Introduction Randomised controlled trials (RCT) provide the highest level of evidence on healthcare interventions. Accurate interpretation of results and meta-analysis requires adequate reporting. The CONSORT statement, first published in 1996 and later revised in 2001 and 2010, aims to improve reporting quality of RCTs. Heart failure with preserved ejection fraction (HFPEF) is a major cause of morbidity and mortality, comparable to HF with reduced ejection fraction. While many studies have investigated the efficacy of various pharmacological agents, scarcity of consensus guidelines reflects the conflicting results generated from such RCTs.

The aims of this study are to: 1) systematically identify RCTs investigating the efficacy of pharmacological therapies in HFPEF, 2) assess reporting quality with CONSORT 2010 statement and 3) identify trends in reporting quality.

Methods Medline, EMBASE and CENTRAL were systematically searched from 1 January 1996 to 1 September 2015 using pre-specified inclusion-exclusion criteria (Table 1). Titles and abstracts were assessed for relevance. Full texts were independently assessed by three authors for inclusion. Reporting quality was independently assessed by two authors using the CONSORT 2010 score, with each item scored and weighted equally. An overall reporting quality score was calculated for each study. Statistical analysis was performed with SPSS.

Results Initial search identified 3426 studies; 32 were included in the final analysis. There was inter-observer agreement for 1078 out of 1184 values (91.1%); Cohen’s kappa score for inter-observer variability was 0.85. The mean score was 55.3% (range 23.3--93.8%, SD 17.3%). Under half of published studies (41%, 13/32) did not adequately report more than half of all relevant sections. The best reported criteria, where applicable, were protocol referencing (criterion 24) (100%, 12/12), interim analysis (7b) (100%, 6/6), changes to methods (3b) (100%, 6/6), statistical methods (12a) (96.9%, 31/32) and result interpretation (22) (96.9%, 31/32). The worst reported were reporting of binary outcomes (17b) (0%, 0/7), abstract (1b) (9.4%, 3/32), allocation concealment mechanism (9) (12.5%, 4/32) and trial ending (14b) (12.5%, 4/32) (Figure 1). The reporting scores showed highly significant correlation with journal impact factor (r = 0.54, P < 0.01), 5-year impact factor (r = 0.49, P < 0.01) and publication year (r = 0.49, P < 0.01). Mean scores increased after publication of each updated CONSORT statement: from 41.0% (range 38.7–43.3%, SD 3.3) to 48.6% (30.0--77.8%, SD 14.5) after CONSORT 2001 (p = 0.53), and from 48.6% to 63.3% (23.3–93.8% SD 17.6) after CONSORT 2010 (P = 0.02) (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Abstract 3 Figure 1

Percentage of studies adequately reporting each CONSORT 2010 checklist item where applicable

• Download figure
• Open in new tab
• Download powerpoint

Abstract 3 Figure 2

Individual study CONSORT 2010 score (open) grouped by available CONSORT statement (1996, 2001 or 2010) at time of study publication. Mean score (filled) shows increasing mean scores with revisions to CONSORT statement. Increase after introduction of CONSORT 2010 was significant (difference +13.1%, P = 0.02)

Conclusion This study identified all RCTs involving pharmacological interventions in HFPEF, and demonstrated that while reporting standards have improved with time, at present the majority of studies do not meet the CONSORT standards for reporting of RCTs. Better compliance is needed.