Start creating an alpha for "download XML to find patients for this measure in your EMIS/SystmOne"

[sebbacon]

Sounds like the first step is to make a version of our measure definition inspector that presents every measure in a tabular format with last updated dates for each definition.

This can then be consumed by third parties wishing to synchronise with our definitions.

Let Dustyn know (via Seb/Brian) when we've done this.

[brianmackenna]

@inglesp has quickly run of two measures to inform thinking, one "easy" and "hard". We may just want to dump the dm+d results to people with the EHRs in the first instance but some thoughts on the following

Lidocaine Measure gives the following dm+d results

376491004 34178211000001102 11052611000001103 11052811000001104 10570111000001104 34178311000001105 11053411000001105 11052711000001107

• This is a mix of VMP, AMP and AMPP. The electronic prescription service implementation guidance for GP developers does not permit prescribing at VMPP or AMPP level so we may wish to only include AMP or VMP levels

• We could (I think) just provide a VMP with definition to included all AMPs as true / false.

Total Oral Morphine Equivalence gives 3776 dm+d results. Comments as above but further thoughts

• This is an example of a measure that ideally would be calculated at a "patient level". For example a person might be taking three opioid products so to stratify patients you would want their total morphine
• Extra logic could be imposed to calculate the total OME per day per patient (but might take a bit of work and difficult for us to do without being able to hold the results in our hands to check for accuracy)
• If patient level data is made available to us it is likely that we would need to think this through at some point

[inglesp]

Here you go: measure_numerator_snomed_codes.zip

code

``` import csv from dmd.models import VMP, AMP, VMPP, AMPP from frontend.models import Measure for m in Measure.objects.all(): print(m.id) with open("/tmp/measure_numerator_snomed_codes/{}.csv".format(m.id), "w") as f: writer = csv.writer(f) for cls in [VMP, AMP, VMPP, AMPP]: for obj in cls.objects.filter(bnf_code__in=m.numerator_bnf_codes).order_by("id"): _ = writer.writerow([obj.id, cls.__name__.lower()]) ```

[brianmackenna]

Reviewing the measures we could implement the following structure

• Patient is a currently registered with the practice (there are a few SnoMed codes for this so we will need to check implementation in core systems)
• Patient has had one of the following medicines in the last six months
• List of VMPs
• Include AMPs = true/false
• Include "why it matters"
• Last updated date

Measures where above will not apply

• GBG excess spend : SystmOne already have searches in their system. They could be moved to the same folder. We could generate a list of AMPs from dm+d where the listed reimbursement price is different from the tariff.

• Seven Day Prescribing for Long Term Conditions This will require quantity logic to be added.

• Cost of milk formulae for Cow's Milk Protein Allergy (CMPA) per listed patients aged 0-4 We could add an age filter.

Other fields In EMIS (I'm sure SystmOne is similar) when you run a rule based search you have the option of returning other pertinent information. For example Antibiotic stewardship: three-day courses for uncomplicated UTIs you may return the patient unique identifier, their name, the drug that triggered the rule but also the gender of the patient, any other antibioptics they have had etc. I propose we do not add these fields at present until further discussion with people who implement directly in SystmOne, EMIS etc

[inglesp]

@brianmackenna, you're going to work out what measure metadata we want to send to the vendors, then I'll dump this and add to the zip file.

[brianmackenna]

Following review of the measure definition here are fields from the metadata that I think should be included and some comments:

id / analyse_url : Not specifically this field but we should think about how a url could be generated in the search report so people can go straight to relevant page on OpenPrescribing. SystmOne have automated via their toolbar to connect to practice dashboards. name description why_it_matters tags : This could be used to organise an OpenPrescribing folder of searches much the same way measures can be prirotised on dahboard for certain clinical areas/topics. (num_or_denom)_type : I think this will be useful but not 100% sure how yet. For example it could be used to automate cdompleting the rest of the logic in the search. When it is bnf_items the search is straightforward search of VMPs etc and should be default. If the type is bnf_quantity the more logic in the searches would require quantity to be added (see comment above about 7 day Rx)

Not in the wiki of measure definition but we also need to include date_reviewed

[inglesp]

Here you go:

measure_defs.zip

code

``` import json from collections import OrderedDict from frontend.models import Measure keys = ['id', 'name', 'description', 'why_it_matters', 'tags', 'numerator_type', 'denominator_type'] for m in Measure.objects.all(): record = OrderedDict((k, getattr(m, k)) for k in keys) with open("measure_definitions/{}.json".format(m.id)) as f: record["date_reviewed"] = json.load(f).get("date_reviewed") with open("/tmp/measure_defs/{}.json".format(m.id), "w") as f: json.dump(record, f, indent=2) ```

[inglesp]

Removing the top-three label because we're waiting for a response from vendors.