Non-dissemination and dissemination Bias in qualitative research: Empirical evidence and conceptualisation

Media type: E-Book

Title: Non-dissemination and dissemination Bias in qualitative research: Empirical evidence and conceptualisation

Contributor: Töws, Ingrid [Verfasser]; Antes, Gerd [Akademischer Betreuer]; Meerpohl, Jörg J. [Akademischer Betreuer]; Antes, Gerd [Sonstige]; Downe, Soo [Sonstige]

Corporation: Albert-Ludwigs-Universität Freiburg, Medizinische Fakultät

imprint: Freiburg: Universität, 2023

Extent: Online-Ressource

Language: English

DOI: 10.6094/UNIFR/233829

Identifier:

Keywords: Qualitative Methode ; Empirische Forschung ; Qualitative Sozialforschung ; Forschungsmethode ; Psychologie ; Umfrage ; Forschung ; Bias ; Gruppendiskussion ; Evidenz-basierte Medizin ; (local)doctoralThesis

Origination:

University thesis: Dissertation, Universität Freiburg, 2022

Footnote:

Description: Abstract: Background: Treatment effects in health-care are best studied by quantitative studies that use quantitative methods to assess the effects of an intervention and report their results as numerical data. Qualitative research in health-care is used to understand factors that affect whether interventions are acceptable and feasible and whether interventions are fully and effectively implemented and how people rate the importance of different health outcomes. In addition, qualitative findings can influence the conceptualization, development, and implementation of interventions in health-care. Findings from qualitative research can inform decisions on the use of evidence-based health-care interventions and contribute to policy decisions in these fields. Lastly, through qualitative research, theories can be developed and explanatory hypotheses can be created. Systematic reviews are a way of synthesising evidence from individual studies and summarise the available scientific evidence for a specific question by following a strict scientific methodology. In order to assess the certainty of the findings from reviews of quantitative studies of the effectiveness of interventions, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach was developed. Qualitative evidence syntheses are a way of systematically summarising qualitative evidence. In qualitative evidence synthesis, evidence from individual qualitative research studies addressing similar research questions or phenomena of interest is analysed and interpreted jointly. The GRADE-CERQual (Confidence in the Evidence from Reviews of Qualitative research) approach was designed specifically to assess how much confidence to place in findings from qualitative evidence syntheses. Dissemination bias describes a systematic error occurring from non-dissemination of studies and individual findings and the key concern with that is the (non-)dissemination of studies due to the nature of their content and message. Dissemination bias is one concern that potentially weakens the assumption of “high confidence”. Time-lag bias, language bias, grey literature bias, and truncation bias are underlying mechanisms that fuel dissemination bias. Empirical evidence has shown that a considerable proportion of quantitative research in clinical effectiveness research remains unpublished once completed. Because statistically significant and positive results have a higher probability of publication, it has been found that dissemination bias led to an overestimation of the reported effects of health interventions. Aim: First, to explore the extent of non-dissemination in qualitative research and stakeholders’ reasons for non-dissemination of qualitative research studies and individual findings. Second, to explore stakeholders’ views and experiences about the relevance of dissemination bias in qualitative research and how dissemination bias might be detected and interventions that might decrease dissemination bias in qualitative research. This includes reflection on the definition and delineation of dissemination bias in qualitative research and consideration of how dissemination bias might occur and how it can be detected. Methods: To address these aims, research with four main components was conducted. These include 1) a cross-sectional survey, 2) a follow-up cohort study of qualitative conference abstracts, 3) group discussions and 4) expert consultations. A cross-sectional survey approach was used to gather data regarding non-dissemination and dissemination bias in qualitative research from qualitative researchers, authors of qualitative evidence syntheses, peer reviewers and editors of scientific journals. Overall, the survey comprised 30 closed and open-ended questions. In order to identify suitable participants with experience in qualitative research, authors of published qualitative studies were searched for. A validated search strategy for qualitative studies was used in Ovid MEDLINE. Participants were also invited through various relevant mailing lists which contained researchers who were involved in qualitative research. The final survey was delivered online using Survey Monkey. All available responses were included in the final analyses and it was planned to report the overall response rate. A descriptive statistical analysis of frequencies and proportions was automatically generated by the survey software and used for the analysis of the survey responses. To explore relations between selected participants’ answer options, a bivariate Pearson correlation analysis was conducted in R Studio for the normally distributed data. Qualitative survey data were analysed using an inductive thematic analysis in MAXQDA. Empirical evidence for this thesis stems from a follow-up study of qualitative abstracts. The study sample was drawn from the conference proceedings of the United Kingdom Royal College of Nursing Research Conferences in 2015 and 2016. All abstracts accepted for oral presentations and posters were potentially eligible. To be included in this cohort study, all abstracts were scrutinised for the following criteria: i) description of a qualitative study, ii) description of a mixed-methods study or a programme description with at least one qualitative element iii) evidence syntheses of qualitative research findings. Subsequent full publications of the studies presented in abstracts were searched for in MEDLINE via OVID, in the Cumulative Index to Nursing & Allied Health Literature (CINAHL) and Google Scholar. Data from the cohort study were summarised with univariate descriptive statistics. Proportions and 95% confidence intervals (CIs) were calculated. Fisher’s exact test was calculated to examine associations between study characteristics and proportions of abstracts with a full publication. Logistic regression was used to evaluate an association between numerical study characteristics and a subsequent full publication. Time to publication was visualised with a Kaplan-Meier curve. Factors potentially associated with time to publication were evaluated with Cox regression analysis. All analyses were conducted in R Studio. Group discussions were selected to understand experts’ behaviours, views, and experiences concerning non-dissemination of qualitative research and dissemination bias in qualitative research. Researchers with knowledge in the field of qualitative research and/ or dissemination bias were consulted through two group discussions held at international scientific conferences which were attended by health researchers, experts in health research methods, and systematic review authors. Additionally, a pragmatic research approach was taken to develop and advance concepts concerning dissemination bias in qualitative research. This entailed iterative consultations with experts in dissemination bias and qualitative evidence synthesis, and developing consensus through multiple face-to-face meetings and teleconferences with these international experts. Protocols from group discussions and expert consultations were transcribed and coded by one researcher. An inductive thematic analysis of all available notes was conducted following the same method as for the qualitative survey responses. Data were documented and structured in a way that allows easy access and reusability of the data by third parties. Original, complete data are stored in the existing local server environment of the Medical Center - University of Freiburg, which allows secure long-term storage of data. Results: Overall, 1032 people initiated the survey of whom 756 (73.3%) completed the survey. In the survey, more than 50% of the respondents reported that one or more of their qualitative studies had not been published in a peer-reviewed journal (≥1 study not published: 68.1%) or another publicly accessible format (≥1 study not published: 51.7%). One-third reported that important individual findings were missing in one or more of their published reports (n=288, 35.6%). Researchers reported that the most important reasons for non-dissemination were that publication was still planned, (n=147, 35.7%), resource constraints, (n=146, 35.4%), and rejection from journals (n=134, 32.5%). Peer reviewers and editors reported that the methodological quality and reporting quality of qualitative studies were the primary reasons for rejecting manuscripts. Most respondents reported that they agreed that non-dissemination of qualitative research has negative impacts on health research (n=614, 76.6%), health policy (n=581, 72.6%) and health-care (n=572, 71.2%). Researchers reported having mixed views about whether publicly accessible repositories for registration of qualitative studies and dissemination of qualitative study findings might decrease dissemination bias. For the cohort study a total of 111 abstracts (96 oral presentations and 15 poster presentations) presented in the year 2015 were included. From the conference in 2016, 159 abstracts (114 oral presentations and 45 poster presentations) were included. The cohort study found that almost one-third, that are 91 of 270 (33.70%; 95% CI 28.09 to 39.68%), of qualitative studies presented at an international nursing research conference did not result in full publication five years after presentation. Studies presented as oral presentations were associated with subsequent full publication when compared to poster presentations. Studies presented by first authors with university affiliations were also associated with subsequent full publication when compared to studies presented by authors without university affiliation. Median time to full publication was 11 months (95% CI 9 to 14 months). Only presentation format impacted on the time to publ

Access State: Open Access

Search in field:

Recently searched for: