The CANadian Pediatric Weight management Registry (CANPWR): lessons learned from developing and initiating a national, multi-centre study embedded in pediatric clinical practice

Our study began as a pilot (2010–2012) that was funded by the Canadian Institutes of Health Research (CIHR) through the Canadian Network and Centre for Trials Internationally program (details available at: www.​cannectin.​ca). As a pilot, CANPWR included five Canadian tertiary-level, multidisciplinary, pediatric weight management clinics (BC Children’s Hospital in Vancouver, BC; Stollery Children’s Hospital in Edmonton, AB; Children’s Hospital of Eastern Ontario in Ottawa, ON; CHU-Sainte Justine in Montreal, QC; McMaster Children’s Hospital in Hamilton, ON). Data collection was supported by a central coordinating site (Population Health Research Institute, McMaster University, Hamilton, ON). Our pilot study was designed to assess a number of factors that would inform a larger-scale study, including acceptability (e.g., Can sites agree on a core set of variables and measurement protocols for data collection? Does data collection for research purposes burden or complement data capture for clinical purposes?) and feasibility (e.g., Are sites able to successfully enrol participants into CANPWR from the sample of boys and girls referred to the clinics? Are sites able to retain participants to enable longitudinal research data collection and analysis, regardless of whether individuals discontinue clinical care for weight management?). Data collected during our pilot were compared and contrasted with normative Canadian data, which highlighted the increased health risks present in children referred for weight management [4].

Building on our pilot experience and supported by a CIHR operating grant received in 2012 (Principal Investigator: KMM), CANPWR expanded to eight sites (original five sites, plus The Hospital for Sick Children in Toronto, ON; ICAN clinic in Toronto, ON; Montreal Children’s Hospital in Montreal, QC). Two newly established clinics (Alberta Children’s Hospital in Calgary, AB; Trillium Health Partners in Mississauga, ON) joined CANPWR in 2016, which brought the total number of sites to 10; subsequently, one site (ICAN clinic) became inactive in 2016 due to changes in the clinical practice.

Monthly teleconferences with research coordinators and investigators were recorded and summarized in minutes. For this manuscript, the research coordinator reviewed these minutes, extracted key themes for each call and summarized the main themes. Based on this, a survey was developed and sent to investigators and research coordinators at each site asking if all key processes and challenges of conducting the study were included (Additional file 1). The final edited survey was circulated electronically to team members at the CANPWR clinics between June and August, 2016. Responses were received from team members at all nine active sites. With these data (see Table 1) and supplemented by information from quality control processes referred to in Lesson 4, our manuscript authorship group summarized the findings as lessons. They used an online communication tool (www.​slack.​com; SLACK Offices, Vancouver, BC) to facilitate communication and host virtual meetings to author our manuscript. A summary of our results, categorized as lessons learned, is provided below.

The observed challenges of conducting multi-site clinical trials [10] are also relevant to observational studies such as CANPWR. As highlighted, specific requirements and length of time for approval from research ethics boards can vary site-to-site. As recommended, for our pilot study, we first sought and received research ethics approval at our central coordinating site in Hamilton, ON. The time that each site required to prepare and submit the application to their respective Research Ethics Board (REB) varied considerably (maximum difference of by up to 40 weeks). This delay occurred, in part, due to the requirement of one REB that the documents be translated into both English and French. This step was completed after the wording of the consent form and case report form (CRF) in English had been approved by the other REBs. After approval at the central coordinating site, the time from submission to approval at the other pilot sites varied from 5 to 17 weeks. As we transitioned from the pilot to main study, sites that were part of the original CANPWR pilot study had their REB applications approved in 2 to 10 weeks. The five sites that joined CANPWR for the full study had initial review periods of 1.5 to 12 weeks.

In addition to timelines, REB requirements influenced other study aspects including the type of data that could be collected and transmitted electronically to the central data centre. This was a result of varying definitions of personal identifying information and varying procedures for accessing clinical data for research. REB requirements also influenced study recruitment as three sites required clinicians to serve as intermediaries between families and our research team members, which included clinicians administering a consent to contact form with families. In other words, families were required to provide their written consent to be contacted by the research team. Because some site leaders perceived that this step might enhance recruitment, the consent to contact step was added at two additional sites. Other site-specific institutional regulations introduced variability in start-up time, including contract negotiations for accessing clinical data, data ownership, and budget. For instance, CANPWR research funds were received from the CIHR and held at our coordinating site in Hamilton. Individual contracts were then prepared with each site to transfer funds based on site-specific study activities (e.g., recruitment, data collection). At one site, contracts were required with both the academic institution and the regional health authority where the clinic was located, a requirement that delayed study initiation.

One of the operational aspects of CANPWR includes monthly teleconference meetings that are led by investigators from our coordinating site in Hamilton. Each month, separate coordinator and investigator teleconferences (duration: 1 h) are held to discuss practical, day-to-day issues as well as broader, academic topics, respectively. This communication has been complemented by an average of one site visit per site so far, made by coordinating site team members to support individual sites and encourage adherence to study policies and procedures. A second site visit is envisioned. These meetings also provided contextual information, serving to highlight the clinic- and research-related variability between sites. Teleconferences and site visits also enabled data management strategies and kept sites accountable to and engaged with all study team members.

The sites varied in their access to a trained research coordinator, especially at study start-up and staff turnover has been common. Three sites enlisted the help of students as either volunteer helpers or research assistants. This created exceptional learning and teaching opportunities as well as reduced costs. Student activities were always overseen by research staff and the implemented quality control measures ensured high quality data. Some challenges introduced by the use of students included frequent turn-over and less flexible schedules. Thus research staff had to organize recurrent training sessions. Ongoing communication with an available central coordinator through both scheduled and ad hoc communications helped to minimize the impact of staff and student turnover.

An initial goal of CANPWR was to be the first harmonized, evidence-based registry to identify the key determinants of weight change in pediatric weight management clinics across Canada. The CANPWR investigators designed measures based on the best available evidence at the time. Where strong empirical evidence was lacking, expert, group-level consensus was necessary for measure development. The initial CANPWR measures set was designed along these principles [3]. Through implementation in the pilot sites, we realized that several questions were burdensome and impractical. For example, questions relating to puberty assessment by physician evaluation were discontinued due to challenges in collecting these data. Another example included a narrow focus on cardiometabolic health outcomes in our pilot, so the Main Study included additional data collection on mental health and health-related quality of life. Consensus on unclear definitions of data elements was reached through discussion with research and clinical team members across study sites, facilitated by our monthly teleconferences. A record of all discussions and decisions was maintained by our central research coordinator (PM).

Although integrating research with clinical practice can be beneficial, there are challenges that can arise. For instance, clinicians have busy schedules. Because they are funded to deliver health services, their interest in research can vary. If there is a lack of interest or attention to details, participant data may be missed. In some clinics, having physical space dedicated to research staff can be limited, which can be sub-optimal for participants and families when collecting data. To mitigate these challenges, CANPWR was embedded into clinical practices, which minimized the time needed by clinicians to identify potential participants, enhanced the feasibility of collecting outcome data, and allowed clinicians to focus on caring for their patients while researchers collected the required data [11].

As CANPWR did not require participating sites to modify their clinical practices, variation in the timing and pace of data collection occurred. The 6-month visit was scheduled 6 months after the date the clinical team considered the beginning of the intervention. This may have been beyond 6 months from study recruitment if there were delays in commencing group sessions. To enable flexibility for study participants, the window for conducting the annual study visits was set broadly (6 months on either side of the calculated date).

To ensure the data used to support the research efforts were valid and secure, CANPWR required a robust data management system and related practices. The CANPWR pilot sites learned that integrating the CRF into routine practice improved data collection. Integration of CANPWR data elements into the workflow of clinic visits was accomplished using paper or electronic data collection forms that were shared among sites and modified by each site to meet their local needs. The study sites that integrated the study case report forms into their clinical practice found that, data elements were more consistently collected and available. The requisite data was then extracted from the clinical record and transferred (online) into the electronic case report form. The specific location of the clinical data extracted by each site for each variable on the study CRF was identified at study start-up and reviewed by the central coordinator during the site visit. Sites were expected to use a paper copy to extract the data from the clinical chart as an intermediate step in order that this could be used for verification if the data was questioned through the routine quality control checks either at the time of data entry if outside the variable limits set or in monthly quality control reports.. Using this system, few challenges were identified in entering data into the study web-based data application.

The coordinating centre implemented monthly performance reports to track incomplete data. Reports were used to identify performance gaps, monitor changes over time, and support continuous quality assurance. These reports were valuable to ensure data completeness.

CANPWR was designed to follow participants for three years, whether or not they were still engaged in clinical care to determine how health outcomes changed once care was completed and to reduce biases introduced by the high attrition rates often seen from weight management clinics [12]. Participants who discontinued attending the clinic, but remained enrolled in CANPWR, were seen at times that accommodated families and the location depended on family preference and space availability (in clinic, in clinic space but out of clinic time or in research space). They were not seen by the clinical team. Continued study participation did not preclude families from re-engaging in clinical care, and our anecdotal experience revealed that for a small number of families, their research engagement facilitated their re-starting clinical care. All laboratory values were shared with the families and their primary care provider. When designing CANPWR, we recognized that including follow-up data collection for three years would be challenging due to high attrition [13] and because follow-up for most childhood obesity treatment studies is ≤24 months [14]. In our communication with families, we were explicit when explaining that their CANPWR participation was separate from their clinical health services; however, at many sites, when families discontinued attending appointments for weight management, it was challenging to engage them in attending CANPWR study visits. When we surveyed our sites, the majority highlighted difficulties in tracking participants longitudinally, noting that increased research personnel time was required as the study progressed to maintain tracking, largely due to attrition from the clinical programs. To mitigate loss to follow-up, the CANPWR investigators gave families unable to attend an in-person visit, the option to complete follow-up study visits by telephone for self-reported measures only (e.g., current health, medication use, health behaviours). While it was intended that this practical option would be effective for collecting data from participants who discontinued clinical care, to date, less than 10% of follow-up visits were carried out that way. Despite having this mode of data collection available for families, attrition remains a substantial issue in our clinics, which highlights the challenges of this issue in successfully managing pediatric obesity [13].