People and organizations tend to respond to incentives. Monetary or not, they can be important determinants of their behaviour. One of the most prominent examples in the field of economics is the downward slopping demand curve. People tend to consume more of a product when its price goes down in the same way a firm seeks to employ more workers when the wage level decreases. Yet the nature of the incentive and the way in which agents respond to it may vary substantially. Therefore, it is important to identify the existence of an incentive and then evaluate whether there was a response in the behaviour of the underlying agents.
By comparing the pre and the post tariff behaviour of trusts, we evaluate whether it was affected by the tariff removal.
If trusts had previously responded to such an incentive, removal of it would result in decreasing the probability of a procedure being cancelled at the last minute in the post-tariff reform period.
To explore the above hypothesis, we used an extract of the Hospital Episode Statistics database covering all the admitted patients for elective procedures in 262 NHS trusts during the period between January of 2009 and December of 2011 in order to have equal time spans before and after the tariff reform.
slightly decreased after June of 2010. The cancellation rate appears to maintain the same pattern when it is calculated for different sub-samples, i.e. for patients with length of stay shorter or equal than one day, since patients with greater lengths of stay are usually special cases and according to the data they are not usually discharged to their usual place of residence.
However, in order to perform a more formal evaluation of whether the tariff reform affected the behaviour of trusts and to examine whether the hints provided by the graphical inspection survive after controlling for patient and trust-level observed heterogeneity, a series of models were estimated using different sub-samples.
Given the fact that this reform was applied to all NHS trusts, there are no distinct treatment and control groups. However, depending on their overall performance regarding last minute cancellations, the effect of tariff reform would have a heterogeneous impact across trusts. The elimination of the fixed reimbursement should have affected trusts with high cancellation rates, weakening their incentives to admit high over low fee work and/or cancel after admitting in the post-reform period. Therefore, trusts with the lowest/zero levels of cancellations were used as the control group, i.e. indifferent to the "treatment" and those with higher levels of mean monthly cancellation rates as the treatment group in order to perform a series of difference-in-difference style analyses.
Such identification strategies have been adopted for the evaluation of internal market NHS reforms but they are also popular in other fields of economics, such as evaluating the employment effects of minimum wage increases. The models also include a range of provider and patient level covariates in order to control for the background risk of cancellation since socio-economic status and co-morbidities have been proven to be significant determinants of last-minute cancellations.
The estimations indicate a small, yet persistent and statistically significant effect of the tariff reform. The cancellation rate has declined during the 'policy-on' period and this result remained robust after performing a large set of estimations using alternative definitions of the control group, different samples and various model specifications, resulting in approximately 600 models.
According to the preferred model specification, patients admitted for an elective procedure in an NHS trust with non-zero monthly cancellation rate were more likely to be cancelled at the last minute prior to the with the tariff in place, however this probability has been declining since the tariff’s removal.
Therefore there is evidence that, perverse or not, the cancellation tariff changed behaviour. Its removal has improved the chances of procedures going ahead at planned.... .assuming of course that there are enough hospital beds, staff and working equipment.