Exploring local government public health grant spending by health indicators, time and deprivation strata: an ecological study in England

This ecological study in England reveals that despite significant cuts to public health grants, spending allocations across local authorities show little correlation with population health needs or deprivation levels, suggesting that financial decisions are driven by factors other than strategic need-based planning.

The Gist

The Big Picture: A Shrinking Pie and a Hungry Neighborhood

Imagine England is made up of 146 different neighborhoods (Local Authorities). Each neighborhood has a Public Health Team whose job is to keep everyone healthy. They run programs like:

• Checking kids' weight and fitness.
• Helping people quit smoking.
• Providing sexual health services.
• Treating alcohol and drug addiction.
• Offering health checks for adults.

To do this work, the government gives each neighborhood a fixed-size pie (the Public Health Grant). The team has to slice this pie up and give a piece to every program.

The Problem:
Over the last decade, the government has been cutting the size of this pie. Since 2015, the pie has shrunk by 26% per person. It's like the government telling the neighborhood teams, "You have to feed the whole neighborhood, but here is 26% less food."

The Big Question:
When the pie gets smaller, how do the teams decide who gets a slice?

• Do they look at who is hungriest (the areas with the most health problems) and give them the biggest piece?
• Or do they just cut everyone's slice equally, or maybe cut the slices of the people who are already struggling the most?

What the Researchers Did

The researchers acted like detectives. They looked at the spending records of these 146 neighborhoods over several years (2017 to 2023).

They compared two things:

1. The "Need": How sick was the neighborhood? (e.g., How many people were obese? How many people were smoking? How many drug overdoses happened?)
2. The "Spending": How much money was actually spent on fixing those specific problems?

They used a special mathematical tool (called Compositional Regression) which is like a balancing scale. Since the total pie size is fixed, if you give more to the "Smoking" program, you must take away from the "Obesity" program. This tool helped them see the trade-offs without getting confused by the math.

What They Found: The "Hungriest" Didn't Get the Most Food

The results were surprising and a bit worrying.

1. The "Need" Didn't Drive the Decisions
You might expect that if a neighborhood had a huge spike in alcohol deaths, they would spend more money on alcohol treatment.

• Reality: The study found no clear link. The money wasn't being moved around based on who was getting sicker.
• Analogy: Imagine a family dinner where one person has a broken leg and needs a wheelchair, and another has a broken arm. You'd expect the family to buy the wheelchair first. But in this study, the family kept buying the same things they always bought, regardless of who was injured.

2. The Poor Got Hit Hardest
The neighborhoods that were already the poorest (the most deprived areas) had the biggest health problems. When the pie shrank, these areas lost the most money in absolute terms.

• Analogy: It's like two people sharing a taxi. One is rich and has a light suitcase; the other is poor and has a heavy suitcase. When the taxi driver says, "We need to take off 20% of the weight," he takes the heavy suitcase off the poor person, leaving them stranded, while the rich person barely notices.

3. What Actually Drives Spending?
If not "need," then what? The study found that spending decisions were driven by:

• History: "We always spent money on this, so we keep spending it."
• Local Politics: "The mayor wants to focus on this specific issue."
• Financial Pressure: "We have to cut costs somewhere, so we cut the easiest thing."
• Urbanization: Whether the area is a city or a village mattered more than how sick the people were.

4. The Only Things That Were Protected
There were only three areas where spending stayed safe across all neighborhoods, regardless of how poor they were:

• Services for babies and toddlers (0–5 years old).
• Health protection (like stopping disease outbreaks).
• Public mental health.

The "So What?" (Why This Matters)

The researchers conclude that the current system is broken. The formula used to decide who gets how much money hasn't been updated since 2013. It's like using a 2013 map to navigate a 2026 city.

Because the money isn't flowing to where the health problems are worst, the gaps between rich and poor neighborhoods are getting wider. The people who need help the most are getting the least help.

The Bottom Line

The Verdict: The government is cutting the budget, and local teams are trying to survive. But they aren't using the "health needs" of their population to decide where to cut. They are cutting based on old habits and financial stress.

The Recommendation: The government needs to rewrite the rules (the formula) for handing out this money. They need to make sure that if a neighborhood is sicker and poorer, they get a bigger slice of the pie to fix it, rather than getting a smaller slice because the whole pie got smaller.

In short: We are trying to fix a leaky roof with a bucket that keeps getting smaller, and we aren't even aiming the bucket at the biggest hole.

Technical Summary

1. Problem Statement and Background

• Context: Local authorities in England receive a ring-fenced public health grant from the central government to fund services aimed at improving population health and reducing inequalities (e.g., sexual health, obesity prevention, drug/alcohol treatment).
• The Issue: Since 2015/16, the real-terms value of this grant has declined by 26% per person. Crucially, these cuts have been applied as a flat percentage reduction without adjusting for local population needs. Consequently, more deprived local authorities (which historically had higher per-capita budgets due to higher health needs) have suffered deeper absolute cuts compared to less deprived areas.
• Knowledge Gap: While previous studies have linked grant cuts to negative health outcomes, no study has systematically investigated how local authorities allocate a diminishing budget across competing public health functions and whether these allocation decisions correlate with specific indicators of population health need or deprivation levels.

2. Methodology

The study employs a longitudinal ecological design analyzing data from 146 upper-tier and unitary local authorities in England.

• Data Sources & Period:

  • Expenditure Data: Revenue expenditure and financing data from the Ministry of Housing, Communities, and Local Government (MHCLG) for financial years 2017/18, 2018/19, 2019/20, and 2022/23. (2020/21 and 2021/22 were excluded due to pandemic-related ungeneralizable spending).
  • Need Indicators: Publicly available health indicators (e.g., obesity prevalence, STI rates, smoking prevalence, mortality rates) serving as proxies for demand.
  • Deprivation: Indexed by the 2015 Index of Multiple Deprivation (IMD), categorized into quintiles.

• Analytical Approach:

  • Compositional Data Analysis: Recognizing that the grant is a fixed sum distributed across multiple categories (summing to 100%), the authors treated the data as compositional. An increase in spending in one category necessitates a decrease in another.
  • Transformation: Data was transformed using isometric log-ratio (ilr) coordinates to handle the mathematical constraints of compositional data and avoid spurious correlations.
  • Regression Model: A compositional regression model was developed.
    • Outcome: The composition of the grant spend across 21 reporting categories.
    • Predictors: Health need indicators (lagged by 2 years to account for decision-making delays), IMD quintiles, urbanization, and population demographics.
  • Statistical Testing: Multivariate Analysis of Variance (MANOVA) was used to test associations, utilizing Pillai's Trace statistics to measure effect sizes.
  • Sensitivity: Analyses were run with 1-year and 3-year lags, and thematic sub-groupings (e.g., children's health, substance misuse) were tested separately.

3. Key Contributions

• Methodological Innovation: This is the first study to apply compositional regression to public health grant allocation. Traditional methods often analyze spending categories in isolation, ignoring the zero-sum nature of a fixed budget. This approach correctly models the interdependence of spending decisions.
• Comprehensive Scope: It is the first study to simultaneously examine multiple indicators of health need, deprivation strata, and time lags to determine if spending decisions are driven by population health profiles.
• Granular Analysis: The study moves beyond aggregate spending to analyze specific reporting categories (e.g., distinguishing between prescribed and non-prescribed children's services).

4. Results

• Descriptive Findings:

  • Spending Dominance: Children's 0–5 services (including health visitors) accounted for the largest share of spending.
  • Impact of Deprivation: More deprived authorities generally spent more per capita than less deprived ones. However, between 2017 and 2023, absolute cuts were deeper in more deprived areas.
  • Protected Areas: Only three areas saw spending protected across all IMD quintiles: Children's 0–5 non-prescribed functions, Health protection, and Public mental health.
  • Misalignment: In several categories (e.g., alcohol misuse, obesity), population health needs (e.g., rising mortality or prevalence) increased while spending decreased or stagnated.

• Regression Findings (MANOVA):

  • Weak Link to Need: While most health need indicators showed statistical significance, they collectively explained only 23% of the variation in grant composition ($R^2 = 0.23$).
  • Dominant Drivers: The largest effect sizes (Pillai's Trace) were driven by Urbanization and IMD Quintile, not by specific health outcome indicators (e.g., smoking rates or obesity prevalence).
  • Thematic Sub-groups: Even when analyzing thematic groups (e.g., sexual health vs. substance misuse), the allocation was more strongly influenced by structural factors (deprivation/urbanization) than by the specific health indicators relevant to those themes. For instance, sexual health spending was not significantly driven by STI diagnosis rates.
  • Lag Sensitivity: Changing the time lag (1, 2, or 3 years) did not alter the findings, suggesting a persistent lack of responsiveness to health needs regardless of timing.

5. Significance and Interpretation

• Core Conclusion: There is no clear relationship between specific indicators of population health need and how local authorities allocate the public health grant. Spending decisions appear to be driven by historical baselines, wider local priorities, and financial pressures rather than current health needs.
• Policy Implications:
  • The current funding formula fails to ensure that resources are directed to areas of highest need, potentially exacerbating health inequalities.
  • The study calls for a review of the public health grant allocation formula to better align funding with local population health needs.
  • It highlights the risk that without a needs-based formula, cuts will continue to disproportionately harm the most deprived communities, where the burden of preventable ill-health is highest.
• Limitations: The study cannot quantify absolute spending changes per category due to the nature of compositional coefficients, and it does not account for external spending (e.g., housing, social care) that might influence local health decisions.

In summary, the study provides robust evidence that the current mechanism for distributing public health funds in England is not responsive to changing health needs, necessitating a structural reform of the funding formula to prevent the widening of health inequalities.