Background The majority of Clinical Commissioning Groups (CCGs) in England have policies for thresholds of weight loss or body mass index for intervention prior to knee replacement surgery. There is significant geographical variation in intervention content, reflecting the current inadequate evidence-base for their effects. Potential unintended consequences include increasing health inequalities through differential impact on vulnerable sociodemographic groups’ access to surgery.

Our aim was to assess the impact of these policies on access to knee replacement surgery in England.

Methods Natural experimental study to compare the rate of primary knee replacement surgery and patient characteristics over time, between intervention policy clinical commissioning groups (CCGs) which introduced policies for patients with overweight or obesity between Jan 2013 and June 2018, and control group CCGs without a policy introduction. Data from the National Joint Registry (NJR) for England for 481,555 patients who had primary knee replacement between Jan 2009 and Dec 2019 in England were analysed using interrupted time series and difference-in differences analyses. Control and intervention CCGs were randomly matched. CCG level data were pooled relative to the policy introduction date.

Results Rate of primary knee replacement surgery per 100,000 population aged 40+ increased over time in all CCGs before policy introduction. While rates continued to increase in control regions, a sustained fall was observed in intervention regions after policy introduction (trend change -0.98 operations per quarter per 100,000 aged 40+ years, 95% confidence interval -1.156 to -0.803, P<0.001). For illustration, after 3 years at this trend, there are 11.8 fewer operations per quarter (0.98 x (3x4)) representing a fall of 17.5% from the rate at the time of policy introduction (67.2). These figures are 19.6 and 29.2% at 5 years. Rates of surgery fell in all patient groups, including non-obese patients. The proportion of independently-funded operations and patients living in the most affluent areas increased after policy introduction.

Conclusion Weight loss and body mass index policy introduction was associated with concerning decreases in the rates of knee replacement surgery. This study’s powerful quasi-experimental design examining policies introduced over a wide timeframe, and control group comparison, strongly reduces the likelihood of this effect being due to secular trends. Lack of outcome data for non-surgical patients is a limitation. The rate decrease affected all patient groups, not just the obese patients policies were targeted at. Changes in patient demographics seen after policy introduction suggest these policies have increased health inequalities and need urgent reconsideration.