This week the Health Foundation hosted a panel discussion which looked back on the Competition & Markets Authority (CMA) review of the planned merger between Ashford & St Peter’s Hospitals NHS Foundation Trust (ASP) and Royal Surrey County Hospital NHS Foundation Trust (RSC).

The CMA, Monitor and the two Trusts reflected on the case and discussed key learning points. I won’t replicate the discussion as the Health Foundation will no doubt produce a summary at some point.

However, having had the prompt of this event, it seems like a good time to start writing about some of the more obscure, but important, aspects of this case. And, it doesn’t get much more obscure than the topic for this column: competition analysis of GP referral patterns for acute elective care.

The CMA’s approach to GP referral analysis

The CMA uses GP referral data to help reach a view on the strength of competition between Acute Trusts in relation to elective care services.

If two Trusts are strong competitors for referrals at the same GP practices, then a merger between those two Trusts may, by reducing the need to compete for patient referrals, reduce the incentive for the merged Trust to deliver a high quality service that attracts patients.

So, how does the CMA measure whether the merging Trusts are strong competitors for referrals from a GP practice? It uses the share of referrals to an Acute Trust at a GP practice as a proxy for how strongly that Acute Trust competes for referrals from that practice. That is, an Acute Trust that gets a large share of patient referrals from a GP practice will be considered a strong competitor for referrals from that practice, and an Acute Trust with only a small share of referrals will be considered a weak competitor. (There are several reasons why this may not be a good proxy, but let’s leave that for another day.)

The CMA then converts a Trust’s share of referrals at all of the different GP practices into a measure of how many patients at, say, ASP can be implied as having RSC as their second preference and vice versa. If this figure is higher than 30-40%, then the CMA starts getting concerned about the effect of the merger on competition.

The CMA carries out this analysis at the specialty level, and across different care settings within each specialty. For example, it will separately look at outpatients, day case patients, and elective inpatients in Urology, and in every other specialty offered by a Trust.

It is reasonable for the CMA to want to examine competition at the specialty level as different acute providers may only offer services in some specialties and not others.

(That said, in Germany the starting point for looking at hospital mergers is at the hospital, rather than specialty, level, and when looking at supermarket mergers, the CMA looks at competition at the store, rather than individual product, level.)

Similarly, it is reasonable for the CMA to want to look at competition between providers not only in each specialty, but also in the different settings in which care is offered in that specialty.

This is because some providers may only offer, for example, outpatient and day-case services in a specialty, but not elective inpatient services. If all care settings are grouped together for analysis, it could overstate the extent of competition in some care settings where not all providers are present.

So far, so good.

So, what’s the problem?

A problem arises when the CMA uses referral figures to analyse competition in day case and/or elective inpatient services in each specialty.

Patients (and their GPs) when choosing an acute trust for their referral for a first outpatient appointment do not know, for certain, what treatment will be required and whether this will involve day case or elective inpatient services.

The possibilities facing a patient at the point of being referred by their GP may include: (i) treatment in an outpatient setting; (ii) admission for day case surgery; (iii) admission for treatment involving an overnight stay (i.e. as an elective inpatient); or (iv) discharge without treatment. (Of course, all of these options are not always relevant – in medical specialties, for example, there may not be any admission for treatment.)

However, focusing on specialties where an admission may be one possible outcome, it can be assumed that all patients at the point of being referred (and their GPs) will take into account the trust’s performance in relation to all possible services the patient might require, including both outpatient and inpatient services.

In other words, when looking at GP referral data for outpatients in a particular specialty, the choices made by these patients are going to reflect their view on the quality of both outpatient and inpatient services in that specialty.

This is because these patients do not know, at the time of choosing their acute trust provider, whether or not they will need inpatient services. As a result, these patients will take information about inpatient services into account in making their choice because there is a possibility that they may use these services.

The CMA, however, tries to gain an insight into the strength of competition between providers of day case and elective inpatient services by looking at the choice of provider made by those patients that have been admitted for day case or elective inpatient treatment.

That is, the CMA takes all patient referral data from GP practices, and isolates those patient referrals where the patient has subsequently been admitted as a day case patient or elective inpatient. It then looks at each provider’s share of these patients at each GP practice, and says that a high share of these patients means that a provider is a strong competitor in day case, or elective inpatient, services in that specialty.

The problem with this approach is that day case patients and elective inpatients do not know, at the time they are choosing their provider, that they will end up being day case or elective inpatients. As a result, there is no reason why their choices should systematically differ from those of all the other patients where some form of admission is also a possibility at the time of their referral.

Take Breast Surgery patients for example …

Take a Breast Surgery patient for example. Many women are referred to hospital to be checked out for possible breast cancer. Fortunately, only a small proportion of those that are referred actually have cancer and require an admission for treatment. Nevertheless, at the point of being referred, the patient has no way of knowing whether or not they have cancer that will need to be operated on.

As a result, all patients that are referred in this specialty can be expected to base their choice of hospital on the possibility that they may require treatment.

Separately analysing the choice of provider by those patients that have, in the end, required treatment is meaningless. These patients did not know that they would require an admission for treatment at the time of being referred (and choosing their hospital), and it is unreasonable to believe that their choices will systematically differ from other patients that did not have cancer.

To the extent that the choices made by admitted patients vary from the total cohort of referred patients, it will be completely random variation.

The problem, however, is that the CMA continues to assign a level of meaningfulness to this random variation.

The consequence is that there is a possibility that in a future hospital merger the CMA will find a competition problem in day-case or elective inpatient services in a particular specialty that is based, not on a genuine competition issue, but on this random variation.

A good example of how an obscure area of analysis might lead to a real problem for two acute trusts planning to merge.