Corporate report

Business critical models for Department of Health and Social Care and its arm’s length bodies

Published 19 December 2024

A list of all business critical models held within the Department of Health and Social Care (DHSC) and its arm’s length bodies (ALBs) as at December 2024.

DHSC models

Adult social care demand models

A suite of models is used to project user numbers and expenditure for adult social care over both the short, medium and longer term.

A shorter-term model is used to estimate the total expenditure necessary to maintain the existing publicly-funded adult social care system given demographic and unit cost pressures over the time frame of a spending review period.

Over longer time periods than a spending review, expenditure projections for adult social care are estimated using a long-term demand model, which is consistent with the shorter-term modelling.

To produce these expenditure projections, both of these models make use of externally commissioned projections of adult social care demand (in particular user numbers) from the Care Policy and Evaluation Centre at the London School of Economics. 

Average cost model (reciprocal healthcare)

This model estimates the average annual cost of healthcare in the UK for different age groups. Calculations are based on expenditure and population data for each UK nation, adjusted for age groups using age-cost curves. The inputs to the model are updated annually and the new results are submitted to the EU each year to inform charging under reciprocal healthcare arrangements.

Calorie model

This is a simulation model used to model the long-term impacts of policies that affect calorie intake at a population level. It uses estimates of change in calorie intake, along with other assumptions, to estimate the effect on obesity-related health outcomes, NHS treatment costs, social care costs and changes in economic output.

Clinical negligence reform modelling

This models the costs of clinical negligence over the next 70 years under the current system, as well as under various changes to the system to consider the impact of policy changes.

Community Pharmacy Contractual Framework fees model

Model to estimate annual spend on remuneration fees made to community pharmacies in England as part of the Community Pharmacy Contractual Framework.

Comprehensive investment appraisal model

It provides the basis on which a decision is made to invest in a proposed development and to select the option that provides the best value for money to the public sector. The model appraises shortlisted options for capital investment by calculating the economic benefits, costs and risks of a range of shortlisted options.

CVDPREVENT

CVDPREVENT data supports the delivery of the CVDPREVENT audit, which aims to use monitoring to improve cardiovascular disease (CVD) prevention and treatment of CVD and high-risk conditions in primary care. The data is used to generate indicators of prevalence, treatment and outcomes, broken down by health geographies and health inequalities.

Drug Tariff for category M, category A and special drugs

A suite of models that adjust generic drug reimbursement prices as reflected in the Drug Tariff for generic medicines.

Ethnicity assignment model

The ethnicity assignment model generates the most frequent ethnic code recorded for each individual within Hospital Episode Statistics, based on multiple criteria. It allows us to estimate the ethnicity for individuals when multiple ethnicities are recorded in the data.

Excess mortality

The model estimates the expected number of deaths that are registered each month for multiple population subgroups (all persons, English regions, by age group, by sex, by deprivation quintile, by cause of death and by place of death). It then compares the observed count of registered deaths with those expected deaths to quantify the excess deaths for that period.

Human resources analytics dashboards

A suite of 9 dashboards (some of which are under construction) showing workforce size and shape, diversity and inclusion, sickness and absence and so on. All data presented is aggregated for anonymity and General Data Protection Regulation compliance.

Immigration health surcharge (IHS) fee calculation model

Calculates the average cost of healthcare services available to IHS payers - to determine the annual fee that those within scope of the IHS will be required to pay.

Infected blood support payments costing

The model is used to estimate the DHSC liability for payments under the England Infected Blood Support Scheme (EIBSS) and for future UK wide interim compensation payments (paid to infected individuals and bereaved partners eligible for support payments). This is done through estimating the number of infected beneficiaries and their bereaved partners who may be entitled to financial support over the next 80 years or more.

Medical and non-medical education and training cost model suite

A collection of models that estimate outlay costs associated with the number of medical and non-medical students entering training.

A suite of medicines expenditure models

These models consist of:

• a forecast model of future drug spend in England based on typical lifecycles of expenditure

• a model which supports the voluntary scheme through calculation of sales growth rates and payment percentages.

These are updated annually.

Medicine margin survey

This model estimates the retained medicine margin made by community pharmacies in England. Data from a survey of community pharmacies informs the model, which estimate the medicine margin from the NHS Drug Tariff prices.

National Living Wage pay model

This model estimates the cost to the state of uplifting the National Living Wage in adult social care over a spending review period. It does this using Low Pay Commission projections for the National Living Wage, applied to Skills for Care data on the wage distribution within adult social care.

Pay negotiation and pay strategy modelling suite

Suite of models to:

• estimate the breakdown of historical pay bill by staff group and cost stream
• quantify historical paybill drivers (such as pay drift and its components)
• link with workforce planning models to forecast future paybill
• cost specific proposed pay reforms using off-shoot models (for example junior doctor contract review and review of consultant doctor reward package)
• inform further modelling by NHS Pensions Authority and Government Actuary’s Department on NHS Pension Scheme
• inform internal briefing around internal paybill and workforce forecasting
• support consideration of pay reform

Public health grant (PHG) demand model

Models the required level of funding needed to maintain PHG service provision at current levels. This considers wage, inflation, demand and policy pressures.

PHG return on investment (RoI) model

Estimates the RoI of marginal changes to the PHG budget and allows users to model different funding scenarios to estimate the health and financial impacts of changes to the PHG funding level.

Public Health Outcomes Framework

The Public Health Outcomes Framework presents indicator trends and benchmarks to enable assessment of outcome measures at upper tier and lower tier local authorities.

Smoke-free generation Markov model

Model used to estimate the impact of the smoke-free generation policy on smoking prevalence, and the consequent long-term impact on smoking related diseases, costs and mortality. The core of the model is based on transition probabilities from the University of Sheffield’s tobacco model.

Adult social care relative needs formula

This model aims to estimate the relative needs for local authority supported adult social care and is used to distribute adult social care funding. The model includes variables that are comparable between local authorities and proxies local authority supported eligibility factors that are legitimately outside of local authorities control including:

• population
• impairment levels
• income and wealth
• unpaid care support
• wages and prices
• population sparsity

Vaccine damage payment scheme

Estimates the number of future claims, successful claims and payout of the vaccine damage payment scheme (which provides payments to individuals who have experienced serious adverse events from vaccines).

Arm’s length bodies (ALBs) models

The models used by each ALB are listed below.

UK Health Security Agency (UKHSA)

Avian influenza susceptible, exposed, infected, recovered (SEIR) model

This model uses a SEIR structure to estimate the number of avian influenza:

• infections
• hospitalisations
• intensive care unit admissions
• deaths

in England under different disease scenarios and vaccination strategies.

Combined R statistical model (CrystalCast)

During COVID-19, the Defence Science and Technology Laboratory and UKHSA used CrystalCast to combine a number of epidemiological nowcast and forecast estimates, including R, as well as combined medium-term projections of hospital admissions, bed occupancy, and mortality. It combines estimates and projections from multiple epidemiological models to produce a consensus, allowing decision-makers (ministers and senior officials) to weigh up uncertain conditions and parameters.

Additional information on the process and the usage of CrystalCast to produce nowcast consensus R and MTPs are on the National Library of Medicine and the Royal Society websites.

COVID-19 vaccine contingent liability model

The model estimates the maximum potential liability that may be incurred through the administration of COVID-19 vaccines.

COVID-19 vaccine provision model

The model estimates the cost of the provision following the administration of the COVID-19 vaccine.

COVID_Hosp_IBM

Individual-based model of SARS-CoV-2 transmission between staff and patients in English hospitals.

Carbapenemase-producing enterobacterales hospital transmission individual-based model

Individual-based model of carbapenemase-producing enterobacterales transmission within an acute NHS hospital trust.

COVID-19 Vaccine Unit inventory forecasting model (scenario modelling tool)

Models demand and supply to understand inventory, expiry and unmet demand for the COVID-19 vaccine.

Dose estimate

Software for estimation of eternal ionising radiation dose and circumstances of exposure, based on International Organization for Standardization standard methods using chromosome aberrations in blood.

Elderly pneumococcal model

Markov model that evaluates the impact and cost-effectiveness of different immunisation strategies against pneumococcal disease in the elderly.

Flu evidence synthesis model

Used to model the cost-effectiveness of influenza (flu) vaccination. This consists of a Susceptible, Exposed, Exposed, Infected, Infected, Recovered (SEEIIR) Ordinary Differential Equations (ODE) model. It also contains components to model the potential impact of updated vaccine programmes, including calculating the averted burden and quality-adjusted life years (QALYs) as a result of implemented vaccine programmes.

Frameworks – Python generic ODE modelling (PyGOM)

PyGOM is a framework for production of ODE based compartmental models and has been used for work on enclosed populations including cruise ships, prisons, UK population-wide model and work on repatriations.

General Register Office (GRO) daily deaths

This is an all-cause mortality model for England. It produces a daily calculation using the date of death, with a baseline estimate based on the past 5 years. This model can help determine likely excess deaths due to factors such as COVID-19, influenza, heatwaves and cold snaps. This model uses data from GRO.

Hepatitis C virus burden model

Estimation of hepatitis C virus prevalence, burden and treatment impact using multiple sources of data within a Bayesian model.

Human immunodeficiency virus (HIV) backcalculation model

Estimation of HIV incidence on the basis of new HIV diagnosis data and CD4 counts at diagnosis, using a Bayesian backcalculation model.

HIV multi-parameter evidence synthesis

Estimation of HIV prevalence based on multiple sources of data, using a Bayesian evidence synthesis model.

Hospital onset length of stay attribution

Causal inference model which uses inverse probability weighting to correctly account for time-dependent confounding. Estimates excess length of stay attributable to hospital-onset infections.

Hospital onset mortality attribution

Causal inference model which uses inverse probability weighting and Cox models to correctly account for time-dependent confounding. Estimates excess mortality attributable to hospital-onset infections.

Hyades

Internal radiation dose estimates for clinical trials with radio-labelled compounds.

Integrated Modules for Bioassay Analysis (IMBA)

Internal radiation dose assessment software.

Joint Agency Modelling

Joint Agency Modelling is a multi-agency tool and supporting process. It enables rapid assessment of the impacts of atmospheric releases during radiation emergencies, based on an estimate of the radioactive material that has (or could be) released. It aims to support strategic decision-making in central government.

Mass assessment suite of codes

A suite of codes based on IMBA and Taurus that can perform automated internal radiation dose assessments for large groups of individuals.

Measles risk model

Mathematical model of measles in England regions.

Mpox transmission and vaccination model

The model consists of 2 parts:

• transmission dynamics (calibrated to outbreak data and to evaluate the impact of different control measures)
• investigates the cost-effectiveness of different vaccination strategies

Pay model suite

Suite of models used for internal pay and HR assessments and cost impacts.

Personal computer - consequences of releases to the environment assessment methodology (PC-CREAM)

PC-CREAM is a model used to assess the radiological impacts of planned discharges of radioactive material to the atmosphere or to water bodies.

Pertussis model

Deterministic model of pertussis transmission dynamics and cost-effectiveness model of maternal pertussis vaccination programme in England.

Pneumococcal model

Deterministic and individual-based models, and cost-effectiveness models of pneumococcal transmission dynamics and pneumococcal vaccination programmes in England and Wales.

Real-time transmission model for pandemic respiratory pathogens

SEIR-based transmission modelling framework using multiple sources that can include prevalence data (both serological and virological) as well as time series data on infection sequelae (such as symptom onset, hospitalisation, death). The model can be very flexibly parameterised to account for the specific characteristics of the pathogen under consideration and the data available.

ReverseEpi

The model locates the source of high-impact biological agents based on cases admitted to hospital.

SynthesisCOVID

It estimates the changes in transmissibility and severity of COVID-19 in UK and vaccine efficacy by fitting a SEEIIR compartmental transmission model to the data of confirmed cases, deaths, and recovery. During the pandemic, the model was calibrated with available data and was used to predict a 2 week moving window.

SynthesisFlu

It estimates the changes in transmissibility and severity of flu in the UK by fitting a SEEIIR compartmental transmission model to the data of confirmed cases, hospitalisations, and intensive care units.

During the winter pressures season, the model was calibrated with available data, and used to predict a 2 week moving window.

Taurus

Internal radiation dose assessment software (International Commission on Radiological Protection, ICRP 2007 recommendations).

Medicines and Healthcare products Regulatory Agency

Safety signals detection analytical models

These models are used to conduct statistical analysis of adverse incidents to medicines, vaccines and device reports to identify possible safety signals for more in depth clinical and non-clinical assessment, potentially leading to a full safety review.

Safety, quality and efficacy assessment model

This model is made up of national legislation, guidance and appropriate international standards. It’s within numerous disciplines (pre-clinical, clinical, pharmacological quality, toxicological, clinical pharmacological and statistical). It assesses applications for individual products to assess the quality, safety and efficacy of the product, as well as the ability for it to be manufactured, supplied appropriately and monitored in use. On the basis of this analysis, a marketing authorisation decision will be made to approve or refuse the medicine to be placed on the market in the UK.

NHS England

Accident and emergency (A&E) admissions forecasting tool

The tool aims to provide forecasts of admissions from A&E at the trust level, one to 3 weeks ahead.

A&E performance model 

A multiple regression model designed to predict A&E 4 hour performance based on:

• beds
• bed occupancy
• A&E attendances
• conversion rate
• admissions

Allocations model

The allocations model is used to determine the ‘fair shares’ allocation for commissioners (integrated care systems). The formulae are derived, in the most part, through patient-level regression models.

Alzheimer’s disease mild cognitive impairment disease modifying treatments model  

Resource impact of new disease modifying treatments (lecanemab and donanemab) with full patient pathway analysis.

Ambulance response model    

Model to understand ambulance resource requirements, forecast response times.

Cardiovascular disease - respiratory Long Term Plan commitments modelling

Modelling of potential benefits - in heart attacks and strokes avoided, lives saved, admissions reduced - from various prevention, rehabilitation and other initiatives.

COVID-19 expected vaccinations

The COVID-19 expected vaccinations model, estimates the weekly expected number of vaccinations for each cohort by comparing demand, capacity and supply.

Elective activity projections  

Calculations on national activity recovery trajectories (in terms of value weighted activity and referral to treatment) and convert between these and point of delivery (POD) activity.

Elective recovery fund

The model calculates financial rewards to systems (sustainability and transformation partnership and integrated care systems) based on if they have achieved expected levels of elective care activity.

Enhanced access funding allocations

In 2022 to 2023 2 budgets for extended access in primary care (GP practices opening longer in the evenings and at weekends) were combined into a single budget for primary care networks (PCNs) with a change in the required opening hours. The 2 previous budgets were distributed to different local organisations using different formulae. The model assessed different methods for allocating the new single budget to PCNs and calculated the 2022 to 2023 allocations to PCNs based on the preferred method.

GP referral analysis

Analysing historical GP referral patterns to gain an indication of which hospitals’ general practices (and their patients) have chosen to go to and whether 2 or more providers who are merging are commonly chosen alternatives to each other. 

Greener NHS carbon footprint model

The greener NHS carbon footprint model produces estimates of the total NHS carbon footprint and carbon footprint plus as defined in the Delivering a net zero NHS report.

High-level allocations model

The model is used to set high-level allocations, the total quantum of funding available for each of the different commissioning streams.

Long Term Workforce Plan model

The Long Term Workforce Plan model has been built to project NHS workforce supply and demand from 2021 to 2022 until 2036 to 2037, including modelling of policy interventions to reduce the gap between counterfactual demand and supply.

Market force factor model

The market force factor is a measure of unavoidable cost differences between healthcare providers, and a means of offsetting the financial implications of these cost differences. Each NHS provider is assigned an individual market force factor value, which is used to adjust national prices and/or unit prices for the services trusts provide. As well as being part of the National Tariff and NHS Payment Scheme, it is also used to adjust allocations.

Medicines spend forecast

Forecast of spend on medicines by NHS in England.

New Hospital Programme demand model 

A Monte Carlo based simulation of demand in acute hospitals, using baseline Hospital Episode Statistics data and a number of change factors (demographic growth, non-demographic growth, activity mitigation, and others) to project demand for hospital care into the future.

New Hospitals Programme integrated economic and financial model

Projects benefits against costs for New Hospitals Programme.

New Hospitals Programme should cost model

Department-based cost model to give an early indication of the New Hospitals Programme costs.

Patient satisfaction with GP service analysis  

Ordered logit regression model to identify drivers of patient satisfaction with their GP and their experience of making an appointment with their GP from the GP patient survey.

Obesity management - tirzepatide   

Model to understand resource and cost implications of a national roll-out of tirzepatide for obese adult patients. Estimated impact on existing services.

Referral to treatment (RTT) demand scenario model      

Scenario model that projects RTT clock starts.

RTT waiting list model        

Stock and flow based model for projecting the total RTT waiting list and number of patients waiting in 4 week time bands.

Tariff production models

This is a suite of models with the function of developing prices to inform or for use in funding non-primary health care delivery. 

Unavoidable costs v2

The unavoidable cost model, based on the financial year 2018 to 2019 national cost collection patient level information and costing system dataset, identifies the unavoidable costs incurred by small departments, sites, and providers in the provision of admitted patient care as a result of their small size.

Workforce-activity model

The model calculates a staffing ratio for each activity type in the acute sector.

Care Quality Commission (CQC)

Fees model

Models for each sector regulated by CQC to calculate appropriate regulatory fees based on the costs of regulation in that sector.

Indicators for evidence

A suite of indicators that measure aspects of quality of care for CQC regulated health and adult social care services. These indicators use data from a wide range of national data collections, including nationally agreed measures of quality. These indicators are mapped to the assessment framework used to carry out assessments and ratings of services. The results of the indicators are considered alongside other relevant evidence to inform judgements.

Risk categorisation

A risk categorisation is used to inform the order in which assessment activity across registered health and adult social care services is carried out. Each service is assigned one of 3 categories. The categories are driven by a rules set and, where they are in place, any risk scores from service specific risk models.

Risk model: adult social care services

Models to support prioritisation of services most at risk of providing poor quality of care. The adult social care risk model combines results from a set of indicators relevant to these services to create a risk score. The risk score is used to calculate an overall risk category for each service. The model covers residential and community adult social care services.

Risk model: general practice

Models to support prioritisation of services most at risk of providing poor quality of care. The GP risk model combines results from a set of indicators relevant to these services to create a risk score. The risk score is used to calculate an overall risk category for each service.

Risk model: independent health services

Models to support prioritisation of services most at risk of providing poor quality of care. The independent health risk model combines results from a set of indicators relevant to these services to create a risk score. The risk score is used to calculate an overall risk category for each service. The risk model covers a wide range of services, across diverse settings (such as hospital, clinics, diagnosis services, dialysis units and so on).

Risk model: urgent and emergency care

Models to support prioritisation of services most at risk of providing poor quality of care. The NHS urgent and emergency risk model combines results from a set of indicators relevant to these units to create a risk score. The risk score is used to calculate an overall risk category for each service.

Human Tissue Authority

License fee model

This model is used to calculate the annual fees of licensed establishments regulated by the Human Tissue Authority by allocating forecast Human Tissue Authority organisational costs against expected regulatory activity for different establishments and sectors.

National Institute for Health and Care Excellence (NICE)

Guideline models

NICE guideline models are used to estimate the cost-effectiveness of various interventions (including treatment, care, testing and follow-up) across clinical, public health and social care settings. The models are used to support recommendations for affordable best practice within the NHS and personal social services, and sometimes the wider public sector. The models use cost utility analysis with benefits measured using QALYs, and allow probabilistic analyses. External academic or commercial models may sometimes be used if relevant to the decision.

The numbers of models developed by the programme vary by year, but for 2024 to 2025 an approximate estimate is 15 models.

HealthTech guidance models

The NICE HealthTech programme produces guidance on interventional procedures, diagnostics, devices and digital health technologies. HealthTech guidance models are used to estimate the cost-effectiveness of technologies and are used to support recommendations for the use of the technologies within the NHS and personal social services. The models use cost utility analysis with benefits measured using QALYs and allow probabilistic analyses. Cost-comparison models may also be used, which undertake analysis of the costs and resource use associated with the health technology compared with that of the comparators.

The numbers of models considered by the programme vary by year, but for 2024 to 2025 an approximate estimate is 25 models.

Medicines guidance models

NICE guidance on medicines is developed through the technology appraisals and the highly specialised technologies programmes. Models are used to estimate the cost-effectiveness of medicines (and sometimes other health technologies) and are used to support recommendations for the use and funding of the medicines within the NHS and personal social services in England. The models are predominately developed externally by the medicine technology manufacturers, health economic consultancies or academic assessment groups. The models use cost utility analysis with benefits measured using QALYs and allow probabilistic analyses. Cost-comparison models may also be used, which analyse the costs and resource use associated with the health technology compared with that of the comparators for circumstances when it is reasonable to assume equivalency of health outcomes.

The number of models used within the programmes varies each year, but an approximate estimate for 2024 to 2025 is 80 to 85 models. 

NHS Blood and Transplant

Blood demand forecast

A statistical model is used to forecast demand for blood components over a period of zero to 5 years, using historic demand trends overlaid with other sources of insight such as population and demographic data.

Blood supply and stock forecasting

Factors such as booked blood donor appointments and forecast attendance rates are used to forecast expected blood collection levels, which when combined with anticipated demand and wastage can be used to produce blood stock forecasts.

Blood pack type selection

Using inputs of booked donation appointments (by blood group and gender) and anticipated product demand or stocks, the optimal pack type that each unit of blood should be collected into is derived. This is important because some blood products can only be manufactured from certain pack types.

Blood platelet stock management algorithm

A tool that makes platelet product inter-site distribution decisions, based on product availability, current local stocks at 14 distribution centres and anticipated demand. Also factors in product characteristics and shelf life.

Organ allocation algorithms

Statistical models are used in the development of organ allocation algorithms for the UK. These models identify and evaluate factors associated with waiting times to transplant and post-transplant outcomes. In turn, they inform algorithms designed to appropriately match and prioritise patients on national waiting lists for organ transplantation.

NHS Business Services Authority (NHSBSA)

England Infected Blood Support Scheme (EIBSS) forecasting

Forecasting the annual number of scheme applications, and their related payment value for the EIBSS. This informs the annual EIBSS payment budget for the service, set by DHSC.

Forecast out-turn prescribing monitoring document

The forecast out-turn is a profile to help forecast the primary care prescribing spend for each practice for the financial year, the profile is calculated using the previous 5 years of data for England. The forecast out-turn is provided by DHSC at the beginning of each financial year. 

The forecast out-turn is used by the NHSBSA in the prescribing monitoring document accessible by GP practices in ePACT2, alternatively the forecast template can be used by GP practices locally to reflect any local initiatives or adjustments. 

Healthy Start forecasting

Forecasting of the number of Healthy Start cards, beneficiaries, top-ups and spend. This helps to inform the card funding annual budget for DHSC. Operational forecasting and spend analysis is conducted. Drawdown requests are made against the budget and DHSC portal is used for funding requests. 

Nursery Milk Scheme forecasting

Forecasting the number of milk and formula portion claims submitted by local authorities, childcare settings, and agents, as well as the average cost per portion. This informs the reimbursement costs for the Nursery Milk Scheme. Forecasts are updated quarterly or as needed in response to any anticipated changes in milk or formula prices, as agreed upon with suppliers.

Student services forecasting

Forecasting the expenditure for NHS Learning Support Fund, NHS bursary and social work bursary. This helps to informs the annual budget for the service to pay students and health education institutions, set by DHSC. 

Vaccine damage payment forecasting

Forecasting of the number of future claims, cost of medical assessments, NHSBSA administration costs, potential payments to successful claimants and administration payments to healthcare providers. This helps to informs the annual budget for the service, set by DHSC. Further monthly operational forecasting takes place, which is reviewed monthly between DHSC and NHSBSA to monitor spend.

NHS Resolution

Cashflow projection models

Range of models and sub-models to derive future year cashflow projections for each scheme.

Contribution setting models

Three main models and range of sub-models to calculate member level contributions.

Incurred but not reported models

Over 100 models and sub-models to calculate provisions for incurred but not reported claims by agreed claims segmentation.

Known claims models

Ten models (one for each scheme) calculate known claims provisions.

Ogden model

Calculates reserve for reported potential high value claims (period payment orders (PPOs)).

Settled PPOs model

Calculates provision for settled PPOs.

Settled PPO recalculation model

Calculates payments for settled PPOs.