--- http_interactions: - request: method: get uri: https://rogue-scholar.org/api/posts/6179ad80-cc7f-4904-9260-0ecb3c3a90ba body: encoding: UTF-8 string: '' headers: Connection: - close Host: - rogue-scholar.org User-Agent: - http.rb/5.1.1 response: status: code: 200 message: OK headers: Age: - '0' Cache-Control: - public, max-age=0, must-revalidate Content-Length: - '14725' Content-Type: - application/json; charset=utf-8 Date: - Thu, 15 Jun 2023 05:34:55 GMT Etag: - '"12ntu9by009bd1"' Server: - Vercel Strict-Transport-Security: - max-age=63072000 X-Matched-Path: - "/api/posts/[slug]" X-Vercel-Cache: - MISS X-Vercel-Id: - fra1::iad1::svxc9-1686807294657-7dc3d3c6f0b6 Connection: - close body: encoding: UTF-8 string: '{"id":"https://www.ideasurg.pub/academic-powerhouse","uuid":"6179ad80-cc7f-4904-9260-0ecb3c3a90ba","url":"https://www.ideasurg.pub/academic-powerhouse/","title":"How to Build an Academic Powerhouse: Let''s Study Who''s Doing it","summary":"A Data Exploration with Public Data from the Academic Surgical Congress","date_published":"2023-06-03T04:54:30Z","date_modified":null,"authors":[{"url":null,"name":"Tejas S. Sathe, MD"}],"image":null,"content_html":"

Students and residents aspiring for a career in academic surgery are looking for training programs that will help jumpstart their careers by exposing them to mentorship and opportunities to conduct research. While the gold standard of academic productivity (for the time being) remains peer-reviewed publications, conference presentations are an important secondary metric. In addition to feeding eventual papers, conference talks allow trainees to practice presentation skills, build their networks, and learn about the work being done at peer institutions and the field more broadly. Residents pursuing professional development are encouraged to apply to and attend conferences, though in the post COVID era the cost of conference attendance and travel are being re-evaluated. Even moreso than publications, conference presentations are typically led by students and residents. For now, academic conference participation can serve as an important albeit limited metric to evaluate an overall institution''s research power – especially as it relates to trainees.

In order to understand how to build a successful research program, we wanted to understand who had already done so. Conceivably, by finding successful examples, we can conduct interviews and focus groups among key stakeholders at these programs to understand what they did, and how these actions can be replicated. We were particularly interested in seeing which programs demonstrated a high number of accepted abstracts and which programs were able to improve their abstract performance from year to year.

The Academic Surgical Congress (ASC) is a joint surgical meeting co-hosted by the Association for Academic Surgery (AAS) and the Society for University Surgeons (SUS) and held annually in February. Anecdotally, it is a popular conference for students and residents to attend. The ASC maintains a public archive of all abstracts dating from 2015-2020. We obtained a machine readable version of this public database through the webmaster (this version of the data spans 2016 to 2023).

For each abstract, we looked at the year and primary institution at which the work was conducted. In order to intuit the primary institution, we searched the institution block string for the first presence of \"university\", \"hospital\", \"institute\", or \"medical center\" or the first expression to occur before a comma. This method was chosen as institutions’ names are written variably (Department of Surgery, University Hospital, New York, NY vs. University Hospital, Division of Colorectal Surgery, Department of Surgery, New York, NY). Unfortunately, our code is unable to merge slight variations in the resultant institution name (University of Michigan vs. University of Michigan Ann Arbor).

In order to do a primary survey of institutional trends in this dataset, we conducted the following analyses. First, we measured the number of abstracts accepted each year from 2016 to 2023. Second, we looked at the number of accepted abstracts by institution over the entire study period. Then, we looked at the number of abstract institutions by institution and year to study changes in institutional trends over time. Finally, we did a subset analysis on post-COVID data (2021 to 2023) to look at absolute and per cent year-over-year changes in accepted abstracts to see which institutions were able to significantly increase their abstract performance in one year.

Accepted Abstracts are increasing but down from peak

We plotted the total number of abstracts accepted by year. The number of accepted abstracts increased from 1125 in 2016 to an all-time high of 1742 in 2020 before declining to 922 and 848 in 2021 and 2022, respectively. Accepted abstracts have since rebounded to 1469 in 2023, or roughly 84% of the maximum amount. Of note, the 2021 and 2022 meetings were held virtually due to the COVID-19 pandemic. Conversely the 2020 meeting was held immediately prior to the institution of lockdowns and other social distancing policies.

Alabama, Michigan lead abstract acceptances during study period

Next, we plotted the total number of accepted abstracts by institution. The graph shows data for the twenty leading institutions. The University of Alabama and the University of Michigan lead in abstract acceptances during the study period (477 and 348, respectively). Given an eight year time-range, the top twenty institutions reflect an average annual abstract acceptance ranging from roughly thirteen to 60.

We also plotted the number of accepted abstracts by year using a separate line plot for each individual. While Michigan has maintained between 30 and 60 abstracts for each year, Alabama went from 6 to 48 abstracts between 2016 and 2017 and has maintained greater than 60 accepted abstracts each year. There is a notable dip in the number of accepted abstracts from 2021 to 2022, consistent with a similar trend in the total yearly acceptance data.

Overall, there appears to be a general clustering of the other eighteen institutions in the top twenty, a trend seen in both the aggregate and the year-to-year data.

MGH, Brigham lead absolute and per cent increase in abstracts in 2023

Next, we measured the absolute and percent change between accepted abstracts in consecutive years for the top twenty institutions. We restricted our analysis to 2021, 2022, and 2023. The greatest absolute increases in accepted abstracts came from Massachusetts General Hospital in 2023 (27), Brigham and Women''s Hospital in 2023 (16), and UCLA in 2022 (16). The greatest percentage increases in accepted abstracts came from Brigham and Women''s Hospital in 2023 (800%), Johns Hopkins in 2023 (500%), and Massachusetts General Hospital in 2023 (386%). Of note, Johns Hopkins and Brigham and Women''s Hospital also had the largest absolute and percentage decrease in accepted abstracts in 2022. Interestingly, University of Alabama (+4 in 2022, +3 in 2023) and University of Michigan (+11 in 2023)  - which rank at the top in all time and yearly abstracts - experienced less extreme swings during the 2021-2023 time period.

Conclusions

In this data exploration, we analyzed abstract acceptance data for the Academic Surgical Congress between 2016 and 2023. We showed that the number of abstracts increased from 2016 to 2020, dipped in 2021 and 2022, and is recovering in 2023 - likely reflecting the variability forced by COVID19 and the transition to virtual conferences in 2021 and 2022. We also showed that the University of Alabama and the University of Michigan have significantly led accepted abstracts throughout the study period with University of Alabama maintaining a top position following an impressive 700 percent increase in abstracts between 2016 and 2017. In 2023, Massachusetts General Hospital and Brigham and Women''s Hospital have led absolute and percent increases in abstract submissions, though this may reflect some degree of reversion following drops in abstract acceptances in 2022.

Limitations and Future Directions

This data exploration has several limitations. First, our study is limited to one conference and conferences themselves are only a limited component of academic productivity. However, the Academic Surgical Congress is an international meeting that is well regarded, popular among trainees, and has consistent multi-year data. Another limitation of our study is our assumption that the first institution listed in the author block is the primary institution–a method which does not take into account multi-institutional abstracts. However, we suspect this reflects the minority of the abstracts. Finally, we have not fully optimized name-matching to correctly combine terms like \"University of Michigan\" and \"University of Michigan Ann Arbor\". However, given that most of the data currently is only for the top 20 institutions, we believe the overall trends will be preserved.

In the future, we can attempt to obtain more granular data on authors and institutions and optimize the capture of this data for aggregated analysis. We are also researching different techniques to perform string matching to fix the institution name issue. We can improve the generalizability of our findings by bringing in similar data from other large meetings – such as the American College of Surgeons Clinical Congress and the Society of American Gastrointestinal and Endoscopic Surgeons (SAGES). We can also use the PubMed API to perform similar analyses in top surgical journals such as JAMA Surgery or the Annals of Surgery.

With regards to the data we have obtained, we can conduct survey-based studies or focus groups with key stakeholders from institutions that have demonstrated strong abstract performance during the study period as well as those who have shown significant improvement. By talking to department chairs, program directors, and research and education leaders we can understand what steps each institution took to improve abstract acceptance (e.g. hiring a new research faculty, creating a mentorship program, providing administrative support for IRBs/stats etc.). Following this, we can map individual interventions to improvements to understand the contribution that each can have on increased academic productivity.

We have also not yet mined the abstract body for themes. Using natural language processing, we can understand how themes and trends of emergent research change from year to year, and even develop models to predict which topics will be of interest in future works. This will likely be the subject of a future data exploration.

Overall, we hope to demonstrate that abstract acceptance data from a single international meeting can provide interesting, meaningful data on which programs have strong research infrastructure and which programs are making strides to achieve it. By studying this data and improving the inputs, we can better understand whose example to follow and ultimately develop a playbook that all institutions can employ to maximize the academic opportunities of students and trainees.

See the Data

You can see the raw data as well as all of our code here, and use this for your own studies with appropriate attribution.

https://deepnote.com/@tejas-sathe-a73f/ASC-Data-Exploration-948ca211-bfc7-459b-9d6d-ef952a5baf32

","tags":["Pre-Print"],"language":"en","references":[],"blog_id":"3cxcm20","blog":{"id":"3cxcm20","title":"I.D.E.A.S.","description":"Innovation and Design Experiments in Academic Surgery (I.D.E.A.S.)","language":"en","favicon":"https://www.ideasurg.pub/favicon.png","feed_url":"https://www.ideasurg.pub/rss/","home_page_url":"https://www.ideasurg.pub/","user_id":"8498eaf6-8c58-4b58-bc15-27eda292b1aa","created_at":"2023-05-31T13:44:29+00:00","indexed_at":"2023-01-04","feed_format":"application/rss+xml","license":"https://creativecommons.org/licenses/by/4.0/legalcode","generator":"Ghost 5.51","category":"Medical and Health Sciences","prefix":"10.59350","modified_at":"2023-06-03T04:54:30+00:00","version":"https://jsonfeed.org/version/1.1","backlog":false}}' recorded_at: Thu, 15 Jun 2023 05:34:55 GMT recorded_with: VCR 6.1.0