OBJECTIVES : Framework synthesis (FS) is a method for synthesis of qualitative research. It can also be used to integrate quantitative and qualitative data, to explore complex healthcare issues. We used FS to explore evidence (identified via systematic review) on alternative causes of diarrhea in patients with neuroendocrine tumors (NETs) and approaches for differential diagnosis (DDx), for which little quantitative data is available.

METHODS : A preliminary framework was developed through a scoping search of the literature and discussion with clinical experts. Any article containing quantitative or qualitative data relevant to DDx of diarrhea in NET patients was eligible, unrestricted by study design. Qualitative data included statements from any section of eligible articles. Two reviewers independently coded and indexed data from each article against the framework using Docear software. Novel themes which emerged from the data were incorporated into the framework iteratively. Due to substantial heterogeneity in study design, formal quality assessment was not conducted.

RESULTS : Data from 47 articles were included in the final framework. Information supporting all predefined themes was identified from observational studies, case reports, guidelines and narrative reviews. Four novel themes emerged, and data were often indexed to more than one theme. A theoretical chronological framework including approaches for DDx of NET diarrhea was developed. Most data were qualitative and sourced from discussion sections; therefore, findings may have been subject to author opinion or lacked supporting evidence. However, findings provided valuable insight on the importance of DDx of diarrhea in NET patients and highlighted diagnostic approaches to further investigate for use in clinical practice.

CONCLUSIONS : We demonstrate a novel use of FS, allowing comprehensive synthesis of heterogeneous data from a systematic review. However, there are limitations associated with evidence quality and assessment. This adaptation of FS may be most appropriate for unexplored or complex research topics, where little high-quality evidence is available.