Linking methodology

Both HES and PROMs data are updated on a cumulative basis and providers can resubmit data for any previous period within the current financial year. Q1 PROMs questionnaires may be administered at any time between a decision to admit being taken and before the episode, and completed Q2 questionnaires received at any point after they are received and returned by patients. This is especially the case for Q2s, which are completed and returned several months after the episode in which the operation occurred.

For this reason, PROMs matching is run on a cumulative basis to ensure that any updated episode or recently returned questionnaire is considered for matching. Linkage rates will therefore increase retrospectively for PROMs data submitted in previous periods.

Patient matching looks at a range of patient identifiable fields, including NHS number, sex and date of birth. These patient identifiable fields from the PROMs questionnaires are used to find a matching patient identifier in the HES data. This identifier is then used to find an associated HES episode. The episode must have the relevant clinical coding and match to the compatible questionnaire i.e. a hip questionnaire must match to a hip replacement episode.

Details of the patient matching ranks can be found in the table below.

The provider matching compares the provider code of the episode in HES to the provider code submitted by the hospital where the Q1 questionnaire was administered. In many cases, the Q1 questionnaire may be completed in one hospital and the operation carried out in another due to patient choice, subcontracting arrangements between providers or shared pathways.

For non-perfect matches, the following groups are used:

• green list - a list of provider codes mapped where a known subcontracting or shared pathway relationship exists between the two organisations
• amber list - organisations where little information is known about the relationship between the two but it is feasible that a relationship could exist

Details of the provider matching ranks (ranks 1 to 3 are classified as allowable matches) can be found in the table below.

Date matching compares the episode start date in HES with the ‘Q1 completed date’. This is the date on which the patient completes and signs the pre-operative questionnaire. Any matches where the completed date is greater than 18 weeks before the operation are deemed unacceptable, as this coincides with the 18-week target for the maximum wait between referral and treatment.

Details of the date matching ranks (ranks 1 to 5 are classified as allowable matches) can be found in the table below.

In cases where a Q1 can potentially link to more than one episode, as a tie breaker, the link with the highest score is chosen thereby ensuring that the highest quality data are selected. A higher weight has been given to the patient match compared to the provider and date matches as the patient match involves matching on multiple fields (NHS number, sex, date of birth, postcode) and is therefore less likely to create a false positive link.

Each match (or potential match) between an episode and a Q1 will have a combined match rank and score, for example a patient match of 1 (score 15), a provider match of 2 (score 8) and a date match of 3 (score 4) will give a combined match rank = 123 and a total score = 27.

We ran the two linking methodologies in parallel (one with HESID and one with Person ID) which confirmed this change had a minimal effect on the linked PROMs-HES data, with around 0.1% of questionnaires being affected. We also saw a small increase in the number questionnaires linking with episodes.

For both financial years, the number of records were the same and the number of final episode matches were similar. More patient matches were found with the HES processing run however these did not return different episode matches compared to the MPS run. Note 2020-21 covers a period where health services were affected by the COVID-19 pandemic. This has directly impacted upon reported volumes of activity pertaining to Hip and Knee replacements reported in PROMS.

All Q1s have a unique number, the PROMs serial number on the front sheet which is scanned when the form is returned. When the Q2 questionnaire is returned it is a simple matter of linking the Q1 and Q2 questionnaires by PROMs serial number to create a match.

An additional step is to look at the patient’s date of birth as submitted on the Q2 questionnaire and compare this with the DOB as submitted on the Q1 questionnaire. The DOB part of the matching algorithm is an additional step to give an indication of extra confidence that the match is correct.

Details of the questionnaire matching ranks can be found in the table below.