Type: Health Service Modelling Associates (HSMA4) project
Title: "The role of Patient Initiated Follow-Up and 'Digital Outpatients' in supporting the elective recovery"
Period: January-July 2022
HSMA4 project members: Martina Fonseca and Xiaochen Ge, NHS England
HSMA4 mentor: Dr Alison Harper, U Exeter
Full documentation: MkDocs repo (private repo)
Context
The pandemic has had a considerable impact on elective care and currently, the waiting list is at its highest on record, with over 6.6 million people waiting for consultant-led elective care [1]. It is estimated that around 6M fewer people completed elective pathways between January 2020 and July 2021 [2]. Operational performance, defined by the proportion of those seen within 18 weeks (target of 92%) has dropped significantly. The proportion seen within 18 weeks has dropped to below 65%[1].
Patient Initiated Follow Up (PIFU) is one of the main tools in the Recovery Plan, alongside other digital solutions such as advice & guidance (A&G) e-referrals.
Advice and guidance allows a clinician to seek advice from another. This may in certain instances mean avoiding an unnecessary GP referral, while seeking advice and guidance from the consultant.
PIFU is where, based on clinical steer, a patient goes on a pathway where they can arrange their own follow-up appointments as-and-when-needed rather than attending regular scheduled appointments. It means for instance that time spent on low clinical value appointments can be repurposed for managing complex patients and upstream care. The interplay between the downstream efficiencies and the upstream referral-to-treatment waiting list, as well as changes to planned levels of resourcing and allocation, can however be hard to grasp.
As such, in this project we propose to use Discrete Event Simulation - where individual patients’ pathway movement through a system with constrained resources is simulated and monitored - to better understand the role that PIFU and other interventions can play in addressing the elective (referral-to-treatment) backlog and supporting patient care.
Project objectives
- Use Discrete Event Simulation to better understand the role patient initiated follow-up (PIFU) can play in addressing the (outpatient and elective) referral-to-treatment (RTT) backlog, with particular emphasis on Rheumatology
- Advocate and demonstrate the value of pathway and behavioural modelling in central and Integrated Care System decision-making for digital pathway redesign, elective recovery and the Long-Term Plan personalisation agenda and Public Health Management Agenda, to support better patient care, experience and system resilience
We focussed on rheumatology given that rheumatology has a good clinical evidence base on PIFU [5-10]; documented pathways [3,7]; PIFU is actively endorsed [3,4,7]; and that it is a mainly outpatient specialty with many chronic patients on long-term follow-up, meaning that the effect of PIFU may be amplified [3].
An overview of the simplified pathway / process map, as employed in the DES simpy model, is shown below.
An intuition of the flow (with changed parameters to improve intuitiveness and clarity of visualisation), is given in the animated process map flow below.
- The main code is found in the
./scrfolder of the repository (see Usage below for more information) - The accompanying documentation can be found here (private repo): MkDocs repo.
To run the model locally you will need to have installed GitHub Desktop (or equivalent like git cli), Python and VSCode (or equivalent editors like Spyper). Below we outline the process with these softwares and based on Windows, but others could be used.
- Search and install.
- Set up with github account
Search and install.
3. Install VS Code
- Search and install.
- Visit: https://github.com/nhsx/HSMA4-12-DES-rheum/
- click the <>Code button and click Open with GitHub Desktop
- click the Clone button to save/clone the code.
- click Current Repository, select HSMA4-12-DES-rheum and right-click on the mouse/trackpad and select Open in Visual Studio Code; this will open VSCode on a separate window.
-
On the VSCode, press the “extensions” button on the left-hand panel (or press Ctrl+Shift+X). Search for “python” and press install.
-
At the top bar, click Terminal > New Terminal (if not in there, click on the three-dot … and click Terminal > New Terminal)
-
On the terminal window, use this command
python -m venv venvpa- Press Shift + Control + P to open the Command Palette and click on the Python: Select Interpreter.
- Select the Python interpreter
venvpa. - Activate the virtual environment using powershell (you may need to change elevation rights):
venvpa\Scripts\activate.ps1 - Install all the required libraries by using the command:
pip install -r requirements.txt
- Or the following:
pip install pandas numpy matplotlib seabornpip install simpy streamlit scipy
The main simpy model implementation is in Batch_rheum_Model.py but the model is run from either a Python script (full flexibility) or an interactive web app (GUI but more limited user parameters).
Run simpy_rheum_v004_run.py in Python (spyder, VS Code) or from simpy_rheum_v004_notebook.ipynb (Jupyter Notebook).
If using the Python notebook, use pip install ipykernel and select the interpreter as venvpa above. The cells can then be run.
A prototype toy interactive app leveraging streamlit was also worked on in Streamlit.
For local machine deployment: the script can be found here streamlit_model_mf_v0031_batch.py.
Overview instructions:
- Navigate to code folder in terminal / powershell by typing
cd src. - Type in terminal / powershell
streamlit run streamlit_model_mf_v0031_batch.py.
Detailed instructions: see this tutorial for guidance.
Some post-processing scripts for the generated logs from the simulation have been created in R, see https://github.com/nhsx/HSMA4-12-DES-rheum/post-processing. These include:
-
Plots of KPIs across scenarios and replications, as shown in example outputs below (with
ggplot2) -
Process mapping outputs (with
bupaR)These will need adapting for new outputs and experiments.
As an example of the model in operation, we aimed to create a hypothetical baseline case for a fictitious trust with realistic parameters and assumptions.
We obtained parameter inputs from GIRFT guidance, NHS Digital, clinical data and publications to build a hypothetical baseline case for a fictitious trust with realistic parameters and assumptions, including reasonable resulting waiting list characteristics.
We then applied what-if modelling to this baseline case based on PIFU or A&G interventions.
Some details of parameters for the use case, both the baseline and what-if, can be found below (for full details within script and - upon request - via project report).
Some outputs are shown below:
nb: simpy output logs themselves not uploaded to repo due to size. Can be regenerated with the code given (the current parameters are set to match scenario A above, for the use case "V3"). The plots were created upon post-processing of simpy logs of appointments and patients across simulated scenarios, with R script 80-postprocessing_plots.R.
See the repo issues for a list of proposed features (and known issues).
Contributions are what make the open source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated.
- Fork the Project
- Create your Feature Branch (
git checkout -b feature/AmazingFeature) - Commit your Changes (
git commit -m 'Add some AmazingFeature') - Push to the Branch (
git push origin feature/AmazingFeature) - Open a Pull Request
See CONTRIBUTING.md for detailed guidance.
Unless stated otherwise, the codebase is released under the MIT Licence. This covers both the codebase and any sample code in the documentation.
See LICENSE for more information.
The documentation is © Crown copyright and available under the terms of the Open Government 3.0 licence.
To find out more about the Digital Analytics and Research Team visit our project website and Data Science Signpost site or get in touch at [email protected].
To find out more about the HSMA programme please visit the HSMA website.
[1] NHS England, “Consultant-led Referral to Treatment Waiting Times,” NHS, 2022. https://www.england.nhs.uk/statistics/statistical-work-areas/rtt-waiting-times/ (accessed Aug. 22, 2022).[2] Health Foundation, "Elective care: how has COVID-19 affected the waiting list?", https://www.health.org.uk/news-and-comment/charts-and-infographics/elective-care-how-has-covid-19-affected-the-waiting-list#:~:text=Between%20January%202020%20and%20July%202021%2C%20more%20than%206%20million,2021%20(January%20to%20July) (accessed December 2021)
[3] A. Macgregor, L. Kay, and P. Lanyon, “Rheumatology GIRFT Programme National Specialty Report,” 2021.
[4] NHS England, “Delivery plan for tackling the COVID-19 backlog of elective care,” 2022.
[5] R. Panduro Poggenborg, O. R. Madsen, L. Dreyer, G. Bukh, and A. Hansen, “Patient-controlled outpatient follow-up on demand for patients with rheumatoid arthritis: a 2-year randomized controlled trial,” # International League of Associations for Rheumatology, p. 2021, doi: 10.1007/s10067-021-05674-y/Published.
[6] S. Hewlett et al., “Patient initiated outpatient follow up in rheumatoid arthritis: Six year randomised controlled trial,” Br Med J, vol. 330, no. 7484, pp. 171–175, Jan. 2005, doi: 10.1136/bmj.38265.493773.8F.
[7] NHS England, “Implementing patient-initiated follow-up (PIFU) in adult rheumatology services,” 2022.
[8] S. Child, V. A. Goodwin, M. G. Perry, C. A. Gericke, and R. Byng, “Implementing a patient-initiated review system in rheumatoid arthritis: A qualitative evaluation,” BMC Health Serv Res, vol. 15, no. 1, Apr. 2015, doi: 10.1186/s12913-015-0837-9.
[9] R. Whear, A. K. Abdul-Rahman, K. Boddy, J. Thompson-Coon, M. Perry, and K. Stein, “The Clinical Effectiveness of Patient Initiated Clinics for Patients with Chronic or Recurrent Conditions Managed in Secondary Care: A Systematic Review,” PLoS One, vol. 8, no. 10, Oct. 2013, doi: 10.1371/journal.pone.0074774.
[10] R. Whear et al., “Patient-initiated appointment systems for adults with chronic conditions in secondary care,” Cochrane Database of Systematic Reviews, vol. 2020, no. 4. John Wiley and Sons Ltd, Apr. 09, 2020. doi: 10.1002/14651858.CD010763.pub2.