Publication Outcomes of Abstracts from the American Society of Hematology Annual Meeting

Researchers presenting abstracts at scientific meetings provide colleagues with early access to their research findings so that they can receive feedback, discuss collaborative science, and explore novel approaches to improving their research. With open access to meeting abstracts, scientists must confirm the scientific validity of their research through manuscript peer-review despite the many strengths of presenting abstracts. Subjective measures of institutional academic success may also include the percentage of abstracts presented at national meetings that lead to subsequent publication of a manuscript, but objective data on this topic is lacking. Thus, we examined American Society of Hematology (ASH) Annual Meeting abstracts to identify predictors of a higher manuscript acceptance rate, and potential reasons for manuscripts not being published. To lay the groundwork for establishing objective measures for manuscript publication success, we compared manuscript acceptance rates for the ASH Annual Meeting to sixteen other Society meetings in thirteen subspecialties.

We conducted a University of Alabama at Birmingham Institutional Review Board approved study of publication outcomes for non-malignant hematology ASH Annual meeting abstracts. (http://www.bloodjournal.org/page/ash-annual-meeting-abstracts). Based on literature suggesting that most abstracts are accepted for manuscript publication within five years of a meeting, we reviewed abstracts only from the 2011 meeting.[1–6] Session topics, category (oral/poster/online only), age group (Pediatric, Adult, both), scientific focus (clinical vs. basic), and country of research were recorded. To identify published manuscripts, a systematic review of PubMed was performed using last names of the primary or senior authors and one to three title keywords. To ensure accuracy for our systematic review, we emailed authors of the abstracts not identified as published; we queried current publication status and the reason their abstract was not published.(Supplementary material) Among published abstracts, we recorded the journal, year of publication, and the “5-Year Impact Factor” of the journal (ISI Web of Knowledge)^SM. Finally, we conducted a systematic review (PubMed) to determine the abstract to manuscript publication rate for other subspecialty meetings.(Supplemental table) Data were analyzed using descriptive statistics, t-test (normally distributed) or Wilcoxon Rank Sums test (not normally distributed), and chi-square (categorical variables). Nominal logistic regression calculated odds ratios (OR) predicting publication success. All data were analyzed using JMP Pro 12 (Cary, NC).

In the nine session topics examined, 685 abstracts were accepted at the ASH 2011 Annual Meeting, with 327 (48%) associated manuscripts indexed for publication in PubMed by April, 2016. (Table I) There was no difference in manuscript publication rate by session topic (p=0.07); platelet disorders (60%) and thalassemia (54%) had the highest while hemostasis (34%) and sickle cell disease (44%) had the lowest rates of manuscript publication. Sessions presented from the podium were associated with higher manuscript publication rates (59%) than those presented as posters (47%) or online alone (10%; p<0.001). Oral abstracts had higher odds of publication than posters (OR=1.6, 95%CI, 1.1–2.4; p=0.009). There was no difference in publication rate by patient age group (pediatric: 46%; adult: 58%; combined: 46%; p=0.2) or country of research (United States: 47%; other countries 48%, p=0.93). Nearly 75% of manuscripts were published within two years of the conference. Scientific category (basic vs clinical) was not associated with a difference in publication rate (clinical vs. basic, p=0.5). In multivariable analysis, presentation format predicted publication success (oral: OR=1.78, 95%CI, 1.2–2.6; p=0.003); scientific category (basic vs clinical, OR=1.4, 95%CI, 0.98–1.96 p=0.06), and session topic (p=0.08) did not predict publication.

Table 1.

Variable Differences for Abstracts Published vs Not Yet Published

	Published (%) n=327	Not Published (%) n=358	p-value
Subject Matter
Antithrombotic Therapy	24 (46.1)	28(53.9)	p=0.07
Bone Marrow Failure	32 (51.6)	30 (48.4)
Erythropoiesis	22 (47.8)	24 (52.2)
Hemostasis	27 (33.8)	53 (66.2)
Lymphocyte	42 (51.8)	39 (48.2)
Platelet Disorder	51(60)	34 (40)
Sickle Cell Disease	81(43.8)	104(56.2)
Thalassemia	27 (54.0)	23 (46.0)
Transfusion	21(47.7)	23(52.3)
Presentation Format
Oral	88 (58.7)	62 (41.3)	p<0.001*
Poster	236 (46.6)	270 (52.7)
Online Publication Only #	3 (10.3)	26 (89.7)
Science Category
Clinical	173 (48.9)	181 (51.1)	p=0.48
Basic	153 (46.9)	178 (53.1)
Clinical Age Group
Pediatric	43 (46.2)	50 (53.8)	p=0.16
Adult	64 (57.7)	47 (42.3)
Combined	56 (46.7)	64 (53.3)
Country of Research
United States	166 (47.4)	184 (52.6)	p=0.93
Other Countries	160 (47.8)	175 (52.2)
Year Published
2011	20 (6.1)	–
2012	116 (35.5)	–
2013	101 (30.9)	–
2014	54 (16.5)	–
2015	29 (8.9)	–
2016	7 (2.1)	–

^*p-value <0.05 considered significant

^#only sickle cell abstracts identified as published online only

The most common journals in which abstracts were published included Blood (n=56, 17%), PLoS One (n=22, 7%), British Journal of Haematology (n=21, 7%), and American Journal of Hematology (n= 20, 6%). The majority of publications were in journals with recorded 5-year impact factors (n=312); the average impact factor was 4.45 (range: 0.8 to 54). Manuscripts based on oral abstracts were published in journals with a higher median impact factor than those based on poster abstracts (6.6 vs 4.2, p<0.0001). Manuscripts from basic science abstracts had a higher median journal impact factor than those from clinical abstracts (5.3 vs 3.9, p<0.0001). Publication year was not associated with a difference in median journal impact factor (p=0.40).

Among the 358 abstracts that were not identified for publication, 74 authors (21%) responded to the email survey. Common reasons for non-publication included: “time commitment of one or more authors changed” (32%) and “research focus changed” (27%). A smaller proportion had either submitted the abstract without acceptance for publication (22%) or were still planning to submit a manuscript (16%).

The 48% ASH manuscript publication acceptance rate was similar to the national mean of 45% at other society meetings. (Supplemental Table) The highest publication rates were from The American College of Rheumatology (59%), American Urogynecologic Society (56%), and American Society of Therapeutic Radiology and Oncology (56%) while the lowest rates were from The European Society for Emergency Medicine (25%) and American Association of Neurological Surgeons (33%). (Supplemental Table) Similar to ASH, oral abstracts at other society meetings had a higher mean publication rate than poster abstracts (52.5% vs 39.8%, p<0.0001); however there were no differences in publication rates for abstracts based on clinical as compared to basic science research (48.1% vs. 48.8%, p=0.7).

In summary, we confirm a correlation between society abstract and journal peer review. ASH podium presentations are associated with a higher manuscript acceptance rate and higher impact factor journals as compared to posters. While one-third of authors are still working on manuscript publication, the majority were not completed due to time commitments or change in research focus. Approximately 45% of national meeting abstracts result in a manuscript publication within five years and institutions can use this as a metric for successful publication rate.

To overcome publication biases in the literature, published abstracts are valuable in providing scientists with access to negative data. However, parents, scientists and clinicians should continue to view these online abstracts as preliminary data and understand that due to the limited description of the methods and results, it is important to avoid over-interpreting the impact of the findings. Full manuscript peer review remains the gold standard for accepting and disseminating medical information.