier, decisions on authorship of the articles should be made when initially negotiating with the various coinvestigators and facilities.

MULTISITE RESEARCH

47

When the study is complete, these will need to be reviewed and revised as necessary. Careful decisions must be made about the presentation of the results, because some conferences will accept only results that have not been previously disseminated in any form. The PI and co-PIs should offer to visit each of the data collection sites and, in conjunction with the coinvestigator at that site, present the results to the appropriate group, including both the presentation of general results and site-specific results. A final and important step is discussion between all the investigators about the implications for further research and the desirability of continuing further collaborative work. As in the initial phase of the study, the coinvestigators’ opinions and insights will provide indispensable direction for the next phase of the research. Clinical insights and variations in viewpoint caused by differing experiences and facility needs will help ensure that the next study will be clinically relevant. During the discussions it will become evident which participants are interested in continuing the collaboration. Consideration of future interest and current effectiveness in the study will facilitate development of the next project. Thus a collaborative research team may be developed between various systems to accomplish the research that could not have been done at a single facility. In the Leukemia Pain Study, subject enrollment has stopped, but the final data points for the last subjects are not yet completed. Data cleaning and entry has been ongoing throughout the study, and content analysis of the interviews is in process, but the statistical analysis

will not be run until all subjects are completed. The sites have already discussed the next step in the program of research, agreed to continue collaboration, and the proposal is in preparation. SUMMARY Incorporating the systems approach provides a method of organizing the thinking, planning, and implementation necessary when conducting a multisite research project. This approach is cyclical and multifaceted, requiring both flexibility and an ongoing awareness of the dynamic nature of the process. Although human nature tends towards the simplistic approach, such as a list of steps needed, a multisite research project demands a more sophisticated approach. In this article the systems approach has been used to examine three aspects of the process: careful structuring of the project, purposeful incorporation of principles that will facilitate the conduct of the project, and collegial closure of the project, all necessary for the success of a multisite study. The systems approach described here provides direction for the facilitation of a complex multisite study. ACKNOWLEDGMENT The authors acknowledge and appreciation is expressed to those who collaborated in the study as coinvestigators, Kathleen Adlard, MN, RN, of Children’s Hospital Orange County and Pauline Beecroft, PhD, RN, FAAN, of Childrens Hospital Los Angeles, and to all who participated in data collection.

REFERENCES American Association of Colleges of Nursing (1997). Education and practice collaboration: Mandate for quality education, practice, and research for health care reform. Journal of Professional Nursing, 13(2), 129-133. Bergstrom,. N., Hansen, B. C., Grant, M., Hanson, R., Kubo, W., Padilla, G., & Wong, H. L. (1984). Collaborative nursing research: Anatomy of a successful consortium. Nursing Research, 33(1), 20-25. Birenbaum, L. K., Hinds, P. S., & Clarke-Steffen, L. (1995). Multisite qualitative nursing research in pediatric oncology. Journal of Pediatric Oncology Nursing, 12,(3), 135-139. Bratt, M. M. (2000). Collaborative institutional research. Journal of Child and Family Nursing, 3(4), 251-258. Broome, M. (1999). Collaborative research: The art of negotiation. Journal of Child and Family Nursing, 2 (1), 1-2. Broome, M., Rehwaldt, M., & Fogg, L. (1998). Relationships between cognitive behavioral techniques, temperament, observed distress, and pain reports in children and adolescents during lumbar puncture. Journal of Pediatric Nursing, 13(1), 48-54. Carlson, K. L., Broome, M., & Vessey, J. A. (2000). Using distraction to reduce reported pain, fear, and behavioral distress

in children and adolescents: A multisite study. Journal of the Society of Pediatric Nursing, 5(2), 75-85. Chenitz, W. C., Sater, B., Davies, H., & Friesen, L. (1990). Developing collaborative research between clinical agencies: A consortium approach. Applied Nursing Research, 3(3), 90-97. Dunn, C. M., & Chadwick, G. (1999). Protecting study volunteers in research; A manual for investigative sites. Boston: CenterWatch. Erlen, J. A., Siminoff, L. A., Sereika, S. M., Sutton, L. B. (1997). Multiple authorship: Issues and recommendations. Journal of Professional Nursing, 13(4), 262-270. Fawcett, J. (1995). Analysis and evaluation of conceptual models of nursing (3rd ed.). Philadelphia: F. A. Davis. Fisher, R., & Ury. W. (1991). Getting to yes: Negotiating agreement without giving in (2nd ed., B. Patton, Ed.). New York: Penguin Books. Friedman, , L. M., Furberg, C. D., & DeMets, D. L. (1996). Fundamentals of clinical trials (3rd ed.). St. Louis, Mosby. Froman, R. D., & Owen, S. V. (1999). American and Korean adolescents’ physical and mental health self-efficacy. Journal of Pediatric Nursing, 14(1), 51-58. Grant, M., Anderson, P., Ashley, M., Dean, G., Ferrell, B., Kagawa-Singer, M., Padilla, G., Robinson, S. B., & Sarna, L.

48

(1998). Developing a team for multicultural, multi-institutional research on fatigue and quality of life. Oncology Nursing Forum, 25(8), 1404-1412. Hagle, M. E., Barbour, L., Flynn, B., Kelley, C., Trippon, M., Braun, D., Beschorner, J., Boxler, J., Hange, P., McGuiore, D., Bressler, L., Kirchhoff, K. (1987). Research collaboration among nurse clinicians. Oncology Nursing Forum, 14(6), 55-59. Hanson, S. A. (1988). Collaborative research and authorship credit: Beginning guidelines. Nursing Research, 37(1), 49-52. Ingersoll, G. L., Brooks, A. M., Fischer, M. S., Hoffere, D. A., Lodge, R. H., Wigsten, K. S., Costello, D., Hartung, D. A., Kiernan, M. E., Parrinello, K. M., Schultz, A. W. (1995). Professional practice model research collaboration. Journal of Nursing Administration, 25(1), 39-46. Koebler, J. A., Miller, K. L., Vojir, C. P., Hester, N. O., Foster, R. L. (1997). Multisite clinical research: A challenge for nursing leaders. Journal of Nursing Administration, 27(7/8), 42-48. Koshar, J. H., Lee, K. A., Goss, G., Heilemann, M. S., & Stinson, J. (1998). The Hispanic teen mother’s origin of birth, use of prenatal care, and maternal and neonatal complications. Journal of Pediatric Nursing, 13(3), 151-157. Krulik, T, Turner-Henson, A., Kanematsu, Y., Al-Ma’aitah, R., Swan, J., & Holaday, B.(1999). Parenting stress and mothers of young children with chronic illness: A cross-cultural study. Journal of Pediatric Nursing, 14(2), 130-140. Minnick, A., Kleinpell, R. M., Micek, W., Dudley, D. (1996). The management of a multisite study. Journal of Professional Nursing, 12(1), 7-15. Nail, L. M., Barsevick, A. M., Meek, P. M., Beck, S. L.,

BOSSERT ET AL

Jones, L. S., Walker, B. L., Whitmer, K. R., Schwartz, A. L., Stephern, S., & King, M. E. (1998). Planning and conducting a multi-institutional project on fatigue. Oncology Nursing Forum, 25(8), 1398-1403. Norton, S. F., & Grady, E. M. (1996). Change Agent Skills. In A. B. Hamric, J. A. Spross, & C. M. Hanson (Eds.), Advanced nursing practice: An integrative approach (pp. 249271). Philadelphia: W. B. Saunders. Putt, A. M. (1978). General systems theory applied to nursing. Boston: Little, Brown, & Company. Smith, C. S., Salyer, J., Geddes, N., & Mark, B. A. (1998). Strategies to enhance internal validity in multi-center longitudinal research. Outcomes Management for Nursing Practice, 2(4), 174-179. Smith, D. (1988). Collaboration in nursing research—A multi-disciplinary approach. International Journal of Nursing Studies, 25(1), 73-78. Thiele, J. A. (1989). Guidelines for collaborative research. Applied Nursing Research, 2(4), 150-153. Tores, G. (1986). Theoretical foundations of nursing. Norwalk, CT: Appleton-Century-Crofts. Van Cleve, L., Bossert, E. A., & Savedra, M. (1997). Children with leukemia: Pain, management and outcomes (RO1NR04201), unpublished data. von Bertalanffy, L. (1972). The history and status of general systems theory. In G. J. Klir (Ed.), Trends in general systems theory (pp 21-42). New York: John Wiley & Sons. Wood, J. T., (1996). Everyday encounters: An introduction to interpersonal communication. Belmont, CA: Wadsworth.