Putting aside the debate over health reform, one thing remains certain, the demand for healthcare will continue to grow. A key question is how do we accurately predict and cost the future demand for healthcare provision?
In fact, the key question is more appropriately this: with the move toward more localised commissioning, how do we predict future demand and cost by individual GP practice area?
Through new methodology and by combining new data sources, we have been able to predict and cost future health needs at a level of geographical detail that will ensure more accurate and localised commissioning for the catchment areas which clinical commissioning groups are expected to cover.
Our research is based around linking personal and individualised health risk with lower levels of geography which means healthcare costs can be mapped more accurately.
To date forecasting models have been built using top down approaches but we would argue that the allocation of commissioning budget to GPs requires a clear understanding of health needs and costs at a detailed individualised level to reflect more accurately the existing and future demand of any catchment area.
The methodology we use combines econometric capabilities with consumer insight and demographic mapping. The forecasting model draws data from various Experian proprietary and public data sources including Hospital Episode Statistics, the Health Survey for England and Indices of Multiple Deprivation to produce detailed, flexible and granular forecasts.
With guidance and help from the British Lung Foundation, we have used the model to forecast and map chronic obstructive pulmonary disease. This is one example which has been chosen for the scale of the cost to the NHS; COPD is the second most common reason for emergency admissions and one of the most costly inpatient conditions treated by the NHS. The same methodology can be applied to many other long-term conditions at any geographical level of detail including coronary heart disease or Type 2 diabetes.
The bottom-up approach begins with building the model at a level of geography based on units of around 750 households. It was felt that this level of detail could meet the flexibility requirements of CCGs where boundaries are likely to be varied.
We have developed a range of possible future scenarios modelling up to 2020. With the most important risk factor to COPD being smoking, our models have been built to predict how a change in future smoking behaviour will impact hospital admissions for COPD.
Different scenarios of COPD cost depend on levels of intervention and assumptions around numbers of future smokers. These scenarios can be adjusted to reflect levels of intervention at any stage according to the current or projected rate of smoking cessation.
Proprietary lifestyle segmentation, Mosaic Public Sector, was used to estimate the proportion of COPD patients that are current smokers. This segments all individuals, households and postcodes in the UK into a set of homogeneous lifestyle types. 146 Mosaic person types aggregate into 69 household types and 15 groups, to create a three-tier classification. The types identify individuals, households or postcodes that are as similar as possible to each other, and as different as possible to any other group.
Based on this, the groups which are much more likely to be admitted with COPD compared to the English average are:
- Group M – elderly people reliant on state support
- Group O – Families in low-rise social housing with high levels of benefit need
- Group L – Active elderly people living in pleasant retirement locations
By applying the methodology we were able to analyse and forecast levels of COPD up to 2020. The table below forecasts the cost of COPD by PCT area in 2015 given the Department of Health’s national ambition to reduce the English smoking population to 18.5 per cent (see table 1 - attached right).
Looking at the top PCTs for cost of COPD we can see that in 2015 Southampton PCT is forecast to have the highest cost of COPD admissions per thousand head of population at £10,643. This is followed by Knowsley (£9,175) and Liverpool (£8,901).
By contrast, areas forecast to have the fewest admissions for COPD and the resultant cost, when compared to total population in 2015 are Richmond & Twickenham (£2,182), Kingston (£2,365) and Berkshire West (£2,400).
The cost of COPD admissions reflects areas where there is forecast to be a higher total smoking population, it typically follows areas experiencing greater levels of deprivation alongside a greater total aging population.
Experian has also looked at the rate of increase in COPD admissions between 2011 and 2015 and the associated cost per head of 1,000 population, again given the reduction in smokers to 18.5 per cent of the population (see table 2 - attached right).
It shows areas likely to see a significant rise in the rate of admission as the Isle of Wight (10.5 per cent), Herefordshire (9.1 per cent) and North East Lincolnshire (5.9 per cent). PCT areas where the greatest reduction in rate of admissions can be found are Tower Hamlets (-10 per cent), Islington (-6.9 per cent) and Camden (-6.7 per cent).
Areas associated with higher rates of increase in COPD admissions are mostly identified as retirement destinations; a pattern of which occurs in coastal locations, predominantly in the West, South and East of Britain. Again, this model suggests that an aging population with more established residents have an associated impact on COPD admission rates.
Southampton City PCT is an interesting discovery; whilst it has the highest cost of COPD per ‘000 population in 2015 the total number of admissions for COPD is forecast to decrease by 5.3 per cent, this drop in admission rates still suggests that there is much work to do.
By developing an understanding of the Mosaic Groups that are more likely to be current smokers and how this will change over time, resources can be targeted effectively down to the individual to drive behaviour change and reduce cost for COPD.
We have also forecast the cost and rate of COPD admission by major town territory up to 2020 taking into account a natural reduction in smokers. It generally shows a prevalence of high risk in towns and cities in the North and Midlands with the exception of Southampton which continues to stand out in the south as having far higher numbers of COPD admissions relative to the total population (see table 3 - attached right).
Bootle and Barnsley are forecast to bear the highest cost of COPD admissions per thousand head of population, but again we see popular retirement destinations such as Newport on the Isle of Wight experiencing a significant increase in COPD admission (up 25.1 per cent) by 2020.
Again these tables show that areas with greater numbers of more deprived, aging populations will bear the highest costs for COPD admissions. These also show that great progress will be made in Southampton and Liverpool although the current scale of the problem is large.
Both PCT and town forecasting models, also break down the population for COPD admissions by heavy, light and moderate smokers, along with ex-regular smokers and those that have never smoked. The benefit of using this approach is that forecasts can be tweaked to reflect results of targeted interventions.
By understanding the impact that smoking cessation interventions have on the current smoking population a more accurate picture of cost and demand can be predicted (see table 4 - attached right).
The British Lung Foundation has used our forecasting at a town level, and is able to target localised interventions (spirometry tests) in areas where the costs and rate of increase in COPD are predicted be the highest. While the BLF are interested in identifying those with mild cases of COPD and reaching out to them before their condition becomes more severe and they ultimately place more of a burden on urgent care services in the future, the forecasts provide the evidence of where the problems are likely to be the greatest in future.
In just two years the BLF have tested over 7,200 people for abnormal lung function in England at 12 specifically targeted locations, with a resultant 19.5 per cent of those tested being referred to their GP for further tests. When the BLF have compared their targeted approach to untargeted campaigns they have seen a 300 per cent uplift in attendances in the most at risk populations. Through early identification of COPD, potential savings to the NHS from these events alone, are estimated at over £300,000 per year.
Experian will be using its health forecasting capability to drive further interventions across a range of health conditions. Providing the intelligence at local level will also significantly improve the accuracy of commissioning for future healthcare services in terms of understanding both future costs and demand. Experian’s expertise is built around an understanding of health risk at person level and it is this understanding that will be fundamental to the localism agenda and the new healthcare structure.