Capacities of Care during COVID: How Hospitals are Handling the virus in 3 New York Boroughs
With the onslaught of COVID-19 cases in the U.S., hospitals are battling the illness on the front lines. New York was handling the largest number of corona related hospitalizations, upwards of 20,000 people a week during the virus’s first peak in early April. Most recently, Florida is the new epicenter for the virus with nearly 400,000 total cases.
To grasp the impact of the virus now, it is integral that we understand how equipped and prepared our hospitals were for the onset of COVID-19.
Through our study, we seek to answer questions surrounding facility-based care during this pandemic in order to understand the factors impacting patient experience. More specifically:
- What are factors that contribute to the lowest and highest performing hospitals?
- How can we leverage these factors to better prepare for a reality post-Covid?
Methodology
In this research, we focus on data collected in 2018 for three boroughs: Manhattan, the Bronx, and Brooklyn. Data was obtained from the New York Department of Health, ProPublica, Census data, and other sources. Due to data availability, our final dataset merged data from various years in order to understand the relationships between factors associated with hospital performance and quality. Data was merged based on hospital location and zip code.
Data was analyzed via statistical methods, such as mean standard deviation, min and max values, and correlations. In order to assess the relative “importance” of which features contribute most to patient satisfaction, our main target variable, machine learning models were employed using H2O, an open-source machine learning and predictive analytics platform.
By mapping patient satisfaction and emergency department timeliness data from the NYS Health Profiles as well as Hospital General Information data using Carto, we were able to illustrate where NY medical facilities are located in relation to one another, visualizing these locations based on feature importance.
Readmission Rate & Patient Satisfaction
A strong negative correlation between the rate of readmission and the percentage of patients highly satisfied indicates that people being readmitted into hospitals are less satisfied with care. Building on this finding, we utilize patient satisfaction as an indicator of overall hospital quality in order to examine the factors that impact the patient experience.
An H2O analysis of variable importance indicates that the top factors influencing patient satisfaction are hospital population, admit decision time and the number of beds. Categorizing these variables, we can more clearly see the relationships between influential factors, disparities between boroughs, and overall satisfaction ratings.
By evaluating patient satisfaction as an indicator of overall hospital quality, we can quantify and compare qualitative elements that influence hospital performance during the COVID-19 pandemic. We additionally factor in hospital deaths, number of beds, median household income, and the type of hospital in order to understand how the availability of resources and funds impact performance.
For a Hospital, What Goes Into Patient Satisfaction?
Defining patient satisfaction is essential to understanding its implications as a critical element of overall hospital quality. For our analysis, we employed the percent of patients highly satisfied across various NYC hospitals from the New York State Department of Health NYS Health Profiles. In this survey, patients are asked to rate their hospital experience on a scale of 0 to 10, with 0 being the worst and 10 being the best.
This scale follows the Agency for Healthcare Research and Quality guidelines, based on the Consumer Assessment of Healthcare Providers and Systems (HCAHPS) Hospital Survey. The HCAHPS defines low hospital quality as a rating from 1–6, mid-level hospital quality a rating of 7–8, and top-level quality a rating of 9–10. Ratings are representative of facility-based care quality measures such as communication between doctors and nurses, responsiveness, cleanliness, and willingness to recommend the hospital.
To begin exploring the external and internal factors contributing to hospital performance, we mapped the highest and lowest performing hospitals in Manhattan, Brooklyn, and the Bronx.
When mapping ratings of hospitals alongside the median household income by borough, we see that Manhattan hospitals generally perform better according to patients, have large supplies of beds, and are located among affluent communities. This raises the following question:
What are the factors that influence hospital performance AND how can these factors be contextualized within larger disparities by neighborhood or borough?
Comparison by Borough: Population, Admit Decision Time, Beds, and Income
Manhattan
Manhattan has the largest GDP among the three boroughs we analyzed. Manhattan facilities had a higher supply of beds with approximately 1.15 beds per patient, higher than facilities in both the Bronx and Brooklyn which had an average bed to patient ratio of 1. In addition, facilities in Manhattan were 4.24 minutes quicker on average to admit patients into the hospital compared to facilities in the Bronx. Manhattan is also home to a majority of long-term care hospitals and the only specialized care hospitals in our dataset, which had medium to high satisfaction ratings — compared to most general acute care facilities with low satisfaction ratings.
Brooklyn
One-third of the hospitals in Brooklyn have over 250 COVID related deaths per 100,000 people, despite having 5 more clinical testing centers than Manhattan. It is also important to note that the more affluent areas of Brooklyn (median household incomes of over $64,000) are home to the highest performing hospitals in the borough while lower-performing hospitals tend to populate poorer neighborhoods. While further research is needed to understand why this is the case, during this pandemic, it is troubling to see these kinds of patterns emerge as they impact patient flows and overall trust in the healthcare system — potentially affecting an individual’s decision to go into a hospital and seek care.
Bronx
The Bronx, with a total of 89 testing centers, has the least number of testing sites compared to Manhattan and Brooklyn which have 112 and 117, respectively. The Bronx also has the highest concentration of hospitals with COVID related deaths: over 250 per 100,000. A lack of adequate testing sites, a limited number of beds, and the highest admit decision times are all factors that could explain why more deaths occur in hospitals in the Bronx.
Factor Importance as a Takeaway
After evaluating the importance of several variables, we found that the hospital population, the number of beds, and admit decision time has the greatest impact on patient satisfaction.
Taking this into account, we can begin to not only understand how the patient experience is affected in hospitals during a crisis but also how these factors may influence mortality rate and trust in local healthcare facilities. Our analysis showed hospitals located in wealthier areas tended to have more resources (ie. Beds per capita), fewer deaths per 100,000 patients, and quicker admit decision times. These factors influence overall satisfaction ratings and speak to the capabilities of these hospitals to care for patients during COVID-19 and beyond.
For future study, data on bankruptcies of hospitals, number of employees laid off, number of employees furloughed, and annual incomes of nearby residences would all contribute to a deeper understanding of patterns regarding hospital capabilities and patient experience.
Mitigating Geographic Inequalities in Healthcare
Hospitals, ground zero for dealing with the virus, can tell us a lot about the inequalities that have historically restricted access to adequate healthcare, pinpointing areas that are in need of improvement. Information silos are commonplace in the healthcare sector and are detrimental to preparedness and care, especially in the face of a pandemic.
As policies and procedures change due to COVID-19, patient experience will evolve as well. There is no reason why patient experience shouldn’t be as simple as a Yelp! or Google review — this information not only helps us fix existing inefficiencies but also helps us better prepare for the future. The fact that our high-level analysis can highlight the association between quality of care, resources, median income by borough, and COVID-19 related deaths suggests that larger societal patterns and disparities are at play and must be addressed to prevent unnecessary deaths during this pandemic.
While this article examined access to healthcare through hospitals in NYC, COVID testing centers pose a new challenge in mitigating systemic inequalities in healthcare access. The map and chart above show the total number of testing sites and people per testing site by the borough as of July 2nd, 2020. Brooklyn has the highest population ratio with 21,649 people per testing site, while Queens and Manhattan have the lowest at 12,530 and 14,544 people per site, respectively. On average, Brooklyn has 30% fewer testing sites per capita compared to the other boroughs. This ratio provides insight on the availability of COVID testing resources for each borough and remains consistent with our findings as Brooklyn sites must accommodate more people per site than Manhattan.
As of June 28th, New York’s number of hospitalizations has fallen below 900 and the state reported its lowest single-day death count since mid-March. But the fight isn’t over. While cases across the country are still rising and given the recent spikes in California, Texas, and Florida, the relationships we explored in NYC must be explored nationwide to mitigate the impacts of COVID and improve healthcare for communities going forward.
