NHS England Waiting List Data: Three Ethnic Groups Are Consistently Over-Represented in the 18–52 Week Wait Band.

We looked at eight months of NHS demographic data. One pattern shows up in the 18–52 week wait band. Here is what we found, what the numbers actually mean, and what we still do not know.

NHS England has been publishing a monthly demographic breakdown of the elective waiting list since July 2025. The data covers age, sex, ethnicity, and deprivation across every trust and every region in England. We assembled all eight releases published to date and built an analytical platform on Microsoft Fabric to examine them systematically.

This is the first piece from that analysis. It covers one finding from the ethnicity dimension. A second piece will follow looking at the specialty breakdown, which adds important context.

What we found

Indian, Pakistani and Bangladeshi patients are over‑represented in the 18–52 week wait band relative to their share of the total waiting list in every monthly snapshot from July 2025 to February 2026. The same groups show the same direction of difference in every release.

In February 2026, that translated to approximately 13,000 more patients from these groups waiting in this band than would be expected if their wait times matched the average for the rest of the list.

Figure 1: Headline equity metrics for selected ethnic groups, analysed separately — NHS England RTT waiting list (February 2026, 18-52 week wait band)

Over seven million people are on the NHS elective waiting list. The 18-week referral-to-treatment standard has not been met nationally since 2016. The demographic data makes it possible to examine how the backlog is distributed across different groups. Across eight months of data, these three ethnic groups are consistently over-represented in the longer wait band.

That is what we see in the data. Before discussing interpretation, it is important to handle the statistics carefully.

Reading the numbers properly

At the level of an individual patient, ethnicity is a very weak predictor of waiting time. For any single person, knowing their ethnic group provides very little information about which wait band they will be in. Most patients in every group wait similar lengths of time. That is statistically true. But that truth is about individuals. The pattern we are examining is about groups, repeated month after month. A weak individual predictor can still produce a steady, meaningful disparity across hundreds of thousands of people. Both facts can be true at once.

The effect size measure we use is Cramer’s V, a standardised statistic ranging from 0 (no association) to 1 (perfect association). For ethnicity, it sits between 0.009 and 0.013 across the eight months. Values below 0.10 are conventionally considered small. All our values are well below that. The table below shows all four demographic dimensions for context:

Table 1: Effect sizes across all four demographic dimensions, July 2025 to February 2026.

With around 7 million patients, chi-square tests return p < 0.0001 even for very small differences. We ran 32 tests in total across four dimensions and eight months, and adjusted the p-values using the Benjamini-Hochberg procedure to account for multiple testing. All remain below 0.0001 after correction. So the pattern is statistically unlikely to be due to random variation. The individual‑level effect size is small, which is a different question from whether the group‑level pattern is meaningful.

Why consistency is the point

The individual-level effect being small does not make the finding uninteresting. It changes what the finding is.

A small per-person difference, in the same direction, across the same groups, in every one of eight monthly snapshots covering hundreds of thousands of patients — that is a consistent distributional pattern in aggregate administrative data. Not a dramatic gap, but a steady lean in one direction that does not shift.

The absolute differences between the most and least represented groups are in the range of 0.3 to 1.2 percentage points. Small. But the direction never reverses. Not in July, not in February, not in any month in between.

The size of the effect is one consideration; another is whether the pattern is real and consistent enough to merit further investigation. Eight months of consistent data suggest that further investigation is warranted.

What the numbers mean in practice

The 13,000 figure for February 2026 is not a statistical abstraction. Those are real people waiting longer than expected given their share of the waiting list. A small effect size in statistics does not mean a small problem when the same displacement repeats across hundreds of thousands of patients, month after month. The number helps translate the effect size into practical terms.

Figure 2: Single‑group equity analysis for Indian patients — NHS England RTT waiting list (February 2026, 18-52 week wait band)

A Cramér’s V of 0.009 might look negligible at first glance. But stopping there would be a mistake. The effect size tells us the individual‑level association is weak. It does not tell us what the pattern means at scale. These are different questions and they need different answers.

What we do not know

The data shows a consistent pattern. It does not explain it. The questions that follow are about what happens inside NHS pathways — how patients are referred, how they move through the system, how prioritisation decisions are made. Those are questions for pathway-level investigation and clinical audit. Waiting list counts alone cannot answer them.

What comes next

The specialty-level breakdown in the WLMDS data lets us ask where inside the system this pattern sits. Which treatment functions are driving it? Is it concentrated in the specialties with the highest breach rates — Ear Nose and Throat, Oral Surgery, Plastic Surgery — where patients of all backgrounds tend to wait longest?

The answer is no. This finding adds important context and is examined in the next piece.

About the analysis

All figures are derived from the NHS England WLMDS demographic publication. Management information, not official statistics. We grouped Indian, Pakistani, and Bangladeshi because each showed the same direction of over‑representation individually, and grouping improves statistical stability. The combined excess count in the 18–52 week wait band is approximately 13,000 patients (Indian ~6k, Pakistani ~5k, Bangladeshi ~2k). British and Any other White background are not grouped because they are not part of the finding. The patterns described here are offered as analytical observations. They do not establish causation, assign responsibility, or constitute a performance assessment of any organisation.

The platform is built on Microsoft Fabric. Monthly CSV files load into a Fabric Warehouse landing schema, transform through dbt staging and intermediate models, and materialise as a star schema in the mart layer. A Power BI semantic model connects to the mart via Import mode. Statistical tests use scipy and statsmodels in Python.

Full analysis and interactive dashboard: http://equiwait.verulamblue.com/

Data source: NHS England WLMDS demographic publication, July 2025 to February 2026. 134 NHS acute trusts, 42 ICBs, 7 regions, England summary.

Stack: Microsoft Fabric Warehouse, dbt, Power BI, Python.

Verulam Blue, May 2026. Independent analytical case study. No affiliation to any NHS body or government organisation.