Download as pdf or txt
Download as pdf or txt
You are on page 1of 65

Academic Research Study (ARS) Course

January 11, 2023

Joseph Dhahbi, MD, PhD


Academic Research Study (ARS) Course
Required two-credit research-based course

ARS I (Preparation phase): 2d semester of year 1


- Prepare and submit a Research Proposal to receive a grade
⎯ take advantage of the summer to start you research

ARS II (Execution phase): 1st semester of year 2


- Conduct the proposed research and generate results
- Prepare and submit a Poster to receive a grade

Optional
- Poster presentation at CUSM Research Day / conferences
- Work with your research advisor to publish
Categories suitable for a research project
Basic Science Research
Literature review:
⎯ Systematic vs. Scoping vs. Rapid Literature Reviews – Dr. Green (2/1/23: 11 - 12 am)
Clinical Research:
⎯ Clinical Research at ARMC – Dr. Neeki (1/25/23: 11 - 12 am)
⎯ Clinical Research Overview and Ethical considerations – Dr. Talib (2/1/23: 10 - 11 am)
Medical Education
Global Health
Epidemiology
Public Health
Public clinical data analysis
Mandatory sessions
Introduction to the course
1/11/23 (10 – 12)
Identification/Navigation of public clinical datasets – J. Dhahbi

1/18/23 (10 - 11) Evidence-Based Research – Healthcare Data & Statistics – D. Farber
1/18/23 (11 - 12) Statistics TBA – F. Macciardi
1/25/23 (10 - 11) Implementation Science – Dr. Wakida
1/25/23 (11 - 12) Clinical Research at ARMC – M. Neeki
2/1/23 (10 - 11) Clinical Research Overview and Ethical considerations – Z. Talib
2/1/23 (11 - 12) Systematic vs. Scoping vs. Rapid Literature Reviews – G. Green
ARS Course Design
1) Group Projects
Working in groups will reinforce teamwork and development of collaborative skills

- Members of each college participate in one or two research projects

- Hybrid groups: students from various colleges participate in a research project

2) Individual Projects are accepted


Responsibilities of students during the ARS course

Students are expected to initiate and lead the research project, and demonstrate independence
⎯ ARS course is a Self-directed learning exercise

Step 1: Identify and choose a research project related to your research interests and learning needs

Step 2: Choose an research advisor based on common research interests


CUSM faculty research interest:
https://www.cusm.org/school-of-medicine/faculty/listings.php
Step 3: Work with your research advisor to do background work, plan your research design and
analysis methods, and prepare and submit a research proposal for feedback/grade
Research advisors:
- Your college faculty
- Other CUSM faculty
- ARMC clinician (attend Dr. Neeki presentation)
- Researchers from the community or other institutions
Research advisor guides students to define their learning objectives
and plan the strategy to achieve the objectives
Research Proposal
The project proposal should include the following sections:

⎯ Introduction/justification
⎯ Hypothesis/question
⎯ Specific aims
⎯ Proposed analysis methods
⎯ References

− The proposal should NOT exceed 3 pages (excluding the references list)

− Grading rubric details what to expect for each section


Research Proposal Grading Rubric
(appendix B in the syllabus)

⎯ Use the grading rubric as a guide


when you prepare the proposal

%
100
Deadlines of ARS I

1) Select projects and submit title by February 24th

2) Submit research project proposal by April 27th

3) Faculty provide feedback & grade if proposal meets expectations

4) If not, students address concerns and resubmit corrected proposal for grade

You must add your name to only one research project proposal
(you can not receive more than one grade)
Grades for Research Proposal
2020-2021
Questions?
Public clinical data
1/18/23 (10 - 11) Evidence-Based Research – Healthcare Data & Statistics – D. Farber

Goals for the session by D. Farber


1) Apply a question framework to a search strategy
2) Identify a source of datasets
3) Locate one dataset based on a research question
Public clinical datasets
- Bridging the gap between government & citizens (transparency and accountability)
- Opportunity for researchers to study all aspects of healthcare: quality, performance, delivery systems, cost …
Agencies & Institutions

National

https://data.chhs.ca.gov/

State

https://healthstat.dph.sbcounty.gov/

Local
NYU School of Medicine – SPARCS Project
Statewide Planning And Research Cooperative System

Patient-level Data
- patient demographics
- diagnoses, procedures
> 5 million Patient-level Data - treatments
- Organized for easier use by medical students - services
- Easily analyzed in Excel using basic statistics tests - charges/cost
- Length of stay
- ambulatory surgery
- emergency admission

- To teach students new tools/skills to care for patients


- To learn new approach for informed decision making (diagnosis, treatment, billing, …)
Permission to use SPARCS data in presentation/publication?

- No limitations - cite NYS DOH as the source

- To apply for identifiable data, use Request Form DOH-5132


www.health.ny.gov/statistics/sparcs/forms
Go to: https://navigator.med.nyu.edu/ace/
SPARCS database is arranged by DRG (Diagnosis Related Groups)

Example: Data related to cost

LOS:
Length Of Stay
Numb
patiener of
ts

956 DRGs with many datasets in each DRG


*

* ECMO: ExtraCorporeal Membrane Oxygenation


$ billions spent
per DRG in one
year in one state
Start Literature search
about clinical significance
of this parameter
175/242 = 72.31%
Let’s explore the
datasets associated
with HF
Heart failure data for all NY State Hospitals
(2020, 2019, 2018)

Heart failure data for a selected group of 8 hospitals (2020)


- more manageable dataset
- a good choice for an initial student project

Heart failure data from NYU hospitals and affiliates (2020)


Let’s download this Heart failure dataset for
all NY State Hospitals for the year 2020
Heart failure data for all NY Sate Hospitals (2020)

52,555 rows (Heart Failure patients)


(sufficient for statistical significance)
⎯ Rows are Heart Failure patients
⎯ Columns are the parameters recorded for each patient
Heart failure data for all NY Sate Hospitals (2020)

52,555 rows (Heart Failure patients)


(sufficient for statistical significance)
⎯ Rows are Heart Failure patients
⎯ Columns are the parameters recorded for each patient
1) What parameters were recorded for each patient?

Copy
&
Transpose

32 recorded
parameters

?
?
https://navigator.med.nyu.edu/ace/
APR? CCS?
https://navigator.med.nyu.edu/ace/
1) What parameters were recorded for each patient?
2) What were the possible outcomes for each parameter?
3) How many patients for each outcome?

https://www.statology.org/excel-count-occurrences/

32 recorded
parameters

Occurrences = outcomes for each parameter


Number of occurrences = Number of patients for each outcome
5 outcomes for the
parameter “Age Group”
Parameter

Number of patients in each outcome

Total number of patients

52,555 patients
Gender Race Ethnicity
unique Occurences unique Occurences unique Occurences
F 25330 Black/African American 12561 Not Span/Hispanic 43644
M 27224 Other Race 9491 Spanish/Hispanic 6373
U 1 White 30134 Unknown 2382
Multi-racial 369 Multi-ethnic 156
52,555 Total
52,555 Total 52,555 Total

APR Severity of Illness Description APR Risk of Mortality Payment Typology 1


unique Occurences unique Occurences unique Occurences
Moderate 20497 Moderate 15922 Medicare 39538
Major 21234 Major 22771 Medicaid 7505
Minor 4298 Extreme 10483 Private Health Insurance 2308
Extreme 6526 Minor 3379 Blue Cross/Blue Shield 1700
Self-Pay 507
52,555 Total 52,555 Total Managed Care, Unspecified 475
Federal/State/Local/VA 415
Miscellaneous/Other 89
Department of Corrections 18

52,555 Total
Length of Stay
unique Occurences
CCS Diagnosis Description 1 4447
unique Occurences 2 7451
HEART FAILURE 46229 3 8527
CHRONIC KIDNEY DISEASE 5736 4 7303
PLEURISY, PLEURAL EFFUSION AND PULMONARY COLLAPSE 500 5 5468
OTHER SPECIFIED AND UNSPECIFIED LOWER RESPIRATORY DISEASE 90 6 4296
7 3358
52,555 Total 8 2516
9 1814
10 1371
11 1082
12 833
13 723

91 1
92 1
100 2
101 1
102 1
104 1
106 1
108 2
110 1
112 1
119 1
120 + 16

52555
How to approach creating your clinical question?

1) Read the SPARCS Data Dictionary to better understand the data elements.

2) Make sure your question is testable by the data present in SPARCS.

3) Is there a sufficient number of cases?

4) Would the result be interesting and useful?

5) Could the health care system, or individual providers, act on the result to
make informed decisions to care for patients?
Examples from previous NYU students that are answerable with SPARCS data

Examples of clinical questions related to length of hospital stay:

- Is there an association between race and length-of-stay for patients hospitalized with Schizophrenia?

- Does severity of illness correlate with length-of-stay for patients with Drug and Alcohol dependence?
- How does hospital level case-load relate to length-of-stay for hip replacement?

- Are there differences in length-of-stay for hip surgery based on Payor type?

- Does day of admission correlate with length-of-stay for CHF?

- Does a patient’s race impact the rate of cardiac catheterization among patients admitted with acute MI?

Answers to such questions inform decision making


More clinical datasets
https://navigator.med.nyu.edu/ace/
Datasets in the State of California
GO to: https://data.chhs.ca.gov/

all 419 datasets

There are 419 datasets


Some datasets are arranged into 7 topics
all 419 datasets

Search for datasets with keywords


type brain
Start Literature search
Data at hospital-level (not patient-level) about clinical significance
of these parameters

58 counties in CA
Data for San Bernardino Hospitals
COVID-19 datasets
Datasets in the State of California
GO to: https://data.chhs.ca.gov/

COVID-19 datasets
Search for datasets with keyword “vaccine”
type vaccine
COVID-19 Post-Vaccination Infection Data

Post-vaccination cases (aka vaccine breakthrough cases):


Positive covid test at least 14 days after primary vaccination or
primary vaccination and a booster

− Tracking Post-vaccination cases is important for monitoring the impact of immunization on


COVID-19 infections, hospitalizations, and deaths

− If the number or severity of cases exceeds expected levels, this could be a signal of reduced
protection against a variant or waning protection over time

− Whole genome sequencing of these cases, can help characterize the effectiveness of current
vaccines against variants.
3 Post-Vaccination
datasets

Start here

Data Dictionary defining the parameters (column names) in the dataset


17 Parameters (columns)
in the dataset
Current dataset

Archived dataset
638 data points from 2/1/21 to 10/31/22
Search for datasets with keyword “equity”
type equity
Collecting health equity data helps to identify
health disparities and improve the state’s response

How does the COVID-19 pandemic


impact different communities?
Defines the parameters
(columns) for all tables
2985 data points
calculated on 12/1/22
Covid-19 literature review projects
Datasets can be analyzed with statistical tests in Excel
The Analysis-ToolPak add-in is a third party Excel add-in that provides special analysis tools for statistical analysis

a “Data Analysis” button


will appear in Tools

Statistical tests in Excel


END

You might also like