«

»

Hospital Analytics Case Studies

Share

hospital analytics case studiesIn preparation for the Health Insights Challenge we are running in partnership with EntityThree, the Centre for Health innovation and Deakin University, I wanted to get to grips with non-clinical advanced analytics applied in healthcare. This blog post is therefore a small sample of a literature review of case studies of advanced analytical models applied in hospitals around the world.

There are many examples, case studies and post-graduate research studies of analytics applied on the clinical side of healthcare. However, what interests me is the application of business analytics in healthcare; that is, the application of advanced analytical models that improve patient outcomes by assisting the practitioners and managers of healthcare institutions to run the business better. In this post I focus on case studies from hospitals.

Queueing Models for Emergency Department

Waiting in the ED with a life-threatening injury or a deadly ill child can be one of the most nerve-wracking experiences patients or parents can go through. No wonder that there are studies done to investigate how the wait times can be improved. In one particular study, performed at the Imam Hosein Hospital in Iran, queueing models were used to determine how waiting times can be shortened by analysing the most cost effective level of care, in the appropriate timeframes, with the most efficient use of limited resources.

The data records of 3000 patient records were analysed, where 47.7% were referred for trauma, and the balance being non trauma cases. Patient arrival rates were between 37 and 125 per day. The average length of stay in trauma section was 3 hours, while for non-trauma it was 4 hours. The study was done using ARENA simulation software, operational research methods and waiting times were analysed in SPSS.

Some of the findings discovered by varying the model parameters were:

  • Adding one or more senior residents decreased the length of stay to 3.75 hours.
  • Adding one bed in ICU and or critical care units, reduced occupancy rate for nursing services from 76% to 67%.
  • One of the causes of overcrowding in the ED are those patients who are going to be discharged, but they are waiting for a final paraclinical procedure, such as a CT scan or radiography for the final decision. Adding another clerk to take ECGs, reduced the average time from request to procedure from 26 to 18 minutes.
  • Addition of 50% more staff to lab’s and specialist consultations led to a 90 minute reduction in length of stay.
  • Increasing the discharge capacity by 50% led to a 50% decline in occupancy capacity.

This study proved that the application of queuing theory can be applied to improve movement through ED and therefore reduce waiting times.

Patient Flow and Movement

There have been many studies of patient flows through various parts of various hospitals, with a lot of focus on the patient flow through the Emergency Department.

A study performed for the University of Southern California, Los Angeles, mapped the processes by which patients are served with various performance measures taken at points throughout the process. The focus of the study was on improving interfaces and reducing delays as patients are transferred from one activity or department to another.

A research project by Wayne State University analysed “boarding” delays, where admitted ED patients are held in ED until an inpatient bed is identified and readied in the admit wards. Recent research has suggested that if hospital admissions of ED patients can be predicted more accurately during patient triage, then bed requests and preparations can be triggered early on to reduce the patient boarding time.

The research group developed a tool kit to assist the triage staff to proactively manage ED patient flow, and thereby reduce costs and improve patient satisfaction. The tool kit employs state-of-the-art data mining and machine learning algorithms to:

  •  Automate predictions of patient admission at triage,
  • Prediction of target wards for patients to be admitted,
  • Estimation of patient’s length-of-stay (LOS) in ED, and
  • Cost sensitive bed reservation policies that recommend optimal ward-bed reservation times for patients.

Workforce Requirements

Another very useful application of analytics in hospitals is in workforce planning, optimisation and forecasting. Workforce forms the biggest on-going operational cost item on any hospital’s income statement, in the order of 60 – 75% of costs, depending on how they are allocated. Staff scheduling or allocation can be quite complicated, because not only does under- or over-provision of health service staff affect the cost side of the business, but inappropriate allocation of different cadres of staff can also affect patient outcomes. This is further compounded by increasing financial pressures on both public sector finances and by payer organisations; difficulties in providing adequate resources and facilities to support the workforce; and increasing patient expectations on the quality of health care.

A method called Workload Indicators of Staffing Need (WISN), based on expert opinion, was analysed at three types of hospitals (general hospitals, teaching hospitals and university hospitals) in Turkey and on ten different staff categories (specialists in internal medicine, gynaecology, paediatrics and general surgery, dentists, pharmacists, nurses (paediatric, emergency service, operating theatre, and polyclinic nurses and nurses working in other wards), midwives and laboratory and radiology technicians(, which account for 84.6 of health personnel at such institutions. Activity analysis (activity standards), together with measures of utilisation and workload were used to determine staffing requirements.

The WISN method of determining staff requirements based on the amount and type of work that the institution undertakes has the potential to reduce costs. It quantifies what staff are needed to undertake the likely workload. It relies on the use of historical data (the previous year’s workload) to project what the coming year’s workload will be. This reliance on historical data is a potential weakness of the method, although it is unusual for workload to change dramatically on a year to year basis. Comparisons between actual staffing and required staffing, either as a difference between the two or as a ratio of actual staff to required staff (the WISN ratio) provide a useful mechanism for assessing priorities to address staff overloads or staff under-utilisation

WISN was piloted in a number of countries and culminated with its adoption, publication and promotion by the World Health Organization.

Re-admissions (re-hospitalisation) analytics

Re-admissions are costly for any hospital, and in many cases insurers and other payers are not prepared to pay for re-admissions.

A healthcare analytics research project at Wayne State University uses heterogeneous medical data from various healthcare organisations (payers, providers, pharmaceuticals) and analytical models within a  ‘rehospitalisation analytics’ framework to identify, characterise and reduce the risks of re-admissions for patients using their electronic health records. This project aims to provide a comprehensive, accurate and timely assessment of the risk of re-admissions. It has the potential to direct more aggressive treatments towards identified high-risk patients. The project aims to develop integrated predictive models that can effectively leverage multiple heterogeneous patient information sources and transfer the acquired knowledge about re-admissions between different hospitals and patient groups in the presence of only few patient records.

Providing special care for a targeted group of patients who are at a high risk of re-admission can significantly improve the chances of avoiding re-admissions and reducing overall health care costs by reducing the number of re-admissions.

Note: there are many, many more published papers on re-admission analytics – both applied to particular cases (like heart conditions, diabetes, and many more) as well as for the general case. This short summary does not even start to scratch the surface… Watch this space for more exciting posts on predicting hospital readmissions.

Concluding remarks

This is but a small sample of business analytics applied in healthcare, but it shows the breadth and depth of analytical research and advanced analytical models that are already applied in hospitals. This is such a vast area, where so much more analytical outcomes can still be devised and applied to improve both business and patient outcomes.

1 ping

  1. simpleNewz - Martin's Insights RSS Feed for 2014-12-08

    […] Hospital Analytics Case Studies […]

Leave a Reply