Methods
Literature search
A comprehensive search strategy was developed from systematic reviews in collaboration with a health science librarian. We combined MeSH headings for cancer, children, survivor/long-term and Nordic countries supplemented with free text words. Further details on search strategies are provided in Supplemental Material. Relevant studies were identified by searching the databases PubMed, EMBASE and Web of Science and by manual review of reference lists in full-text articles as well as the clinical trials register, clinicaltrials.gov. Studies were screened based on title and abstract according to inclusion and exclusion criteria as described below.
Study selection criteria
We searched for studies in English published from January 1st 2010 - March 1st 2020 conducted within the Nordic countries (Denmark, Norway, Sweden, Finland and Iceland) and including an invitation to a clinical examination for assessment of late effects after childhood cancer. Participants had to be younger than 18 years at time of the cancer diagnosis, whereas no requirements to the minimum size of the study population was set up. The clinical examination should have been performed at least three years after the cancer diagnosis and could include any type of clinical examination such as echocardiography, endocrinologic or neurocognitive testing or radiologic imaging. Studies were excluded if no clinical examination was performed in the study population, and if data were solely based on questionnaires, blood samples, telephone interviews, online focus group interviews, or routine out-patient follow-up visits. We also excluded studies if part of the study population was above 18 years of age at diagnosis, or if the study was conducted outside the Nordic countries. Furthermore, reviews, surveys, guidelines or editorial letters were excluded in addition to posters and conference abstracts, since they typically do not describe the recruitment process in deltail. If more studies on the same cohort were published, only the first study published on the cohort was included.
Data extraction
Eligible full text studies were included if they met the predefined inclusion criteria. For each study, the following data were abstracted for the study population: participation rate, cancer type, sex distribution, time since diagnosis or stem cell transplantation (SCT), duration of the examination (estimated to more or less than three hours), the person who contacted the participant (doctor, nurse or other) and how the participant was approached (postal letter, mail or telephone). Four predefined diagnostic groups of childhood cancer were used: Hematological malignancies, central nervous system tumors, solid tumors, and mixed tumors defined as more than one diagnostic group. In case of missing data, corresponding or senior authors were contacted to obtain additional information.
Definitions of participation rate
Invited participants were defined as survivors receiving an invitation to participate in a late effect study including a clinical examination. Invited participants were further divided into accepted participants (invited participants who accepted the invitation), study non-participants (invited participants who declined or did not respond to the invitation), study dropouts (invited participants who accepted the invitation but did not complete the clinical examination) and study completers (invited participants who accepted the invitation and completed the clinical examination).
Statistical analysis
The study participation rate (SPR) was calculated as the ratio between the number of accepted and invited participants. In case-control studies, only the SPR for the cases was calculated and not for the controls. To address the potential effect of participant characteristics, we used a random-effect logistic regression model for the logit-transformed (log odds) participation rates. For studies with a 100% participation rate, the number of non-participants was set to 0.5. We investigated cancer type (hematological malignancies, central nervous system tumors, solid tumors or mixed diagnosis), time since diagnosis (more or less than ten years) and duration of the examination (more or less than three hours), separately. The results were presented in forest plots in which the logit-transformed participation rates were transformed back to the participation rate-scale with corresponding 95% confidence intervals (CI).