SELECT admissions.subject_id, diagnosis, insurance, gender
FROM patients
INNER JOIN admissions
ON admissions.subject_id = PATIENTS.subject_id
WHERE diagnosis LIKE '%congestive heart failure%'
OR diagnosis LIKE '%chronic kidney disease%'
OR diagnosis LIKE '%diabetes%'
OR diagnosis LIKE '%respiratory%'
OR diagnosis LIKE '%pneumonia%'
OR diagnosis LIKE '%cardiac%'Analysis Report Two - Data, Data Everywhere
Executive Summary
The three articles from this week focus on how data and new technology are being utilized in healthcare to deliver higher quality care, increase efficiency, and patient safety to improve both operational outcomes and patient outcomes. Rather than just collecting data, the data is being leveraged to make informed decisions and quality improvements.
Introduction
Big data analytics in healthcare identifies trends by analyzing large volumes of data and uses the data to reduce costs and support clinical decision-making. By analyzing this data, informed decisions can be made to improve outcomes. A few mentioned concerns include issues with big data analytics is data quality and with privacy and security. (Raghupathi and Raghupathi 2014) A clinical decision support systems (CDSS) is designed to use evidence-based practice to alert providers with a goal to improve patient care and safety. These alerts do not replace providers’ clinical decision-making but are a stepping stone to guide them. Downfalls to CDSS alerts include alert fatigue and research has found that many providers ignore CDSS alerts. Other struggles CDSS faces are communications with other systems to export the information in, which can leave CDSS alerts inaccurate. (Sutton et al. 2020) Lastly, From Data Overload to Targeted Care: AI’s Role in Health, discusses how AI and devices such as an Apple watch (wearable device) is allowing data to be analyzed in real time. Receiving data sooner allows providers to make clinical decisions sooner. Starting treatment earlier reduces complications and increases patient outcomes. Concerns with AI and wearable devices include issues with data and privacy.
The Healthcare Context
“Electronic Health Records (EHRs) are implemented for various reasons, including to support coordination, collaboration, and shared decision- making, and are considered as a major means to deliver high-value care.” (Vos et al. 2020) EHRs allow data to be shared throughout the different areas of the hospital, primary care or specialty office, within the company itself in real-time. This has many advantages, including increasing patient safety, improve care coordination and increase operational efficiency. Data can be shared with many different departments and staff members. EHRs update in real-time allowing different departments to see changes as they happen, having this data increases quality and patient safety. (Upadhyay and Hu 2022) While this is beneficial, there is still a limitation receiving data from other organizations. Healthcare collaboration between hospitals, specialists, and primary care offices are essential to provide quality care coordination. One disadvantage of an EHR is they do not connect to hospital portals. For example, many times when PCP’s see patients in for a hospital follow up medications were changed at discharge but the primary care provider does not have hospital discharge summaries and unsure of what medications were changed. “One third of in-hospital prescription changes was not or incorrectly documented in the PCP’s record, which likely puts patients at risk of adverse drug events after hospital visits. Such flawed reliability of a routine care process is unacceptable and warrants improvement and close monitoring.” (Poldervaart et al. 2017) It is challenging for the provider to coordinate the best care with limited data and can result in safety issues or readmission.
In conclusion, technology can both help and hinder the quality of care. Having data more quickly increases patient safety, improves care coordination, and increases operational efficiency. Technology such as EHRs can also hinder care, being unable to connect with outside hospitals often leaves the primary care provider without a discharge summary, making it difficult to coordinate the best care.
Data Visualizations
Visualization One - Two Table Join
ggplot(data = medicare,
mapping = aes(insurance, fill = gender)) +
geom_bar() +
coord_flip() +
theme_minimal() + # Cleans up the background grid lines
labs(
title = "Diagnosis Associated with Sepsis Across Insurance Payers",
subtitle = "Data taken from admissions and patients table within MIMIC-III",
x = "Insurance",
y = "Diagnosis Associated with Sepsis",
caption = "Source: MIMIC-III Clinical Database v1.4"
)
“Sepsis is a common reason for hospitalization and can contribute to 1 in 4 hospital deaths. Data shows there is a higher chance of Sepsis hospitalizations for patients > 65 years old with comorbidities. The median age of adult patients with sepsis was 69 years; 127 (52%) were male.” (Novosad 2016) This diagram joined the patients table and the admissions table, patients table was needed to include gender data. I also included a WHERE statement to include diagnosis’ that increases risk of Sepsis such as congestive heart failure, chronic kidney disease, diabetes, respiratory, pneumonia, and cardiac. This data was put together to visualize the association of patients with Medicare insurance and diagnosis’ associated with Sepsis and to look for gender trends. The data includes insurance type, diagnosis’ associated with Sepsis, and gender. Looking at the data, it is evident that there are more patients’ comorbidities with Medicare insurance than with Private insurance. This could be because Medicare patients are > 65 years old and private insurance patients typically are younger. In terms of gender, the data from Mimic III shows that Sepsis was slightly higher in males compared to females, aligning with the research mentioned above.
Visualization Two - Three Table Join
SELECT CAST(LOS AS INT) AS Days, ADMISSIONS.subject_id, diagnosis, insurance, gender, discharge_location, LOS
FROM patients
INNER JOIN admissions
ON admissions.subject_id = PATIENTS.subject_id
INNER JOIN icustays
ON admissions.hadm_id = icustays.hadm_id
WHERE insurance LIKE 'medicare'
AND diagnosis LIKE 'sepsis' #Put your ggplot visualization in this block
ggplot(data = myquery2,
aes(x = gender, y = Days)) +
geom_violin() 
This diagram joined three tables the patients, admissions and icustays table. I also included a WHERE statement to patients with a diagnosis of Sepsis and covered by Medicare. I did use the CAST function to display length of stay in days. This data was put together to visualize the association of patient’s ICU length of stay with Medicare insurance with a sepsis diagnosis and comparing gender . With length of stay in days on the Y axis and gender on the X axis. This data indicates that females length of stay is more concentrated in the 1-2 days, while males are more variable with length of stays as long as 10 days. Comorbidities were not taken into account but could have potentially affected length of stay, but would analyzed further to determine.
Recommendations for Industry
“Sepsis is common and increases utilization and has a mortality rate of 17%. Sepsis is a persistently growing health concern in America, especially in light of the aging population demographics. More than 1.5 million people are affected by sepsis in the USA each year, leading to 250,000 deaths.” (Liu et al. 2020)
The healthcare industry needs to focus on prevention in the primary care setting, prior to a hospital admission. Patients hospitalized for sepsis typically have one or more comorbidities, like diabetes, cardiovascular disease, chronic kidney disease, pneumonia, or COPD. Primary care is typically the first point of contact for patients and should utilize that opportunity to work on prevention on sepsis by providing vaccinations that are clinically appropriate and optimize chronic conditions. “For example, pneumonia is the most common infection causing sepsis, and vaccination is an important and highly effective prevention strategy.” (Novosad 2016)
Primary care offices should use their EHR to pull data and this data should identify patients at high risk for sepsis. This data should include patients over 65 years of age, with one or more comorbidities. Once identified, an alert could be made within the EHR. For example, if working in eClinicalworks, a global alert can be created to easily identify high risk patients. Also, a CDSS alert should be created for pneumonia vaccine, to alert providers that the vaccine is due. Focusing on preventive care is a defensive strategy, by reducing risk of hospitalizations and decreasing utilization.
References
Liu, Andrew C, Krishna Patel, Ramya Dhatri Vunikili, Kipp W Johnson, Fahad Abdu, Shivani Kamath Belman, Benjamin S Glicksberg, et al. 2020. “Sepsis in the Era of Data-Driven Medicine: Personalizing Risks, Diagnoses, Treatments and Prognoses.” Briefings in Bioinformatics 21 (4): 1182–95.
Novosad, Shannon A. 2016. “Vital Signs: Epidemiology of Sepsis: Prevalence of Health Care Factors and Opportunities for Prevention.” MMWR. Morbidity and Mortality Weekly Report 65.
Poldervaart, Judith M, Marije A van Melle, Sanne Willemse, Niek J de Wit, and Dorien LM Zwart. 2017. “In-Hospital Prescription Changes and Documentation in the Medical Records of the Primary Care Provider: Results from a Medical Record Review Study.” BMC Health Services Research 17 (1): 792.
Raghupathi, Wullianallur, and Viju Raghupathi. 2014. “Big Data Analytics in Healthcare: Promise and Potential.” Health Information Science and Systems 2 (1): 3.
Sutton, Reed T, David Pincock, Daniel C Baumgart, Daniel C Sadowski, Richard N Fedorak, and Karen I Kroeker. 2020. “An Overview of Clinical Decision Support Systems: Benefits, Risks, and Strategies for Success.” NPJ Digital Medicine 3 (1): 17.
Upadhyay, Soumya, and Han-fen Hu. 2022. “A Qualitative Analysis of the Impact of Electronic Health Records (EHR) on Healthcare Quality and Safety: Clinicians’ Lived Experiences.” Health Services Insights 15: 11786329211070722.
Vos, Janita FJ, Albert Boonstra, Arjen Kooistra, Marc Seelen, and Marjolein Van Offenbeek. 2020. “The Influence of Electronic Health Record Use on Collaboration Among Medical Specialties.” BMC Health Services Research 20 (1): 676.