Issues and Possible Solutions with the Mental Health Indicator Set of the Dataset and Business Rules v8.5
Dr Gavin Jamie 30 July

1  Introduction

Of all of the areas in the new QOF indicators the one that seems to be giving the most problem with respect to producing rules is the area of mental health. There have now been two releases of the codes and considerable problems have been identified with practices even identifying the patients who should be included in these registers.

In some cases practices have resorted to going back to changing codes on old records. This is certainly not best practice but in the absence of negation codes it has been the only solution.

The timescales involved in turning the codes into searches on practice systems mean that it is likely that the next release of codes will be implemented towards the end of the year. In order that practices have a chance to meet the criteria before April 2007 it is therefore vital that the next release of codes is error free.

Below is my summary of the current problems that I am aware of in this area and potential solutions.

2  Diagnostic codes

2.1  Current problems

Identifying Read codes for psychotic disease is hard. This is largely down to the very poor quality of the Read code hierarchy in this area, particularly under E1. Version 8.5 of the business rules currently says
E1...% (excluding E118., E11z1, E11z2,E140.%, E135.)
Unfortunately the codes under E1 are so varied that the include such concepts as 'single major depressive episode, severe, without psychosis E1123', 'recurrent major depressive episodes, mild E1131' and 'Schizophrenia in remission E1005'. Clearly none of these should be on the register.

2.2  Suggested criteria

This is a longer and more complex list than would be desired. This reflects the underlying lack of structure in the read codes.

E10.% (excluding E1005, E1015, E1025, E1035, E1055, E1075)

E1104

E1124

E1134

E114.% (excluding E1146)

E115.% (excluding E1156)

E116.% (excluding E1166)

E11y.%

E11z0

E13..% (excluding E135)

E14..% (excluding E140)

E2 and Eu codes as listed in version 8.5 of the ruleset.

3  Patients recovered from mental illness

3.1  Structure of the indicators

The individual areas in the mental health section divide in general into those for the whole of the registered population and those only for patients currently taking lithium containing medication. This second group of indicators have no current problems with their query specification as it has remained unchanged since the two previous years of QMAS.

3.2  Identification of denominator populations

For the majority of the other indicators the problems are related to the identification of the population who are eligible for a given indicator. The problems listed about with the read codes for identifying patients with psychotic or bipolar illness become more apparent in these indicators. Additionally as of version 8.5 there is not mechanism to exclude patients who have recovered from psychotic or bipolar illness. These may make up a substantial part of the register currently. The mental health indicators can be onerous for both practice and patients and it is sensible to avoid unnecessary appointment purely for the sake of QOF.

3.3  Proposed solution

For indicators MH 9, 6 and 7 changing of the codes for Mh_Cod are required. In addition a further criteria is needed if patients who have recovered from psychotic or bipolar disease are to be excluded from the register. It would be reasonable for this to mirror rule two, and indeed must exist as part of rule two to ensure that both exclusion criteria are looked for. Fortunately a code exists in Read codes v2 which has exactly the rubric required.

Two new concepts are required.
mh_res 212t Resolution of psychotic of bipolar disorder
mh_res_dat Date of resolution of psychotic or bipolar disorder (most recent occurrence)

Thus the new rule two would be

(If mh_dat ≠null AND mh_res_dat ≠null AND mh_res_dat > mh_dat) AND

(If ron_dat ≠null AND roff_dat ≠null AND roff_dat > ron_dat)

If true - reject

If false - next rule

This would need to be inserted as rule two in MH 9, 6 and 7

4  Mental health reviews (MH 7)

4.1  Current situation

Perhaps the most convoluted area is that of the mental health review. Version 8.0 of the business rules did not come close to assessing this area. Version 8.5 takes several steps towards an effective system but remains fatally flawed. There is probably only one more change to correct problems and leave practices with a realistic chance of adapting their systems to the rules.

The logic of 8.5 is as follows. Firstly the patients who current have a diagnosis of psychotic or bipolar illness or are on the mental health register are identified. This is common to other criteria and will need amendment as discussed in section 3.3.

This group of patients are then searched for a 'Mental health review DNA' code. If this is present then they proceed into the next rule.

Then the Read code for a mental health review is searched for. This is the same code used in MH9 as evidence that a mental health review has taken place. In this context it assumed to mean precisely the opposite. It is then checked to see whether a MH follow up took place within two weeks after the date of this code.

Exceptions are then dealt with.

4.2  Current problems

There is a critical flaw in MH7 and MH9 - the read codes (v2) 6A6 and 8BM0 are in conflict in these indicators; to mean that a review has taken place (MH 9) and to note the date that a review did not take place (MH 7). There is no concept of 'scheduled' within these business rules or Read codes.

As the system currently stands MH 9 will count both patients who have had a review and patients who have merely be scheduled to have a review and not turned up.

This is clearly contrary to the intent of these QOF areas.

It is unclear why the date of this code is taken rather than the data that the DNA is recorded. In the whole of the rest of the QOF the date coded is assumed to be the date of the incidence of the occurrence described in the code. Note 4 in version 8.5 of the rules makes it quite clear that this date is not to be trusted but it is not clear why this should be considered to be the case.

4.3  Possible solutions

4.3.1  Do nothing

To do nothing would leave the problems above intact. Some practice may have developed work arounds and others are likely to. Some will not. The mental health QOF will not deliver quality data and is unlikely to fully achieve its objectives.

4.3.2  A new code 'MH Review Scheduled'

If it were determined that the date that the DNA is recorded is not to be trusted a new code could be produced explicitly scheduling the mental health review. This would have been a useful option in version 8.0 of the codes but as the next public release of the codes is likely to be six months into the QOF year and Read code releases after this it will entail considerable amount of work for practices going back through records and changing or inserting codes. As was discussed above changing past records for the sake of QOF is not best practice.

4.3.3  Make dnarev_dat = mhrdna_dat

The reasons for not trusting the data entered for the DNA are not at all clear, yet they cause almost all of the problems associated with this indicator. This simple change would simplify the rules and remove anomalies. It would also be backward compatible with the current rules meaning very little additional work for practices. This is my recommended solution.

5  Acknowledgements

The GP forum on doctors.net.uk have been a source of information as flaws have been identified. I would also like to thank Dr Gwion Rhys for his suggestions.
This document was translated from LATEX by HEVEA.