GMs rurality - NHS Employers

6 downloads 274 Views 2MB Size Report
Number of large and small practice a certain distance from the next nearest GP ...... During the periods for which the d
Report to NHS Employers Adjusting the General Medical Services Allocation Formula for the unavoidable effects of geographically-dispersed populations on practice sizes and locations March 2006

Contents 1.

Introduction 1.1. Context of this work: Present Funding of General Medical Services 1.2. Objectives for the work described in this report 1.3. Scope of this report 1.4. Status of the material in this report 1.5. Structure of this report

1 1 2 3 4 4

2.

Analytical framework adopted for this work 2.1. Examining what constitutes unavoidable costs 2.2. Defining criteria for assessing whether practices are “appropriately small”. 8 2.3. Analysis of the extent to which practice costs vary with size of Practice 2.4. Analysis of patient travel times an travel costs 2.5. An analytical model of the rurality adjustment 2.6. Exclusions from our analysis 2.7. Differences between the existing approach to allocating funds and the approach considered here

6 6

8 10 12 13 14

3.

Data and Data Sources 3.1. GP Practice Cost Data 3.2. GP Practice Location Data 3.3. GP Practice Population Data 3.4. Matching Practice and locations data 3.5. Model Coverage 3.6. Geographical coverage

16 16 17 17 18 19 19

4.

Modelling of economies of scale 4.1. Overview of the data with summary statistics and graphs 4.2. Variation in the data 4.3. Differences in average costs per patient with list size 4.3.1. Dispensing status 4.3.2. Potential magnitude of additional funding at small list sizes 4.4. Regression analysis 4.4.1. Single variable regression 4.4.2. Summary of results of single variable regression 4.4.3. Multivariate regression 4.5. Typical extra cost per patient 4.6. Comparison with the Carr-Hill analysis underlying the present formula 4.7. Conclusions 5.1. GIS Modelling Methodology 5.1.1. Travel Distances and Time 5.1.2. Travel Costs 5.2. Travel Distance Correlations

21 22 24 25 25 26 30 31 31 35 36 36 37 39 39 39 41

5.3. 6.

Travel Distances and Costs

42

Appropriately Small GP Practices 6.1. Unavoidable Travel Distances 6.2. Unavoidable Net Costs 6.3. The factors influencing net additional travel costs 6.4. Using distance to next nearest GP practice as a criterion for defining whether practices are appropriately small. 6.5. Prevalence of Unavoidably Small GP Practices 6.6. Conclusions

45 45 49 51

7.

Unavoidable Cost Adjustment 7.1. Derivation of the Adjustment 7.2. Comparison with the Current Approach

57 57 58

8.

Conclusions

62

9.

Appendix 1 – List of Variables

65

10.

Appendix 2 – Characteristics of GP Practices

66

11.

Appendix 3 - Precedents on Access from competition authorities

67

12.

Appendix 4 - Analytical solution to rurality adjustment of GMS

69

53 54 56

GMS Funding Formula Rurality Adjustment

Executive Summary

NHS Employers commissioned Deloitte, working in collaboration with SDG, to carry out research to estimate the unavoidable effect of geographically dispersed populations on the sizes and locations of practices providing General Medical Services (GMS). This report describes the results of this research, which was relatively small scale, comprising approximately 11 manweeks of work. Terms of reference and approach The terms of reference for this research indicated that it is well established that there are diseconomies of scale for smaller practices. They required us to consider the extent to which such diseconomies of scale would lead to practices serving rural areas having unavoidably higher costs, and hence a requirement for increased funding, due to their smaller number of patients (i.e. smaller list sizes). The terms of reference also indicated that costs should be regarded as unavoidable if they would be present even with an optimal configuration of practices, but should not be regarded as unavoidable if they reflected sub-optimal configuration of practices. In the light of these terms of reference we have interpreted the unavoidable costs of serving geographically dispersed populations as follows: “The unavoidable costs of serving geographically dispersed populations are the additional costs that practices incur when they are “appropriately small” given their circumstances. A practice is “appropriately small” if the disadvantages to patients resulting from a configuration with larger practice sizes would outweigh the cost savings resulting from larger practice size. Under this definition, practices in rural areas may be “appropriately small”. If they were larger, for example if they merged with neighbouring practices, access for many of their patients might become significantly more inconvenient and costly, outweighing any savings from larger scale. The additional costs arising from small size are thus appropriately incurred. In contrast urban practices are less likely to be appropriately small because it is less likely that the additional costs to patients of travelling to different sites would be larger than cost savings from merger. This definition of when practices are appropriately small requires an assessment of the trade-off between cost savings in practices and additional travel burdens on patients. We have made this assessment by estimating the following. z

How practice costs vary with size of practice, specifically how costs per patient vary with list size.

z

The additional travel costs that are incurred as a result of changes in practice configuration. The additional travel costs include both the value of patients travel time and the cash costs of travel they incur. These costs are used to represent the disadvantages to patients arising from the different travel burdens of different practice configurations.

When practices are appropriately small, the magnitude of the additional funding can be assessed based on the size of the practice. This approach is illustrated in the diagram below.

Confidential

1

GMS Funding Formula Rurality Adjustment

Practice economies of scale analysis

How large are additional costs of appropriately small practice?

How large are potential practice cost savings arising from increased Practice scale?

Are practice cost savings less than additional travel costs?

Is practice appropriately small?

How large are additional travel costs when practice scale is increased?

Yes

No

Practice not appropriately small so no additional funding required Travel cost analysis

Statistical analysis was undertaken to assess how costs varied with list size, and thus the extent of diseconomies of scale. This analysis was based on cost data for existing practices. Any assessment of whether costs of present practices are efficiently incurred is outside the scope of this work. Additional travel costs that might be incurred by patients were modeled by estimating the effect of removing and merging each GP practice and measuring the marginal effects on patients’ travel costs of the new configuration. In carrying out this analysis we have assumed that if a GP practice were merged with another then its registered population at an Output Area (OA) level would be redistributed among other Practices in the same proportion in which residents from the same OA are currently registered. From this we calculated the effect of a change in practice configurations (from the pre-merger to the post- merger configuration) on patient travel distances and thus patient travel costs. Results for the analysis of economies of scale The variation in practice costs with list size is shown in the chart below. The chart shows both cost and expenses, that is with and without GP salaries.

Confidential

2

GMS Funding Formula Rurality Adjustment 300

250

£Cost/Expenses

200

Average cost per patient for band Average Expenses per patient Median cost per patient for band Median Expenses per patient for band

150

100

50

0 0

2000

4000

6000

8000

10000

12000

14000

Average list size

Practices with list sizes below 1900 incur additional costs per patient compared with larger practices. This increase is statistically highly significant. However the precise magnitude of the additional costs at list sizes below 1900 is uncertain, for example because of variation in the data, lack of information on the dispensing costs of small practices and fewer instances of nondispensing practices at small list sizes. This increase can be modelled by a 1/(list size) function. This functional form is be consistent with the data on both practice costs and expenses. In contrast no robust evidence was found of continuing economies of scale as list sizes increased above 1900. There appears to be little, if any, evidence of economies of scale at such list sizes for non-dispensing practices. There is some tentative evidence of lower costs at larger list sizes for the dispensing practices. However, this may simply reflect additional dispensing costs rather than costs funded by the GMS formula. These results are consistent with the hypothesis of a one-GP practice having largely fixed costs, with both the GP’s expenses and total cost, which includes their own salary, largely fixed. At list sizes of up to 1900 these fixed costs are spread over a larger number of patients. There are no such instances in urban areas. At list sizes of greater than 1900, which represents full capacity for a GP, costs cannot be spread over a larger number of patients. Consequently no further economies of scale are realised on non-dispensing costs. Results of analysis of travel costs The simulation work we have carried out shows that areas of low population density contain a number of instances where removing a practice would impose large additional travel costs on patients. Instances of removal of practices resulting in reduced travel costs are found in some cases because patients often are not registered with their nearest GP, especially in urban areas. For many, but not all, practices in rural areas the costs to patients if the practice were not there would exceed the potential savings from economies of scale from larger practices. In many of these cases the practice is small, implying that additional funding is required. Such practices

Confidential

3

GMS Funding Formula Rurality Adjustment

correspond to the definition of “appropriately small” set out above. There are few if any such instances in areas of high population density. The chart below is illustrates the average additional travel costs compared with the savings in GP costs for rural and urban areas.

25

Patient/GP Costs (£)

20 15 10 5 0 -5 -10 -15 Rural

Urban Patient Costs

GP Costs

Factors influencing whether a practice is appropriately small The likelihood that a practice is appropriately small, in that removing it would impose a large additional burden on patients, depends on: z

the distance to the next nearest practice; and, to a lesser extent,

z

the density of the population in the area served.

Population density and distance to the next nearest GP are inversely related, although the relationship is not exact. Typically, in areas of low population density the distance to the nearest GP is greater than in areas of high population density. Although these two variables appear largely to be measuring the same underlying phenomenon, the rurality of a practice, regression results demonstrate that distance to the next nearest GP practice is a more robust indicator and as a result is the most appropriate variable to determine whether a practice is appropriately small. Possible form of additional funding for appropriately small practices The method for allocating additional funding to appropriately small practices could take a variety of forms. Some of the main types of funding adjustment that the review group is likely to wish to consider are shown in the table below.

Confidential

4

GMS Funding Formula Rurality Adjustment

Decisions on whether a rurality adjustment should be granted for a practice Variables driving decision on whether a rurality funding adjustment is required

Variables driving the decision of whether an adjustment is required could include: z

Network distance to next nearest practice

z

Linear distance to next nearest practice

z

Average distance of population, which is used in the present formula

z

Density of population in area from which patients come

Variables involving distance to the next nearest practice could be modified to account for branch surgeries or other medical facilities Whether decision is yes/no or probabilistic

A decision to make a funding adjustment could be binary (yes/no) or probabilistic, weighting the adjustment by the probability that a practice is appropriately small. This might, for example, take the form of an increasing weighting to the adjustment based on how far away the next nearest practice is, with those where the next nearest practice is very distant receiving more additional funding, other things being equal.

Varying the adjustment with list size

More demanding criteria (higher thresholds) could be required to justify additional funding for very small practices. As a purely illustrative example, some rurality adjustment might be granted to all small practices (list size less than 1900) more than 3km from the next nearest practice, but further rurality adjustment for very small practices (for example with list sizes below 1000) might only be granted in cases whether the practice is more than 8 km from the nearest practice. The rationale for this is that the smaller the higher costs of very small list sizes (e.g. 300 patients) may not be justified even in circumstances where small list sizes (e.g. 1500 patients) are justified by additional patient travel times.

Decisions on how large the adjustment should be Degree of variation

To what is it applied

Basis of adjustments

Adjustment could be: z

Uniform for all small practices

z

Banded according to list size

z

Continuously variable as a function of list size

The adjustment could apply to: z

Practice expenses per patient

z

Total practice costs per patient

z

Costs for non-dispensing practices

z

Costs for all practices

Confidential

5

GMS Funding Formula Rurality Adjustment

Number of practices affected and scale of funding adjustment There are few practices which are both small and distant from the next nearest practice. The number of practices in our sample is shown by distance to the next nearest in the Table below. Out of our sample of 7461 practices only 267 both have a list size below 1900 and are more than 1km distant from the next nearest practice, and only 148 have a list size below 1900 and are more than 2km distant from the next nearest practice. These totals would be somewhat larger if the complete population were considered. However many of the practices not in our sample are in London, so would not be far from the next nearest GP. Consequently the total population is likely to contain even fewer practices proportionately than the sample shown here. Number of large and small practice a certain distance from the next nearest GP Next Nearest GP (network km) 1 2 3 4 5 6 7 8 9 10 10+ Total

Number of practices

1,900+ km Dependent Variables (truncated*) Distance > 5km Net Cost > £20 Explanatory Variables (truncated*) 10% lowest density 20% lowest density Nearest GP (km) Nearest GP 0-1km Nearest GP 1-3km Nearest GP 3-5km Nearest GP 5-10km Nearest GP 10>km

* excluding GP practices with a population under 100 (n=285) and the 1st and 99th percentiles (n=164)

Confidential

65

GMS Funding Formula Rurality Adjustment

10.

Appendix 2 – Characteristics of GP Practices Table 10.1 Average Characteristics of Unavoidably Small GP Practices Metric

Population Density Distance to GP (km) Distance Alternative GP (km) Urban Rural Category1 Morphology Score2

Isolated and Small (